From 53b0d778530dcc76c3eb1f7888cf586fa6caf316 Mon Sep 17 00:00:00 2001
From: Benjamin Admin <benjaminadmin@MacBook-Pro.local>
Date: Sat, 11 Apr 2026 14:39:48 +0200
Subject: [PATCH] Multi-page PDF support: create one session per page

When uploading a PDF with > 1 page to the OCR pipeline, each page
now gets its own session (grouped by document_group_id). Previously
only page 1 was processed. The response includes a pages array with
all session IDs so the frontend can navigate between them.

Single-page PDFs and images continue to work as before.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 .../backend/ocr_pipeline_sessions.py          | 103 +++++++++++++++++-
 1 file changed, 100 insertions(+), 3 deletions(-)

diff --git a/klausur-service/backend/ocr_pipeline_sessions.py b/klausur-service/backend/ocr_pipeline_sessions.py
index 99372eb..cd747b8 100644
--- a/klausur-service/backend/ocr_pipeline_sessions.py
+++ b/klausur-service/backend/ocr_pipeline_sessions.py
@@ -71,13 +71,36 @@ async def create_session(
     file: UploadFile = File(...),
     name: Optional[str] = Form(None),
 ):
-    """Upload a PDF or image file and create a pipeline session."""
+    """Upload a PDF or image file and create a pipeline session.
+
+    For multi-page PDFs (> 1 page), each page becomes its own session
+    grouped under a ``document_group_id``.  The response includes a
+    ``pages`` array with one entry per page/session.
+    """
     file_data = await file.read()
     filename = file.filename or "upload"
     content_type = file.content_type or ""
 
-    session_id = str(uuid.uuid4())
     is_pdf = content_type == "application/pdf" or filename.lower().endswith(".pdf")
+    session_name = name or filename
+
+    # --- Multi-page PDF handling ---
+    if is_pdf:
+        try:
+            import fitz  # PyMuPDF
+            pdf_doc = fitz.open(stream=file_data, filetype="pdf")
+            page_count = pdf_doc.page_count
+            pdf_doc.close()
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=f"Could not read PDF: {e}")
+
+        if page_count > 1:
+            return await _create_multi_page_sessions(
+                file_data, filename, session_name, page_count,
+            )
+
+    # --- Single page (image or 1-page PDF) ---
+    session_id = str(uuid.uuid4())
 
     try:
         if is_pdf:
@@ -93,7 +116,6 @@ async def create_session(
         raise HTTPException(status_code=500, detail="Failed to encode image")
 
     original_png = png_buf.tobytes()
-    session_name = name or filename
 
     # Persist to DB
     await create_session_db(
@@ -134,6 +156,81 @@ async def create_session(
     }
 
 
+async def _create_multi_page_sessions(
+    pdf_data: bytes,
+    filename: str,
+    base_name: str,
+    page_count: int,
+) -> dict:
+    """Create one session per PDF page, grouped by document_group_id."""
+    document_group_id = str(uuid.uuid4())
+    pages = []
+
+    for page_idx in range(page_count):
+        session_id = str(uuid.uuid4())
+        page_name = f"{base_name} — Seite {page_idx + 1}"
+
+        try:
+            img_bgr = render_pdf_high_res(pdf_data, page_number=page_idx, zoom=3.0)
+        except Exception as e:
+            logger.warning(f"Failed to render PDF page {page_idx + 1}: {e}")
+            continue
+
+        ok, png_buf = cv2.imencode(".png", img_bgr)
+        if not ok:
+            continue
+        page_png = png_buf.tobytes()
+
+        await create_session_db(
+            session_id=session_id,
+            name=page_name,
+            filename=filename,
+            original_png=page_png,
+            document_group_id=document_group_id,
+            page_number=page_idx + 1,
+        )
+
+        _cache[session_id] = {
+            "id": session_id,
+            "filename": filename,
+            "name": page_name,
+            "original_bgr": img_bgr,
+            "oriented_bgr": None,
+            "cropped_bgr": None,
+            "deskewed_bgr": None,
+            "dewarped_bgr": None,
+            "orientation_result": None,
+            "crop_result": None,
+            "deskew_result": None,
+            "dewarp_result": None,
+            "ground_truth": {},
+            "current_step": 1,
+        }
+
+        h, w = img_bgr.shape[:2]
+        pages.append({
+            "session_id": session_id,
+            "name": page_name,
+            "page_number": page_idx + 1,
+            "image_width": w,
+            "image_height": h,
+            "original_image_url": f"/api/v1/ocr-pipeline/sessions/{session_id}/image/original",
+        })
+
+        logger.info(
+            f"OCR Pipeline: created page session {session_id} "
+            f"(page {page_idx + 1}/{page_count}) from {filename} ({w}x{h})"
+        )
+
+    return {
+        "document_group_id": document_group_id,
+        "filename": filename,
+        "name": base_name,
+        "page_count": page_count,
+        "pages": pages,
+    }
+
+
 @router.get("/sessions/{session_id}")
 async def get_session_info(session_id: str):
     """Get session info including deskew/dewarp/column results for step navigation."""