2 Commits

Include renames

Author	SHA1	Message	Date
OLmatter	905b6e6651	v23.3: upgrade OCR to PP-OCRv6_tiny + fix click-no-captcha 15s wait ... Backend: - Default OCR model PP-OCRv5_server_rec -> PP-OCRv6_tiny_rec (~14x faster, 83ms/img vs 1189ms on 379 real captchas, accuracy still 100%) - Configurable via config.json ocr_model or env CNCAPTCHA_CPU_OCR_MODEL/GLM_OCR_MODEL - GUI OCR model choices updated to v6 (tiny/medium) + v5 server fallback - worker.py: split 3 crops into independent OCR tasks for parallel recognition - ppocr_worker.py: batch forward (predict_batch_*) + single-crop path - server.py: partial results aggregation + OCR_MODEL env passthrough - requirements: paddleocr>=3.7.0 / paddlex>=3.7.0 (v6 requires) Userscript v23.3: - Fix: auto-click subscribe then wait 15s when click didn't trigger captcha. Synthetic click sometimes gets swallowed by Zhipu frontend (button DOM ready but component state machine not ready), no captcha pops up, main loop stuck in WAITING until MODAL_WAIT=15000 timeout. - Fix: use 'iframe got new prompt+bg image' as the signal. iframe increments GM counter glm_captcha_seen_seq each time it sees a new captcha (prompt or bg changed); main loop records baseline at click time, compares in WAITING - counter increased = captcha popped, wait patiently; no increase after 1.5s = click didn't trigger captcha, immediately retry subscribe. Counter is monotonic, no residue, no timestamp race.	2026-06-23 00:07:51 +08:00
sunshineandsunshine	6b052da423	PR A: Pipeline Backend, YOLO to OCR multi-process, CPU auto-allocation (#10) ... * feat(pipeline): multi-process YOLO→OCR pipeline backend - server.py: FastAPI gateway, 3-level mp.Queue pipeline, round-robin dispatch - worker.py: YOLO character detection worker (core-pinned, single-threaded) - ppocr_worker.py: PP-OCRv5 recognition worker with pre-warm cache - evaluate.py: select_fixed3 box selection from scripts/tools/ - Smart CPU core allocation: N_YOLO = cores/4, N_OCR = cores/2 - config.json auto-created on first run (_auto + _cores metadata) - /health endpoint: {status:'starting'|'ok', workers:N, ready_workers:N} - Worker watchdog: auto-restart crashed OCR workers - Shutdown cleanup: _shutdown event + try/finally for orphaned processes - one_click_start.ps1: pipeline dep check (fastapi/uvicorn/psutil) non-blocking - setup_backend.py: smoke test includes pipeline deps - requirements-backend-cpu/gpu.txt: +fastapi, uvicorn[standard], psutil - README: pipeline architecture + auto CPU allocation - .gitignore: +config.json, server logs Minimal verification: python -m py_compile backend/server.py backend/worker.py backend/ppocr_worker.py backend/evaluate.py # OK python backend/server.py → GET /health → {'status':'ok','workers':12,'ready_workers':12} * fix: review feedback - timeout cleanup, watchdog, health, paddleocr - handle_direct/handle_direct_url: clean pending_requests on TimeoutError - _worker_watchdog: always check p.is_alive(), remove ready_count bypass - /health: add alive_workers field (count is_alive) - start_backend.ps1: include paddleocr in import check --------- Co-authored-by: sunshine <qt22260@gmail.com>	2026-06-15 20:57:57 +08:00