Compare commits

...

30 Commits

Author SHA1 Message Date
52f4d02722 Merge pull request 'feat(skills): dolibarr-sandbox-write — host-guarded write skill (V9)' (#21) from claude/dolibarr-sandbox-write into main 2026-06-29 20:50:11 +02:00
d2e8b3a3a4 feat(skills): dolibarr-sandbox-write — host-guarded write skill (V9)
The write-capable companion to the read-only dolibarr* skills, scoped to the
erp-sandbox. Lets an AI agent rehearse bookkeeping writes against a copy of prod
(ADR-0003) before a human promotes the reviewed change to prod.

- scripts/dol-write.sh: write wrapper that REFUSES any host that is not
  erp-sandbox.arcodange.lab (the structural prod-safety guarantee) using the
  ai_agent_sandbox key from a gitignored .env.
- scripts/thirdparty-create.sh: create client/supplier fiches; codes auto-assign
  via the elephant mask (code="-1").
- scripts/invoice-create.sh: customer (/invoices) or supplier (/supplierinvoices)
  invoices with product/service lines + ref_supplier, optional validate.
- scripts/payment-record.sh: record a règlement (VIR/CB/CHQ/LIQ); customer pays
  full + marks paid, supplier needs an amount.
- SKILL.md (safety model + workflows + the human-gated promote flow), .env.example,
  example input.

Proven end-to-end live against the sandbox: client -> invoice (service+product
lines, HT 1100 / TTC 1320) -> validate -> payment (paid); supplier -> supplier
invoice (ref_supplier carried) -> validate. Host guard verified to refuse a prod
URL before sending.

Avoirs (credit notes) and bin/arcodange CLI wiring are planned follow-ups.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 20:49:31 +02:00
c269751422 Merge pull request 'fix(test): grant societe client voir (262) so /thirdparties list works' (#20) from claude/sandbox-provision-run into main 2026-06-29 20:17:20 +02:00
7619a4f358 fix(test): grant societe client voir (262) so /thirdparties list works
Validating ai_agent_sandbox's key against the sandbox API, /thirdparties
returned 404 (the voir_tous ACL trap) while /invoices, /products,
/supplierinvoices returned 200. The missing right is `societe client voir`
(id 262, "see all thirdparties") — prod's ai_agent has it. Added it to
WRITE_IDS so the list endpoint works; other modules' lists are fine with plain
`lire`.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 20:16:59 +02:00
37865c55c5 Merge pull request 'fix(test): persist API key + anchor rights selector + idempotent createUser' (#19) from claude/sandbox-provision-run into main 2026-06-29 14:31:24 +02:00
18c5d0ebda fix(test): persist generated API key + anchor rights selector + idempotent createUser
First real run against the sandbox revealed three issues in userSetup.ts:

1. generateApiKey generated the key client-side and read it into the file but
   never submitted the edit form, so Dolibarr never persisted api_key (DB stayed
   NULL → the key could not authenticate). Now it clicks Save after generating.

2. assignRights matched `rights=<id>` as an href substring, so a short id like
   12 (facture creer) also matched rights=121 / rights=1232 and .first() clicked
   the wrong link — facture creer was never granted. Anchored with a trailing
   "&" (rights=<id>&) for an exact match.

3. createUser was not idempotent: a re-run hit the existing login and failed to
   parse a new id. Added findUserId (look up by login via the user list) and
   return the existing id instead of creating a duplicate.

Verified the symptoms live: ai_agent_sandbox (rowid 4) had api_key NULL and was
missing only facture/creer among the 11 intended rights.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 14:30:36 +02:00
650594bb70 Merge pull request 'fix(sandbox): Dolibarr dev mode (DOLI_PROD=0) + correct install.lock path' (#18) from claude/sandbox-dev-mode into main 2026-06-29 14:11:18 +02:00
416cd807a3 fix(sandbox): run Dolibarr in dev mode (DOLI_PROD=0) + correct install.lock path
After seeding erp-sandbox from prod, the home dashboard rendered a generic
"technical error" banner per box: prod mode ($dolibarr_main_prod=1, the image
default via DOLI_PROD) escalates the seed's minor non-fatal warnings into that
banner. Setting DOLI_PROD=0 for non-prod environments makes Dolibarr render
real errors inline (correct for a rehearsal env) and clears the banners.

config.yaml adds `DOLI_PROD: "0"` only when env != prod, so the prod configmap
is byte-identical (prod keeps the image default DOLI_PROD=1) — verified via
helm template diff. ArgoCD rolls only the sandbox pod.

Also corrects the test/README install.lock path: Dolibarr checks the DATA root
(/var/www/documents, a PVC — persists across restarts), not /var/www/html. And
notes that a prod-seeded sandbox still needs install.lock created (the seed +
documents/mycompany sync don't include it).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 14:09:59 +02:00
2154bf319e Merge pull request 'feat(test): split env config — .env (prod) vs .env.sandbox (sandbox)' (#17) from claude/test-env-sandbox-split into main 2026-06-29 11:26:39 +02:00
e4a7f99333 feat(test): split env config — .env (prod) vs .env.sandbox (sandbox)
provisionSandbox.ts now loads its own .env.sandbox (via @std/dotenv loadSync)
instead of the shared .env, so prod (main.ts → .env) and sandbox
(provisionSandbox.ts → .env.sandbox) configs don't collide. .gitignore widened
to .env* (keeping .env.example tracked). .env.example rewritten to document the
two-file convention + the per-env kubectl secret sources, including the caveat
that a prod-seeded sandbox uses PROD's admin password.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 11:26:05 +02:00
04281f0ab7 Merge pull request 'docs(test): preserve the install.lock step in test/README' (#16) from claude/test-readme-install-lock into main 2026-06-29 07:54:22 +02:00
c010099dae docs(test): preserve the install.lock step in test/README
The pre-existing (untracked) test/README documented creating Dolibarr's
install.lock after a fresh install — a non-obvious operational step missing from
the rewritten README. Preserve it (generalized to the per-env namespace/label,
with a note that a prod-seeded instance doesn't need it).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 07:54:01 +02:00
4cc5ca39ce Merge pull request 'feat(ops): erp-sandbox iso-prod seed + documents sync tooling (ADR-0003 E2)' (#15) from claude/e2-sandbox-lifecycle into main 2026-06-29 07:43:44 +02:00
fbc0cc6962 Merge pull request 'feat(test): provision erp-sandbox via Playwright (REST API + write-scoped ai_agent_sandbox user)' (#14) from claude/poc-provision-sandbox into main 2026-06-29 07:43:32 +02:00
7264f00ed4 feat(ops): erp-sandbox iso-prod seed + documents sync tooling (ADR-0003 E2)
Productionizes the sandbox state-lifecycle mechanisms validated live against
erp-sandbox. `ops/sandbox/sandbox-lifecycle.sh`:
  - refresh-from-prod: read-only pg_dump of prod erp (default_transaction_read_only)
    -> DROP OWNED BY erp_sandbox_role CASCADE -> pg_restore into erp-sandbox, using
    the sandbox's own membership creds (no DROP/CREATE DATABASE, no CREATEDB, no
    superuser). Dumps the full public schema (so app helper functions + triggers
    come over) and filters the provisioner-owned pgbouncer user_lookup function
    from the restore TOC. Scales the pod to 0 for exclusive access; copies prod
    creds into a transient secret that is deleted on exit.
  - sync-documents: tar-pipe the documents/mycompany tree (company logo + uploads)
    prod -> sandbox, since uploaded files live on the PVC, not the DB.

Prod integrity is structural: prod is read-only during dump; the restore can only
write erp-sandbox (erp_sandbox_role owns only the sandbox DB and cannot drop prod
erp/erp_role); the platform's only prod-capable superuser stays behind the
human-gated postgres.yaml CI and is never used here.

README documents the integrity guarantee, the encryption + PVC fidelity caveats,
the BDD reset loop, and the hardening backlog (dedicated read-only dump role,
golden-cache PVC).

Refs ADR-0003 (factory#19). Chart owner-role fix = erp#13.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 07:42:00 +02:00
523f0cf001 feat(test): provision erp-sandbox via Playwright (REST API + write-scoped ai_agent_sandbox user)
Extend the Deno + Playwright UI-automation POC to provision the erp-sandbox
Dolibarr for the AI agent:

- moduleSetup.ts: add enableApiModule(ctx) — toggles the REST API / Web services
  module on /admin/modules.php (kanban). Resilient: tries the fr_FR card label
  "API/Web services REST (serveur)" first, falls back to a /API.*REST|REST.*API/i
  title match if the exact label is absent.
- userSetup.ts (new): createUser (returns the new numeric id), assignRights
  (clicks each addrights link on /user/perms.php, idempotent), generateApiKey
  (triggers Dolibarr's generate control on the user card and reads the value back).
- provisionSandbox.ts (new entrypoint, main.ts untouched): login → enable API →
  create ai_agent_sandbox (non-admin) → grant write rights → generate API key,
  then write the key to test/.ai_agent_sandbox.key (gitignored) instead of
  printing it.
- .gitignore (new), .env.example + README.md: sandbox vars, the
  deno run --allow-all provisionSandbox.ts command, and kubectl one-liners to
  pull DOLI_ADMIN_PASSWORD (secretkv) / DOLI_DB_PASSWORD (vso-db-credentials)
  from the erp-sandbox namespace.

Why UI not SQL: API keys are encrypted with the instance's DOLI_INSTANCE_UNIQUE_ID,
so the key must be generated by the sandbox itself, not INSERTed raw.

deno check passes for provisionSandbox.ts and scripts/admin/userSetup.ts.
NOT run end-to-end: the sandbox Dolibarr is not installed yet (empty DB / fresh
install wizard), so the selectors are best-effort Dolibarr 22 conventions and
must be confirmed on the first real run.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-29 07:34:21 +02:00
cb17332314 Merge pull request 'fix(chart): template Postgres owner role in update_ownership.sql for multi-env' (#13) from claude/fix-chart-owner-role into main 2026-06-28 22:30:54 +02:00
c0d5f2e144 fix(chart): template Postgres owner role in update_ownership.sql for multi-env
The Dolibarr before-start step `chart/scripts/update_ownership.sql` (embedded
into a ConfigMap by `chart/templates/scripts-config.yaml`) hardcoded the
Postgres owner role `erp_role`. It reassigns ownership of all public-schema
objects to that role after install. For any non-prod environment the owner
role differs — by the multi-env elision rule (ADR-0002/0003) it is snake-case
`<app>_role` for prod and `<app>_<env>_role` for non-prod, so the sandbox owner
role is `erp_sandbox_role`. With the literal `erp_role`, installing Dolibarr in
`erp-sandbox` would reassign sandbox tables to prod's `erp_role`, which (a)
breaks the sandbox runtime (its dynamic DB creds are a member of
`erp_sandbox_role`, not `erp_role`) and (b) breaks the ADR-0003 reset
(`DROP OWNED BY erp_sandbox_role`).

Fix: make the owner role env-aware via a new chart value `db.ownerRole`.
- values.yaml: default `ownerRole: erp_role` (prod).
- values-sandbox.yaml: override `ownerRole: erp_sandbox_role`.
- update_ownership.sql: all `'erp_role'` literals → `'{{ .Values.db.ownerRole }}'`.
- scripts-config.yaml: render that one SQL file through `tpl` so the value is
  substituted (the other script has no template vars and stays on `.Files.Get`).
  The SQL's `$$`, `%I`, `format(...)`, `RAISE NOTICE` are not Go-template syntax,
  so `tpl` only substitutes the added `{{ .Values.db.ownerRole }}`.

Verified: the prod ConfigMap render (values.yaml only) is byte-identical to
origin/main (empty diff, still `erp_role`); the sandbox render
(-f values.yaml -f values-sandbox.yaml) now contains `erp_sandbox_role` and no
bare `erp_role`; `helm lint` passes (no worse than origin/main).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-28 22:29:18 +02:00
354b40549d Merge pull request 'feat(multi-env): Phase D3 — erp iac creates erp-sandbox Vault auth + creds + KV' (#12) from claude/phaseD-erp-sandbox-iac into main 2026-06-28 17:30:57 +02:00
e2fa327361 feat(multi-env): Phase D3 — erp iac creates erp-sandbox Vault auth + creds + KV
ADR-0002 Phase D, erp-repo layer. iac/main.tf iterates `envs = ["prod",
"sandbox"]` so the app_roles module + the admin-bootstrap resources are
materialised per environment:
  - module.app_roles["sandbox"] → auth/kubernetes/role/erp-sandbox +
    postgres/creds/erp-sandbox (dynamic role GRANTs erp_sandbox_role — the
    snake-case owner role created in factory#17 — and REVOKEs on DATABASE
    erp-sandbox; token_policies ["default","erp-sandbox"] = the policy from
    tools#3).
  - random_password.admin_initial_password["sandbox"], random_uuid.dolibarr_id
    ["sandbox"], and vault_kv_secret_v2.dolibarr_admin_setup["sandbox"]
    (kvv2/erp-sandbox/config) → the sandbox Dolibarr's own admin password +
    encryption id.

State migration via `moved` blocks: the pre-existing single-env resources are
re-keyed into the for_each map under "prod" so introducing for_each does NOT
destroy+recreate them. Critical for random_uuid.dolibarr_id (prevent_destroy =
true — prod encryption id + paid-module binding): a wrong/absent moved would
HARD-FAIL the apply rather than lose it. The module's internal moved
(role -> role[0]) chains with the module re-key. Verified the exact compound
(old-module state → new module count+moved + for_each, one apply) with two
standalone tofu plans: both show "X moved", sandbox created, 0 destroyed.

env=prod renders byte-identical to the single-env baseline (module elision
rule), so the prod erp Vault auth role, dynamic creds, admin secret + KV are
unchanged.

D3 of Phase D. D1 = factory#17 (DB+role, merged). D2 = tools#3 (Vault
policies, merged). D4 (ArgoCD Application) is next.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-28 17:26:46 +02:00
b4bdbe75df Merge pull request 'chart: Phase C of multi-env evolution — template literals, add sandbox overlay' (#11) from claude/chart-multi-env-prep into main 2026-05-31 23:27:05 +02:00
ec4df4719f chart: template hardcoded single-env literals; add values-sandbox.yaml overlay
Phase C of the multi-env evolution discussed in the runbook design thread
(see PR description). Pure refactor — the prod helm template render is
verified byte-identical (10857 bytes both before and after, diff exit 0).

What was hardcoded, now templated:
- chart/templates/vaultauth.yaml          role: erp                       → role: {{ .Values.vault.k8sRole }}
- chart/templates/vaultdynamicsecret.yaml path: creds/erp                 → path: {{ .Values.vault.dynamicPath }}
- chart/templates/vaultsecret.yaml        path: erp/config                → path: {{ .Values.vault.staticPath }}
- chart/templates/config.yaml             DOLI_DB_NAME: erp               → DOLI_DB_NAME: {{ .Values.db.name }}
                                          DOLI_URL_ROOT: https://erp..lab → DOLI_URL_ROOT: 'https://{{ .Values.host }}'

values.yaml gains a documented multi-env coordinate block with prod defaults
(env, instance, host, db.name, vault.k8sRole, vault.dynamicPath, vault.staticPath).
The elision rule (env=prod → no suffix, env=non-prod → "<app>-<env>" suffix)
guarantees the prod render is unchanged.

chart/values-sandbox.yaml is added as the ready-to-use overlay for Phase D.
It is NOT wired into any helm install / ArgoCD app today — the platform side
(factory/postgres/iac tfvars, tools/hashicorp-vault/iac module signature) is
not yet evolved. The file documents the convention so the Phase D commit can
just `helm install -f values.yaml -f values-sandbox.yaml`.

Also fixes .gitea/workflows/vault.yaml CI typo: the vault_step JWT role was
gitea_cicd_webapp (copy-paste from the template repo) instead of
gitea_cicd_erp. Real bug — the erp CI would have failed JWT auth against
Vault. Fix unrelated to multi-env but bundled here because it's small and
touches the same file family.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-31 23:26:20 +02:00
444886b91a Merge pull request 'arcodange-email-ingest V8.1: filter calendar invites + newsletter senders' (#10) from claude/arcodange-email-ingest-v81 into main 2026-05-31 15:18:58 +02:00
1d38f25c23 arcodange-email-ingest V8.1: filter calendar invites + newsletter senders
email-list.sh gains two hard-exclusion filters (applied before the
candidate test, regardless of attachments):

- EXCLUDE_PATTERN matches subjects starting with Invitation: / Updated
  invitation: / Canceled event: / Accepted: / Declined: / Tentative: /
  Maybe: (after stripping Re:/Fwd:/Tr: prefixes). Filters Google Calendar
  events that always carry an .ics attachment.
- EXCLUDE_SENDER matches updates.<domain>, noreply@*calendar, news@,
  newsletter@. Filters newsletter blast traffic.

Effect on --all-folders --candidates-only baseline: 27 noisy → 12
actionable (calendar invites + the staying-ahead.ai newsletter blast
removed). Real supplier docs intact: Darnis F1042 in /Notification, 3 Free
Mobile factures in /Inbox/abonnements, Mistral + Anthropic in /Inbox/books.

The originally-planned --mark-ingested feature is deferred to V8.2:
flag-set requires the Zoho OAuth scope ZohoMail.messages.UPDATE which our
read-only refresh_token doesn't have. Documented in SKILL.md: once the
user opts in to the wider scope, --mark-ingested becomes a one-line flag
on email-inspect.sh and is_candidate() learns to skip flag_info messages.

Captured the new --all-folders baseline at examples/email-list-all-folders.txt.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-31 15:18:31 +02:00
794aa18d2a Merge pull request 'add arcodange-email-ingest — Zoho Mail → Dolibarr supplier-invoice drafts' (#9) from claude/arcodange-email-ingest into main 2026-05-31 14:56:58 +02:00
c2d8479f5e add arcodange-email-ingest — Zoho Mail → Dolibarr supplier-invoice drafts
V8 — first inbound-side skill. Closes the loop from "bill arrives by email"
to "ready to enter in Dolibarr UI". Read-only at every layer.

What ships:
- arcodange-email-ingest/scripts/zoho-curl.sh   OAuth wrapper with token cache
                                                (50 min TTL, mode 600) — avoids
                                                hitting Zoho OAuth rate limit on
                                                every invocation.
- arcodange-email-ingest/scripts/email-list.sh   List candidates in /Inbox/books
                                                (where the books@ alias auto-
                                                routes mail). --candidates-only
                                                filter on supplier patterns or
                                                attachments. --all-folders to
                                                scan everything.
- arcodange-email-ingest/scripts/email-inspect.sh   Pull message + attachments,
                                                pdftotext on each PDF, heuristic
                                                extract (supplier, ref, dates,
                                                totals, VAT rate), emit Dolibarr
                                                supplier-invoice draft JSON.

Architecture choice — Zoho API (not IMAP):
- books@arcodange.fr is an alias of gabrielradureau@arcodange.fr → one OAuth
  refresh_token covers everything.
- Gmail folded in via forwarding (arcodange@gmail.com → books@) — no Google
  API setup, no app-passwords, no second OAuth flow.
- Token-based auth, no SCA rabbit hole.

V8.0 baseline (in /Inbox/books):
- 3 candidates: Mistral AI facture, Anthropic Stripe receipt (Fwd Gmail),
  INPI payment receipt (Fwd Gmail).
- Heuristic extraction is best-effort: works on amounts/refs for some
  templates, misses others (Mistral PDF format, Stripe receipt layout).
- --save-pdf <DIR> lets the operator grab the PDFs for manual entry when
  the heuristic falls short.

Rate-limit pitfall documented: Zoho OAuth refresh has an aggressive throttle
("too many requests continuously"). The cache file at $TMPDIR/zoho-access-$USER
(mode 600, 50 min TTL) prevents this; on 401 the wrapper auto-refreshes once
and retries.

V8.1+ ideas in SKILL.md out-of-scope:
- mark ingested emails (IMAP flag or Zoho label)
- body text extraction (inline-HTML invoices)
- per-template parsers or LLM-based extraction
- IMAP fallback for non-Zoho mailboxes

CLI: bin/arcodange email {list|inspect|curl} integrated.
Base updates: dolibarr/SKILL.md cross-link, dolibarr/README.md env schema
extended with ZOHO_CLIENT_ID/SECRET/REFRESH_TOKEN/DC.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-31 14:56:15 +02:00
a1042a483b Merge pull request 'arcodange-bank-reco V7: avoir netting + fk_account context + wire-ref matching' (#8) from claude/arcodange-bank-reco-v7 into main 2026-05-31 14:20:37 +02:00
246c7fc5a9 arcodange-bank-reco V7: avoir netting + fk_account context + wire-ref matching
Three improvements that reduce the V6.1 exit-1 signal from 10 to 1 on
the current Arcodange baseline. Every bucket now has a single, clear
purpose; the only entry counted as a failure is a genuine action item.

A. fk_account context on dolibarr-only
   - Fetches /bankaccounts and tags each dolibarr-only with the account
     ref + label (e.g. "CCA1 (G.RADUREAU Compte Courant Asso)").
   - Splits dolibarr-only into "on API-tracked accounts" (QON*/WIS* — real
     gaps) vs "not in API scope" (CCA1 / personal — expected gaps).
   - Personal-account entries no longer count toward the failure verdict.

B. Avoir-cycle netting
   - Pairs AVC entries of -X on socid S with FAC entries of +X on the
     same socid within ±5d.
   - Both surface in a dedicated AVOIR-NETTED bucket and are excluded from
     dolibarr-only, since the bank only sees the net of the cycle.
   - Resolves the V6.1 noise where AVC001-CL0001001 + FAC001-CL00001
     appeared as fake gaps for a 510€ cancel-and-reissue dance.

C. Wire-reference strong matching (--enrich flag, opt-in)
   - When --enrich is passed, bank-match.sh fetches /v1/transfers/{id}
     per Wise TRANSFER and reads the wire `reference` field.
   - References containing a FAC\d+(CL\d+)? pattern strong-match against
     the corresponding Dolibarr customer invoice (annotated [wire-ref]
     vs the loose [amt+date] kind).
   - Verified on FAC002 5100€: KM's wire memo "FOR INVOICE FAC002CL0001002"
     gives an unambiguous match independent of date drift.

Baseline (Jan-May 2026, --enrich on):
  6 matched · 1 internal · 2 avoir-netted · 7 bank-known · 1 bank-UNKNOWN
  0 dol-only-API · 7 dol-only-personal
  → exit-1 count = 1 (just the +2147€ KM Wise 2026-05-29 to record).

The CLI (bin/arcodange) gains --enrich on the match subcommand. The
SKILL.md has a new "V7 bucket structure" section explaining the seven
buckets and a before/after table showing the signal/noise improvement.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-31 14:20:06 +02:00
0f5b6bcbad Merge pull request 'arcodange-bank-reco: known-patterns catalog + annotated bank-only buckets' (#7) from claude/arcodange-bank-reco-patterns into main 2026-05-31 14:07:39 +02:00
4b6a5f7529 arcodange-bank-reco: add known-patterns.json catalog + bank-match annotation
V6.1 follow-up to the bank-reco V6 ship. Splits the BANK-ONLY bucket into
"known patterns" (intentional gaps, documented and classified) vs
"unknown" (real action items).

What the catalog covers today:
- FOUREZ Quentin → capital_deposit (apport en capital 1000 € initial,
  notaire FOUREZ centralisateur du dépôt). Maps to Dolibarr account 1013.
- URSSAF → social_charges (account 645100)
- MISTRAL.AI, CLAUDE.AI → ai_subscription (account 6262)
- Wise *Plan, qonto_fee → bank_fee (account 627)
- BALANCE_DEPOSIT / FEATURE_CHARGE on Wise → internal_topup (self-funding
  pair, often nets to zero)

Effect on the V6 baseline (Jan-May 2026):
- Before catalog: 8 BANK-ONLY mixed entries (noise + signal)
- After catalog:  7 known + 1 UNKNOWN (just the +2147 € KM Wise payment
  2026-05-29 that genuinely needs a Dolibarr entry)

The catalog is JSON (not YAML — stdlib only, no dependency). Schema
documented in SKILL.md. Pattern matches case-insensitive regex against
both bank label AND operation type. Optional filters: bank, side,
amount_min, amount_max.

Exit code now reflects only the UNKNOWN bank-only and dolibarr-only
counts — the verdict is no longer noisy because of intentional gaps.

Edit known-patterns.json as new recurring patterns emerge.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-31 14:06:46 +02:00
40 changed files with 2491 additions and 98 deletions

View File

@@ -124,6 +124,90 @@ Current state (V1 baseline):
- Wise **STANDARD EUR** : 5 308,25 € live - Wise **STANDARD EUR** : 5 308,25 € live
- **Total bank-side** : 9 499,79 € - **Total bank-side** : 9 499,79 €
## Known-patterns catalog ([known-patterns.json](known-patterns.json))
Bank movements that have no Dolibarr counterpart fall into two groups:
1. **Intentional gaps** — operational expenses or one-off events the operator knows about (URSSAF mensuel, AI subs, capital deposit, Wise plan fees). These keep recurring but their accounting treatment is well-understood.
2. **Real action items** — incoming payments not yet entered, expenses missing a supplier invoice, anomalies.
Without a catalog, both look identical in the BANK-ONLY bucket — noise drowns the signal. The catalog is an operator-curated list of patterns; `bank-match.sh` reads it and splits BANK-ONLY into two sub-buckets:
- **BANK-ONLY — known patterns** : annotated with `[classification]` + a one-line note (which Dolibarr account to use, etc.). Don't action; just verify.
- **BANK-ONLY — unknown** : the real signal. Each entry is either a missing supplier invoice, an unrecorded payment, or a new pattern to add to the catalog.
**Schema** (JSON, see [known-patterns.json](known-patterns.json) for current entries):
```json
{
"patterns": [
{
"pattern": "regex (case-insensitive, matched against bank label + operation type)",
"classification": "capital_deposit | social_charges | ai_subscription | bank_fee | internal_topup | personal_apport | needs_classification",
"bank": "qonto | wise (optional, default both)",
"side": "credit | debit (optional, default both)",
"amount_min": 0.0, "amount_max": 99999.0, // optional numeric bounds
"note": "human-readable context — what this is, which Dolibarr account, recurring schedule, etc."
}
]
}
```
**Editing workflow:**
1. Run `bin/arcodange bank match` → look at BANK-ONLY unknown.
2. For each recurring entry that's "expected", add a pattern to `known-patterns.json`.
3. Re-run match → the entry should now appear in the known sub-bucket.
4. For one-off action items (e.g. "+2147 € KM May 29 not in Dolibarr"), don't add a pattern — enter it in Dolibarr instead.
**V6.1 catalog** ships with these patterns for the current Arcodange baseline:
- `FOUREZ Quentin` → capital_deposit (initial 1000 € apport via notaire, 2026-01-21)
- `URSSAF` → social_charges
- `MISTRAL.AI` / `CLAUDE.AI` → ai_subscription
- `Wise *Plan` → bank_fee (Wise account plan billed via Qonto card)
- `qonto_fee` → bank_fee
- `BALANCE_DEPOSIT|For your account plan` → internal_topup (the Wise +50/-50 self-funding pair)
After applying the catalog to the V6 baseline, the **only remaining BANK-UNKNOWN** is the **+2147 € KissMetrics payment on 2026-05-29** that hasn't been entered in Dolibarr — the actual signal.
## V7 bucket structure
V7 adds three improvements that reshape the output buckets:
| Bucket | Meaning | Counts toward exit-1? |
|---|---|---|
| **MATCHED** | Bank ↔ Dolibarr paired. Annotated with match kind: `[wire-ref]` (strong, via `--enrich`) or `[amt+date]` (loose). | No |
| **INTERNAL** | Wise↔Qonto consolidations (5000€ moved between Arcodange's own accounts). | No |
| **AVOIR-NETTED** | Dolibarr AVC + FAC cancellation cycles paired and excluded (the bank only saw the net). | No |
| **BANK-ONLY — known patterns** | Bank movement with a `known-patterns.json` annotation. Intentional gap. | No |
| **BANK-ONLY — unknown** | Bank movement with no Dolibarr counterpart AND no catalog pattern. **Real action item**. | Yes |
| **DOLIBARR-ONLY — on API-tracked accounts** (QON*/WIS*) | Dolibarr payment that the bank should have shown. **Real gap**. | Yes |
| **DOLIBARR-ONLY — not in API scope** (CCA1 perso etc.) | Expected gap — we have no API on those accounts. | No |
Exit code 0 iff the two "real gap" buckets are empty.
### `--enrich` — wire-reference strong matching
`bank-match.sh --enrich` fetches `/v1/transfers/{id}` for each Wise TRANSFER and reads the `reference` field (the wire memo from the sender, e.g. `FROM KISSMETRICS HOLDINGS INC FOR INVOICE FAC002CL0001002/ VENDOR:DEV`). When the reference contains a `FAC\d+(CL\d+)?` pattern matching a Dolibarr customer invoice, that pairing takes precedence over the loose date+amount match. Only the strong-matched ones get `[wire-ref]`; the rest fall through to `[amt+date]`. Cost: 1 extra HTTP call per Wise transfer.
### Avoir cycle netting
When Arcodange cancels and reissues an invoice (FAC001 → AVC001 + FAC001-NEW), the bank sees one net credit but Dolibarr stores 3 payment entries. V7 pairs AVC entries of -X with FAC entries of +X for the same socid within ±5d, surfaces them in **AVOIR-NETTED**, and excludes them from `dolibarr-only`. Removes the V6.1 noise where AVC001 + FAC001-CL00001 appeared as fake gaps.
### fk_account context
`bank-match.sh` now fetches `/bankaccounts` and tags `dolibarr-only` entries with their account ref + label. Splits into API-tracked (QON*/WIS* — real gaps) vs not-in-scope (everything else — expected). The 7 CCA1 personal-account entries that used to look like failures are now correctly classified as expected gaps.
### Effect on the baseline
| | V6 | V6.1 | V7 |
|---|---|---|---|
| MATCHED | 6 (all amt+date) | 6 | 6 (1 wire-ref strong + 5 amt+date when --enrich) |
| BANK-ONLY total | 8 mixed | 7 known + 1 UNKNOWN | 7 known + 1 UNKNOWN |
| AVOIR-NETTED | — | — | 2 (silently absorbed) |
| DOL-only TRUE GAP | 9 (noisy) | 9 (noisy) | **0** |
| DOL-only EXPECTED | — | — | 7 (CCA1 personal) |
| Exit-1 signal count | 17 (noise) | 10 (less noise) | **1** (just the +2147€ KM) |
## Matching heuristic — what's in v1 and what's V7 ## Matching heuristic — what's in v1 and what's V7
Today's match logic: Today's match logic:

View File

@@ -1,36 +1,50 @@
# Bank reconciliation: 2026-01-01 → 2026-05-31 (window ±7d, fees: off) # Bank reconciliation: 2026-01-01 → 2026-05-31 (window ±7d, fees: off, enrich: on)
=== MATCHED (6 bank ↔ Dolibarr) === === MATCHED (6 bank ↔ Dolibarr) ===
Qonto 2026-01-27 - 50.00 card Wise *Plan ↔ supplier FAF2026001 (2026-01-26, Δ-1d) Qonto 2026-01-27 - 50.00 card Wise *Plan ↔[amt+date] supplier FAF2026001 (2026-01-26, Δ-1d)
Wise 2026-02-05 + 510.00 TRANSFER Kissmetrics Holdings Inc ↔ customer FAC001-CL0001001 (2026-02-05, Δ+0d) Wise 2026-02-05 + 510.00 TRANSFER Kissmetrics Holdings Inc ↔[amt+date] customer FAC001-CL0001001 (2026-02-05, Δ+0d)
Wise 2026-03-06 + 5100.00 TRANSFER Kissmetrics Holdings Inc ↔ customer FAC002-CL0001002 (2026-03-12, Δ+6d) Wise 2026-03-06 + 5100.00 TRANSFER Kissmetrics Holdings Inc ↔[wire-ref] customer FAC002-CL0001002 (2026-03-12, Δ+6d)
Qonto 2026-03-13 - 612.00 transfer DARNIS OPERATIONS ↔ supplier FAF2026008 (2026-03-13, Δ+0d) Qonto 2026-03-13 - 612.00 transfer DARNIS OPERATIONS ↔[amt+date] supplier FAF2026008 (2026-03-13, Δ+0d)
Wise 2026-04-20 + 2550.00 TRANSFER Kissmetrics Holdings Inc ↔ customer FAC003-CL0001003 (2026-04-20, Δ+0d) Wise 2026-04-20 + 2550.00 TRANSFER Kissmetrics Holdings Inc ↔[amt+date] customer FAC003-CL0001003 (2026-04-20, Δ+0d)
Qonto 2026-05-10 - 306.00 transfer DARNIS OPERATIONS ↔ supplier FAF2026009 (2026-05-10, Δ+0d) Qonto 2026-05-10 - 306.00 transfer DARNIS OPERATIONS ↔[amt+date] supplier FAF2026009 (2026-05-10, Δ+0d)
=== INTERNAL (Wise↔Qonto consolidations, 1) === === INTERNAL (Wise↔Qonto consolidations, 1) ===
Wise 2026-03-13 - 5000.00 TRANSFER ARCODANGE ↔ Qonto 2026-03-13 +5000.00 Wise 2026-03-13 - 5000.00 TRANSFER ARCODANGE ↔ Qonto 2026-03-13 +5000.00
=== BANK-ONLY (8 bank movements without Dolibarr counterpart) === === AVOIR-NETTED (2 Dolibarr entries pairing AVC↔FAC cancellation cycles) ===
Qonto 2026-01-16 + 5.22 qonto_fee Qonto customer 2026-02-05 -510.00 AVC001-CL0001001 ↔ netted against FAC001-CL00001
Qonto 2026-01-21 + 1000.00 income FOUREZ Quentin customer 2026-02-05 510.00 FAC001-CL00001 ↔ netted against AVC001-CL0001001
Wise 2026-01-26 - 50.00 FEATURE_CHARGE For your account plan
Wise 2026-01-26 + 50.00 BALANCE_DEPOSIT To EUR === BANK-ONLY — known patterns (7, intentional gaps documented in known-patterns.json) ===
Qonto 2026-04-03 - 172.68 card MISTRAL.AI Qonto 2026-01-16 + 5.22 qonto_fee Qonto [bank_fee]
Qonto 2026-04-13 - 180.00 card CLAUDE.AI SUBSCRIPTION └─ Qonto fees ou refunds. Petites valeurs. Dolibarr: account 627.
Qonto 2026-05-22 - 493.00 direct_debit URSSAF D ILE DE FRANCE Qonto 2026-01-21 + 1000.00 income FOUREZ Quentin [capital_deposit]
└─ Apport en capital social initial 1000 €. Maître FOUREZ Quentin, notaire centralisateur du dépôt. Date typique : 2026-01-21. Dolibarr: account 1013.
Wise 2026-01-26 - 50.00 FEATURE_CHARGE For your account plan [internal_topup]
└─ Solde Wise rechargé pour couvrir un frais immédiat (souvent net zéro avec le FEATURE_CHARGE du même jour).
Wise 2026-01-26 + 50.00 BALANCE_DEPOSIT To EUR [internal_topup]
└─ Solde Wise rechargé pour couvrir un frais immédiat (souvent net zéro avec le FEATURE_CHARGE du même jour).
Qonto 2026-04-03 - 172.68 card MISTRAL.AI [ai_subscription]
└─ Mistral AI API subscription. Récurrent mensuel. Dolibarr: account 6262 + supplier 'Mistral AI'.
Qonto 2026-04-13 - 180.00 card CLAUDE.AI SUBSCRIPTION [ai_subscription]
└─ Claude AI subscription (Anthropic). Récurrent mensuel. Dolibarr: account 6262 + supplier 'Anthropic'.
Qonto 2026-05-22 - 493.00 direct_debit URSSAF D ILE DE FRANCE [social_charges]
└─ Cotisations sociales URSSAF (régime mensuel/trimestriel). Dolibarr: account 645100 (charges de sécurité sociale).
=== BANK-ONLY — unknown (1, NEEDS attention: missing supplier invoice / unrecorded payment / new pattern) ===
Wise 2026-05-29 + 2147.00 TRANSFER Kissmetrics Holdings Inc Wise 2026-05-29 + 2147.00 TRANSFER Kissmetrics Holdings Inc
=== DOLIBARR-ONLY (9 Dolibarr payments without bank movement) === === DOLIBARR-ONLY — on API-tracked accounts (0, REAL GAP: bank should have shown this) ===
supplier 2026-01-04 1.99 FAF2026003 (fk_account=3)
supplier 2026-01-06 202.80 FAF2026005 (fk_account=3)
supplier 2026-01-09 55.93 FAF2026002 (fk_account=3)
supplier 2026-01-09 148.80 FAF2026004 (fk_account=3)
supplier 2026-01-12 8.43 FAF2026006 (fk_account=3)
supplier 2026-01-15 1.30 FAF2026002 (fk_account=3)
supplier 2026-01-17 3.20 FAF2026007 (fk_account=3)
customer 2026-02-05 -510.00 AVC001-CL0001001 (fk_account=2)
customer 2026-02-05 510.00 FAC001-CL00001 (fk_account=2)
-------------------------------------------------------------------------------- === DOLIBARR-ONLY — on accounts NOT in API scope (7, expected gap: CCA1 perso etc.) ===
# 6 matched, 1 internal, 8 bank-only, 9 dolibarr-only supplier 2026-01-04 1.99 FAF2026003 (CCA1 (G.RADUREAU Compte Courant Asso))
supplier 2026-01-06 202.80 FAF2026005 (CCA1 (G.RADUREAU Compte Courant Asso))
supplier 2026-01-09 55.93 FAF2026002 (CCA1 (G.RADUREAU Compte Courant Asso))
supplier 2026-01-09 148.80 FAF2026004 (CCA1 (G.RADUREAU Compte Courant Asso))
supplier 2026-01-12 8.43 FAF2026006 (CCA1 (G.RADUREAU Compte Courant Asso))
supplier 2026-01-15 1.30 FAF2026002 (CCA1 (G.RADUREAU Compte Courant Asso))
supplier 2026-01-17 3.20 FAF2026007 (CCA1 (G.RADUREAU Compte Courant Asso))
----------------------------------------------------------------------------------------------------
# 6 matched, 1 internal, 2 avoir-netted, 7 bank-known, 1 bank-UNKNOWN, 0 dol-only-API, 7 dol-only-personal
# patterns loaded from /Users/gabrielradureau/Work/Arcodange/erp/.claude/worktrees/happy-wilson-ee5645/.claude/skills/arcodange-bank-reco/scripts/../known-patterns.json: 7 pattern(s)

View File

@@ -0,0 +1,60 @@
{
"_schema": "v1",
"_description": "Operator-curated catalogue of known recurring/intentional bank movements. Used by bank-match.sh to annotate the BANK-ONLY bucket so the operator can immediately tell 'needs Dolibarr entry' from 'documented intentional gap'. Edit this file as new recurring patterns emerge.",
"_match_rules": "Pattern matched case-insensitively as a regex against the bank label. Optional filters: bank (qonto|wise), side (credit|debit), amount_min, amount_max, type (Wise activity type). All present filters must match.",
"_classifications": {
"capital_deposit": "Apport en capital social. Dolibarr account 1013 (capital souscrit appelé versé).",
"social_charges": "URSSAF, retraite complémentaire, etc. Dolibarr account 645x.",
"ai_subscription": "Claude / Mistral / OpenAI / similar. Dolibarr account 6262 (frais télécom / abonnements logiciels).",
"bank_fee": "Plan bancaire, frais d'opération, refunds. Dolibarr account 627 (services bancaires).",
"internal_topup": "Solde Wise/Qonto rechargé pour couvrir un frais immédiat. Often nets out.",
"personal_apport": "Apport en compte courant d'associé (Gabriel finançant Arcodange depuis son perso). Dolibarr account 4551.",
"needs_classification": "Pattern catched but no Dolibarr account assignment defined yet; surface for review."
},
"patterns": [
{
"pattern": "FOUREZ.*Quentin",
"classification": "capital_deposit",
"bank": "qonto",
"side": "credit",
"note": "Apport en capital social initial 1000 €. Maître FOUREZ Quentin, notaire centralisateur du dépôt. Date typique : 2026-01-21. Dolibarr: account 1013."
},
{
"pattern": "URSSAF",
"classification": "social_charges",
"bank": "qonto",
"side": "debit",
"note": "Cotisations sociales URSSAF (régime mensuel/trimestriel). Dolibarr: account 645100 (charges de sécurité sociale)."
},
{
"pattern": "MISTRAL\\.AI",
"classification": "ai_subscription",
"side": "debit",
"note": "Mistral AI API subscription. Récurrent mensuel. Dolibarr: account 6262 + supplier 'Mistral AI'."
},
{
"pattern": "CLAUDE\\.AI",
"classification": "ai_subscription",
"side": "debit",
"note": "Claude AI subscription (Anthropic). Récurrent mensuel. Dolibarr: account 6262 + supplier 'Anthropic'."
},
{
"pattern": "Wise.*Plan",
"classification": "bank_fee",
"side": "debit",
"note": "Wise account plan billed via card. Wise's internal fee for keeping the BUSINESS profile active."
},
{
"pattern": "qonto_fee",
"classification": "bank_fee",
"bank": "qonto",
"note": "Qonto fees ou refunds. Petites valeurs. Dolibarr: account 627."
},
{
"pattern": "BALANCE_DEPOSIT|For your account plan",
"classification": "internal_topup",
"bank": "wise",
"note": "Solde Wise rechargé pour couvrir un frais immédiat (souvent net zéro avec le FEATURE_CHARGE du même jour)."
}
]
}

View File

@@ -24,7 +24,7 @@ SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
BANK_CURL="${SCRIPT_DIR}/bank-curl.sh" BANK_CURL="${SCRIPT_DIR}/bank-curl.sh"
DOL_CURL="${SCRIPT_DIR}/../../dolibarr/scripts/dol-curl.sh" DOL_CURL="${SCRIPT_DIR}/../../dolibarr/scripts/dol-curl.sh"
SINCE=""; UNTIL=""; MONTH=""; WINDOW=7; INCLUDE_FEES=0 SINCE=""; UNTIL=""; MONTH=""; WINDOW=7; INCLUDE_FEES=0; ENRICH=0
while [[ $# -gt 0 ]]; do while [[ $# -gt 0 ]]; do
case "$1" in case "$1" in
--since) SINCE="$2"; shift 2 ;; --since) SINCE="$2"; shift 2 ;;
@@ -32,6 +32,7 @@ while [[ $# -gt 0 ]]; do
--month) MONTH="$2"; shift 2 ;; --month) MONTH="$2"; shift 2 ;;
--window-days) WINDOW="$2"; shift 2 ;; --window-days) WINDOW="$2"; shift 2 ;;
--include-fees) INCLUDE_FEES=1; shift ;; --include-fees) INCLUDE_FEES=1; shift ;;
--enrich) ENRICH=1; shift ;;
-h|--help) sed -n '2,18p' "$0" | sed 's/^# \{0,1\}//'; exit 0 ;; -h|--help) sed -n '2,18p' "$0" | sed 's/^# \{0,1\}//'; exit 0 ;;
*) echo "bank-match.sh: unknown arg: $1" >&2; exit 2 ;; *) echo "bank-match.sh: unknown arg: $1" >&2; exit 2 ;;
esac esac
@@ -66,9 +67,10 @@ QURL="/v2/transactions?bank_account_id=${QONTO_ACCT}&settled_at_from=${SINCE}T00
# --- 2. Pull Wise activities --- # --- 2. Pull Wise activities ---
"${BANK_CURL}" wise "/v1/profiles/${WISE_PROFILE_ID}/activities?size=100&since=${SINCE}T00:00:00.000Z&until=${UNTIL}T23:59:59.999Z" > "${WORK}/wise.json" "${BANK_CURL}" wise "/v1/profiles/${WISE_PROFILE_ID}/activities?size=100&since=${SINCE}T00:00:00.000Z&until=${UNTIL}T23:59:59.999Z" > "${WORK}/wise.json"
# --- 3. Pull Dolibarr customer + supplier invoices and their payments --- # --- 3. Pull Dolibarr customer + supplier invoices, payments, and bank accounts ---
"${DOL_CURL}" '/invoices?limit=500&sortfield=t.datef&sortorder=ASC' > "${WORK}/dol_inv.json" "${DOL_CURL}" '/invoices?limit=500&sortfield=t.datef&sortorder=ASC' > "${WORK}/dol_inv.json"
"${DOL_CURL}" '/supplierinvoices?limit=500' > "${WORK}/dol_sup.json" "${DOL_CURL}" '/supplierinvoices?limit=500' > "${WORK}/dol_sup.json"
"${DOL_CURL}" '/bankaccounts' > "${WORK}/dol_acct.json"
mkdir -p "${WORK}/dol_pay" "${WORK}/dol_supay" mkdir -p "${WORK}/dol_pay" "${WORK}/dol_supay"
for id in $(python3 -c "import json,sys; print(' '.join(str(r['id']) for r in json.load(open(sys.argv[1])) if r.get('id')))" "${WORK}/dol_inv.json"); do for id in $(python3 -c "import json,sys; print(' '.join(str(r['id']) for r in json.load(open(sys.argv[1])) if r.get('id')))" "${WORK}/dol_inv.json"); do
@@ -78,11 +80,26 @@ for id in $(python3 -c "import json,sys; print(' '.join(str(r['id']) for r in js
"${DOL_CURL}" "/supplierinvoices/${id}/payments" > "${WORK}/dol_supay/${id}.json" 2>/dev/null || echo "[]" > "${WORK}/dol_supay/${id}.json" "${DOL_CURL}" "/supplierinvoices/${id}/payments" > "${WORK}/dol_supay/${id}.json" 2>/dev/null || echo "[]" > "${WORK}/dol_supay/${id}.json"
done done
# --- 3b. Optional: enrich Wise TRANSFER activities with wire references ---
if [[ "${ENRICH}" == "1" ]]; then
mkdir -p "${WORK}/wise_refs"
for tid in $(python3 -c "
import json, sys
acts = json.load(open(sys.argv[1])).get('activities') or []
for a in acts:
r = a.get('resource') or {}
if r.get('type')=='TRANSFER' and r.get('id'): print(r['id'])
" "${WORK}/wise.json"); do
"${BANK_CURL}" wise "/v1/transfers/${tid}" > "${WORK}/wise_refs/${tid}.json" 2>/dev/null || true
done
fi
# --- 4. Match in python --- # --- 4. Match in python ---
python3 - "${WORK}" "${SINCE}" "${UNTIL}" "${WINDOW}" "${INCLUDE_FEES}" <<'PY' PATTERNS_FILE="${SCRIPT_DIR}/../known-patterns.json"
python3 - "${WORK}" "${SINCE}" "${UNTIL}" "${WINDOW}" "${INCLUDE_FEES}" "${PATTERNS_FILE}" "${ENRICH}" <<'PY'
import json, sys, os, re, datetime, collections import json, sys, os, re, datetime, collections
work, since, until, window_days, include_fees = sys.argv[1:6] work, since, until, window_days, include_fees, patterns_file, enrich = sys.argv[1:8]
window = int(window_days); include_fees = include_fees == "1" window = int(window_days); include_fees = include_fees == "1"; enrich = enrich == "1"
since_d = datetime.date.fromisoformat(since); until_d = datetime.date.fromisoformat(until) since_d = datetime.date.fromisoformat(since); until_d = datetime.date.fromisoformat(until)
def strip(s): return re.sub(r'<[^>]+>', '', s or '').strip() def strip(s): return re.sub(r'<[^>]+>', '', s or '').strip()
@@ -110,7 +127,22 @@ for a in (json.load(open(os.path.join(work,"wise.json"))).get("activities") or [
m = re.search(r'([\d,.]+)\s*([A-Z]{3})', pa) m = re.search(r'([\d,.]+)\s*([A-Z]{3})', pa)
amt = float(m.group(1).replace(",", "")) if m else 0.0 amt = float(m.group(1).replace(",", "")) if m else 0.0
title = strip(a.get("title") or "")[:40] title = strip(a.get("title") or "")[:40]
wise_movs.append({"bank":"Wise", "date":dt, "sign":sign, "amount":amt, "label":title, "op":typ, "matched_dol":None, "matched_internal":False}) res = a.get("resource") or {}
resource_id = str(res.get("id")) if res.get("type") == "TRANSFER" else None
wise_movs.append({"bank":"Wise", "date":dt, "sign":sign, "amount":amt, "label":title, "op":typ, "matched_dol":None, "matched_internal":False, "wise_resource_id":resource_id, "wire_ref":""})
# 4b'. If --enrich, load per-transfer wire references and attach to Wise movs
if enrich:
ref_dir = os.path.join(work, "wise_refs")
if os.path.isdir(ref_dir):
for m in wise_movs:
if not m["wise_resource_id"]: continue
p = os.path.join(ref_dir, f"{m['wise_resource_id']}.json")
if not os.path.isfile(p): continue
try:
t = json.load(open(p))
m["wire_ref"] = (t.get("reference") or "")
except Exception: pass
bank_movs = qonto_movs + wise_movs bank_movs = qonto_movs + wise_movs
@@ -121,7 +153,7 @@ for w in [m for m in bank_movs if m["bank"]=="Wise" and m["sign"]=="-"]:
w["matched_internal"] = q; q["matched_internal"] = w w["matched_internal"] = q; q["matched_internal"] = w
break break
# 4d. Normalize Dolibarr payments # 4d. Normalize Dolibarr payments — carry socid too for avoir netting
dol_pays = [] dol_pays = []
inv_by_id = {str(r["id"]): r for r in json.load(open(os.path.join(work,"dol_inv.json")))} inv_by_id = {str(r["id"]): r for r in json.load(open(os.path.join(work,"dol_inv.json")))}
for fn in os.listdir(os.path.join(work,"dol_pay")): for fn in os.listdir(os.path.join(work,"dol_pay")):
@@ -131,7 +163,9 @@ for fn in os.listdir(os.path.join(work,"dol_pay")):
d = datetime.datetime.strptime(p["date"], "%Y-%m-%d %H:%M:%S").date() d = datetime.datetime.strptime(p["date"], "%Y-%m-%d %H:%M:%S").date()
if d < since_d or d > until_d: continue if d < since_d or d > until_d: continue
amt = float(p.get("amount") or 0) amt = float(p.get("amount") or 0)
dol_pays.append({"side":"customer", "ref":inv["ref"], "date":d, "amount":amt, "fk_account":inv.get("fk_account"), "matched_bank":None}) dol_pays.append({"side":"customer", "ref":inv["ref"], "date":d, "amount":amt,
"fk_account":inv.get("fk_account"), "socid":inv.get("socid"),
"matched_bank":None, "netted_against":None})
sup_by_id = {str(r["id"]): r for r in json.load(open(os.path.join(work,"dol_sup.json")))} sup_by_id = {str(r["id"]): r for r in json.load(open(os.path.join(work,"dol_sup.json")))}
for fn in os.listdir(os.path.join(work,"dol_supay")): for fn in os.listdir(os.path.join(work,"dol_supay")):
@@ -141,37 +175,136 @@ for fn in os.listdir(os.path.join(work,"dol_supay")):
d = datetime.datetime.strptime(p["date"], "%Y-%m-%d %H:%M:%S").date() d = datetime.datetime.strptime(p["date"], "%Y-%m-%d %H:%M:%S").date()
if d < since_d or d > until_d: continue if d < since_d or d > until_d: continue
amt = float(p.get("amount") or 0) amt = float(p.get("amount") or 0)
dol_pays.append({"side":"supplier", "ref":sup["ref"], "date":d, "amount":amt, "fk_account":sup.get("fk_account"), "matched_bank":None}) dol_pays.append({"side":"supplier", "ref":sup["ref"], "date":d, "amount":amt,
"fk_account":sup.get("fk_account"), "socid":sup.get("socid"),
"matched_bank":None, "netted_against":None})
# 4e. Match: each bank movement (non-internal) tries to find a Dolibarr counterpart # 4d.1. AVOIR cycle netting: an AVC (credit note) for -X on socid S cancels out
for m in [x for x in bank_movs if not x["matched_internal"]]: # a FAC for +X on the same socid, within a small date window. Bank sees the NET
bank_signed = m["amount"] if m["sign"]=="+" else -m["amount"] # of the cycle (typically +X for the reissued FAC with the new ref scheme).
# For customer payments (Dol records them as positive amounts): +bank credit matches +dol customer payment # Pair an AVC with a FAC of opposite sign + equal abs(amount) + same socid +
# For supplier payments: -bank debit matches +dol supplier payment (positive in Dol since it's the amount paid out) # within ±5d. Mark both as "netted" so they're excluded from matching and
# Heuristic: match abs(amount) within 0.01 and date within window. # excluded from the dolibarr-only failure count.
candidates = [p for p in dol_pays if p["matched_bank"] is None and abs(p["amount"] - m["amount"]) < 0.01 and abs((p["date"] - m["date"]).days) <= window] avcs = [p for p in dol_pays if p["side"]=="customer" and p["ref"].startswith("AVC") and p["amount"] < 0]
for avc in avcs:
candidates = [p for p in dol_pays
if p is not avc
and p["side"]=="customer"
and p["socid"] == avc["socid"]
and abs(p["amount"] + avc["amount"]) < 0.01 # opposite signs equal magnitude
and abs((p["date"] - avc["date"]).days) <= 5
and p["netted_against"] is None
and p["matched_bank"] is None]
if candidates:
# Prefer the OLDEST (the original cancelled FAC), not the reissue.
# Heuristic: refs with shorter / older numbering scheme. If multiple,
# pick smallest date delta.
candidates.sort(key=lambda p: (abs((p["date"] - avc["date"]).days), p["ref"]))
partner = candidates[0]
avc["netted_against"] = partner["ref"]
partner["netted_against"] = avc["ref"]
# 4e. Match — two-pass:
# PASS 1 (strong) : Wise transfers with an --enrich'd wire reference containing
# a "FAC***" pattern try to match the Dolibarr invoice with
# that exact ref. This is the highest-confidence match.
# PASS 2 (loose) : remaining bank movements use the date+amount heuristic.
# Netted Dolibarr entries (avoir cycle) are excluded from both passes.
# Build customer ref -> dol payment index (only un-netted, un-matched entries)
ref_index = collections.defaultdict(list)
for p in dol_pays:
if p["matched_bank"] is None and p["netted_against"] is None:
# Strip trailing dash/suffix variants — FAC002CL0001002 vs FAC002-CL0001002 are equivalent
normalized = re.sub(r'[^A-Z0-9]', '', p["ref"].upper())
ref_index[normalized].append(p)
# Pass 1: strong match on wire references
for m in [x for x in bank_movs if not x["matched_internal"] and x.get("wire_ref")]:
refs_in_wire = re.findall(r'FAC\d+(?:CL\d+)?', (m["wire_ref"] or "").upper().replace("-",""))
for r in refs_in_wire:
if r in ref_index and ref_index[r]:
p = ref_index[r].pop(0)
m["matched_dol"] = p; m["match_kind"] = "wire-ref"
p["matched_bank"] = m
break
# Pass 2: loose date+amount match for remaining bank movements
for m in [x for x in bank_movs if not x["matched_internal"] and not x["matched_dol"]]:
candidates = [p for p in dol_pays
if p["matched_bank"] is None and p["netted_against"] is None
and abs(p["amount"] - m["amount"]) < 0.01
and abs((p["date"] - m["date"]).days) <= window]
if candidates: if candidates:
# Pick smallest date delta
candidates.sort(key=lambda p: abs((p["date"] - m["date"]).days)) candidates.sort(key=lambda p: abs((p["date"] - m["date"]).days))
p = candidates[0] p = candidates[0]
m["matched_dol"] = p; p["matched_bank"] = m m["matched_dol"] = p; m["match_kind"] = "amt+date"
p["matched_bank"] = m
# 4f. Annotate non-matched movements with known-patterns catalog
patterns = []
if os.path.isfile(patterns_file):
try: patterns = json.load(open(patterns_file)).get("patterns", [])
except Exception as e: print(f" /!\\ failed to load {patterns_file}: {e}", file=sys.stderr)
def match_pattern(mov):
# Match against both the bank label AND the operation type — different
# banks surface useful info in different fields (Qonto puts the operation
# type in `op`, e.g. "qonto_fee"; Wise puts the activity type in `op`,
# e.g. "BALANCE_DEPOSIT", and the human title in `label`).
haystack = (mov.get("label") or "") + " | " + (mov.get("op") or "")
for pat in patterns:
if pat.get("bank") and pat["bank"] != mov["bank"].lower(): continue
if pat.get("side") and pat["side"] != ("credit" if mov["sign"]=="+" else "debit"): continue
amin = pat.get("amount_min"); amax = pat.get("amount_max")
if amin is not None and mov["amount"] < amin: continue
if amax is not None and mov["amount"] > amax: continue
if re.search(pat["pattern"], haystack, re.IGNORECASE):
return pat
return None
for m in bank_movs:
if m["matched_dol"] or m["matched_internal"]: continue
m["known"] = match_pattern(m)
# --- 5. Render --- # --- 5. Render ---
# Load Dolibarr bank accounts (for fk_account context on dolibarr-only)
dol_accts = {}
try:
for a in json.load(open(os.path.join(work, "dol_acct.json"))):
dol_accts[str(a["id"])] = {"ref": a.get("ref","-"), "label": a.get("label","-"), "country": a.get("country_code","")}
except Exception: pass
# Heuristic: which Dolibarr accounts are NOT covered by Qonto/Wise API today?
# Convention: CCA = Compte Courant d'Associé (personal). Anything not QON*/WIS*
# is treated as "API-invisible" and tagged as such.
def account_kind(fk_account):
if not fk_account: return ("unknown", "fk_account=None")
a = dol_accts.get(str(fk_account))
if not a: return ("unknown", f"fk_account={fk_account} (not in /bankaccounts)")
ref = (a["ref"] or "").upper()
if ref.startswith(("QON", "WIS")):
return ("api_tracked", f"{a['ref']} ({a['label']})")
return ("personal_or_other", f"{a['ref']} ({a['label']})")
def fmt_bank(m): def fmt_bank(m):
return f" {m['bank']:<5} {m['date']} {m['sign']:<2}{m['amount']:>9.2f} {m['op'][:18]:<18} {m['label']}" return f" {m['bank']:<5} {m['date']} {m['sign']:<2}{m['amount']:>9.2f} {m['op'][:18]:<18} {m['label']}"
print(f"# Bank reconciliation: {since} → {until} (window ±{window}d, fees: {'on' if include_fees else 'off'})") print(f"# Bank reconciliation: {since} → {until} (window ±{window}d, fees: {'on' if include_fees else 'off'}, enrich: {'on' if enrich else 'off'})")
print() print()
matched = [m for m in bank_movs if m["matched_dol"]] matched = [m for m in bank_movs if m["matched_dol"]]
internal = [m for m in bank_movs if m["matched_internal"] and m["sign"]=="-"] internal = [m for m in bank_movs if m["matched_internal"] and m["sign"]=="-"]
bank_only = [m for m in bank_movs if not m["matched_dol"] and not m["matched_internal"]] bank_only = [m for m in bank_movs if not m["matched_dol"] and not m["matched_internal"]]
dol_only = [p for p in dol_pays if p["matched_bank"] is None] netted_dol_pairs = [p for p in dol_pays if p["netted_against"]]
dol_only = [p for p in dol_pays if p["matched_bank"] is None and p["netted_against"] is None]
print(f"=== MATCHED ({len(matched)} bank ↔ Dolibarr) ===") print(f"=== MATCHED ({len(matched)} bank ↔ Dolibarr) ===")
for m in sorted(matched, key=lambda m: m["date"]): for m in sorted(matched, key=lambda m: m["date"]):
p = m["matched_dol"] p = m["matched_dol"]
delta = (p["date"] - m["date"]).days delta = (p["date"] - m["date"]).days
print(fmt_bank(m) + f" ↔ {p['side']:<8} {p['ref']:<24} ({p['date']}, Δ{delta:+d}d)") kind = m.get("match_kind", "?")
print(fmt_bank(m) + f" ↔[{kind}] {p['side']:<8} {p['ref']:<24} ({p['date']}, Δ{delta:+d}d)")
print() print()
print(f"=== INTERNAL (Wise↔Qonto consolidations, {len(internal)}) ===") print(f"=== INTERNAL (Wise↔Qonto consolidations, {len(internal)}) ===")
@@ -180,19 +313,51 @@ for m in sorted(internal, key=lambda m: m["date"]):
print(fmt_bank(m) + f" ↔ {other['bank']} {other['date']} {other['sign']}{other['amount']:.2f}") print(fmt_bank(m) + f" ↔ {other['bank']} {other['date']} {other['sign']}{other['amount']:.2f}")
print() print()
print(f"=== BANK-ONLY ({len(bank_only)} bank movements without Dolibarr counterpart) ===") # Avoir cycles netted out (informational; bank correctly sees only the net)
for m in sorted(bank_only, key=lambda m: m["date"]): if netted_dol_pairs:
print(f"=== AVOIR-NETTED ({len(netted_dol_pairs)} Dolibarr entries pairing AVC↔FAC cancellation cycles) ===")
for p in sorted(netted_dol_pairs, key=lambda p: (p["date"], p["ref"])):
print(f" {p['side']:<8} {p['date']} {p['amount']:>9.2f} {p['ref']:<24} ↔ netted against {p['netted_against']}")
print()
bank_known = [m for m in bank_only if m.get("known")]
bank_unknown = [m for m in bank_only if not m.get("known")]
print(f"=== BANK-ONLY — known patterns ({len(bank_known)}, intentional gaps documented in known-patterns.json) ===")
for m in sorted(bank_known, key=lambda m: m["date"]):
k = m["known"]
cls = k.get("classification","?")
print(fmt_bank(m) + f" [{cls}]")
print(f" └─ {k.get('note','')}")
print()
print(f"=== BANK-ONLY — unknown ({len(bank_unknown)}, NEEDS attention: missing supplier invoice / unrecorded payment / new pattern) ===")
for m in sorted(bank_unknown, key=lambda m: m["date"]):
print(fmt_bank(m)) print(fmt_bank(m))
print() print()
print(f"=== DOLIBARR-ONLY ({len(dol_only)} Dolibarr payments without bank movement) ===") # Split dolibarr-only by whether the fk_account is API-tracked (real gap)
for p in sorted(dol_only, key=lambda p: p["date"]): # or personal_or_other (expected gap — we have no API on those accounts)
print(f" {p['side']:<8} {p['date']} {p['amount']:>9.2f} {p['ref']} (fk_account={p['fk_account']})") dol_only_api = [p for p in dol_only if account_kind(p["fk_account"])[0] == "api_tracked"]
dol_only_personal = [p for p in dol_only if account_kind(p["fk_account"])[0] != "api_tracked"]
print(f"=== DOLIBARR-ONLY — on API-tracked accounts ({len(dol_only_api)}, REAL GAP: bank should have shown this) ===")
for p in sorted(dol_only_api, key=lambda p: p["date"]):
_, ctx = account_kind(p["fk_account"])
print(f" {p['side']:<8} {p['date']} {p['amount']:>9.2f} {p['ref']:<24} ({ctx})")
print() print()
# Verdict print(f"=== DOLIBARR-ONLY — on accounts NOT in API scope ({len(dol_only_personal)}, expected gap: CCA1 perso etc.) ===")
fails = len(bank_only) + len(dol_only) for p in sorted(dol_only_personal, key=lambda p: p["date"]):
print("-" * 80) _, ctx = account_kind(p["fk_account"])
print(f"# {len(matched)} matched, {len(internal)} internal, {len(bank_only)} bank-only, {len(dol_only)} dolibarr-only") print(f" {p['side']:<8} {p['date']} {p['amount']:>9.2f} {p['ref']:<24} ({ctx})")
print()
# Verdict: only UNKNOWN bank-only AND dol-only-on-API-tracked count as failures.
# Avoir-netted pairs and personal-account dolibarr entries are intentional/expected.
fails = len(bank_unknown) + len(dol_only_api)
print("-" * 100)
print(f"# {len(matched)} matched, {len(internal)} internal, {len(netted_dol_pairs)} avoir-netted, {len(bank_known)} bank-known, {len(bank_unknown)} bank-UNKNOWN, {len(dol_only_api)} dol-only-API, {len(dol_only_personal)} dol-only-personal")
print(f"# patterns loaded from {patterns_file}: {len(patterns)} pattern(s)")
sys.exit(0 if fails == 0 else 1) sys.exit(0 if fails == 0 else 1)
PY PY

View File

@@ -0,0 +1,113 @@
---
name: arcodange-email-ingest
description: Scrape supplier-invoice emails from the Arcodange Zoho mailbox (`gabrielradureau@arcodange.fr` + its `books@arcodange.fr` alias + forwarded Gmail) via the Zoho Mail OAuth API, list candidates matching supplier patterns, download PDF attachments, run pdftotext + heuristic extract, and emit Dolibarr-ready supplier-invoice draft JSON for the operator to paste into the Dolibarr UI. Two workflows — (1) list candidates in a folder (default `/Inbox/books` where the alias auto-routes mail); (2) inspect one message by id, download + parse PDFs, propose draft entries. Surfaces concrete data: supplier name guess (first PDF line), invoice ref, invoice date, total HT/TVA/TTC, VAT rate. Read-only at every layer (Zoho scopes are READ-only; no write to Dolibarr). Use when the user asks "list pending supplier invoices in mail", "ingest invoices from email", "draft Dolibarr entry from this email", "audit cohort supplier docs from mail". Depends on `dolibarr` for the shared `.env`. SKIP for write-side Dolibarr operations (V9 candidate), for non-Zoho mailboxes (use IMAP fallback in a future skill if needed), and for attachments that aren't PDFs (only PDF text extraction is wired today).
requires:
bins: ["curl", "jq", "python3", "pdftotext"]
auth: true
---
# arcodange-email-ingest — supplier-invoice emails → Dolibarr draft
Close the inbound side of the accounting loop: bills land in `books@arcodange.fr`, this skill turns them into Dolibarr-ready draft entries for the operator to validate + create.
Depends on the [dolibarr](../dolibarr/SKILL.md) base skill (shared `.env`).
**CLI shortcuts:** `bin/arcodange email list | inspect | curl`
## Architecture choice — Zoho API, not IMAP
We chose the Zoho Mail OAuth API over IMAP because:
- **Richer metadata** — folder paths, attachment IDs, search operators, threads.
- **One account covers everything** — `books@arcodange.fr` is an alias of `gabrielradureau@arcodange.fr`. One refresh_token + the `/accounts` endpoint exposes both, plus all the other aliases (`contact@`, `bonjour@`, etc.).
- **Gmail folded in via forwarding** — `arcodange@gmail.com` forwards incoming to `books@` (configured in Gmail UI). No Google API setup, no app-passwords, no second OAuth flow.
- **Token-only auth** — no app-password fragility, no SCA dance (unlike Wise).
The single canonical inbox path: **`/Inbox/books`** — Zoho's auto-filter routes incoming mail to the `books@` alias into this sub-folder. Scan it first; widen with `--all-folders` only if needed.
## Prerequisites
1. Base skill set up ([dolibarr/README.md](../dolibarr/README.md)).
2. Zoho OAuth Self-Client created and a refresh_token generated. The `.env` extension:
```
ZOHO_CLIENT_ID=<from api-console.zoho.com self-client>
ZOHO_CLIENT_SECRET=<same>
ZOHO_REFRESH_TOKEN=<exchanged from one-time code>
ZOHO_DC=eu # eu | com | in | au
```
Setup walkthrough is in the V8 prep section of the cohort review notes.
3. Gmail forwarding to `books@arcodange.fr` enabled (Gmail Settings → Forwarding and POP/IMAP).
4. `pdftotext` (`brew install poppler` on macOS).
## Workflows
### 1. List candidates
```bash
bin/arcodange email list # default: /Inbox/books, last 30 msgs, no filter
bin/arcodange email list --candidates-only # filter to subjects/attachments matching supplier patterns
bin/arcodange email list --folder /Inbox/contact --limit 50
bin/arcodange email list --all-folders --candidates-only # scan everything (slower, more API calls)
```
Captured at [examples/email-list.txt](examples/email-list.txt). The candidate filter matches subjects against `facture|invoice|receipt|reçu|payment|paiement|abonnement|subscription|order|commande|bill` OR any message with an attachment.
**Hard exclusions** (V8.1) — applied before the candidate test, regardless of attachments:
- Subjects starting with `Invitation:` / `Updated invitation:` / `Canceled event:` / `Accepted:` / `Declined:` / `Tentative:` / `Maybe:` (after stripping `Re:` / `Fwd:` / `Tr:` prefixes) → filters calendar events that always carry an `.ics` attachment.
- Senders matching newsletter/marketing patterns (`updates.<domain>`, `noreply@*calendar*`, `news@`, `newsletter@`, etc.).
The `[*]` column marks candidates, `[Y]` marks emails with attachments. Compared to V8.0, V8.1 cuts the `--all-folders --candidates-only` baseline from ~27 noisy entries down to ~12 actionable ones.
### 2. Inspect one email + draft Dolibarr entry
```bash
bin/arcodange email inspect 1775141901205014300
bin/arcodange email inspect 1775141901205014300 --folder /Inbox/books # default
bin/arcodange email inspect 1775141901205014300 --save-pdf ~/Documents/factures-2026-Q2/
bin/arcodange email inspect 1775141901205014300 --json # machine-readable
```
The script:
1. Fetches the email metadata (subject / from / date) via `/messages/view`.
2. Lists attachments via `/messages/{mid}/attachmentinfo`.
3. Downloads each attachment via `/messages/{mid}/attachments/{aid}`.
4. For each `.pdf`, runs `pdftotext -layout`, applies regex heuristics to extract:
- Supplier name guess (first non-empty PDF line — often the supplier letterhead).
- Invoice reference (`facture/invoice n° XXX`).
- Invoice date.
- Total HT / TVA / TTC + VAT rate %.
5. Emits a draft JSON record per attachment — paste into the Dolibarr UI manually.
Heuristics are intentionally conservative (regex-based, no LLM dependency). For PDF templates where the heuristic fails, the raw `pdftotext` output is on disk in the work dir; rerun with `--save-pdf` to grab the PDF for manual entry.
Captured at [examples/email-inspect.txt](examples/email-inspect.txt) for the V8 baseline (Mistral AI receipt).
## What it doesn't do (V8.0 scope)
- **Does not write to Dolibarr.** The supplier invoice is still created manually in the Dolibarr UI from the draft JSON. V9 candidate: automate via `/supplierinvoices` POST.
- **Does not mark emails as ingested.** Each run re-emits the same candidates. Implementing this requires extending the OAuth scope: the current refresh_token only has READ scopes (`ZohoMail.messages.READ` etc.). The flag-set endpoint (`PUT /api/accounts/{aid}/updatemessage`) requires `ZohoMail.messages.UPDATE`, which would force the user to regenerate the refresh_token. **V8.2 candidate** — once the user opts in to the wider scope, `--mark-ingested` becomes a one-line flag on `email-inspect.sh` and `is_candidate()` in `email-list.sh` learns to skip messages with `flagid == flag_info`.
- **No body extraction yet.** We only parse PDF attachments. Inline-HTML invoices (rare — most suppliers send PDFs) would need body fetch via `/content`.
- **Heuristic extraction is best-effort.** Different supplier PDF templates yield different field-extraction reliability. The draft JSON is a starting point, not ground truth.
## Token cache
`zoho-curl.sh` caches the OAuth access_token in `$TMPDIR/zoho-access-$USER` (mode 600, TTL 50 min). Avoids hitting Zoho's OAuth refresh rate-limit on every invocation. On 401, the wrapper auto-refreshes once and retries.
## API endpoints used (Zoho Mail)
| Endpoint | Purpose |
|---|---|
| `POST /oauth/v2/token` (accounts.zoho.{dc}) | Refresh access_token from refresh_token |
| `GET /accounts` | Discover accountId + aliases on the account |
| `GET /accounts/{aid}/folders` | List folders (with paths like `/Inbox/books`) |
| `GET /accounts/{aid}/messages/view?folderId=&limit=&start=` | List messages in a folder |
| `GET /accounts/{aid}/folders/{fid}/messages/{mid}/attachmentinfo` | List attachments metadata |
| `GET /accounts/{aid}/folders/{fid}/messages/{mid}/attachments/{aid}` | Download attachment bytes |
## Out of scope
- **Writing to Dolibarr** (V9 candidate — would lift the read-only constraint on the API key, or use a separate write-scoped key).
- **Marking ingested emails** (V8.1 trivial follow-up).
- **Non-PDF attachments** (heuristics are PDF-specific).
- **Body-text extraction** (would need `/content` endpoint, deferred).
- **IMAP fallback** for non-Zoho mailboxes (deferred — Gmail forwarding to books@ covers the only known external mailbox today).
- **LLM-based extraction** (deferred — regex covers the current set of supplier templates well enough).

View File

@@ -0,0 +1,31 @@
================================================================================
Email 1775141901205014300
================================================================================
subject : Votre facture nº MSTRL-API-814045-001 de Mistral AI SAS
from : no-reply@mistral.ai
date : 2026-04-02
attached : True
-- Attachment 1: invoice-MSTRL-API-814045-001.pdf (74377 bytes, 1771 chars extracted) --
pdf_top_line = 'Facture'
invoice_ref = 'API-814045-001'
invoice_date_raw = None
total_ht = None
total_tva = None
total_ttc = None
vat_rate_pct = '20.0'
Suggested Dolibarr supplier-invoice draft entries:
[
{
"supplier_hint": "Facture",
"invoice_ref": "API-814045-001",
"invoice_date": null,
"total_ht": null,
"total_tva": null,
"total_ttc": null,
"vat_rate_pct": "20.0",
"source_email": "1775141901205014300",
"source_attachment": "invoice-MSTRL-API-814045-001.pdf"
}
]

View File

@@ -0,0 +1,16 @@
date cand att messageId folder from subject
----------------------------------------------------------------------------------------------------------------------------------
2026-05-20 [*] [Y] 1779312401677014300 /clients/KissMetrics rsirvent@digitalocean.com Re: VM not running despite status=active, after volume
2026-05-20 [*] [Y] 1779298419301014300 /clients/KissMetrics tdziuba@kissmetrics.io Re: VM not running despite status=active, after volume
2026-05-20 [*] [Y] 1779285954272004400 /clients/KissMetrics tdziuba@kissmetrics.io Re: VM not running despite status=active, after volume
2026-05-05 [*] [ ] 1777970798248014300 /Inbox/abonnements freemobile@free-mobile.fr Votre facture mobile Free est disponible
2026-04-21 [*] [Y] 1776785469477004300 /Notification noreply@hiway.fr Darnis Operations - Facture F1042
2026-04-12 [*] [Y] 1776017238960014300 /Inbox/books arcodange@gmail.com Fwd: Your receipt from Anthropic Ireland, Limited #2109
2026-04-04 [*] [ ] 1775264759983014300 /Inbox/abonnements freemobile@free-mobile.fr Votre facture mobile Free est disponible
2026-04-02 [*] [Y] 1775141901205014300 /Inbox/books no-reply@mistral.ai Votre facture nº MSTRL-API-814045-001 de Mistral AI SAS
2026-03-05 [*] [ ] 1772689535069004400 /Inbox/helloworld freemobile@free-mobile.fr Votre facture mobile Free est disponible
2026-02-08 [*] [Y] 1770582421208004400 /Inbox/bureaux ne-pas-repondre@portailpro.gouv.fr Valider votre espace personnel sur Portailpro.gouv
2026-01-09 [*] [ ] 1767989744791004400 /Inbox/books gabrielradureau@gmail.com Fwd: INPI - Votre paiement pour la commande Réf. 181876
2026-01-06 [*] [Y] 1767710535894005600 /Inbox gabrielradureau@gmail.com Statuts
----------------------------------------------------------------------------------------------------------------------------------
# 12 message(s) (candidates only)

View File

@@ -0,0 +1,7 @@
date cand att messageId folder from subject
----------------------------------------------------------------------------------------------------------------------------------
2026-04-12 [*] [Y] 1776017238960014300 /Inbox/books arcodange@gmail.com Fwd: Your receipt from Anthropic Ireland, Limited #2109
2026-04-02 [*] [Y] 1775141901205014300 /Inbox/books no-reply@mistral.ai Votre facture nº MSTRL-API-814045-001 de Mistral AI SAS
2026-01-09 [*] [ ] 1767989744791004400 /Inbox/books gabrielradureau@gmail.com Fwd: INPI - Votre paiement pour la commande Réf. 181876
----------------------------------------------------------------------------------------------------------------------------------
# 3 message(s) (candidates only)

View File

@@ -0,0 +1,256 @@
#!/usr/bin/env bash
# Inspect one email by id and propose a Dolibarr supplier-invoice draft.
#
# Usage:
# email-inspect.sh <messageId> [--folder PATH] # default folder: /Inbox/books
# [--save-pdf DIR] # save PDF attachments under DIR/
# [--json] # emit a single JSON object on stdout
#
# Pipeline (read-only):
# 1. Find the message (in the given folder, default /Inbox/books).
# 2. List attachments via /attachmentinfo.
# 3. For each PDF attachment: download, run pdftotext, extract supplier-side
# heuristics (name, totals, dates, ref).
# 4. Emit a draft "Dolibarr-ready" record per attachment so the operator can
# hand-create the supplier invoice in the Dolibarr UI.
#
# This skill DOES NOT write to Dolibarr. Auto-creation of supplier invoices is
# V9 candidate.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
ZOHO_CURL="${SCRIPT_DIR}/zoho-curl.sh"
if [[ $# -lt 1 ]]; then
echo "email-inspect.sh: missing <messageId>" >&2
echo " Hint: bin/arcodange email list to see candidate ids." >&2
exit 2
fi
MID="$1"; shift || true
FOLDER="/Inbox/books"; SAVE_PDF_DIR=""; FMT="text"
while [[ $# -gt 0 ]]; do
case "$1" in
--folder) FOLDER="$2"; shift 2 ;;
--save-pdf) SAVE_PDF_DIR="$2"; shift 2 ;;
--json) FMT="json"; shift ;;
-h|--help) sed -n '2,18p' "$0" | sed 's/^# \{0,1\}//'; exit 0 ;;
*) echo "email-inspect.sh: unknown arg: $1" >&2; exit 2 ;;
esac
done
command -v pdftotext >/dev/null || { echo "email-inspect.sh: pdftotext not found (brew install poppler)" >&2; exit 2; }
WORK="$(mktemp -d -t emailinspect.XXXXXX)"
trap 'rm -rf "${WORK}"' EXIT
# 1. accountId + folderId
"${ZOHO_CURL}" /accounts > "${WORK}/accounts.json"
AID=$(python3 -c "import json,sys; d=json.load(open(sys.argv[1])); print((d.get('data') or [{}])[0].get('accountId',''))" "${WORK}/accounts.json")
"${ZOHO_CURL}" "/accounts/${AID}/folders" > "${WORK}/folders.json"
FID=$(python3 -c "
import json, sys
d = json.load(open(sys.argv[1]))
target = sys.argv[2]
for f in (d.get('data') or []):
if f.get('path') == target:
print(f.get('folderId')); break" "${WORK}/folders.json" "${FOLDER}")
[[ -z "${FID}" ]] && { echo "email-inspect.sh: folder '${FOLDER}' not found" >&2; exit 2; }
# 2. Find the message in the folder listing (to grab metadata: subject, from, date)
"${ZOHO_CURL}" "/accounts/${AID}/messages/view?folderId=${FID}&limit=100&sortorder=false&start=1" > "${WORK}/folder_msgs.json"
python3 - "${WORK}/folder_msgs.json" "${MID}" > "${WORK}/meta.json" <<'PY'
import json, sys
d = json.load(open(sys.argv[1]))
mid = sys.argv[2]
for m in (d.get("data") or []):
if str(m.get("messageId")) == mid:
json.dump(m, sys.stdout); sys.exit(0)
sys.exit(f"messageId {mid} not found in this folder")
PY
# 3. Attachment metadata
"${ZOHO_CURL}" "/accounts/${AID}/folders/${FID}/messages/${MID}/attachmentinfo" > "${WORK}/attachinfo.json"
# 4. Download each attachment — needs raw bytes (Accept: */*), not the JSON
# wrapper's default. We bypass zoho-curl.sh for the attachment download but
# reuse the cached access_token it wrote.
set -a; source "${SCRIPT_DIR}/../../dolibarr/.env"; set +a
: "${ZOHO_DC:=eu}"
TOKEN_CACHE="${TMPDIR:-/tmp}/zoho-access-$(whoami)"
if [[ ! -s "${TOKEN_CACHE}" ]]; then
echo "email-inspect.sh: missing access token cache — run any zoho-curl call first to populate it" >&2
exit 2
fi
ACCESS_TOKEN=$(cat "${TOKEN_CACHE}")
MAIL_BASE="https://mail.zoho.${ZOHO_DC}/api"
mkdir -p "${WORK}/atts" "${WORK}/text"
ATT_IDS=$(python3 -c "
import json, sys
d = json.load(open(sys.argv[1]))
data = d.get('data') or {}
for a in (data.get('attachments') or []):
print(f\"{a.get('attachmentId')}|{a.get('attachmentName','-')}\")" "${WORK}/attachinfo.json")
while IFS='|' read -r aid aname; do
[[ -z "${aid}" ]] && continue
outpath="${WORK}/atts/${aname}"
curl -sS \
-H "Authorization: Zoho-oauthtoken ${ACCESS_TOKEN}" \
-H "Accept: */*" \
--max-time 60 \
-o "${outpath}" \
"${MAIL_BASE}/accounts/${AID}/folders/${FID}/messages/${MID}/attachments/${aid}" || true
# If pdf, extract text (bash 3.2 compatible — no ${var,,})
aname_lc=$(echo "${aname}" | tr '[:upper:]' '[:lower:]')
if [[ "${aname_lc}" == *.pdf ]]; then
pdftotext -layout "${outpath}" "${WORK}/text/${aname%.pdf}.txt" 2>/dev/null || true
fi
done <<< "${ATT_IDS}"
# Optional save
if [[ -n "${SAVE_PDF_DIR}" ]]; then
mkdir -p "${SAVE_PDF_DIR}"
cp "${WORK}/atts/"*.pdf "${SAVE_PDF_DIR}/" 2>/dev/null || true
fi
# 5. Heuristic extract + render
python3 - "${WORK}" "${FMT}" <<'PY'
import json, sys, os, re, datetime, glob
work, fmt = sys.argv[1:3]
meta = json.load(open(os.path.join(work,"meta.json")))
ts = int(meta.get("sentDateInGMT") or meta.get("receivedTime") or 0) // 1000
mail_date = datetime.datetime.fromtimestamp(ts).strftime("%Y-%m-%d") if ts else None
mail_from = (meta.get("fromAddress") or meta.get("sender") or "-").replace("&lt;","<").replace("&gt;",">").replace("<","").replace(">","")
mail_subject = meta.get("subject") or "-"
# Heuristics on PDF text
def extract(text):
out = {}
# First non-empty line is often the supplier name (or the address block first line)
lines = [l.strip() for l in text.splitlines() if l.strip()]
out["pdf_top_line"] = lines[0] if lines else None
# Total TTC / HT / TVA — try multiple French/English patterns
def first_match(*patterns):
for p in patterns:
for line in lines:
m = re.search(p, line, re.IGNORECASE)
if m: return m.group(1).replace(",", ".").replace(" ", "")
return None
def parse_amount(s):
if not s: return None
clean = s.replace(",", ".").replace(" ", "")
try:
v = float(clean)
# Money amounts < 1M EUR; filters out VAT-number false positives (FR12345678901)
return v if 0 <= v < 1_000_000 else None
except: return None
def first_amount(*patterns):
for p in patterns:
for line in lines:
m = re.search(p, line, re.IGNORECASE)
if m:
v = parse_amount(m.group(1))
if v is not None: return f"{v:.2f}"
return None
out["total_ht"] = first_amount(r'(?:total\s*ht|montant\s*ht|net\s*amount|subtotal)[^\d-]*([\d \.,]+)')
# TVA: require currency suffix to avoid matching VAT-number digits
out["total_tva"] = first_amount(r'(?:tva|vat)[^\d-]*([\d \.,]+)\s*(?:€|eur)\b')
out["total_ttc"] = first_amount(r'(?:total\s*ttc|amount\s*due|total\s*due|grand\s*total|montant\s*total|amount\s*paid)[^\d-]*([\d \.,]+)')
# Invoice ref — must contain a digit (filters "umber", "Invoice", etc.)
m = re.search(r'(?:facture|invoice|receipt|reçu)\s*(?:n[°o]?|number|#|:)\s*([A-Za-z0-9][\w\d/-]{2,})', text, re.IGNORECASE)
if m and any(c.isdigit() for c in m.group(1)):
out["invoice_ref"] = m.group(1)
else:
# Fallback: any reasonable ref-shaped token after "Invoice" / "Facture" header
m = re.search(r'\b([A-Z]{2,}[-/]?\d[\w\d/-]{2,})\b', text)
out["invoice_ref"] = m.group(1) if m else None
# Invoice date — try ISO, French DD/MM/YYYY, English MM/DD/YYYY, French long form
out["invoice_date_raw"] = None
for p in (
r'\b(\d{4}-\d{2}-\d{2})\b',
r'(?:date|émise\s*le|invoice\s*date|date\s*de\s*facturation)[:\s]*(\d{1,2}[\s/.-]\d{1,2}[\s/.-]\d{2,4})',
r'(?:date|émise\s*le|invoice\s*date)[:\s]*(\d{1,2}\s+\w{3,9}\.?\s+\d{4})',
):
m = re.search(p, text, re.IGNORECASE)
if m: out["invoice_date_raw"] = m.group(1).strip(); break
# VAT rate (e.g. "20%") — restrict to 0-25% so "100%" / page footers don't match.
vrate = None
for line in lines:
m = re.search(r'\b(\d{1,2}([.,]\d+)?)\s*%', line)
if m:
v = float(m.group(1).replace(",", "."))
if 0 <= v <= 25:
vrate = m.group(1).replace(",", "."); break
out["vat_rate_pct"] = vrate
return out
pdfs = []
for pdf in sorted(glob.glob(os.path.join(work,"atts","*.pdf")) +
glob.glob(os.path.join(work,"atts","*.PDF"))):
name = os.path.basename(pdf)
txt_path = os.path.join(work,"text", os.path.splitext(name)[0] + ".txt")
text = open(txt_path).read() if os.path.isfile(txt_path) else ""
h = extract(text)
h["attachment_name"] = name
h["pdf_size_bytes"] = os.path.getsize(pdf)
h["pdf_text_len"] = len(text)
pdfs.append(h)
result = {
"email": {
"messageId": meta.get("messageId"),
"subject": mail_subject,
"from": mail_from,
"date": mail_date,
"hasAttachment": str(meta.get("hasAttachment","")) == "1",
},
"attachments": pdfs,
"dolibarr_draft_suggestions": [
{
"supplier_hint": p.get("pdf_top_line"),
"invoice_ref": p.get("invoice_ref"),
"invoice_date": p.get("invoice_date_raw"),
"total_ht": p.get("total_ht"),
"total_tva": p.get("total_tva"),
"total_ttc": p.get("total_ttc"),
"vat_rate_pct": p.get("vat_rate_pct"),
"source_email": meta.get("messageId"),
"source_attachment": p.get("attachment_name"),
} for p in pdfs
]
}
if fmt == "json":
print(json.dumps(result, indent=2, ensure_ascii=False))
sys.exit(0)
print("=" * 80)
print(f" Email {meta.get('messageId')}")
print("=" * 80)
print(f" subject : {mail_subject}")
print(f" from : {mail_from}")
print(f" date : {mail_date}")
print(f" attached : {result['email']['hasAttachment']}")
print()
if not pdfs:
print(" (no PDF attachments — try inspecting body or other types)")
for i, p in enumerate(pdfs, 1):
print(f" -- Attachment {i}: {p['attachment_name']} ({p['pdf_size_bytes']} bytes, {p['pdf_text_len']} chars extracted) --")
for k in ("pdf_top_line","invoice_ref","invoice_date_raw","total_ht","total_tva","total_ttc","vat_rate_pct"):
v = p.get(k)
print(f" {k:<16} = {v!r}")
print()
print(" Suggested Dolibarr supplier-invoice draft entries:")
print(json.dumps(result["dolibarr_draft_suggestions"], indent=4, ensure_ascii=False))
PY

View File

@@ -0,0 +1,141 @@
#!/usr/bin/env bash
# List candidate supplier-invoice emails from the books@ Zoho mailbox.
#
# Usage:
# email-list.sh [--folder PATH] # default: /Inbox/books (the books@ alias-filtered folder)
# [--limit N] # default: 30
# [--candidates-only] # filter by subject pattern OR attachment
# [--all-folders] # scan every folder (slow, lots of API calls)
#
# Output: table with mid, date, from, subject, hasAttachment.
# A "candidate" is a message whose subject matches a supplier-like pattern
# (facture/invoice/receipt/reçu/payment/paiement/abonnement/order/commande)
# OR which has an attachment.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
ZOHO_CURL="${SCRIPT_DIR}/zoho-curl.sh"
FOLDER="/Inbox/books"
LIMIT=30
CANDIDATES_ONLY=0
ALL_FOLDERS=0
while [[ $# -gt 0 ]]; do
case "$1" in
--folder) FOLDER="$2"; shift 2 ;;
--limit) LIMIT="$2"; shift 2 ;;
--candidates-only) CANDIDATES_ONLY=1; shift ;;
--all-folders) ALL_FOLDERS=1; shift ;;
-h|--help) sed -n '2,12p' "$0" | sed 's/^# \{0,1\}//'; exit 0 ;;
*) echo "email-list.sh: unknown arg: $1" >&2; exit 2 ;;
esac
done
WORK="$(mktemp -d -t emailist.XXXXXX)"
trap 'rm -rf "${WORK}"' EXIT
# 1. Discover accountId
"${ZOHO_CURL}" /accounts > "${WORK}/accounts.json"
AID=$(python3 -c "import json,sys; d=json.load(open(sys.argv[1])); print((d.get('data') or [{}])[0].get('accountId',''))" "${WORK}/accounts.json")
[[ -z "${AID}" ]] && { echo "email-list.sh: no accountId in /accounts response" >&2; exit 1; }
# 2. Resolve folder path → folderId
"${ZOHO_CURL}" "/accounts/${AID}/folders" > "${WORK}/folders.json"
# Build list of (folderId, path) tuples to scan
if [[ "${ALL_FOLDERS}" == "1" ]]; then
FOLDER_IDS=$(python3 -c "
import json, sys
d = json.load(open(sys.argv[1]))
for f in (d.get('data') or []):
fid = f.get('folderId'); path = f.get('path') or f.get('folderName','-')
# Skip noisy system folders
if path in ('/Drafts','/Templates','/Snoozed','/Sent','/Spam','/Trash','/Outbox'): continue
print(f\"{fid}|{path}\")" "${WORK}/folders.json")
else
FOLDER_IDS=$(python3 -c "
import json, sys
d = json.load(open(sys.argv[1]))
target = sys.argv[2]
for f in (d.get('data') or []):
if f.get('path') == target:
print(f\"{f.get('folderId')}|{f.get('path')}\")
break" "${WORK}/folders.json" "${FOLDER}")
if [[ -z "${FOLDER_IDS}" ]]; then
echo "email-list.sh: folder '${FOLDER}' not found. Available:" >&2
python3 -c "import json,sys; [print(f' {f.get(\"path\",\"-\")}') for f in json.load(open(sys.argv[1])).get('data',[])]" "${WORK}/folders.json" >&2
exit 2
fi
fi
# 3. Fetch messages per folder
mkdir -p "${WORK}/msgs"
COUNT=0
while IFS='|' read -r fid fpath; do
[[ -z "${fid}" ]] && continue
COUNT=$((COUNT+1))
out="${WORK}/msgs/$(printf '%03d' "${COUNT}").json"
"${ZOHO_CURL}" "/accounts/${AID}/messages/view?folderId=${fid}&limit=${LIMIT}&sortorder=false&start=1" > "${out}" 2>/dev/null || echo '{"data":[]}' > "${out}"
echo "${fpath}" > "${out}.path"
done <<< "${FOLDER_IDS}"
# 4. Render
python3 - "${WORK}/msgs" "${CANDIDATES_ONLY}" <<'PY'
import json, sys, os, re, datetime, glob
msgs_dir, candidates_only_str = sys.argv[1:3]
candidates_only = candidates_only_str == "1"
CANDIDATE_PATTERN = re.compile(
r'facture|invoice|receipt|re[cç]u|payment|paiement|abonnement|subscription|order|commande|invoice|bill',
re.IGNORECASE,
)
# Subjects that look like calendar invites / event updates / generic notifications
# get filtered out of --candidates-only — they always have a .ics attachment so
# the "has-attachment" heuristic alone catches them as false positives.
EXCLUDE_PATTERN = re.compile(
r'^(?:re:\s*|fwd:\s*|tr:\s*)*' # strip Re:/Fwd:/Tr: prefixes
r'(?:invitation|updated\s+invitation|canceled\s+event|accepted|declined|tentative|maybe)\s*:',
re.IGNORECASE,
)
# Senders that are pure noise — newsletter/marketing patterns.
EXCLUDE_SENDER = re.compile(
r'(updates\.|noreply@.*calendar|@calendar\.|news@|newsletter@|@updates\.)',
re.IGNORECASE,
)
def is_candidate(m):
subj = m.get("subject","") or ""
sender = m.get("fromAddress","") or m.get("sender","") or ""
# Hard exclusions take precedence over inclusions
if EXCLUDE_PATTERN.match(subj.strip()): return False
if EXCLUDE_SENDER.search(sender): return False
if str(m.get("hasAttachment","")) == "1": return True
if CANDIDATE_PATTERN.search(subj): return True
return False
rows = []
for f in sorted(glob.glob(os.path.join(msgs_dir, "*.json"))):
fpath = open(f + ".path").read().strip()
try: data = json.load(open(f)).get("data") or []
except: continue
for m in data:
if candidates_only and not is_candidate(m): continue
ts = int(m.get("sentDateInGMT") or m.get("receivedTime") or 0) // 1000
dt = datetime.datetime.fromtimestamp(ts).strftime("%Y-%m-%d") if ts else "-"
frm = (m.get("fromAddress") or m.get("sender") or "-").replace("&lt;","<").replace("&gt;",">").replace("<","").replace(">","")[:36]
subj = (m.get("subject") or "-")[:55]
has = "Y" if str(m.get("hasAttachment","")) == "1" else " "
cand = "*" if is_candidate(m) else " "
rows.append((dt, fpath, cand, has, m.get("messageId","-"), frm, subj))
rows.sort(key=lambda r: r[0], reverse=True)
print(f"{'date':<10} {'cand':<4} {'att':<3} {'messageId':<22} {'folder':<22} {'from':<36} subject")
print("-" * 130)
for dt, fpath, cand, has, mid, frm, subj in rows:
print(f"{dt:<10} [{cand}] [{has}] {mid:<22} {fpath[:22]:<22} {frm:<36} {subj}")
print("-" * 130)
print(f"# {len(rows)} message(s)" + (" (candidates only)" if candidates_only else ""))
PY

View File

@@ -0,0 +1,126 @@
#!/usr/bin/env bash
# Read-only curl wrapper for the Zoho Mail API.
#
# Usage:
# zoho-curl.sh <path> # e.g. zoho-curl.sh /accounts
# zoho-curl.sh -i <path> # include curl's -i (response headers)
# zoho-curl.sh -o file.json <path> # write body to file
#
# Reads credentials from ../../dolibarr/.env (the shared canonical file).
# Required vars:
# ZOHO_CLIENT_ID, ZOHO_CLIENT_SECRET, ZOHO_REFRESH_TOKEN, ZOHO_DC
#
# Token strategy: each invocation refreshes a short-lived access_token from
# the refresh_token (Zoho access_tokens live 1h; the cost of refreshing on
# every call is ~150 ms and avoids state on disk). On 401 from the mail API
# we re-refresh once and retry (covers refresh-token rotation cases).
#
# Exits non-zero on HTTP >= 400 and writes body to stdout + a short message
# to stderr — same shape as dol-curl.sh / bank-curl.sh.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
ENV_FILE="${SCRIPT_DIR}/../../dolibarr/.env"
if [[ ! -f "${ENV_FILE}" ]]; then
echo "zoho-curl.sh: missing ${ENV_FILE}" >&2
echo " Required vars: ZOHO_CLIENT_ID, ZOHO_CLIENT_SECRET, ZOHO_REFRESH_TOKEN, ZOHO_DC." >&2
echo " See arcodange-email-ingest/SKILL.md for the OAuth setup." >&2
exit 2
fi
set -a; source "${ENV_FILE}"; set +a
: "${ZOHO_CLIENT_ID:?zoho-curl.sh: ZOHO_CLIENT_ID not set in .env}"
: "${ZOHO_CLIENT_SECRET:?zoho-curl.sh: ZOHO_CLIENT_SECRET not set in .env}"
: "${ZOHO_REFRESH_TOKEN:?zoho-curl.sh: ZOHO_REFRESH_TOKEN not set in .env}"
: "${ZOHO_DC:=eu}"
ACCOUNTS_BASE="https://accounts.zoho.${ZOHO_DC}"
MAIL_BASE="https://mail.zoho.${ZOHO_DC}/api"
# Parse pass-through curl args (everything before the last positional)
PASSTHRU=()
while [[ $# -gt 1 ]]; do
PASSTHRU+=("$1"); shift
done
if [[ $# -lt 1 ]]; then
echo "zoho-curl.sh: missing API path. Example: zoho-curl.sh /accounts" >&2
exit 2
fi
API_PATH="$1"
# Cache access_token in tmpfs to avoid hitting OAuth rate limits on every
# zoho-curl invocation. Zoho access_tokens live 1h; we refresh after 50 min.
CACHE_FILE="${TMPDIR:-/tmp}/zoho-access-$(whoami)"
CACHE_TTL_SECONDS=$((50 * 60))
get_access_token() {
if [[ -f "${CACHE_FILE}" ]]; then
local age
age=$(( $(date +%s) - $(stat -f %m "${CACHE_FILE}" 2>/dev/null || stat -c %Y "${CACHE_FILE}") ))
if [[ ${age} -lt ${CACHE_TTL_SECONDS} ]]; then
cat "${CACHE_FILE}"
return 0
fi
fi
local token
if ! token=$(curl -sS -X POST "${ACCOUNTS_BASE}/oauth/v2/token" \
--max-time 15 \
-d "grant_type=refresh_token" \
-d "client_id=${ZOHO_CLIENT_ID}" \
-d "client_secret=${ZOHO_CLIENT_SECRET}" \
-d "refresh_token=${ZOHO_REFRESH_TOKEN}" \
| python3 -c "
import json, sys
try: d = json.load(sys.stdin)
except: sys.exit('failed to parse OAuth response')
if 'access_token' not in d:
sys.exit(f'OAuth refresh failed: {d}')
print(d['access_token'])"); then
return 1
fi
if [[ -z "${token}" ]]; then
return 1
fi
# Store cache (mode 600) only on success
printf '%s' "${token}" > "${CACHE_FILE}"
chmod 600 "${CACHE_FILE}"
printf '%s' "${token}"
}
do_call() {
local token="$1"
local body_file="$2"
local headers_file="$3"
curl -sS \
-H "Authorization: Zoho-oauthtoken ${token}" \
-H "Accept: application/json" \
--max-time 30 \
-o "${body_file}" \
-D "${headers_file}" \
-w "%{http_code}" \
${PASSTHRU[@]+"${PASSTHRU[@]}"} \
"${MAIL_BASE}${API_PATH}"
}
ACCESS_TOKEN=$(get_access_token)
[[ -z "${ACCESS_TOKEN}" ]] && { echo "zoho-curl.sh: empty access_token" >&2; exit 1; }
BODY_FILE="$(mktemp -t zohocurl.XXXXXX)"
HEADERS_FILE="$(mktemp -t zohohdr.XXXXXX)"
trap 'rm -f "${BODY_FILE}" "${HEADERS_FILE}"' EXIT
HTTP_CODE=$(do_call "${ACCESS_TOKEN}" "${BODY_FILE}" "${HEADERS_FILE}")
# Retry once on 401 with a fresh token (handles edge cases of refresh-token rotation)
if [[ "${HTTP_CODE}" == "401" ]]; then
ACCESS_TOKEN=$(get_access_token)
HTTP_CODE=$(do_call "${ACCESS_TOKEN}" "${BODY_FILE}" "${HEADERS_FILE}")
fi
cat "${BODY_FILE}"
if [[ "${HTTP_CODE}" -ge 400 ]]; then
echo "zoho-curl.sh: HTTP ${HTTP_CODE} on ${API_PATH}" >&2
exit 1
fi

View File

@@ -0,0 +1,10 @@
# Copy to .env (mode 600 — gitignored). Never commit a real key.
DOLIBARR_SANDBOX_URL=https://erp-sandbox.arcodange.lab
DOLIBARR_SANDBOX_API_KEY=
# Populate the key from the Playwright provisioner output (repo test/):
# printf 'DOLIBARR_SANDBOX_API_KEY=%s\n' "$(cat ../../../test/.ai_agent_sandbox.key)" >> .env
#
# The host guard only allows the sandbox FQDN. Override the pattern ONLY if the
# sandbox host changes — never widen it to a prod host.
# DOLIBARR_SANDBOX_HOST_RE=^https://erp-sandbox\.arcodange\.lab(/|$)

View File

@@ -0,0 +1,117 @@
---
name: dolibarr-sandbox-write
description: >-
WRITE operations against the Arcodange Dolibarr SANDBOX (erp-sandbox.arcodange.lab)
— the rehearsal environment where an AI agent records thirdparties, invoices and
payments before any change is promoted to prod. Create client/supplier fiches
(auto-coded), customer + supplier invoices with product/service lines and the
supplier's own reference, validate them, and record règlements (payments). Every
write goes through dol-write.sh, which REFUSES any host that is not the sandbox —
the structural guarantee (ADR-0003) that this skill can never mutate production.
Use when the user asks to "create a thirdparty / supplier / client fiche", "saisir
une facture", "record an invoice with lines", "enregistrer un règlement / paiement",
or to rehearse a write before promoting it to prod. SKIP for production writes
(prod stays read-only via the `dolibarr` skill's `ai_agent` key; promotion is a
separate, human-gated replay), and for credit notes/avoirs (a planned follow-up).
Depends on the write-scoped `ai_agent_sandbox` Dolibarr user + its API key.
requires:
bins: [bash, curl, python3]
auth: ".env with DOLIBARR_SANDBOX_URL + DOLIBARR_SANDBOX_API_KEY (mode 600, gitignored)"
---
# dolibarr-sandbox-write
Write-capable companion to the read-only `dolibarr*` skills, scoped to the
**sandbox**. It exists so an AI agent can *rehearse* bookkeeping writes against a
faithful copy of prod (see ADR-0003 + the `ops/sandbox/` seed tooling), then a
human promotes the reviewed change to prod.
## The safety model (read this first)
- **Host guard.** `scripts/dol-write.sh` reads `DOLIBARR_SANDBOX_URL` from `.env`
and refuses to send any request unless it matches `erp-sandbox.arcodange.lab`.
Point it at `erp.arcodange.lab` (prod) and it exits non-zero *before* the
request. This is the structural reason the skill cannot write prod.
- **Credential scope.** The key is `ai_agent_sandbox`'s — valid only on the
sandbox host, with create+read rights on thirdparties / invoices / supplier
invoices / products / contacts (+ `societe client voir`). Prod's `ai_agent`
key is read-only and lives in a different skill's `.env`.
- **Resettable.** Anything written here is wiped by `ops/sandbox/sandbox-lifecycle.sh
refresh-from-prod`, so mistakes cost a reset, not data.
- **Promotion to prod is NOT in this skill.** Rehearse here → capture a reviewable
diff with the read-only `dolibarr-data-snapshot` skill (before/after) → a human
approves → the same operations are replayed against prod under a separate,
human-held prod-write credential. Never wire a prod-write key into this skill.
## Setup
Create `.env` (mode 600, gitignored) next to `scripts/`:
```sh
cd .claude/skills/dolibarr-sandbox-write
umask 077
{ echo "DOLIBARR_SANDBOX_URL=https://erp-sandbox.arcodange.lab"
printf 'DOLIBARR_SANDBOX_API_KEY=%s\n' "$(cat /path/to/.ai_agent_sandbox.key)"; } > .env
```
The key is produced by the Playwright provisioner in the repo's `test/`
(`provisionSandbox.ts` → `.ai_agent_sandbox.key`). Verify: `scripts/dol-write.sh
GET /status` should return HTTP 200 with `"environment":"non-production"`.
## Workflows
All three read a JSON object on **stdin** (or a file path as `$1`) and emit ids.
### 1 · Thirdparty (fiche client/fournisseur) — `scripts/thirdparty-create.sh`
```sh
echo '{"name":"KissMetrics","role":"client","tva_intra":"US.."}' | scripts/thirdparty-create.sh
echo '{"name":"OVH","role":"supplier","siret":"..."}' | scripts/thirdparty-create.sh
```
`role`: `client` | `supplier` | `both`. Codes auto-assign from the mask
(`CL{0000}` / `FO{0000}`) via the `-1` sentinel; pass `client_code`/`supplier_code`
to override. Optional: `country_id` (default 1=FR), `siret`, `tva_intra`,
`address`, `zip`, `town`, `email`, `phone`, `idprof1`. Emits the new id.
### 2 · Invoice (facture) — `scripts/invoice-create.sh`
```sh
echo '{"socid":42,"kind":"customer","validate":true,
"lines":[{"desc":"Conseil","qty":2,"price_ht":500,"tva":20,"type":"service"},
{"desc":"Licence","qty":1,"price_ht":100,"tva":20,"type":"product"}]}' \
| scripts/invoice-create.sh
# supplier invoice carrying the supplier's own reference:
echo '{"socid":7,"kind":"supplier","ref_supplier":"INV-2026-042","validate":true,
"lines":[{"desc":"Hosting","qty":1,"price_ht":80,"tva":20,"type":"service"}]}' \
| scripts/invoice-create.sh
```
`kind`: `customer` (`/invoices`) | `supplier` (`/supplierinvoices`). Lines carry
`desc, qty, price_ht, tva, type` (product|service) and optional `product_id`
(`fk_product`) to link a catalogue product. Totals + TVA are computed by Dolibarr.
`validate:true` turns the draft (`PROV…`) into a final numbered invoice; omit it
to leave a draft. Emits `{id, ref, ref_supplier, total_ht, total_ttc, statut}`.
### 3 · Payment (règlement) — `scripts/payment-record.sh`
```sh
echo '{"invoice_id":19,"mode":"VIR","account_id":1}' | scripts/payment-record.sh
echo '{"invoice_id":13,"kind":"supplier","mode":"VIR","account_id":1,"amount":96}' \
| scripts/payment-record.sh
```
The invoice must be **validated** first. `mode`: `VIR|CB|CHQ|LIQ`. Customer
payments settle the full remaining amount and mark the invoice paid; **supplier**
payments require an explicit `amount`. `account_id` is the bank account id (the
read-only `dolibarr-payments-state` skill lists them; `ai_agent_sandbox` does not
yet have `banque lire`, so pass the id). Emits the new payment id.
## Gotchas
- **Validate before paying.** A draft (`statut=0`, ref `PROV…`) cannot be paid.
- **Codes.** The thirdparty code module is `mod_codeclient_elephant` (auto). The
REST create needs `code_client`/`code_fournisseur = "-1"` to trigger it — the
script does this; without it the API errors `ErrorCustomerCodeRequired`.
- **Dates** are sent as Unix epochs; pass `date:"YYYY-MM-DD"` or omit for today.
- **`banque lire`** isn't granted yet → `GET /bankaccounts` returns empty. Add it
to the provisioner's rights set if account discovery from this skill is needed.
- **Avoirs (credit notes)** are a planned follow-up (a customer invoice with
`type=2` referencing the original).

View File

@@ -0,0 +1,9 @@
{
"socid": 42,
"kind": "customer",
"validate": true,
"lines": [
{ "desc": "Conseil (jours)", "qty": 2, "price_ht": 500, "tva": 20, "type": "service" },
{ "desc": "Licence annuelle", "qty": 1, "price_ht": 100, "tva": 20, "type": "product" }
]
}

View File

@@ -0,0 +1,82 @@
#!/usr/bin/env bash
# Host-guarded WRITE wrapper for the Arcodange Dolibarr SANDBOX API.
#
# THE GUARANTEE (ADR-0003): this wrapper REFUSES to run against anything that is
# not the erp-sandbox host. It is the structural reason an AI agent driving this
# skill can never mutate production — prod is a different host, and the guard
# below rejects it before any request is sent.
#
# Usage:
# dol-write.sh <METHOD> <path> [json-body|@file]
# dol-write.sh POST /thirdparties '{"name":"Acme","client":"1"}'
# dol-write.sh POST /invoices @invoice.json
# dol-write.sh PUT /thirdparties/42 '{"phone":"+33..."}'
# dol-write.sh GET /thirdparties?limit=5 # reads are allowed too
#
# Reads DOLIBARR_SANDBOX_URL + DOLIBARR_SANDBOX_API_KEY from the sibling .env
# (.claude/skills/dolibarr-sandbox-write/.env), mode 600, gitignored.
# Prints the response body to stdout; exits non-zero on HTTP >= 400.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
ENV_FILE="${SCRIPT_DIR}/../.env"
if [[ ! -f "${ENV_FILE}" ]]; then
echo "dol-write.sh: missing ${ENV_FILE}" >&2
echo " Create it with DOLIBARR_SANDBOX_URL + DOLIBARR_SANDBOX_API_KEY. See README.md." >&2
exit 2
fi
# shellcheck disable=SC1090
set -a; source "${ENV_FILE}"; set +a
: "${DOLIBARR_SANDBOX_URL:?dol-write.sh: DOLIBARR_SANDBOX_URL not set in .env}"
: "${DOLIBARR_SANDBOX_API_KEY:?dol-write.sh: DOLIBARR_SANDBOX_API_KEY not set in .env}"
# ---------------------------------------------------------------------------
# HOST GUARD — the structural safety invariant. Only the sandbox host passes.
# Override the allowed pattern only via DOLIBARR_SANDBOX_HOST_RE in .env if the
# sandbox FQDN ever changes; never widen it to include a prod host.
# ---------------------------------------------------------------------------
ALLOWED_RE="${DOLIBARR_SANDBOX_HOST_RE:-^https://erp-sandbox\.arcodange\.lab(/|$)}"
if [[ ! "${DOLIBARR_SANDBOX_URL}" =~ ${ALLOWED_RE} ]]; then
echo "dol-write.sh: REFUSING to write — DOLIBARR_SANDBOX_URL='${DOLIBARR_SANDBOX_URL}'" >&2
echo " is not the erp-sandbox host (allowed: ${ALLOWED_RE})." >&2
echo " This skill only writes the sandbox. Promotion to prod is a separate, human-gated step." >&2
exit 3
fi
if [[ $# -lt 2 ]]; then
echo "dol-write.sh: usage: dol-write.sh <METHOD> <path> [json-body|@file]" >&2
exit 2
fi
METHOD="$1"; API_PATH="$2"; BODY="${3-}"
case "${METHOD}" in
GET|POST|PUT|DELETE|PATCH) ;;
*) echo "dol-write.sh: unsupported method '${METHOD}'" >&2; exit 2 ;;
esac
CURL_ARGS=(
-sS -X "${METHOD}"
-H "DOLAPIKEY: ${DOLIBARR_SANDBOX_API_KEY}"
-H "Accept: application/json"
--max-time 30
)
if [[ -n "${BODY}" ]]; then
# curl --data supports '@file' to read a JSON body from a file.
CURL_ARGS+=( -H "Content-Type: application/json" --data "${BODY}" )
fi
BODY_FILE="$(mktemp -t dolwrite.XXXXXX)"
trap 'rm -f "${BODY_FILE}"' EXIT
HTTP_CODE=$(curl "${CURL_ARGS[@]}" \
-o "${BODY_FILE}" -w "%{http_code}" \
"${DOLIBARR_SANDBOX_URL}/api/index.php${API_PATH}")
cat "${BODY_FILE}"
if [[ "${HTTP_CODE}" -ge 400 ]]; then
echo "" >&2
echo "dol-write.sh: HTTP ${HTTP_CODE} on ${METHOD} ${API_PATH}" >&2
exit 1
fi

View File

@@ -0,0 +1,68 @@
#!/usr/bin/env bash
# Create a customer or supplier invoice (facture) with product/service lines in
# the SANDBOX, optionally validating it.
#
# Input: a JSON object on stdin (or a file path in $1):
# socid (required) thirdparty id
# kind "customer" | "supplier" (default "customer")
# date "YYYY-MM-DD" (default today)
# ref_supplier supplier's own invoice ref (supplier invoices)
# validate true|false (default false = leave draft)
# lines: [ { desc, qty, price_ht, tva, type: "product"|"service", product_id? } ]
#
# Emits {id, ref, ref_supplier, total_ht, total_ttc, statut} on stdout.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
W="${SCRIPT_DIR}/dol-write.sh"
SRC="${1:-}"
if [[ -n "${SRC}" && "${SRC}" != "-" ]]; then INPUT="$(cat "${SRC}")"; else INPUT="$(cat)"; fi
PYF="$(mktemp -t dolpy.XXXXXX)"; trap 'rm -f "${PYF}"' EXIT
cat > "${PYF}" <<'PY'
import json, sys, datetime
d = json.loads(sys.stdin.read())
if not d.get("socid"):
sys.exit("invoice-create.sh: 'socid' is required")
supplier = d.get("kind", "customer").lower() in ("supplier", "fournisseur")
endpoint = "/supplierinvoices" if supplier else "/invoices"
ds = d.get("date")
epoch = int((datetime.datetime.strptime(ds, "%Y-%m-%d") if ds
else datetime.datetime.now()).timestamp())
lines = []
for ln in d.get("lines", []):
is_product = ln.get("type", "service").lower() in ("product", "produit")
L = {
"desc": ln.get("desc", ""),
"subprice": str(ln.get("price_ht", ln.get("subprice", 0))),
"qty": str(ln.get("qty", 1)),
"tva_tx": str(ln.get("tva", ln.get("tva_tx", 20))),
"product_type": "0" if is_product else "1",
}
if ln.get("product_id"):
L["fk_product"] = str(ln["product_id"])
lines.append(L)
body = {"socid": d["socid"], "date": epoch, "type": 0, "lines": lines}
if supplier and d.get("ref_supplier"):
body["ref_supplier"] = d["ref_supplier"]
print(endpoint)
print(json.dumps(body))
print("1" if d.get("validate") else "0")
PY
MAPPED="$(printf '%s' "${INPUT}" | python3 "${PYF}")"
ENDPOINT="$(sed -n 1p <<<"${MAPPED}")"
BODY="$(sed -n 2p <<<"${MAPPED}")"
VALIDATE="$(sed -n 3p <<<"${MAPPED}")"
ID="$("${W}" POST "${ENDPOINT}" "${BODY}")"
if [[ ! "${ID}" =~ ^[0-9]+$ ]]; then
echo "invoice-create.sh: create did not return an id: ${ID}" >&2
exit 1
fi
if [[ "${VALIDATE}" == "1" ]]; then
"${W}" POST "${ENDPOINT}/${ID}/validate" '{}' >/dev/null
fi
"${W}" GET "${ENDPOINT}/${ID}" | python3 -c "import json,sys
d=json.load(sys.stdin)
print(json.dumps({k:d.get(k) for k in ('id','ref','ref_supplier','total_ht','total_ttc','statut')}))"

View File

@@ -0,0 +1,59 @@
#!/usr/bin/env bash
# Record a payment (règlement) on a validated invoice in the SANDBOX.
#
# Input: a JSON object on stdin (or a file path in $1):
# invoice_id (required) the invoice to pay
# kind "customer" | "supplier" (default "customer")
# mode "VIR" | "CB" | "CHQ" | "LIQ" (default "VIR")
# account_id (required) the bank account id receiving/paying
# date "YYYY-MM-DD" (default today)
# amount (REQUIRED for supplier; customer pays the full remaining)
# num, comment (optional)
#
# The invoice must be VALIDATED first (invoice-create.sh ... "validate":true).
# Emits the new payment id on stdout.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
W="${SCRIPT_DIR}/dol-write.sh"
SRC="${1:-}"
if [[ -n "${SRC}" && "${SRC}" != "-" ]]; then INPUT="$(cat "${SRC}")"; else INPUT="$(cat)"; fi
PYF="$(mktemp -t dolpy.XXXXXX)"; trap 'rm -f "${PYF}"' EXIT
cat > "${PYF}" <<'PY'
import json, sys, datetime
d = json.loads(sys.stdin.read())
if not d.get("invoice_id"):
sys.exit("payment-record.sh: 'invoice_id' is required")
if not d.get("account_id"):
sys.exit("payment-record.sh: 'account_id' is required")
# Stable Dolibarr c_paiement ids (sandbox seeded from prod / standard defaults).
MODE = {"VIR": 2, "CB": 6, "CHQ": 7, "LIQ": 4}
mode = MODE.get(str(d.get("mode", "VIR")).upper())
if mode is None:
sys.exit("payment-record.sh: unknown mode (use VIR|CB|CHQ|LIQ)")
supplier = d.get("kind", "customer").lower() in ("supplier", "fournisseur")
ds = d.get("date")
epoch = int((datetime.datetime.strptime(ds, "%Y-%m-%d") if ds
else datetime.datetime.now()).timestamp())
inv = d["invoice_id"]
if supplier:
if d.get("amount") is None:
sys.exit("payment-record.sh: supplier payments require an 'amount'")
endpoint = "/supplierinvoices/%s/payments" % inv
body = {"datepaye": epoch, "paymentid": mode, "accountid": d["account_id"],
"amount": str(d["amount"]), "num_payment": d.get("num", ""),
"comment": d.get("comment", "")}
else:
endpoint = "/invoices/%s/payments" % inv
body = {"datepaye": epoch, "paymentid": mode, "closepaidinvoices": "yes",
"accountid": d["account_id"], "num_payment": d.get("num", ""),
"comment": d.get("comment", "")}
print(endpoint)
print(json.dumps(body))
PY
MAPPED="$(printf '%s' "${INPUT}" | python3 "${PYF}")"
ENDPOINT="$(sed -n 1p <<<"${MAPPED}")"
BODY="$(sed -n 2p <<<"${MAPPED}")"
"${W}" POST "${ENDPOINT}" "${BODY}"

View File

@@ -0,0 +1,50 @@
#!/usr/bin/env bash
# Create a client and/or supplier thirdparty (fiche tiers) in the SANDBOX.
#
# Input: a JSON object on stdin (or a file path in $1). Fields:
# name (required)
# role "client" | "supplier" | "both" (default "client")
# country_id numeric, default 1 (France)
# client_code / supplier_code default "-1" = auto-generate via the code mask
# siret, tva_intra, address, zip, town, email, phone, idprof1 (optional)
#
# Emits the new thirdparty id on stdout. All writes go through dol-write.sh,
# which refuses any host that is not the sandbox.
#
# Examples:
# echo '{"name":"KissMetrics","role":"client","tva_intra":"US.."}' | thirdparty-create.sh
# thirdparty-create.sh fournisseur.json
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
W="${SCRIPT_DIR}/dol-write.sh"
SRC="${1:-}"
if [[ -n "${SRC}" && "${SRC}" != "-" ]]; then INPUT="$(cat "${SRC}")"; else INPUT="$(cat)"; fi
PYF="$(mktemp -t dolpy.XXXXXX)"; trap 'rm -f "${PYF}"' EXIT
cat > "${PYF}" <<'PY'
import json, sys
d = json.loads(sys.stdin.read())
if not d.get("name"):
sys.exit("thirdparty-create.sh: 'name' is required")
role = d.get("role", "client").lower()
is_client = role in ("client", "both")
is_supp = role in ("supplier", "fournisseur", "both")
body = {
"name": d["name"],
"client": "1" if is_client else "0",
"fournisseur": "1" if is_supp else "0",
"country_id": str(d.get("country_id", 1)),
# "-1" => Dolibarr auto-assigns the next code from the mask
# (COMPANY_ELEPHANT_MASK_CUSTOMER / _SUPPLIER); "0" when that role is off.
"code_client": (d.get("client_code", "-1") if is_client else "0"),
"code_fournisseur": (d.get("supplier_code", "-1") if is_supp else "0"),
}
for k in ("siret", "tva_intra", "address", "zip", "town", "email", "phone", "idprof1"):
if d.get(k):
body[k] = d[k]
print(json.dumps(body))
PY
BODY="$(printf '%s' "${INPUT}" | python3 "${PYF}")"
"${W}" POST /thirdparties "${BODY}"

View File

@@ -19,6 +19,12 @@ WISE_API_TOKEN=<from wise.com/settings/api-tokens>
WISE_PROFILE_ID=<numeric id of the BUSINESS profile — bank probe prints it> WISE_PROFILE_ID=<numeric id of the BUSINESS profile — bank probe prints it>
# Optional: only needed if Wise ever opens the EU statement endpoint # Optional: only needed if Wise ever opens the EU statement endpoint
WISE_SCA_KEY_PATH=~/.config/arcodange-erp/wise-sca-private.pem WISE_SCA_KEY_PATH=~/.config/arcodange-erp/wise-sca-private.pem
# Required by arcodange-email-ingest only
ZOHO_CLIENT_ID=<from api-console.zoho.com self-client>
ZOHO_CLIENT_SECRET=<same>
ZOHO_REFRESH_TOKEN=<exchanged from one-time code via /oauth/v2/token>
ZOHO_DC=eu # eu | com | in | au
EOF EOF
chmod 600 .claude/skills/dolibarr/.env chmod 600 .claude/skills/dolibarr/.env
``` ```

View File

@@ -151,7 +151,8 @@ Not available on this account (intentionally): `/setup/modules` (admin-only), `/
- Workflow skill for supplier-side TVA déductible (CA3 lignes 19 / 20 / 17+24): [dolibarr-tva-deductible](../dolibarr-tva-deductible/SKILL.md). - Workflow skill for supplier-side TVA déductible (CA3 lignes 19 / 20 / 17+24): [dolibarr-tva-deductible](../dolibarr-tva-deductible/SKILL.md).
- Workflow skill for composite CA3-ready TVA summary (collectée + déductible + net): [dolibarr-tva-summary](../dolibarr-tva-summary/SKILL.md). - Workflow skill for composite CA3-ready TVA summary (collectée + déductible + net): [dolibarr-tva-summary](../dolibarr-tva-summary/SKILL.md).
- **Bank-side reconciliation** (Qonto + Wise ↔ Dolibarr matching): [arcodange-bank-reco](../arcodange-bank-reco/SKILL.md). - **Bank-side reconciliation** (Qonto + Wise ↔ Dolibarr matching): [arcodange-bank-reco](../arcodange-bank-reco/SKILL.md).
- Future workflow skills follow the `dolibarr-<topic>` convention (ERP-internal) or `arcodange-<topic>` (cross-system, like bank reconciliation). Each one depends on this skill for connection + permissions + endpoint reference; each one keeps its triggers focused on its specific business workflow. - **Email ingestion** (Zoho Mail → supplier-invoice draft for Dolibarr): [arcodange-email-ingest](../arcodange-email-ingest/SKILL.md).
- Future workflow skills follow the `dolibarr-<topic>` convention (ERP-internal) or `arcodange-<topic>` (cross-system). Each one depends on this skill for connection + permissions + endpoint reference; each one keeps its triggers focused on its specific business workflow.
## Out of scope ## Out of scope

View File

@@ -22,7 +22,7 @@ concurrency:
url: https://vault.arcodange.lab url: https://vault.arcodange.lab
caCertificate: ${{ secrets.HOMELAB_CA_CERT }} caCertificate: ${{ secrets.HOMELAB_CA_CERT }}
jwtGiteaOIDC: ${{ needs.gitea_vault_auth.outputs.gitea_vault_jwt }} jwtGiteaOIDC: ${{ needs.gitea_vault_auth.outputs.gitea_vault_jwt }}
role: gitea_cicd_webapp role: gitea_cicd_erp
method: jwt method: jwt
path: gitea_jwt path: gitea_jwt
secrets: | secrets: |

View File

@@ -72,10 +72,15 @@ COMMANDS
probe Auth + discovery (org slug, profile id, balance ids) probe Auth + discovery (org slug, profile id, balance ids)
qonto-transactions [--month|--since|--until] Qonto transactions table (incoming + outgoing) qonto-transactions [--month|--since|--until] Qonto transactions table (incoming + outgoing)
wise-transactions [--month|--since|--until|--type|--enrich] Wise activities (incoming + outgoing) wise-transactions [--month|--since|--until|--type|--enrich] Wise activities (incoming + outgoing)
match [--month|--since|--until|--window-days N] Match bank ↔ Dolibarr (3 buckets) match [--month|--since|--until|--window-days N|--enrich] Match bank ↔ Dolibarr (split buckets)
balance Live balances + Dolibarr cross-check per fk_account balance Live balances + Dolibarr cross-check per fk_account
curl <qonto|wise> <path> Raw read-only curl through bank-curl.sh curl <qonto|wise> <path> Raw read-only curl through bank-curl.sh
email Supplier-invoice emails from the Zoho mailbox
list [--folder|--limit|--candidates-only|--all-folders] List candidates
inspect <messageId> [--folder|--save-pdf|--json] Parse PDFs + draft Dolibarr entry
curl <path> Raw read-only curl through zoho-curl.sh
whoami GET /users/info — confirm auth whoami GET /users/info — confirm auth
ping GET /status — liveness + Dolibarr version ping GET /status — liveness + Dolibarr version
curl <path> Raw read-only curl through dol-curl.sh curl <path> Raw read-only curl through dol-curl.sh
@@ -236,6 +241,30 @@ EOF
esac esac
;; ;;
email)
sub="${1:-help}"; shift || true
case "${sub}" in
list) exec "${SKILLS}/arcodange-email-ingest/scripts/email-list.sh" "$@" ;;
inspect) exec "${SKILLS}/arcodange-email-ingest/scripts/email-inspect.sh" "$@" ;;
curl) exec "${SKILLS}/arcodange-email-ingest/scripts/zoho-curl.sh" "$@" ;;
help|-h|--help)
cat <<'EOF'
arcodange email — supplier-invoice ingestion from the Zoho mailbox.
list [--folder PATH|--limit N|--candidates-only|--all-folders]
List messages (default: /Inbox/books)
inspect <messageId> [--folder PATH|--save-pdf DIR|--json]
Parse PDF attachments, propose Dolibarr supplier-invoice draft
curl <path> Raw read-only call through zoho-curl.sh
Requires ZOHO_CLIENT_ID, ZOHO_CLIENT_SECRET, ZOHO_REFRESH_TOKEN, ZOHO_DC in .env.
See arcodange-email-ingest/SKILL.md for OAuth setup.
EOF
;;
*) echo "arcodange email: unknown subcommand '${sub}' (try 'arcodange email help')" >&2; exit 2 ;;
esac
;;
whoami) whoami)
exec "${DOLC}" /users/info exec "${DOLC}" /users/info
;; ;;

View File

@@ -10,12 +10,12 @@ BEGIN
WHERE schemaname = 'public' WHERE schemaname = 'public'
LIMIT 1; LIMIT 1;
-- Si le propriétaire actuel est différent de erp_role -- Si le propriétaire actuel est différent de {{ .Values.db.ownerRole }}
IF current_schema_owner <> 'erp_role' THEN IF current_schema_owner <> '{{ .Values.db.ownerRole }}' THEN
-- Construire et exécuter la requête REASSIGN OWNED BY -- Construire et exécuter la requête REASSIGN OWNED BY
EXECUTE format('REASSIGN OWNED BY %I TO %I', current_schema_owner, 'erp_role'); EXECUTE format('REASSIGN OWNED BY %I TO %I', current_schema_owner, '{{ .Values.db.ownerRole }}');
RAISE NOTICE 'Ownership of all objects in schema "public" has been reassigned from % to %', current_schema_owner, 'erp_role'; RAISE NOTICE 'Ownership of all objects in schema "public" has been reassigned from % to %', current_schema_owner, '{{ .Values.db.ownerRole }}';
ELSE ELSE
RAISE NOTICE 'No change needed; the owner of schema "public" is already %', 'erp_role'; RAISE NOTICE 'No change needed; the owner of schema "public" is already %', '{{ .Values.db.ownerRole }}';
END IF; END IF;
END $$; END $$;

View File

@@ -10,8 +10,8 @@ data:
DOLI_DB_HOST_PORT: !!str 5432 DOLI_DB_HOST_PORT: !!str 5432
# DOLI_DB_USER: root # DOLI_DB_USER: root
# DOLI_DB_PASSWORD: root # DOLI_DB_PASSWORD: root
DOLI_DB_NAME: erp DOLI_DB_NAME: {{ .Values.db.name }}
DOLI_URL_ROOT: 'https://erp.arcodange.lab' DOLI_URL_ROOT: 'https://{{ .Values.host }}'
# DOLI_ADMIN_LOGIN: 'admin' # DOLI_ADMIN_LOGIN: 'admin'
# DOLI_ADMIN_PASSWORD: 'admininitialpassword' # DOLI_ADMIN_PASSWORD: 'admininitialpassword'
DOLI_ENABLE_MODULES: Societe,Facture DOLI_ENABLE_MODULES: Societe,Facture
@@ -20,4 +20,9 @@ data:
PHP_INI_DATE_TIMEZONE: Europe/Paris PHP_INI_DATE_TIMEZONE: Europe/Paris
DOLI_AUTH: dolibarr DOLI_AUTH: dolibarr
DOLI_CRON: !!str 0 DOLI_CRON: !!str 0
{{- if ne .Values.env "prod" }}
# Non-prod (sandbox): dev mode — render real errors inline instead of the generic
# prod "technical error" banner. Prod keeps the image default (DOLI_PROD=1).
DOLI_PROD: !!str 0
{{- end }}
# DOLI_INSTANCE_UNIQUE_ID: <random salt for encryption> # DOLI_INSTANCE_UNIQUE_ID: <random salt for encryption>

View File

@@ -7,4 +7,4 @@ data:
{{- .Files.Get "scripts/update_conf_db_credentials.sh" | nindent 4 }} {{- .Files.Get "scripts/update_conf_db_credentials.sh" | nindent 4 }}
update_table_ownership.sql: | update_table_ownership.sql: |
{{- .Files.Get "scripts/update_ownership.sql" | nindent 4 }} {{- tpl (.Files.Get "scripts/update_ownership.sql") . | nindent 4 }}

View File

@@ -7,7 +7,7 @@ spec:
method: kubernetes method: kubernetes
mount: kubernetes mount: kubernetes
kubernetes: kubernetes:
role: erp role: {{ .Values.vault.k8sRole }}
serviceAccount: {{ include "erp.serviceAccountName" . }} serviceAccount: {{ include "erp.serviceAccountName" . }}
audiences: audiences:
- vault - vault

View File

@@ -9,7 +9,7 @@ spec:
mount: postgres mount: postgres
# Path to the secret # Path to the secret
path: creds/erp path: {{ .Values.vault.dynamicPath }}
# Where to store the secrets, VSO will create the secret # Where to store the secrets, VSO will create the secret
destination: destination:

View File

@@ -10,7 +10,7 @@ spec:
mount: kvv2 mount: kvv2
# path of the secret # path of the secret
path: erp/config path: {{ .Values.vault.staticPath }}
# dest k8s secret # dest k8s secret
destination: destination:

40
chart/values-sandbox.yaml Normal file
View File

@@ -0,0 +1,40 @@
# Sandbox overlay — to be combined with values.yaml:
# helm install erp-sandbox chart/ -f chart/values.yaml -f chart/values-sandbox.yaml \
# --namespace erp-sandbox --create-namespace
#
# Activates Phase D of the multi-env evolution (cf. PR thread). Prerequisites:
# - factory/postgres/iac/terraform.tfvars: erp has envs = ["prod", "sandbox"]
# - tools/hashicorp-vault/iac/modules/app_roles: env parameter applied
# - arcodange-org/erp/iac/main.tf: for_each over local.envs (Phase D commit)
# - ArgoCD: Application "erp-sandbox" registered (Phase E)
#
# Derived names follow the elision rule: env=sandbox → suffix "-sandbox".
env: sandbox
instance: erp-sandbox
host: erp-sandbox.arcodange.lab
db:
name: erp-sandbox
ownerRole: erp_sandbox_role
vault:
k8sRole: erp-sandbox
dynamicPath: creds/erp-sandbox
staticPath: erp-sandbox/config
# Ingress annotations + hosts — override to point at the sandbox FQDN
ingress:
enabled: true
annotations:
traefik.ingress.kubernetes.io/router.entrypoints: websecure
traefik.ingress.kubernetes.io/router.tls: "true"
traefik.ingress.kubernetes.io/router.tls.certresolver: letsencrypt
traefik.ingress.kubernetes.io/router.tls.domains.0.main: arcodange.lab
traefik.ingress.kubernetes.io/router.tls.domains.0.sans: erp-sandbox.arcodange.lab
traefik.ingress.kubernetes.io/router.middlewares: localIp@file
hosts:
- host: erp-sandbox.arcodange.lab
paths:
- path: /
pathType: Prefix

View File

@@ -2,6 +2,27 @@
# This is a YAML-formatted file. # This is a YAML-formatted file.
# Declare variables to be passed into your templates. # Declare variables to be passed into your templates.
# ----------------------------------------------------------------------------
# Multi-environment coordinates (default = prod, elision rule applies).
# Override in values-<env>.yaml for any non-prod instance — see SKILL.md
# of the factory runbook (doc/runbooks/new-web-app/conventions.md).
# By the elision rule, env=prod produces names identical to single-env apps;
# env=sandbox produces "<app>-sandbox" everywhere except the Postgres owner
# role which uses snake-case "<app>_sandbox_role".
# ----------------------------------------------------------------------------
env: prod
instance: erp # derived id: env=prod → erp, else <app>-<env>
host: erp.arcodange.lab # internal hostname for this instance
db:
name: erp # PostgreSQL database name (matches factory tfvars)
ownerRole: erp_role # Postgres owner role; snake-case <app>_role for prod / <app>_<env>_role for non-prod (matches factory/postgres/iac)
vault:
k8sRole: erp # VaultAuth role (postgres/iac issues this per instance)
dynamicPath: creds/erp # path under postgres/ mount for short-lived DB creds
staticPath: erp/config # path under kvv2/ mount for the static admin config
replicaCount: 1 replicaCount: 1
image: image:

View File

@@ -3,30 +3,64 @@ locals {
name = "erp" name = "erp"
product_name = "dolibarr" # unused product_name = "dolibarr" # unused
} }
# Environments this app is deployed to. By the elision rule (factory runbook
# conventions.md / ADR-0002) env=prod renders identical to the single-env
# baseline; non-prod envs get a "<name>-<env>" instance id from the module.
envs = toset(["prod", "sandbox"])
} }
module "app_roles" { module "app_roles" {
source = "git::ssh://git@192.168.1.202:2222/arcodange-org/tools.git//hashicorp-vault/iac/modules/app_roles?depth=1&ref=main" source = "git::ssh://git@192.168.1.202:2222/arcodange-org/tools.git//hashicorp-vault/iac/modules/app_roles?depth=1&ref=main"
for_each = local.envs
name = local.app.name name = local.app.name
env = each.key
} }
resource "random_password" "admin_initial_password" { resource "random_password" "admin_initial_password" {
for_each = local.envs
length = 32 length = 32
} }
resource "random_uuid" "dolibarr_id" { # used for encryption as well as when buying modules resource "random_uuid" "dolibarr_id" { # used for encryption as well as when buying modules
for_each = local.envs
lifecycle { lifecycle {
prevent_destroy = true prevent_destroy = true
} }
} }
resource "vault_kv_secret_v2" "dolibarr_admin_setup" { resource "vault_kv_secret_v2" "dolibarr_admin_setup" {
mount = module.app_roles.mount_paths.kvv2 for_each = local.envs
name = format("%sconfig", module.app_roles.kvv2_path_prefix) mount = module.app_roles[each.key].mount_paths.kvv2
name = format("%sconfig", module.app_roles[each.key].kvv2_path_prefix)
data_json = jsonencode( data_json = jsonencode(
{ {
DOLI_ADMIN_LOGIN = "admin", DOLI_ADMIN_LOGIN = "admin",
DOLI_ADMIN_PASSWORD = random_password.admin_initial_password.result DOLI_ADMIN_PASSWORD = random_password.admin_initial_password[each.key].result
DOLI_INSTANCE_UNIQUE_ID = random_uuid.dolibarr_id.result DOLI_INSTANCE_UNIQUE_ID = random_uuid.dolibarr_id[each.key].result
} }
) )
} }
# State migration (ADR-0002 Phase D): re-key the pre-existing single-env resources
# into the for_each map under the "prod" key so that introducing for_each does NOT
# plan a destroy+create. This is critical for random_uuid.dolibarr_id — it carries
# prevent_destroy = true (it is the prod Dolibarr encryption id + paid-module
# binding), so a missing moved block would HARD-FAIL the apply rather than silently
# losing it. The module's own internal moved (role -> role[0]) chains with the
# module re-key here.
moved {
from = module.app_roles
to = module.app_roles["prod"]
}
moved {
from = random_password.admin_initial_password
to = random_password.admin_initial_password["prod"]
}
moved {
from = random_uuid.dolibarr_id
to = random_uuid.dolibarr_id["prod"]
}
moved {
from = vault_kv_secret_v2.dolibarr_admin_setup
to = vault_kv_secret_v2.dolibarr_admin_setup["prod"]
}

60
ops/sandbox/README.md Normal file
View File

@@ -0,0 +1,60 @@
# erp-sandbox lifecycle ops
Tooling to make `erp-sandbox` **iso-prod** and to reset it, implementing
[ADR-0003](https://gitea.arcodange.lab/arcodange-org/factory/src/branch/main/vibe/ADR/0003-sandbox-state-lifecycle.md)
(sandbox state lifecycle). The sandbox exists so AI agents can rehearse Dolibarr
**write** operations against a faithful copy of prod, with a structural guarantee
that the rehearsal path can never mutate prod.
## The prod-integrity guarantee (why this is safe)
| Layer | Enforcement |
| --- | --- |
| prod is **read-only** during a refresh | `pg_dump` runs with `default_transaction_read_only=on` |
| the restore can only write the sandbox | it uses the sandbox's own dynamic creds — a member of `erp_sandbox_role`, which **owns only `erp-sandbox`** |
| no database is dropped/created | wipe is `DROP OWNED BY erp_sandbox_role CASCADE`; reload is `pg_restore` (no `CREATEDB`, no superuser) |
| prod is structurally undroppable | `DROP DATABASE` needs ownership; `erp_sandbox_role` does not own prod `erp` (owned by `erp_role`) |
The only prod-capable credential on the platform is the `superuser=true` provider
in `factory postgres/iac/providers.tf`, used **only** in the human-gated
`postgres.yaml` CI. This tooling never touches it.
## Usage
```sh
./sandbox-lifecycle.sh refresh-from-prod # clone prod DB (data + config) into erp-sandbox
./sandbox-lifecycle.sh sync-documents # copy mycompany/ uploads (company logo, PDFs)
./sandbox-lifecycle.sh refresh # both, in order
```
`refresh-from-prod` scales the sandbox pod to 0, dumps the full prod `public`
schema (read-only), wipes the sandbox's app objects, restores, and scales back
up. It dumps the **whole** schema (not just `llx_*`) so app helper functions and
their triggers (e.g. `update_modified_column_tms()`) come over; it filters out
the provisioner-owned `user_lookup` pgbouncer function from the restore TOC
because that object already exists per-environment and is not app data.
## Two fidelity caveats (by design — see ADR-0003)
1. **Encryption.** Dolibarr ties some encrypted fields (notably API keys) to
`DOLI_INSTANCE_UNIQUE_ID`. The sandbox has its **own** uuid, so prod-encrypted
values won't decrypt there. This is why the write-scoped `ai_agent_sandbox`
API key must be **generated inside the sandbox** (see `../../test/` POC),
not copied from prod. Most data is plaintext and unaffected.
2. **Uploaded files live on the PVC, not the DB.** A DB refresh copies the logo
*const* (`MAIN_INFO_SOCIETE_LOGO`) but not the image; `sync-documents` copies
the `documents/mycompany` tree so the logo + attachments actually render.
## BDD reset loop (E4)
For repeated rehearsals, `refresh-from-prod` is the "reset to prod state". A
faster checkpoint/reset that avoids re-reading prod each time (cache a golden
dump on a small PVC, then `DROP OWNED + pg_restore` from it) is the documented
next optimization — see ADR-0003 §Decision/Consequences.
## Hardening backlog
- Replace the transient copy of prod's read+write creds with a **dedicated
read-only Postgres role** (issued via a Vault dynamic role) so the dump path is
least-privilege by construction, not just by `default_transaction_read_only`.
- Provision a golden-cache PVC for fast BDD resets.

134
ops/sandbox/sandbox-lifecycle.sh Executable file
View File

@@ -0,0 +1,134 @@
#!/usr/bin/env bash
#
# sandbox-lifecycle.sh — seed / refresh the erp-sandbox Dolibarr from prod, and
# sync its uploaded documents, with prod integrity guaranteed structurally.
#
# Implements ADR-0003 (factory vibe/ADR/0003-sandbox-state-lifecycle.md):
# - prod is read ONLY (pg_dump runs in a default_transaction_read_only session);
# - the restore writes ONLY to erp-sandbox, using the sandbox's own dynamic
# credentials (a member of erp_sandbox_role, which owns only the sandbox DB),
# so it is structurally incapable of touching prod 'erp' (owned by erp_role);
# - no DROP/CREATE DATABASE, no CREATEDB, no superuser — wipe is
# `DROP OWNED BY erp_sandbox_role CASCADE`, reload is `pg_restore`.
#
# The only prod-capable credential on the platform is the superuser provider in
# factory postgres/iac, exercised solely in the human-gated postgres.yaml CI —
# this script never uses it.
#
# Requires: kubectl (context on the lab cluster), and a postgres:16 image
# reachable by the cluster. Run from anywhere.
#
# Usage:
# ./sandbox-lifecycle.sh refresh-from-prod # iso-prod seed: prod DB -> erp-sandbox
# ./sandbox-lifecycle.sh sync-documents # copy mycompany/ uploads (logo, PDFs)
# ./sandbox-lifecycle.sh refresh # refresh-from-prod + sync-documents
#
set -euo pipefail
PROD_NS="erp"
SB_NS="erp-sandbox"
PROD_DB="erp"
SB_DB="erp-sandbox"
SB_ROLE="erp_sandbox_role" # snake-case owner role (ADR-0002 elision rule)
PGHOST="192.168.1.202" # direct Postgres (NOT pgbouncer — pooler breaks pg_dump)
PG_IMAGE="postgres:16-alpine"
DOC_ROOT="/var/www/documents" # dolibarr_main_data_root
TMP_PROD_SECRET="prod-db-ro-temp" # transient copy of prod creds, deleted on exit
log() { printf '\033[1;36m==>\033[0m %s\n' "$*"; }
die() { printf '\033[1;31mABORT:\033[0m %s\n' "$*" >&2; exit 1; }
sb_pod() { kubectl get pod -n "$SB_NS" -l app.kubernetes.io/instance=erp-sandbox -o name 2>/dev/null | head -1; }
prod_pod() { kubectl get pod -n "$PROD_NS" -l app.kubernetes.io/instance=erp -o name 2>/dev/null | head -1; }
cleanup_secret() { kubectl delete secret "$TMP_PROD_SECRET" -n "$SB_NS" --ignore-not-found >/dev/null 2>&1 || true; }
refresh_from_prod() {
command -v python3 >/dev/null || die "python3 required to copy the prod secret without exposing it"
trap cleanup_secret EXIT
log "Copying prod DB creds into a transient, read-only-intent secret in $SB_NS (values stay base64)"
kubectl get secret vso-db-credentials -n "$PROD_NS" -o json \
| python3 -c "import json,sys; d=json.load(sys.stdin); d['metadata']={'name':'$TMP_PROD_SECRET','namespace':'$SB_NS'}; d.pop('status',None); d['data']={k:d['data'][k] for k in ('username','password')}; print(json.dumps(d))" \
| kubectl apply -f - >/dev/null
log "Scaling erp-sandbox to 0 (exclusive DB access for the restore)"
kubectl scale deploy erp-sandbox -n "$SB_NS" --replicas=0 >/dev/null
kubectl wait --for=delete pod -l app.kubernetes.io/instance=erp-sandbox -n "$SB_NS" --timeout=120s >/dev/null 2>&1 || true
log "Running the seed Job (pg_dump prod read-only -> DROP OWNED -> pg_restore into sandbox)"
kubectl delete job sandbox-seed -n "$SB_NS" --ignore-not-found >/dev/null 2>&1 || true
kubectl apply -f - >/dev/null <<EOF
apiVersion: batch/v1
kind: Job
metadata: { name: sandbox-seed, namespace: $SB_NS }
spec:
backoffLimit: 0
ttlSecondsAfterFinished: 900
template:
spec:
restartPolicy: Never
containers:
- name: seed
image: $PG_IMAGE
env:
- { name: PROD_PGUSER, valueFrom: { secretKeyRef: { name: $TMP_PROD_SECRET, key: username } } }
- { name: PROD_PGPASSWORD, valueFrom: { secretKeyRef: { name: $TMP_PROD_SECRET, key: password } } }
- { name: SB_PGUSER, valueFrom: { secretKeyRef: { name: vso-db-credentials, key: username } } }
- { name: SB_PGPASSWORD, valueFrom: { secretKeyRef: { name: vso-db-credentials, key: password } } }
- { name: PGHOST, value: "$PGHOST" }
- { name: PGSSLMODE, value: "disable" }
command: ["/bin/sh","-c"]
args:
- |
set -eu
SBDB=\$(PGPASSWORD=\$SB_PGPASSWORD psql -h "\$PGHOST" -U "\$SB_PGUSER" -d $SB_DB -tAc 'select current_database()')
[ "\$SBDB" = "$SB_DB" ] || { echo "ABORT: target is '\$SBDB' not $SB_DB"; exit 1; }
echo "source=$PROD_DB (read-only) target=$SB_DB ok"
# 1. dump prod — full public schema (incl. helper functions + triggers), read-only session
PGPASSWORD=\$PROD_PGPASSWORD PGOPTIONS='-c default_transaction_read_only=on' \\
pg_dump -h "\$PGHOST" -U "\$PROD_PGUSER" -d $PROD_DB -n public -Fc -f /tmp/golden.dump
# drop provisioner-owned infra (pgbouncer user_lookup) from the TOC: it already
# exists in the sandbox and is not app data, so restoring it conflicts.
pg_restore -l /tmp/golden.dump | grep -vi 'user_lookup' > /tmp/golden.toc
echo "dump: \$(ls -l /tmp/golden.dump | awk '{print \$5}') bytes, tables=\$(grep -c 'TABLE DATA ' /tmp/golden.toc)"
# 2. wipe sandbox app objects (everything owned by the app role; infra untouched)
PGPASSWORD=\$SB_PGPASSWORD psql -h "\$PGHOST" -U "\$SB_PGUSER" -d $SB_DB -v ON_ERROR_STOP=1 \\
-c "DROP OWNED BY $SB_ROLE CASCADE;"
# 3. restore golden, owned by the sandbox role
PGPASSWORD=\$SB_PGPASSWORD \\
pg_restore -L /tmp/golden.toc --no-owner --role=$SB_ROLE -d $SB_DB /tmp/golden.dump \\
&& echo "restore: clean" || echo "restore: completed with ignorable warnings"
# 4. verify
Q() { PGPASSWORD=\$SB_PGPASSWORD psql -h "\$PGHOST" -U "\$SB_PGUSER" -d $SB_DB -tAc "\$1"; }
echo "llx tables=\$(Q "select count(*) from pg_tables where schemaname='public' and tablename like 'llx_%'") company=\$(Q "select value from llx_const where name='MAIN_INFO_SOCIETE_NOM'") lang=\$(Q "select value from llx_const where name='MAIN_LANG_DEFAULT'") owner=\$(Q "select tableowner from pg_tables where tablename='llx_societe'")"
echo "DONE."
EOF
kubectl wait --for=condition=complete job/sandbox-seed -n "$SB_NS" --timeout=300s >/dev/null 2>&1 \
|| die "seed Job did not complete — see: kubectl logs -n $SB_NS job/sandbox-seed"
kubectl logs -n "$SB_NS" job/sandbox-seed | sed 's/^/ /'
kubectl delete job sandbox-seed -n "$SB_NS" --ignore-not-found >/dev/null 2>&1 || true
log "Scaling erp-sandbox back to 1"
kubectl scale deploy erp-sandbox -n "$SB_NS" --replicas=1 >/dev/null
cleanup_secret; trap - EXIT
log "Refresh complete. Run 'sync-documents' to also copy the company logo + uploads."
}
sync_documents() {
local pp sp
pp=$(prod_pod); sp=$(sb_pod)
[ -n "$pp" ] || die "no prod erp pod found"
[ -n "$sp" ] || die "no erp-sandbox pod found"
log "Syncing $DOC_ROOT/mycompany (logo + uploads) ${pp##*/} -> ${sp##*/} via tar pipe"
kubectl exec -n "$PROD_NS" "${pp#pod/}" -- tar -C "$DOC_ROOT" -cf - mycompany 2>/dev/null \
| kubectl exec -i -n "$SB_NS" "${sp#pod/}" -- tar -C "$DOC_ROOT" -xf -
log "Documents synced. (For a one-shot logo only, scope the tar to mycompany/logos.)"
}
case "${1:-}" in
refresh-from-prod) refresh_from_prod ;;
sync-documents) sync_documents ;;
refresh) refresh_from_prod; sync_documents ;;
*) echo "usage: $0 {refresh-from-prod|sync-documents|refresh}" >&2; exit 2 ;;
esac

View File

@@ -1,5 +1,27 @@
# Copy this template to one of:
# .env — production target, loaded by main.ts
# .env.sandbox — sandbox target, loaded by provisionSandbox.ts
# Both are gitignored. Never commit real secret values.
# --- Target ---
# prod: https://erp.arcodange.lab (.env)
# sandbox: https://erp-sandbox.arcodange.lab (.env.sandbox)
DOLIBARR_ADDRESS=https://erp.arcodange.lab DOLIBARR_ADDRESS=https://erp.arcodange.lab
DOLI_DB_PASSWORD=
DOLI_ADMIN_LOGIN=admin DOLI_ADMIN_LOGIN=admin
DOLI_ADMIN_PASSWORD="" DOLI_ADMIN_PASSWORD=""
DOLI_DB_PASSWORD=""
ROOT_FOLDER=$HOME/erp ROOT_FOLDER=$HOME/erp
# Populate the passwords from the cluster secrets, e.g. (prod shown):
# DOLI_ADMIN_PASSWORD <- kubectl get secret secretkv -n erp -o jsonpath='{.data.DOLI_ADMIN_PASSWORD}' | base64 -d
# DOLI_DB_PASSWORD <- kubectl get secret vso-db-credentials -n erp -o jsonpath='{.data.password}' | base64 -d
#
# NOTE for a sandbox SEEDED from prod (ops/sandbox/sandbox-lifecycle.sh): the seed
# clones prod's admin password into the sandbox, so .env.sandbox's
# DOLI_ADMIN_PASSWORD must be PROD's admin password (-n erp), not the sandbox
# secretkv. The DB password is the sandbox's own (-n erp-sandbox).
# Optional: fix the provisioned user's password (else one is generated and only
# the API key is emitted to .ai_agent_sandbox.key).
# AI_AGENT_SANDBOX_PASSWORD=""

6
test/.gitignore vendored Normal file
View File

@@ -0,0 +1,6 @@
# Secrets — never commit. Covers .env (prod, main.ts) and .env.sandbox
# (sandbox, provisionSandbox.ts), plus any generated *.key.
.env*
!.env.example
.ai_agent_sandbox.key
*.key

106
test/README.md Normal file
View File

@@ -0,0 +1,106 @@
# test — Dolibarr UI automation (Deno + Playwright)
A small Deno + Playwright POC that drives the Dolibarr admin UI in the `fr-FR`
locale. Playwright fills the same forms a human admin would, so the automation
works even where the REST API can't (e.g. generating an API key, which is
encrypted with the instance's own `DOLI_INSTANCE_UNIQUE_ID`).
## Layout
- `main.ts` — original entrypoint (first install, company/display/module setup).
- `provisionSandbox.ts` — entrypoint that provisions the `erp-sandbox` instance
for the AI agent (enable REST API, create a write-scoped user, generate its
API key).
- `scripts/login.ts` — admin login / logout / whoami helpers.
- `scripts/forms.ts``fillForm`, `toggleOnOff`, CKEditor/ACE helpers.
- `scripts/admin/moduleSetup.ts``configureModule`, `enableApiModule`.
- `scripts/admin/userSetup.ts``createUser`, `assignRights`, `generateApiKey`.
## Configure
Copy `.env.example` to `.env` and fill it in. `.env`, `*.key`, and
`.ai_agent_sandbox.key` are gitignored — never commit secrets.
```sh
cp .env.example .env
```
## Lock the installer (after a fresh install via `main.ts`)
Dolibarr keeps its web installer reachable until an `install.lock` file exists.
After a fresh install (the `main.ts` flow), create it in the target pod — for the
sandbox:
```sh
kubectl -n erp-sandbox exec \
"$(kubectl get pod -n erp-sandbox -l app.kubernetes.io/instance=erp-sandbox -o name)" -- \
/bin/sh -c 'touch /var/www/documents/install.lock && chown www-data:www-data /var/www/documents/install.lock'
```
The path is the Dolibarr **data root** (`/var/www/documents`, a PVC) — that's where
Dolibarr checks, and being on the PVC the lock persists across pod restarts. For
prod, swap to `-n erp -l app.kubernetes.io/instance=erp`. A sandbox **seeded** from
prod still needs this: the seed (see `../ops/sandbox/`) copies the DB +
`documents/mycompany`, not `install.lock`.
## Provision the sandbox
Provisions `erp-sandbox.arcodange.lab`: enables the REST API module, creates the
write-scoped `ai_agent_sandbox` user, grants it its write rights, and has
Dolibarr generate the user's API key. The key is written to
`test/.ai_agent_sandbox.key` (gitignored) — it is never printed.
```sh
cd test
deno run --allow-all provisionSandbox.ts
```
Populate `.env` from the `erp-sandbox` namespace secrets first. `secretkv`
carries the app env (including `DOLI_ADMIN_PASSWORD`); `vso-db-credentials`
carries the database password:
```sh
# Admin password (key DOLI_ADMIN_PASSWORD inside the secretkv secret)
kubectl get secret secretkv -n erp-sandbox \
-o jsonpath='{.data.DOLI_ADMIN_PASSWORD}' | base64 -d
# Database password (key `password` inside vso-db-credentials)
kubectl get secret vso-db-credentials -n erp-sandbox \
-o jsonpath='{.data.password}' | base64 -d
```
Set in `.env`:
```sh
DOLIBARR_ADDRESS=https://erp-sandbox.arcodange.lab
DOLI_ADMIN_LOGIN=admin
DOLI_ADMIN_PASSWORD="<from secretkv above>"
DOLI_DB_PASSWORD="<from vso-db-credentials above>"
# Optional — otherwise a random password is generated and only the API key emitted:
# AI_AGENT_SANDBOX_PASSWORD="<choose one>"
```
### After it runs
The generated API key lands in `test/.ai_agent_sandbox.key`. Next step (not
automated by this POC): load it into the `dolibarr` skill's sandbox config /
Vault at `kvv2/erp-sandbox/ai_agent`.
> [!IMPORTANT]
> The sandbox Dolibarr is not installed/provisioned yet (empty DB, fresh install
> wizard). Until the install wizard has been completed against the sandbox,
> `provisionSandbox.ts` will not have a UI to drive, and the selectors in
> `moduleSetup.ts` / `userSetup.ts` are best-effort (Dolibarr 22 conventions,
> not verified live). Confirm them on the first real run.
### Write rights granted
The `ai_agent_sandbox` user is created non-admin and granted read + create on:
| Module | rights ids |
| ---------------- | ---------------------------------- |
| facture | lire=11, creer=12 |
| societe | lire=121, creer=122 |
| societe contact | lire=281, creer=282 |
| fournisseur | lire=1181, facture lire=1231, facture creer=1232 |
| produit | lire=31, creer=32 |

26
test/deno.lock generated
View File

@@ -1,7 +1,8 @@
{ {
"version": "4", "version": "5",
"specifiers": { "specifiers": {
"jsr:@std/dotenv@*": "0.225.2", "jsr:@std/dotenv@*": "0.225.2",
"npm:@types/node@*": "24.2.0",
"npm:playwright@^1.48.2": "1.48.2" "npm:playwright@^1.48.2": "1.48.2"
}, },
"jsr": { "jsr": {
@@ -10,18 +11,33 @@
} }
}, },
"npm": { "npm": {
"@types/node@24.2.0": {
"integrity": "sha512-3xyG3pMCq3oYCNg7/ZP+E1ooTaGB4cG8JWRsqqOYQdbWNY4zbaV0Ennrd7stjiJEFZCaybcIgpTjJWHRfBSIDw==",
"dependencies": [
"undici-types"
]
},
"fsevents@2.3.2": { "fsevents@2.3.2": {
"integrity": "sha512-xiqMQR4xAeHTuB9uWm+fFRcIOgKBMiOBP+eXiyT7jsgVCq1bkVygt00oASowB7EdtpOHaaPgKt812P9ab+DDKA==" "integrity": "sha512-xiqMQR4xAeHTuB9uWm+fFRcIOgKBMiOBP+eXiyT7jsgVCq1bkVygt00oASowB7EdtpOHaaPgKt812P9ab+DDKA==",
"os": ["darwin"],
"scripts": true
}, },
"playwright-core@1.48.2": { "playwright-core@1.48.2": {
"integrity": "sha512-sjjw+qrLFlriJo64du+EK0kJgZzoQPsabGF4lBvsid+3CNIZIYLgnMj9V6JY5VhM2Peh20DJWIVpVljLLnlawA==" "integrity": "sha512-sjjw+qrLFlriJo64du+EK0kJgZzoQPsabGF4lBvsid+3CNIZIYLgnMj9V6JY5VhM2Peh20DJWIVpVljLLnlawA==",
"bin": true
}, },
"playwright@1.48.2": { "playwright@1.48.2": {
"integrity": "sha512-NjYvYgp4BPmiwfe31j4gHLa3J7bD2WiBz8Lk2RoSsmX38SVIARZ18VYjxLjAcDsAhA+F4iSEXTSGgjua0rrlgQ==", "integrity": "sha512-NjYvYgp4BPmiwfe31j4gHLa3J7bD2WiBz8Lk2RoSsmX38SVIARZ18VYjxLjAcDsAhA+F4iSEXTSGgjua0rrlgQ==",
"dependencies": [ "dependencies": [
"fsevents",
"playwright-core" "playwright-core"
] ],
"optionalDependencies": [
"fsevents"
],
"bin": true
},
"undici-types@7.10.0": {
"integrity": "sha512-t5Fy/nfn+14LuOc2KNYg75vZqClpAiqscVvMygNnlsHBFpSXdJaYtXMcdNLpl/Qvc3P2cB3s6lOV51nqsFq4ag=="
} }
}, },
"workspace": { "workspace": {

152
test/provisionSandbox.ts Normal file
View File

@@ -0,0 +1,152 @@
import { loadSync } from "jsr:@std/dotenv";
// Sandbox provisioning loads its OWN .env.sandbox; prod config stays in .env (main.ts).
loadSync({ envPath: ".env.sandbox", export: true });
import { chromium } from "playwright";
import path from "node:path";
import login from "./scripts/login.ts";
import moduleSetup from "./scripts/admin/moduleSetup.ts";
import userSetup from "./scripts/admin/userSetup.ts";
/*
provisionSandbox.ts — separate entrypoint (does NOT touch main.ts).
Provisions the erp-sandbox Dolibarr so the AI agent can write to it:
1. enable the REST API / Web services module,
2. create a write-scoped `ai_agent_sandbox` user (non-admin),
3. grant it the write rights it needs,
4. have Dolibarr generate its API key and emit it safely to
test/.ai_agent_sandbox.key (gitignored).
Run:
cd test && deno run --allow-all provisionSandbox.ts
CANNOT be run end-to-end yet: the sandbox Dolibarr is not installed/provisioned
(empty DB, fresh install wizard). Selector correctness in moduleSetup.ts /
userSetup.ts is therefore best-effort and must be verified on the first real
run, AFTER the install wizard has been completed against the sandbox.
NEXT STEP (not implemented here): load the generated key into the dolibarr
skill's sandbox config / Vault at kvv2/erp-sandbox/ai_agent. We intentionally
do NOT write to Vault from this POC.
*/
/*
Write rights to grant `ai_agent_sandbox` (stable Dolibarr rights ids, verified
against prod Dolibarr 22.0.4). Each module's read (lire) + create (creer):
facture: lire=11, creer=12
societe: lire=121, creer=122, client voir (see-all)=262
societe contact: lire=281, creer=282
fournisseur: lire=1181, facture lire=1231, facture creer=1232
produit: lire=31, creer=32
*/
const WRITE_IDS = [
11, // facture lire
12, // facture creer
121, // societe lire
122, // societe creer
262, // societe client "voir" (see-all) — REQUIRED for the /thirdparties LIST
// endpoint; without it the list 404s (the voir_tous ACL trap, cf. the
// dolibarr skill SKILL.md). Other modules' lists work with plain `lire`.
281, // societe contact lire
282, // societe contact creer
1181, // fournisseur lire
1231, // fournisseur facture lire
1232, // fournisseur facture creer
31, // produit lire
32, // produit creer
];
const KEY_FILE = ".ai_agent_sandbox.key";
/*
Initialisation — mirrors main.ts but targeted at the sandbox. The default
address points at the sandbox; admin creds come from env (the same
DOLI_ADMIN_LOGIN / DOLI_ADMIN_PASSWORD vars main.ts uses, populated from the
erp-sandbox namespace secrets — see test/README.md "Provision the sandbox").
*/
const dolibarrAddress = Deno.env.get("DOLIBARR_ADDRESS") ||
"https://erp-sandbox.arcodange.lab";
const debug = true;
const DBpassword = Deno.env.get("DOLI_DB_PASSWORD") || "undefined";
const adminCredentials = {
username: Deno.env.get("DOLI_ADMIN_LOGIN") || "admin",
password: Deno.env.get("DOLI_ADMIN_PASSWORD") || "undefined",
};
// The new user's login is fixed; its password comes from env or is generated.
const AI_AGENT_LOGIN = "ai_agent_sandbox";
const aiAgentPassword = Deno.env.get("AI_AGENT_SANDBOX_PASSWORD") ||
generatePassword();
const rootFolderPath = Deno.env.get("ROOT_FOLDER") ||
path.join(Deno.cwd(), "..");
const imgFolderPath = Deno.env.get("IMG_FOLDER") ||
path.join(rootFolderPath, "static/img");
const configFolderPath = Deno.env.get("CONFIG_FOLDER") ||
path.join(rootFolderPath, "static/config");
/* Generate a reasonably strong random password (used only if none provided). */
function generatePassword(): string {
const bytes = new Uint8Array(18);
crypto.getRandomValues(bytes);
// URL-safe base64 → no shell-hostile characters.
return btoa(String.fromCharCode(...bytes))
.replace(/\+/g, "-")
.replace(/\//g, "_")
.replace(/=+$/, "");
}
const browser = await chromium.launch({
headless: false,
logger: {
isEnabled: (_name, _severity) => debug,
log: (name, severity, message, args) =>
console.warn(`${severity}| ${name} :: ${message} __ ${args}`),
},
});
const context = await browser.newContext({ locale: "fr-FR" });
const page = await context.newPage();
const globalCtx = {
dolibarrAddress,
DBpassword,
adminCredentials,
debug,
rootFolderPath,
imgFolderPath,
configFolderPath,
browser,
page,
context,
};
try {
await login.doAdminLogin(globalCtx);
console.log(`connected as ${await login.whoAmI(globalCtx)}`);
await moduleSetup.enableApiModule(globalCtx);
console.log("REST API module enabled");
const userId = await userSetup.createUser(globalCtx, {
login: AI_AGENT_LOGIN,
password: aiAgentPassword,
lastname: "AI Agent (sandbox)",
admin: false,
});
console.log(`created user '${AI_AGENT_LOGIN}' (id=${userId})`);
await userSetup.assignRights(globalCtx, userId, WRITE_IDS);
console.log(`granted ${WRITE_IDS.length} write rights to id=${userId}`);
const apiKey = await userSetup.generateApiKey(globalCtx, userId);
// Emit the key safely: write it to a gitignored file rather than printing it.
await Deno.writeTextFile(KEY_FILE, apiKey + "\n");
console.log(`AI_AGENT_SANDBOX API KEY WRITTEN TO test/${KEY_FILE}`);
} catch (error) {
console.error(error);
} finally {
await browser.close();
}

View File

@@ -1,19 +1,20 @@
// info-box // info-box
import { Page } from "playwright"; import { Locator, Page } from "playwright";
import forms from "../forms.ts"; import forms from "../forms.ts";
import login from "../login.ts"; import login from "../login.ts";
// span.info-box-title has-text /moduleName/ // span.info-box-title has-text /moduleName/
// span.fa-cog[title="Configuration"] // span.fa-cog[title="Configuration"]
async function configureModule( type ModuleCtx = {
globalCtx: {
page: Page; page: Page;
dolibarrAddress: string; dolibarrAddress: string;
adminCredentials: { username: string; password: string }; adminCredentials: { username: string; password: string };
}, };
async function configureModule(
globalCtx: ModuleCtx,
{ {
moduleName, moduleName,
enabled, enabled,
@@ -42,6 +43,61 @@ async function configureModule(
} }
} }
/*
Enable Dolibarr's REST API / Web services module.
Activation maps to llx_const.MAIN_MODULE_API=1; in the UI this is the module
card on /admin/modules.php (kanban mode). In fr_FR the card title is
"API/Web services REST (serveur)" (family "Interfaces").
NOTE: confirm the exact card title on the first real run against an installed
sandbox — the label has not been verified live here. To stay resilient we try
the known fr_FR label first (anchored exact match, same as configureModule),
and if that card is not present we fall back to matching ANY module card whose
title contains "API ... REST" / "REST ... API" in either order.
*/
async function enableApiModule(globalCtx: ModuleCtx): Promise<void> {
const {page, dolibarrAddress}= globalCtx;
// fr_FR label as observed in the module catalogue. Confirm on first real run.
const knownLabel = "API/Web services REST (serveur)";
await login.doAdminLogin(globalCtx);
await page.goto(`${dolibarrAddress}/admin/modules.php?mode=commonkanban`);
const exactCard = page.locator('.info-box', {
has: page.locator('.info-box-title',{
hasText: new RegExp('^'+knownLabel+'\n?$','i')
})
});
if(await exactCard.count() > 0) {
await forms.toggleOnOff(exactCard, true);
return;
}
// Fallback: the exact fr_FR label was not found (different Dolibarr version,
// locale, or wording). Match any card whose title looks like the REST API
// module regardless of word order.
const fuzzyCard = page.locator('.info-box', {
has: page.locator('.info-box-title',{
hasText: /API.*REST|REST.*API/i
})
});
if(await fuzzyCard.count() === 0) {
throw new Error(
`REST API module card not found on ${dolibarrAddress}/admin/modules.php `+
`(tried exact label "${knownLabel}" and /API.*REST|REST.*API/i). `+
`Confirm the card title in the running sandbox.`,
);
}
// If several cards match the fuzzy pattern, toggle the first one only.
await forms.toggleOnOff(fuzzyCard.first() as Locator, true);
}
export default { export default {
configureModule, configureModule,
enableApiModule,
} }

View File

@@ -0,0 +1,297 @@
/*
User provisioning for the Dolibarr admin UI, driven by Playwright.
Why the UI and not raw SQL:
- Passwords use MAIN_SECURITY_HASH_ALGO=password_hash (bcrypt), so we let
Dolibarr hash them by submitting the create form.
- API keys are stored ENCRYPTED with the instance key DOLI_INSTANCE_UNIQUE_ID.
Each instance (incl. the sandbox) has its own uuid, so a key can only be
produced correctly by that instance — we MUST generate it via the UI and
read back the resulting value, never INSERT a raw key.
IMPORTANT — selectors below are best-effort and have NOT been verified against
a live, installed sandbox (the sandbox Dolibarr is not provisioned yet). Field
names / icon controls are taken from Dolibarr 22 conventions and must be
confirmed on the first real run. Spots that are guessed are marked GUESS:.
*/
import type { Page } from "playwright";
import forms from "../forms.ts";
import login from "../login.ts";
type UserCtx = {
page: Page;
dolibarrAddress: string;
adminCredentials: { username: string; password: string };
// imgFolderPath is required by forms.fillForm's signature even though no
// file input is used here; provisionSandbox.ts supplies it from globalCtx.
imgFolderPath: string;
};
type NewUser = {
login: string;
password: string;
lastname?: string;
admin?: boolean;
};
// The text fields filled via forms.fillForm (the admin flag is a checkbox we
// handle separately, so it is intentionally not part of this type).
type UserFormFields = {
login: string;
password: string;
lastname?: string;
};
// fr_FR Dolibarr user create form (/user/card.php?action=create&mode=create).
// GUESS: confirm these input names on first real run. `login` and `lastname`
// are stable across Dolibarr versions; the password field has been `password`
// for the manual-entry case (the "generate automatically" option is a separate
// radio/checkbox we leave untouched so our explicit value is used).
const newUserInputNames = new Map<keyof UserFormFields, string>([
["login", "login"],
["lastname", "lastname"],
["password", "password"],
]);
/*
Create a user via /user/card.php?action=create and return its numeric id.
The admin flag is a checkbox (name="admin"); fillForm has no checkbox handler
so we toggle it explicitly. After submit Dolibarr redirects to the user card
whose URL carries ?id=<n> — we parse the id from there (falling back to a
hidden input on the page).
*/
async function createUser(
globalCtx: UserCtx,
{ login: userLogin, password, lastname, admin = false }: NewUser,
): Promise<number> {
const { page, dolibarrAddress, imgFolderPath } = globalCtx;
await login.doAdminLogin(globalCtx);
// Idempotent: if the login already exists (e.g. a re-run), return its id rather
// than submitting a duplicate create (which Dolibarr rejects).
const existingId = await findUserId(globalCtx, userLogin);
if (existingId !== undefined) return existingId;
await page.goto(`${dolibarrAddress}/user/card.php?action=create&mode=create`);
// Fill text inputs (login, lastname, password) without submitting yet.
const formFields: UserFormFields = {
login: userLogin,
password,
lastname,
};
await forms.fillForm(
{ page, imgFolderPath },
formFields,
newUserInputNames,
0,
);
// Admin checkbox — only present when the logged-in user is themselves admin.
// GUESS: input[name="admin"]; on some setups it's a <select> instead.
const adminCheckbox = page.locator('input[name="admin"]');
if (await adminCheckbox.count() > 0) {
if (admin) {
await adminCheckbox.check();
} else {
// Leave unchecked (default), but be explicit/defensive.
if (await adminCheckbox.isChecked()) await adminCheckbox.uncheck();
}
}
// Submit the create form. The create page has a single primary submit button
// ("Créer utilisateur"); :nth-match index 1 mirrors the fillForm convention.
await page.click(':nth-match(input[type="submit"], 1)');
// After creation Dolibarr lands on the user card. Parse the id.
await page.waitForLoadState("domcontentloaded");
const id = await readUserIdFromPage(page);
if (id === undefined) {
throw new Error(
`Could not determine the new user id for login "${userLogin}" after ` +
`submitting the create form (current URL: ${page.url()}).`,
);
}
return id;
}
/*
Read the numeric user id from the current user card: first from the URL
(?id=<n>), then from a hidden input[name="id"] as a fallback.
*/
async function readUserIdFromPage(page: Page): Promise<number | undefined> {
const fromUrl = new URL(page.url()).searchParams.get("id");
if (fromUrl && /^\d+$/.test(fromUrl)) return Number(fromUrl);
// GUESS: a hidden input carrying the id (common on Dolibarr card pages).
const hidden = page.locator('input[name="id"]');
if (await hidden.count() > 0) {
const val = await hidden.first().getAttribute("value");
if (val && /^\d+$/.test(val)) return Number(val);
}
return undefined;
}
/*
Find an existing user's numeric id by login via the user list, so createUser is
idempotent across re-runs. Matches the table row containing the login, then reads
the id from that row's user-card link.
*/
async function findUserId(
globalCtx: UserCtx,
userLogin: string,
): Promise<number | undefined> {
const { page, dolibarrAddress } = globalCtx;
await page.goto(
`${dolibarrAddress}/user/list.php?search_login=${encodeURIComponent(userLogin)}`,
);
await page.waitForLoadState("domcontentloaded");
const cardLink = page
.locator("tr", { hasText: userLogin })
.locator('a[href*="/user/card.php?id="]')
.first();
if (await cardLink.count() === 0) return undefined;
const href = (await cardLink.getAttribute("href")) ?? "";
const m = href.match(/[?&]id=(\d+)/);
return m ? Number(m[1]) : undefined;
}
/*
Grant a set of rights to a user on /user/perms.php?id=<userId>.
Dolibarr renders each grantable permission as a link whose href carries
action=addrights&rights=<id> (and a matching action=delrights to revoke).
A right that is already granted shows the delrights link instead, so we treat
the presence of an addrights link as "not yet granted" and click it; if it's
absent we assume the right is already on and skip it (defensive/idempotent).
*/
async function assignRights(
globalCtx: UserCtx,
userId: number,
rightIds: number[],
): Promise<void> {
const { page, dolibarrAddress } = globalCtx;
await login.doAdminLogin(globalCtx);
for (const rightId of rightIds) {
// Re-navigate per right: Dolibarr reloads the perms table after each grant,
// which would otherwise stale the locators.
await page.goto(`${dolibarrAddress}/user/perms.php?id=${userId}`);
// The grant control is an <a> linking to action=addrights&rights=<id>. The
// trailing "&" anchors the id so rights=12 does NOT substring-match rights=121
// / rights=1232 etc. (Dolibarr hrefs are ...&rights=<id>&confirm=yes...).
const addLink = page.locator(
`a[href*="action=addrights"][href*="rights=${rightId}&"]`,
);
if (await addLink.count() === 0) {
// Either already granted (only a delrights link is shown) or the row is
// not visible for this user. Skip rather than fail the whole run.
continue;
}
await addLink.first().click();
await page.waitForLoadState("domcontentloaded");
}
}
/*
Generate (or read) the user's API key on /user/card.php?id=<userId> in edit
mode and RETURN it as a string.
With the API module enabled the user card edit form shows an api_key text
input plus a "generate" control (a dice/refresh icon-link, typically
id="generate_api_key" or an <a> with action=...generate...apikey). We trigger
generation, then read the value back out of the api_key input. If a key is
already present we read and return it without regenerating (idempotent).
The caller is responsible for handling the returned secret safely — DO NOT log
it at debug verbosity.
*/
async function generateApiKey(
globalCtx: UserCtx,
userId: number,
): Promise<string> {
const { page, dolibarrAddress } = globalCtx;
await login.doAdminLogin(globalCtx);
await page.goto(`${dolibarrAddress}/user/card.php?id=${userId}&action=edit`);
// GUESS: the API key field is input[name="api_key"].
const apiKeyInput = page.locator('input[name="api_key"]');
if (await apiKeyInput.count() === 0) {
throw new Error(
`API key input (input[name="api_key"]) not found on the user card for ` +
`id=${userId}. Is the REST API module enabled? Confirm the field name ` +
`on the running sandbox.`,
);
}
// If a key is already set, reuse it.
const existing = (await apiKeyInput.first().inputValue()).trim();
if (existing.length > 0) return existing;
// Trigger generation. GUESS: a generate control next to the field. We try a
// few likely selectors in order and click the first that exists.
const generateSelectors = [
"#generate_api_key",
'a[href*="action=generateapikey"]',
'a[href*="generate"][href*="apikey"]',
// Icon-only fallbacks commonly used by Dolibarr "generate" buttons.
"span.fa-dice",
"a.fa-refresh",
];
let clicked = false;
for (const sel of generateSelectors) {
const ctrl = page.locator(sel);
if (await ctrl.count() > 0) {
await ctrl.first().click();
clicked = true;
break;
}
}
if (!clicked) {
throw new Error(
`No API-key generate control found near input[name="api_key"] for ` +
`id=${userId} (tried ${generateSelectors.join(", ")}). Confirm the ` +
`generate control on the running sandbox.`,
);
}
// The generate control fills the input client-side; poll the input value
// (via the locator, no browser globals) until it is non-empty or we time out.
let key = "";
const deadline = Date.now() + 10_000;
while (Date.now() < deadline) {
key = (await apiKeyInput.first().inputValue()).trim();
if (key.length > 0) break;
await page.waitForTimeout(200);
}
if (key.length === 0) {
throw new Error(
`API key field was still empty after triggering generation for ` +
`id=${userId}.`,
);
}
// Persist: the generate control only fills the field client-side. Submit the
// edit form so Dolibarr saves api_key to the DB — otherwise it stays NULL and
// the generated key never authenticates.
const saveBtn = page.locator('input[name="save"], button[name="save"]');
if (await saveBtn.count() > 0) {
await saveBtn.first().click();
} else {
await page.click(':nth-match(input[type="submit"], 1)');
}
await page.waitForLoadState("domcontentloaded");
return key;
}
export default {
createUser,
assignRights,
generateApiKey,
};