[split-required] Split 500-850 LOC files (batch 2)
backend-lehrer (10 files): - game/database.py (785 → 5), correction_api.py (683 → 4) - classroom_engine/antizipation.py (676 → 5) - llm_gateway schools/edu_search already done in prior batch klausur-service (12 files): - orientation_crop_api.py (694 → 5), pdf_export.py (677 → 4) - zeugnis_crawler.py (676 → 5), grid_editor_api.py (671 → 5) - eh_templates.py (658 → 5), mail/api.py (651 → 5) - qdrant_service.py (638 → 5), training_api.py (625 → 4) website (6 pages): - middleware (696 → 8), mail (733 → 6), consent (628 → 8) - compliance/risks (622 → 5), export (502 → 5), brandbook (629 → 7) studio-v2 (3 components): - B2BMigrationWizard (848 → 3), CleanupPanel (765 → 2) - dashboard-experimental (739 → 2) admin-lehrer (4 files): - uebersetzungen (769 → 4), manager (670 → 2) - ChunkBrowserQA (675 → 6), dsfa/page (674 → 5) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Benjamin Admin
337
klausur-service/backend/grid_editor_api_grid.py
Normal file
337
klausur-service/backend/grid_editor_api_grid.py
Normal file
@@ -0,0 +1,337 @@
|
||||
"""
|
||||
Grid Editor API — grid build, save, and retrieve endpoints.
|
||||
"""
|
||||
|
||||
import logging
|
||||
import time
|
||||
from typing import Any, Dict
|
||||
|
||||
from fastapi import APIRouter, HTTPException, Query, Request
|
||||
|
||||
from grid_build_core import _build_grid_core
|
||||
from ocr_pipeline_session_store import (
|
||||
get_session_db,
|
||||
update_session_db,
|
||||
)
|
||||
from ocr_pipeline_common import (
|
||||
_cache,
|
||||
_load_session_to_cache,
|
||||
_get_cached,
|
||||
)
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(prefix="/api/v1/ocr-pipeline", tags=["grid-editor"])
|
||||
|
||||
|
||||
@router.post("/sessions/{session_id}/build-grid")
|
||||
async def build_grid(
|
||||
session_id: str,
|
||||
ipa_mode: str = Query("auto", pattern="^(auto|all|de|en|none)$"),
|
||||
syllable_mode: str = Query("auto", pattern="^(auto|all|de|en|none)$"),
|
||||
enhance: bool = Query(True, description="Step 3: CLAHE + denoise for degraded scans"),
|
||||
max_cols: int = Query(0, description="Step 2: Max column count (0=unlimited)"),
|
||||
min_conf: int = Query(0, description="Step 1: Min OCR confidence (0=auto)"),
|
||||
):
|
||||
"""Build a structured, zone-aware grid from existing Kombi word results.
|
||||
|
||||
Requires that paddle-kombi or rapid-kombi has already been run on the session.
|
||||
Uses the image for box detection and the word positions for grid structuring.
|
||||
|
||||
Query params:
|
||||
ipa_mode: "auto" (only when English IPA detected), "all" (force), "none" (skip)
|
||||
syllable_mode: "auto" (only when original has dividers), "all" (force), "none" (skip)
|
||||
|
||||
Returns a StructuredGrid with zones, each containing their own
|
||||
columns, rows, and cells — ready for the frontend Excel-like editor.
|
||||
"""
|
||||
session = await get_session_db(session_id)
|
||||
if not session:
|
||||
raise HTTPException(status_code=404, detail=f"Session {session_id} not found")
|
||||
|
||||
try:
|
||||
result = await _build_grid_core(
|
||||
session_id, session,
|
||||
ipa_mode=ipa_mode, syllable_mode=syllable_mode,
|
||||
enhance=enhance,
|
||||
max_columns=max_cols if max_cols > 0 else None,
|
||||
min_conf=min_conf if min_conf > 0 else None,
|
||||
)
|
||||
except ValueError as e:
|
||||
raise HTTPException(status_code=400, detail=str(e))
|
||||
|
||||
# Save automatic grid snapshot for later comparison with manual corrections
|
||||
# Lazy import to avoid circular dependency with ocr_pipeline_regression
|
||||
from ocr_pipeline_regression import _build_reference_snapshot
|
||||
|
||||
wr = session.get("word_result") or {}
|
||||
engine = wr.get("ocr_engine", "")
|
||||
if engine in ("kombi", "rapid_kombi"):
|
||||
auto_pipeline = "kombi"
|
||||
elif engine == "paddle_direct":
|
||||
auto_pipeline = "paddle-direct"
|
||||
else:
|
||||
auto_pipeline = "pipeline"
|
||||
auto_snapshot = _build_reference_snapshot(result, pipeline=auto_pipeline)
|
||||
|
||||
gt = session.get("ground_truth") or {}
|
||||
gt["auto_grid_snapshot"] = auto_snapshot
|
||||
|
||||
# Persist to DB and advance current_step to 11 (reconstruction complete)
|
||||
await update_session_db(session_id, grid_editor_result=result, ground_truth=gt, current_step=11)
|
||||
|
||||
logger.info(
|
||||
"build-grid session %s: %d zones, %d cols, %d rows, %d cells, "
|
||||
"%d boxes in %.2fs",
|
||||
session_id,
|
||||
len(result.get("zones", [])),
|
||||
result.get("summary", {}).get("total_columns", 0),
|
||||
result.get("summary", {}).get("total_rows", 0),
|
||||
result.get("summary", {}).get("total_cells", 0),
|
||||
result.get("boxes_detected", 0),
|
||||
result.get("duration_seconds", 0),
|
||||
)
|
||||
|
||||
return result
|
||||
|
||||
|
||||
@router.post("/sessions/{session_id}/rerun-ocr-and-build-grid")
|
||||
async def rerun_ocr_and_build_grid(
|
||||
session_id: str,
|
||||
ipa_mode: str = Query("auto", pattern="^(auto|all|de|en|none)$"),
|
||||
syllable_mode: str = Query("auto", pattern="^(auto|all|de|en|none)$"),
|
||||
enhance: bool = Query(True, description="Step 3: CLAHE + denoise for degraded scans"),
|
||||
max_cols: int = Query(0, description="Step 2: Max column count (0=unlimited)"),
|
||||
min_conf: int = Query(0, description="Step 1: Min OCR confidence (0=auto)"),
|
||||
vision_fusion: bool = Query(False, description="Step 4: Vision-LLM fusion for degraded scans"),
|
||||
doc_category: str = Query("", description="Document type for Vision-LLM prompt context"),
|
||||
):
|
||||
"""Re-run OCR with quality settings, then rebuild the grid.
|
||||
|
||||
Unlike build-grid (which only rebuilds from existing words),
|
||||
this endpoint re-runs the full OCR pipeline on the cropped image
|
||||
with optional CLAHE enhancement, then builds the grid.
|
||||
|
||||
Steps executed: Image Enhancement -> OCR -> Grid Build
|
||||
"""
|
||||
session = await get_session_db(session_id)
|
||||
if not session:
|
||||
raise HTTPException(status_code=404, detail=f"Session {session_id} not found")
|
||||
|
||||
import time as _time
|
||||
t0 = _time.time()
|
||||
|
||||
# 1. Load the cropped/dewarped image from cache or session
|
||||
if session_id not in _cache:
|
||||
await _load_session_to_cache(session_id)
|
||||
cached = _get_cached(session_id)
|
||||
|
||||
dewarped_bgr = cached.get("cropped_bgr") if cached.get("cropped_bgr") is not None else cached.get("dewarped_bgr")
|
||||
if dewarped_bgr is None:
|
||||
raise HTTPException(status_code=400, detail="No cropped/dewarped image available. Run preprocessing steps first.")
|
||||
|
||||
import numpy as np
|
||||
img_h, img_w = dewarped_bgr.shape[:2]
|
||||
ocr_input = dewarped_bgr.copy()
|
||||
|
||||
# 2. Scan quality assessment
|
||||
scan_quality_info = {}
|
||||
try:
|
||||
from scan_quality import score_scan_quality
|
||||
quality_report = score_scan_quality(ocr_input)
|
||||
scan_quality_info = quality_report.to_dict()
|
||||
actual_min_conf = min_conf if min_conf > 0 else quality_report.recommended_min_conf
|
||||
except Exception as e:
|
||||
logger.warning(f"rerun-ocr: scan quality failed: {e}")
|
||||
actual_min_conf = min_conf if min_conf > 0 else 40
|
||||
|
||||
# 3. Image enhancement (Step 3)
|
||||
is_degraded = scan_quality_info.get("is_degraded", False)
|
||||
if enhance and is_degraded:
|
||||
try:
|
||||
from ocr_image_enhance import enhance_for_ocr
|
||||
ocr_input = enhance_for_ocr(ocr_input, is_degraded=True)
|
||||
logger.info("rerun-ocr: CLAHE enhancement applied")
|
||||
except Exception as e:
|
||||
logger.warning(f"rerun-ocr: enhancement failed: {e}")
|
||||
|
||||
# 4. Run dual-engine OCR
|
||||
from PIL import Image
|
||||
import pytesseract
|
||||
|
||||
# RapidOCR
|
||||
rapid_words = []
|
||||
try:
|
||||
from cv_ocr_engines import ocr_region_rapid
|
||||
from cv_vocab_types import PageRegion
|
||||
full_region = PageRegion(type="full_page", x=0, y=0, width=img_w, height=img_h)
|
||||
rapid_words = ocr_region_rapid(ocr_input, full_region) or []
|
||||
except Exception as e:
|
||||
logger.warning(f"rerun-ocr: RapidOCR failed: {e}")
|
||||
|
||||
# Tesseract
|
||||
pil_img = Image.fromarray(ocr_input[:, :, ::-1])
|
||||
data = pytesseract.image_to_data(pil_img, lang='eng+deu', config='--psm 6 --oem 3', output_type=pytesseract.Output.DICT)
|
||||
tess_words = []
|
||||
for i in range(len(data["text"])):
|
||||
text = (data["text"][i] or "").strip()
|
||||
conf_raw = str(data["conf"][i])
|
||||
conf = int(conf_raw) if conf_raw.lstrip("-").isdigit() else -1
|
||||
if not text or conf < actual_min_conf:
|
||||
continue
|
||||
tess_words.append({
|
||||
"text": text, "left": data["left"][i], "top": data["top"][i],
|
||||
"width": data["width"][i], "height": data["height"][i], "conf": conf,
|
||||
})
|
||||
|
||||
# 5. Merge OCR results
|
||||
from ocr_pipeline_ocr_merge import _split_paddle_multi_words, _merge_paddle_tesseract, _deduplicate_words
|
||||
rapid_split = _split_paddle_multi_words(rapid_words) if rapid_words else []
|
||||
if rapid_split or tess_words:
|
||||
merged_words = _merge_paddle_tesseract(rapid_split, tess_words)
|
||||
merged_words = _deduplicate_words(merged_words)
|
||||
else:
|
||||
merged_words = tess_words
|
||||
|
||||
# 6. Store updated word_result in session
|
||||
cells_for_storage = [{"text": w["text"], "left": w["left"], "top": w["top"],
|
||||
"width": w["width"], "height": w["height"], "conf": w.get("conf", 0)}
|
||||
for w in merged_words]
|
||||
word_result = {
|
||||
"cells": [{"text": " ".join(w["text"] for w in merged_words),
|
||||
"word_boxes": cells_for_storage}],
|
||||
"image_width": img_w,
|
||||
"image_height": img_h,
|
||||
"ocr_engine": "rapid_kombi",
|
||||
"word_count": len(merged_words),
|
||||
"raw_paddle_words": rapid_words,
|
||||
}
|
||||
# 6b. Vision-LLM Fusion (Step 4) — correct OCR using Vision model
|
||||
vision_applied = False
|
||||
if vision_fusion:
|
||||
try:
|
||||
from vision_ocr_fusion import vision_fuse_ocr
|
||||
category = doc_category or session.get("document_category") or "vokabelseite"
|
||||
logger.info(f"rerun-ocr: running Vision-LLM fusion (category={category})")
|
||||
merged_words = await vision_fuse_ocr(ocr_input, merged_words, category)
|
||||
vision_applied = True
|
||||
# Rebuild storage from fused words
|
||||
cells_for_storage = [{"text": w["text"], "left": w["left"], "top": w["top"],
|
||||
"width": w["width"], "height": w["height"], "conf": w.get("conf", 0)}
|
||||
for w in merged_words]
|
||||
word_result["cells"] = [{"text": " ".join(w["text"] for w in merged_words),
|
||||
"word_boxes": cells_for_storage}]
|
||||
word_result["word_count"] = len(merged_words)
|
||||
word_result["ocr_engine"] = "vision_fusion"
|
||||
except Exception as e:
|
||||
logger.warning(f"rerun-ocr: Vision-LLM fusion failed: {e}")
|
||||
|
||||
await update_session_db(session_id, word_result=word_result)
|
||||
|
||||
# Reload session with updated word_result
|
||||
session = await get_session_db(session_id)
|
||||
|
||||
ocr_duration = _time.time() - t0
|
||||
logger.info(
|
||||
"rerun-ocr session %s: %d words (rapid=%d, tess=%d, merged=%d) in %.1fs "
|
||||
"(enhance=%s, min_conf=%d, quality=%s)",
|
||||
session_id, len(merged_words), len(rapid_words), len(tess_words),
|
||||
len(merged_words), ocr_duration, enhance, actual_min_conf,
|
||||
scan_quality_info.get("quality_pct", "?"),
|
||||
)
|
||||
|
||||
# 7. Build grid from new words
|
||||
try:
|
||||
result = await _build_grid_core(
|
||||
session_id, session,
|
||||
ipa_mode=ipa_mode, syllable_mode=syllable_mode,
|
||||
enhance=enhance,
|
||||
max_columns=max_cols if max_cols > 0 else None,
|
||||
min_conf=min_conf if min_conf > 0 else None,
|
||||
)
|
||||
except ValueError as e:
|
||||
raise HTTPException(status_code=400, detail=str(e))
|
||||
|
||||
# Persist grid
|
||||
await update_session_db(session_id, grid_editor_result=result, current_step=11)
|
||||
|
||||
# Add quality info to response
|
||||
result["scan_quality"] = scan_quality_info
|
||||
result["ocr_stats"] = {
|
||||
"rapid_words": len(rapid_words),
|
||||
"tess_words": len(tess_words),
|
||||
"merged_words": len(merged_words),
|
||||
"min_conf_used": actual_min_conf,
|
||||
"enhance_applied": enhance and is_degraded,
|
||||
"vision_fusion_applied": vision_applied,
|
||||
"document_category": doc_category or session.get("document_category", ""),
|
||||
"ocr_duration_seconds": round(ocr_duration, 1),
|
||||
}
|
||||
|
||||
total_duration = _time.time() - t0
|
||||
logger.info(
|
||||
"rerun-ocr+build-grid session %s: %d zones, %d cols, %d cells in %.1fs",
|
||||
session_id,
|
||||
len(result.get("zones", [])),
|
||||
result.get("summary", {}).get("total_columns", 0),
|
||||
result.get("summary", {}).get("total_cells", 0),
|
||||
total_duration,
|
||||
)
|
||||
|
||||
return result
|
||||
|
||||
|
||||
@router.post("/sessions/{session_id}/save-grid")
|
||||
async def save_grid(session_id: str, request: Request):
|
||||
"""Save edited grid data from the frontend Excel-like editor.
|
||||
|
||||
Receives the full StructuredGrid with user edits (text changes,
|
||||
formatting changes like bold columns, header rows, etc.) and
|
||||
persists it to the session's grid_editor_result.
|
||||
"""
|
||||
session = await get_session_db(session_id)
|
||||
if not session:
|
||||
raise HTTPException(status_code=404, detail=f"Session {session_id} not found")
|
||||
|
||||
body = await request.json()
|
||||
|
||||
# Validate basic structure
|
||||
if "zones" not in body:
|
||||
raise HTTPException(status_code=400, detail="Missing 'zones' in request body")
|
||||
|
||||
# Preserve metadata from the original build
|
||||
existing = session.get("grid_editor_result") or {}
|
||||
result = {
|
||||
"session_id": session_id,
|
||||
"image_width": body.get("image_width", existing.get("image_width", 0)),
|
||||
"image_height": body.get("image_height", existing.get("image_height", 0)),
|
||||
"zones": body["zones"],
|
||||
"boxes_detected": body.get("boxes_detected", existing.get("boxes_detected", 0)),
|
||||
"summary": body.get("summary", existing.get("summary", {})),
|
||||
"formatting": body.get("formatting", existing.get("formatting", {})),
|
||||
"duration_seconds": existing.get("duration_seconds", 0),
|
||||
"edited": True,
|
||||
}
|
||||
|
||||
await update_session_db(session_id, grid_editor_result=result, current_step=11)
|
||||
|
||||
logger.info("save-grid session %s: %d zones saved", session_id, len(body["zones"]))
|
||||
|
||||
return {"session_id": session_id, "saved": True}
|
||||
|
||||
|
||||
@router.get("/sessions/{session_id}/grid-editor")
|
||||
async def get_grid(session_id: str):
|
||||
"""Retrieve the current grid editor state for a session."""
|
||||
session = await get_session_db(session_id)
|
||||
if not session:
|
||||
raise HTTPException(status_code=404, detail=f"Session {session_id} not found")
|
||||
|
||||
result = session.get("grid_editor_result")
|
||||
if not result:
|
||||
raise HTTPException(
|
||||
status_code=404,
|
||||
detail="No grid editor data. Run build-grid first.",
|
||||
)
|
||||
|
||||
return result
|
||||
Reference in New Issue
Block a user