Compare commits

...

28 Commits

Author SHA1 Message Date
Benjamin Admin
65177d3ff7 fix: robust PaddleOCR init with multiple fallback strategies
Some checks failed
CI / go-lint (pull_request) Failing after 2s
CI / python-lint (pull_request) Failing after 11s
CI / nodejs-lint (pull_request) Failing after 2s
CI / test-go-consent (pull_request) Failing after 2s
CI / test-python-voice (pull_request) Failing after 14s
CI / test-bqas (pull_request) Failing after 11s
CI / deploy-hetzner (pull_request) Has been skipped
Deploy to Coolify / deploy (push) Has been cancelled
PaddleOCR 3.x removed show_log param and lang='latin'. Try multiple
init strategies in order until one succeeds.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 11:09:33 +01:00
Benjamin Admin
559d6a351c fix: resolve stash conflict
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
CI / go-lint (pull_request) Failing after 2s
CI / python-lint (pull_request) Failing after 14s
CI / nodejs-lint (pull_request) Failing after 3s
CI / test-go-consent (pull_request) Failing after 3s
CI / test-python-voice (pull_request) Failing after 11s
CI / test-bqas (pull_request) Failing after 10s
CI / deploy-hetzner (pull_request) Has been skipped
2026-03-13 10:59:30 +01:00
Benjamin Admin
8fd11998e4 merge: resolve docker-compose.coolify.yml conflict (accept remote) 2026-03-13 10:56:36 +01:00
Benjamin Admin
4ce649aa71 fix: upgrade PaddleOCR to 3.x for PP-OCRv5 and stability
Old paddlepaddle==2.6.2 + paddleocr==2.8.1 caused hangs on first OCR
request. Upgrading to paddlepaddle>=3.0.0 + paddleocr>=2.9.0 enables
native PP-OCRv5 support and fixes stability issues.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:53:18 +01:00
Benjamin Admin
5ee3cc0104 fix: load PaddleOCR model in background thread
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
The import and model loading can take minutes and was blocking
the startup event, causing health checks to timeout. Now loads
in a background thread — health endpoint returns 200 immediately
with status 'loading' until model is ready.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:21:59 +01:00
Benjamin Admin
b36712247b fix: add detailed logging for PaddleOCR model loading debug
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:19:10 +01:00
Benjamin Admin
86b11c7e5f fix: catch all exceptions in PaddleOCR version fallback
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
PaddleOCR 2.8.1 throws a generic Exception (not ValueError) when
ocr_version='PP-OCRv5' is used. Broadened except clause to catch
any error and fall back to lang='latin' for older versions.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:12:32 +01:00
Benjamin Admin
8003dcac39 fix: PaddleOCR 3.4.0 compatibility — use lang=en with PP-OCRv5
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
PaddleOCR 3.4.0 removed 'latin' language support, causing ValueError
at startup. Now uses lang='en' with ocr_version='PP-OCRv5' and falls
back to lang='latin' for older PaddleOCR versions.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:54:52 +01:00
Benjamin Admin
778c44226e fix: expose port 8095 directly (bypass Traefik 60s timeout)
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 14:16:04 +01:00
Benjamin Admin
79891063dd fix: pin PaddlePaddle 2.6.2 + PaddleOCR 2.8.1 (stable, no PIR bug)
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
PaddlePaddle 3.x hat oneDNN/PIR Executor Bug. Zurueck auf 2.6.2
mit bewaeherter ocr() API statt predict().

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 13:32:54 +01:00
Benjamin Admin
2c9b0dc448 fix: disable oneDNN (FLAGS_use_mkldnn=0) for PaddlePaddle compat
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 13:25:36 +01:00
Benjamin Admin
3133615044 fix: add libgomp1 (OpenMP) + remove unused lang parameter
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
PaddlePaddle braucht libgomp.so.1 fuer Inferenz.
lang wird ignoriert bei explizitem model_name.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 13:19:47 +01:00
Benjamin Admin
2bc0f87325 fix: PaddleOCR model pre-load at startup + 5min healthcheck grace
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Model wird beim Container-Start geladen (nicht erst beim ersten Request).
Health-Check start_period auf 300s erhoeht fuer initialen Download.
/health gibt "loading" zurueck bis Modell bereit ist.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 13:12:14 +01:00
Benjamin Admin
4ee38d6f0b fix: remove show_log (unknown in PaddleOCR v3 API)
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 12:52:52 +01:00
Benjamin Admin
992d4f2a6b fix: PaddleOCR v3 API — explicit model name + predict() statt ocr()
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
lang="latin" braucht text_recognition_model_name in PP-OCRv5.
Neue API nutzt predict() statt ocr(), Ergebnis-Format angepasst.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 12:47:07 +01:00
Benjamin Admin
8f5f9641c7 fix: libgl1-mesa-glx → libgl1 (Debian bookworm)
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 10:33:28 +01:00
Benjamin Admin
7cdb53051f feat: PaddleOCR Service (PP-OCRv5 Latin auf x86_64)
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Microservice fuer PaddleOCR auf Hetzner. FastAPI mit /ocr und /health
Endpoints, API-Key Auth, 4GB Memory Limit, Modell-Cache Volume.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 10:20:41 +01:00
Sharang Parnerkar
d834753a98 Remove services not needed by SDK from Coolify deployment
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Remove backend-core, billing-service, night-scheduler, and admin-core
as they are not used by any compliance/SDK service. Update
health-aggregator CHECK_SERVICES to reference consent-service instead.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-10 13:29:23 +01:00
Sharang Parnerkar
395011d0f4 Add healthchecks to backend-core, consent-service, billing-service, admin-core
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Coolify/Traefik requires healthchecks to route traffic to containers.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-08 00:57:43 +01:00
Sharang Parnerkar
9e1660f954 Remove Traefik labels from coolify compose — Coolify handles routing
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-08 00:05:26 +01:00
Sharang Parnerkar
13ff930b5e Fix admin-core build: ensure public directory exists before build
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-07 23:44:30 +01:00
Sharang Parnerkar
5d1c837f49 Fix backend-core TARGETARCH: auto-detect instead of hardcoded arm64
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
The Dockerfile hardcoded TARGETARCH=arm64 for Mac Mini. Coolify server
is x86_64, causing exit code 126 (wrong binary arch). Now uses Docker
BuildKit's auto-detected TARGETARCH with dpkg fallback.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-07 23:37:59 +01:00
Sharang Parnerkar
1dd9662037 Add QDRANT_API_KEY support to rag-service
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
- Add QDRANT_API_KEY to config.py (empty string = no auth)
- Pass api_key to QdrantClient constructor (None when empty)
- Add QDRANT_API_KEY to coolify compose and env example

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-07 23:35:11 +01:00
Sharang Parnerkar
4626edb232 Sync coolify compose with main: remove voice-service, update rag/embedding
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
- Remove voice-service (removed in main branch)
- Remove voice_session_data volume
- Add OLLAMA_URL and OLLAMA_EMBED_MODEL to rag-service
- Update embedding-service default model to BAAI/bge-m3, memory 4G→8G
- Update health-aggregator CHECK_SERVICES (remove voice-service)
- Update .env.coolify.example accordingly

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-07 23:23:52 +01:00
Sharang Parnerkar
3c29b621ac Merge remote-tracking branch 'origin/main' into coolify 2026-03-07 23:10:41 +01:00
Sharang Parnerkar
755570d474 fix: use Alpine-compatible addgroup/adduser flags in Dockerfiles
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Replace --system/--gid/--uid (Debian syntax) with -S/-g/-u (BusyBox/Alpine).
Coolify ARG injection causes exit code 255 with Debian-style flags.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-06 22:38:31 +01:00
Sharang Parnerkar
e890b1490a refactor(coolify): externalize postgres, qdrant, S3; remove jitsi/synapse
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
- Remove PostgreSQL, Qdrant, MinIO services (managed separately in Coolify)
- Remove Jitsi stack (web, xmpp, jicofo, jvb) and Synapse/synapse-db
- Add POSTGRES_HOST, QDRANT_URL, S3_ENDPOINT/S3_ACCESS_KEY/S3_SECRET_KEY env vars
- Remove Traefik labels from internal-only services
- Health aggregator no longer checks external services
- Core now has 10 services: valkey + 9 application services

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-03 09:23:20 +01:00
Sharang Parnerkar
d15de16c47 feat: add Coolify deployment configuration
Some checks failed
Deploy to Coolify / deploy (push) Has been cancelled
Add docker-compose.coolify.yml (17 services), .env.coolify.example,
and Gitea Action workflow for Coolify API deployment. Removes nginx,
vault, gitea, woodpecker, mailpit, and dev-only services. Adds Traefik
labels for *.breakpilot.ai domain routing with Let's Encrypt SSL.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-25 10:43:02 +01:00
4 changed files with 169 additions and 1 deletions

View File

@@ -15,6 +15,7 @@ networks:
volumes:
valkey_data:
embedding_models:
paddleocr_models:
services:
@@ -141,6 +142,35 @@ services:
networks:
- breakpilot-network
# =========================================================
# OCR SERVICE (PaddleOCR PP-OCRv5 Latin)
# =========================================================
paddleocr-service:
build:
context: ./paddleocr-service
dockerfile: Dockerfile
container_name: bp-core-paddleocr
expose:
- "8095"
environment:
PADDLEOCR_API_KEY: ${PADDLEOCR_API_KEY:-}
FLAGS_use_mkldnn: "0"
volumes:
- paddleocr_models:/root/.paddleocr
deploy:
resources:
limits:
memory: 4G
healthcheck:
test: ["CMD", "curl", "-f", "http://127.0.0.1:8095/health"]
interval: 30s
timeout: 10s
start_period: 300s
retries: 5
restart: unless-stopped
networks:
- breakpilot-network
# =========================================================
# HEALTH AGGREGATOR
# =========================================================
@@ -153,7 +183,7 @@ services:
- "8099"
environment:
PORT: 8099
CHECK_SERVICES: "valkey:6379,consent-service:8081,rag-service:8097,embedding-service:8087"
CHECK_SERVICES: "valkey:6379,consent-service:8081,rag-service:8097,embedding-service:8087,paddleocr-service:8095"
healthcheck:
test: ["CMD", "curl", "-f", "http://127.0.0.1:8099/health"]
interval: 30s

View File

@@ -0,0 +1,16 @@
FROM python:3.11-slim
WORKDIR /app
RUN apt-get update && apt-get install -y --no-install-recommends \
libgl1 libglib2.0-0 libgomp1 curl \
&& rm -rf /var/lib/apt/lists/*
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
EXPOSE 8095
HEALTHCHECK --interval=30s --timeout=10s --start-period=120s --retries=3 \
CMD curl -f http://127.0.0.1:8095/health || exit 1
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8095"]

115
paddleocr-service/main.py Normal file
View File

@@ -0,0 +1,115 @@
"""PaddleOCR Remote Service — PP-OCRv5 Latin auf x86_64."""
import io
import logging
import os
import threading
import numpy as np
from fastapi import FastAPI, File, Header, HTTPException, UploadFile
from PIL import Image
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
app = FastAPI(title="PaddleOCR Service")
_engine = None
_ready = False
_loading = False
API_KEY = os.environ.get("PADDLEOCR_API_KEY", "")
def _load_model():
"""Load PaddleOCR model in background thread."""
global _engine, _ready
try:
logger.info("Importing paddleocr...")
from paddleocr import PaddleOCR
logger.info("Import done. Loading PaddleOCR model...")
# Try multiple init strategies for different PaddleOCR versions
inits = [
# PaddleOCR 3.x (no show_log)
dict(lang="en", ocr_version="PP-OCRv5", use_angle_cls=True),
# PaddleOCR 3.x with show_log
dict(lang="en", ocr_version="PP-OCRv5", use_angle_cls=True, show_log=False),
# PaddleOCR 2.8+ (latin)
dict(lang="latin", use_angle_cls=True, show_log=False),
# PaddleOCR 2.8+ (en, no version)
dict(lang="en", use_angle_cls=True, show_log=False),
]
for i, kwargs in enumerate(inits):
try:
_engine = PaddleOCR(**kwargs)
logger.info(f"PaddleOCR init succeeded with strategy {i}: {kwargs}")
break
except Exception as e:
logger.info(f"PaddleOCR init strategy {i} failed: {e}")
else:
raise RuntimeError("All PaddleOCR init strategies failed")
_ready = True
logger.info("PaddleOCR model loaded successfully — ready to serve")
except Exception as e:
logger.error(f"Failed to load PaddleOCR model: {e}")
@app.on_event("startup")
def startup_load_model():
"""Start model loading in background so health check passes immediately."""
global _loading
_loading = True
thread = threading.Thread(target=_load_model, daemon=True)
thread.start()
logger.info("Model loading started in background thread")
@app.get("/health")
def health():
if _ready:
return {"status": "ok", "model": "PP-OCRv5-latin"}
if _loading:
return {"status": "loading"}
return {"status": "error"}
@app.post("/ocr")
async def ocr(
file: UploadFile = File(...),
x_api_key: str = Header(default=""),
):
if API_KEY and x_api_key != API_KEY:
raise HTTPException(status_code=401, detail="Invalid API key")
if not _ready:
raise HTTPException(status_code=503, detail="Model still loading")
img_bytes = await file.read()
img = Image.open(io.BytesIO(img_bytes)).convert("RGB")
img_np = np.array(img)
result = _engine.ocr(img_np)
words = []
for line in result[0] or []:
box, (text, conf) = line[0], line[1]
x_min = min(p[0] for p in box)
y_min = min(p[1] for p in box)
x_max = max(p[0] for p in box)
y_max = max(p[1] for p in box)
words.append(
{
"text": text.strip(),
"left": int(x_min),
"top": int(y_min),
"width": int(x_max - x_min),
"height": int(y_max - y_min),
"conf": round(conf * 100, 1),
}
)
return {
"words": words,
"image_width": img_np.shape[1],
"image_height": img_np.shape[0],
}

View File

@@ -0,0 +1,7 @@
paddlepaddle>=3.0.0
paddleocr>=2.9.0
fastapi>=0.110.0
uvicorn>=0.25.0
python-multipart>=0.0.6
Pillow>=10.0.0
numpy>=1.24.0