Files
gomag-vending/CLAUDE.md
Claude Agent 419464a62c feat: add CI/CD testing infrastructure with test.sh orchestrator
Complete testing system: pyproject.toml (pytest markers), test.sh
orchestrator with auto app start/stop and colorful summary,
pre-push hook, Gitea Actions workflow.

New QA tests: API health (7 endpoints), responsive (3 viewports),
log monitoring (ERROR/ORA-/Traceback detection), real GoMag sync,
PL/SQL package validation, smoke prod (read-only).

Converted test_app_basic.py and test_integration.py to pytest.
Added pytestmark to all existing tests (unit/e2e/oracle).
E2E conftest upgraded: console error collector, screenshot on
failure, auto-detect live app on :5003.

Usage: ./test.sh ci (30s) | ./test.sh full (2-3min)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-24 10:40:25 +00:00

3.8 KiB

CLAUDE.md

Project Overview

System: Import Comenzi Web GoMag → Sistem ROA Oracle Stack: FastAPI + Jinja2 + Bootstrap 5.3 + Oracle PL/SQL + SQLite

Documentatie completa: README.md

Implementare cu TeamCreate

OBLIGATORIU: Folosim TeamCreate + TaskCreate, NU Agent tool cu subagenti paraleli. Skill-ul superpowers:dispatching-parallel-agents NU se aplica in acest proiect.

  • Team lead citeste TOATE fisierele implicate, creeaza planul
  • ASTEAPTA aprobare explicita de la user inainte de implementare
  • Task-uri pe fisiere non-overlapping (evita conflicte)
  • Cache-bust static assets (?v=N) la fiecare schimbare UI

Development Commands

# INTOTDEAUNA via start.sh (seteaza Oracle env vars)
./start.sh
# NU folosi uvicorn direct — lipsesc LD_LIBRARY_PATH si TNS_ADMIN

Testing & CI/CD

# Teste rapide (unit + e2e, ~30s, fara Oracle)
./test.sh ci

# Teste complete (totul inclusiv Oracle + sync real + PL/SQL, ~2-3 min)
./test.sh full

# Smoke test pe productie (read-only, dupa deploy)
./test.sh smoke-prod --base-url http://79.119.86.134/gomag

# Doar un layer specific
./test.sh unit          # SQLite CRUD, imports, routes
./test.sh e2e           # Browser tests (Playwright)
./test.sh oracle        # Oracle integration
./test.sh sync          # Sync real GoMag → Oracle
./test.sh qa            # API health + responsive + log monitor
./test.sh logs          # Doar log monitoring

# Validate prerequisites
./test.sh --dry-run

Flow zilnic:

  1. Lucrezi pe branch fix/* sau feat/*
  2. git push → pre-push hook ruleaza ./test.sh ci automat (~30s)
  3. Inainte de PR → ./test.sh full manual (~2-3 min)
  4. Dupa deploy pe prod → ./test.sh smoke-prod --base-url http://79.119.86.134/gomag

Output: qa-reports/ — health score, raport markdown, screenshots, baseline comparison.

Markers pytest: unit, oracle, e2e, qa, sync

Reguli critice (nu le incalca)

Flux import comenzi

  1. Download GoMag API → JSON → parse → validate SKU-uri → import Oracle
  2. Ordinea: parteneri (cauta/creeaza) → adresecomandafactura cache
  3. SKU lookup: ARTICOLE_TERTI (mapped) are prioritate fata de NOM_ARTICOLE (direct)
  4. Complex sets (kituri/pachete): un SKU → multiple CODMAT-uri cu cantitate_roa; preturile se preiau din lista de preturi Oracle
  5. Comenzi anulate (GoMag statusId=7): verifica daca au factura inainte de stergere din Oracle

Statusuri comenzi

IMPORTED / ALREADY_IMPORTED / SKIPPED / ERROR / CANCELLED / DELETED_IN_ROA

  • Upsert: IMPORTED existent NU se suprascrie cu ALREADY_IMPORTED
  • Recovery: la fiecare sync, comenzile ERROR sunt reverificate in Oracle

Parteneri

  • Prioritate: companie (PJ, cod_fiscal + registru) daca exista in GoMag, altfel persoana fizica cu shipping name
  • Adresa livrare: intotdeauna GoMag shipping
  • Adresa facturare: daca shipping ≠ billing person → shipping pt ambele; altfel → billing din GoMag

Preturi

  • Dual policy: articolele sunt rutate la id_pol_vanzare sau id_pol_productie pe baza contului contabil (341/345 = productie)
  • Daca pretul lipseste, se insereaza automat pret=0

Dashboard paginare

  • Contorul din paginare arata totalul comenzilor din perioada selectata (ex: "378 comenzi"), NU doar cele filtrate
  • Butoanele de filtru (Importat, Omise, Erori, Facturate, Nefacturate, Anulate) arata fiecare cate comenzi are pe langa total
  • Aceasta este comportamentul dorit: userul vede cate comenzi totale sunt, din care cate importate, cu erori etc.

Invoice cache

  • Coloanele factura_* pe orders (SQLite), populate lazy din Oracle (vanzari WHERE sters=0)
  • Refresh complet: verifica facturi noi + facturi sterse + comenzi sterse din ROA

Deploy Windows

Vezi README.md