Fix: Update all old-style imports inside packages to new paths

65 files in klausur-service packages + 3 in backend-lehrer packages
had stale imports referencing deleted shim modules.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

backend-lehrer/messenger/conversations.py

@@ -266,7 +266,7 @@ async def send_message(conversation_id: str, message: MessageBase):
 
             if contact and contact.get("email"):
                 try:
-                    from email_service import email_service
+                    from services.email import email_service
 
                     result = email_service.send_messenger_notification(
                         to_email=contact["email"],

backend-lehrer/units/learning_api.py

@@ -363,7 +363,7 @@ def api_generate_story(unit_id: str, payload: StoryGeneratePayload):
         raise HTTPException(status_code=404, detail="Lerneinheit nicht gefunden.")
 
     try:
-        from story_generator import generate_story
+        from services.story_generator import generate_story
         result = generate_story(
             vocabulary=payload.vocabulary,
             language=payload.language,

backend-lehrer/vocabulary/api.py

@@ -22,7 +22,7 @@ from .db import (
     get_all_pos,
     VocabularyWord,
 )
-from learning_units import (
+from units.learning import (
     LearningUnitCreate,
     create_learning_unit,
     get_learning_unit,
@@ -343,7 +343,7 @@ async def api_translate_words(payload: TranslateRequest):
     Uses local LLM (Ollama) for translation. Results are cached in the
     vocabulary_words.translations JSONB field.
     """
-    from translation_service import translate_and_store
+    from services.translation import translate_and_store
 
     if payload.target_language not in {"tr", "ar", "uk", "ru", "pl", "fr", "es"}:
         raise HTTPException(status_code=400, detail=f"Sprache '{payload.target_language}' nicht unterstuetzt")

klausur-service/backend/admin/__init__.py

@@ -2,5 +2,5 @@
 admin package — admin APIs for NiBiS, RAG, templates.
 
 Backward-compatible re-exports: consumers can still use
-``from admin_api import ...`` etc. via the shim files in backend/.
+``from admin.api import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/admin/api.py

@@ -7,7 +7,7 @@ This module was split into:
   - admin_templates.py  (Legal templates ingestion, search)
 
 The `router` object is assembled here by including all sub-routers.
-Importers that did `from admin_api import router` continue to work.
+Importers that did `from admin.api import router` continue to work.
 """
 
 from fastapi import APIRouter

klausur-service/backend/admin/rag.py

@@ -28,7 +28,7 @@ except ImportError:
     MINIO_AVAILABLE = False
 
 try:
-    from metrics_db import (
+    from metrics.db import (
         init_metrics_tables, store_feedback, log_search, log_upload,
         calculate_metrics, get_recent_feedback, get_upload_history
     )

klausur-service/backend/compliance/__init__.py

@@ -2,5 +2,5 @@
 compliance package — compliance pipeline, RBAC/ABAC policy engine.
 
 Backward-compatible re-exports: consumers can still use
-``from compliance_models import ...`` etc. via the shim files in backend/.
+``from compliance.models import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/grid/__init__.py

@@ -2,7 +2,7 @@
 Grid package — restructured from grid_* flat modules.
 
 Backward-compatible re-exports: consumers can still use
-``from grid_build_core import ...`` etc. via the shim files in backend/.
+``from grid.build.core import ...`` etc. via the shim files in backend/.
 
 Sub-packages:
   - grid.build   — grid construction pipeline (_build_grid_core and phases)

klausur-service/backend/grid/build/cell_ops.py

@@ -9,7 +9,7 @@ import logging
 import re
 from typing import Any, Dict, List, Tuple
 
-from cv_ocr_engines import (
+from ocr.engines.engines import (
     _words_to_reading_order_text, _group_words_into_lines, _lookup_ipa,
 )
 

klausur-service/backend/grid/build/cleanup.py

@@ -10,7 +10,7 @@ import logging
 import re
 from typing import Any, Dict, List
 
-from cv_ocr_engines import _words_to_reading_order_text
+from ocr.engines.engines import _words_to_reading_order_text
 
 logger = logging.getLogger(__name__)
 

klausur-service/backend/grid/build/finalize.py

@@ -69,14 +69,14 @@ def _finalize_grid(
 
     # --- Word-gap merge ---
     try:
-        from cv_syllable_detect import merge_word_gaps_in_zones
+        from ocr.detect.syllable.detect import merge_word_gaps_in_zones
         merge_word_gaps_in_zones(zones_data, session_id)
     except Exception as e:
         logger.warning("Word-gap merge failed: %s", e)
 
     # --- Pipe auto-correction ---
     try:
-        from cv_syllable_detect import autocorrect_pipe_artifacts
+        from ocr.detect.syllable.detect import autocorrect_pipe_artifacts
         autocorrect_pipe_artifacts(zones_data, session_id)
     except Exception as e:
         logger.warning("Pipe autocorrect failed: %s", e)
@@ -132,10 +132,10 @@ def _detect_dictionary(
     margin_strip_detected: bool,
 ) -> Dict[str, Any]:
     """Run dictionary detection on the assembled grid."""
-    from cv_layout import _score_dictionary_signals
+    from ocr.layout.layout import _score_dictionary_signals
     dict_detection: Dict[str, Any] = {"is_dictionary": False, "confidence": 0.0}
     try:
-        from cv_vocab_types import ColumnGeometry
+        from ocr.types import ColumnGeometry
         for z in zones_data:
             zone_cells = z.get("cells", [])
             zone_cols = z.get("columns", [])
@@ -222,7 +222,7 @@ def _insert_syllable_dividers(
 
     if _syllable_eligible:
         try:
-            from cv_syllable_detect import insert_syllable_dividers
+            from ocr.detect.syllable.detect import insert_syllable_dividers
             force_syllables = (syllable_mode in ("all", "de", "en"))
             syllable_insertions = insert_syllable_dividers(
                 zones_data, img_bgr, session_id,
@@ -241,7 +241,7 @@ def _split_merged_words(
 ) -> None:
     """Split merged words using dictionary lookup."""
     try:
-        from cv_review import _try_split_merged_word, _SPELL_AVAILABLE
+        from ocr.review.review import _try_split_merged_word, _SPELL_AVAILABLE
         if not _SPELL_AVAILABLE:
             return
         split_count = 0
@@ -307,7 +307,7 @@ def _run_spell_checker(
 ) -> None:
     """Run SmartSpellChecker on all cells."""
     try:
-        from smart_spell import SmartSpellChecker
+        from ocr.spell.smart_spell import SmartSpellChecker
         _ssc = SmartSpellChecker()
         spell_fix_count = 0
 

klausur-service/backend/grid/build/text_ops.py

@@ -10,8 +10,8 @@ import logging
 import re
 from typing import Any, Dict, List, Optional, Set, Tuple
 
-from cv_color_detect import detect_word_colors
-from cv_ocr_engines import (
+from ocr.detect.color_detect import detect_word_colors
+from ocr.engines.engines import (
     fix_cell_phonetics, fix_ipa_continuation_cell, _text_has_garbled_ipa,
     _lookup_ipa,
 )
@@ -207,7 +207,7 @@ def _run_ipa_correction(
 
     # --- German IPA (wiki-pronunciation-dict + epitran) ---
     if de_ipa_target_cols:
-        from cv_ipa_german import insert_german_ipa
+        from ocr.ipa_german import insert_german_ipa
         insert_german_ipa(all_cells, de_ipa_target_cols)
 
     ipa_target_cols = en_ipa_target_cols | de_ipa_target_cols

klausur-service/backend/grid/build/zones.py

@@ -11,11 +11,11 @@ from typing import Any, Dict, List, Optional
 import cv2
 import numpy as np
 
-from cv_box_detect import detect_boxes, split_page_into_zones
-from cv_graphic_detect import detect_graphic_elements
-from cv_color_detect import recover_colored_text
-from cv_vocab_types import PageZone
-from ocr_pipeline_session_store import get_session_image
+from ocr.detect.box_detect import detect_boxes, split_page_into_zones
+from ocr.detect.graphic_detect import detect_graphic_elements
+from ocr.detect.color_detect import recover_colored_text
+from ocr.types import PageZone
+from ocr.pipeline.session_store import get_session_image
 
 from grid.editor.filters import (
     _filter_border_strip_words,

klausur-service/backend/grid/editor/api.py

@@ -8,7 +8,7 @@ The actual endpoints live in:
   - grid_editor_api_unified.py (build-unified-grid, unified-grid)
 
 This module re-exports the combined router and key symbols so that
-existing `from grid_editor_api import router` / `from grid_editor_api import _build_grid_core`
+existing `from grid.editor.api import router` / `from grid.editor.api import _build_grid_core`
 continue to work unchanged.
 """
 
@@ -20,7 +20,7 @@ from .api_box import router as _box_router
 from .api_unified import router as _unified_router
 
 # Re-export _build_grid_core so callers that do
-# `from grid_editor_api import _build_grid_core` keep working.
+# `from grid.editor.api import _build_grid_core` keep working.
 from grid.build.core import _build_grid_core  # noqa: F401
 
 # Merge all sub-routers into one combined router

klausur-service/backend/grid/editor/api_box.py

@@ -7,7 +7,7 @@ import logging
 from fastapi import APIRouter, HTTPException, Request
 
 from .filters import _words_in_zone
-from ocr_pipeline_session_store import (
+from ocr.pipeline.session_store import (
     get_session_db,
     update_session_db,
 )
@@ -76,7 +76,7 @@ async def build_box_grids(session_id: str, request: Request):
         pass
     layout_overrides = body.get("overrides", {})
 
-    from cv_box_layout import build_box_zone_grid
+    from ocr.detect.box_layout import build_box_zone_grid
 
     img_w = grid_data.get("image_width", 0) or word_result.get("image_width", 0)
     img_h = grid_data.get("image_height", 0) or word_result.get("image_height", 0)
@@ -119,7 +119,7 @@ async def build_box_grids(session_id: str, request: Request):
 
         # Apply SmartSpellChecker to all box cells
         try:
-            from smart_spell import SmartSpellChecker
+            from ocr.spell.smart_spell import SmartSpellChecker
             ssc = SmartSpellChecker()
             for cell in box_grid.get("cells", []):
                 text = cell.get("text", "")

klausur-service/backend/grid/editor/api_grid.py

@@ -7,11 +7,11 @@ import logging
 from fastapi import APIRouter, HTTPException, Query, Request
 
 from grid.build.core import _build_grid_core
-from ocr_pipeline_session_store import (
+from ocr.pipeline.session_store import (
     get_session_db,
     update_session_db,
 )
-from ocr_pipeline_common import (
+from ocr.pipeline.common import (
     _cache,
     _load_session_to_cache,
     _get_cached,
@@ -60,7 +60,7 @@ async def build_grid(
 
     # 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
+    from ocr.pipeline.regression import _build_reference_snapshot
 
     wr = session.get("word_result") or {}
     engine = wr.get("ocr_engine", "")
@@ -134,7 +134,7 @@ async def rerun_ocr_and_build_grid(
     # 2. Scan quality assessment
     scan_quality_info = {}
     try:
-        from scan_quality import score_scan_quality
+        from ocr.pipeline.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
@@ -146,7 +146,7 @@ async def rerun_ocr_and_build_grid(
     is_degraded = scan_quality_info.get("is_degraded", False)
     if enhance and is_degraded:
         try:
-            from ocr_image_enhance import enhance_for_ocr
+            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:
@@ -159,8 +159,8 @@ async def rerun_ocr_and_build_grid(
     # RapidOCR
     rapid_words = []
     try:
-        from cv_ocr_engines import ocr_region_rapid
-        from cv_vocab_types import PageRegion
+        from ocr.engines.engines import ocr_region_rapid
+        from ocr.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:
@@ -182,7 +182,7 @@ async def rerun_ocr_and_build_grid(
         })
 
     # 5. Merge OCR results
-    from ocr_pipeline_ocr_merge import _split_paddle_multi_words, _merge_paddle_tesseract, _deduplicate_words
+    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)
@@ -207,7 +207,7 @@ async def rerun_ocr_and_build_grid(
     vision_applied = False
     if vision_fusion:
         try:
-            from vision_ocr_fusion import vision_fuse_ocr
+            from ocr.pipeline.vision_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)

klausur-service/backend/grid/editor/api_gutter.py

@@ -6,7 +6,7 @@ import logging
 
 from fastapi import APIRouter, HTTPException, Request
 
-from ocr_pipeline_session_store import (
+from ocr.pipeline.session_store import (
     get_session_db,
     update_session_db,
 )
@@ -35,7 +35,7 @@ async def gutter_repair(session_id: str):
             detail="No grid data. Run build-grid first.",
         )
 
-    from cv_gutter_repair import analyse_grid_for_gutter_repair
+    from ocr.gutter.repair import analyse_grid_for_gutter_repair
 
     image_width = grid_data.get("image_width", 0)
     result = analyse_grid_for_gutter_repair(grid_data, image_width=image_width)
@@ -86,7 +86,7 @@ async def gutter_repair_apply(session_id: str, request: Request):
     # Allows the user to pick a different correction from the alternatives list
     text_overrides = body.get("text_overrides", {})
 
-    from cv_gutter_repair import apply_gutter_suggestions
+    from ocr.gutter.repair import apply_gutter_suggestions
 
     suggestions = gutter_result.get("suggestions", [])
 

klausur-service/backend/grid/editor/api_unified.py

@@ -6,7 +6,7 @@ import logging
 
 from fastapi import APIRouter, HTTPException
 
-from ocr_pipeline_session_store import (
+from ocr.pipeline.session_store import (
     get_session_db,
     update_session_db,
 )
@@ -32,7 +32,7 @@ async def build_unified_grid_endpoint(session_id: str):
     if not grid_data:
         raise HTTPException(status_code=400, detail="No grid data. Run build-grid first.")
 
-    from unified_grid import build_unified_grid
+    from grid.unified import build_unified_grid
 
     result = build_unified_grid(
         zones=grid_data.get("zones", []),

klausur-service/backend/grid/editor/headers.py

@@ -8,7 +8,7 @@ import logging
 import re
 from typing import Dict, List, Optional
 
-from cv_ocr_engines import _text_has_garbled_ipa
+from ocr.engines.engines import _text_has_garbled_ipa
 
 logger = logging.getLogger(__name__)
 
@@ -394,7 +394,7 @@ def _detect_colspan_cells(
     if len(columns) < 2 or not zone_words or not rows:
         return cells
 
-    from cv_words_first import _assign_word_to_row
+    from ocr.words_first import _assign_word_to_row
 
     # Column boundaries (midpoints between adjacent columns)
     col_boundaries = []

klausur-service/backend/grid/editor/helpers.py

@@ -2,7 +2,7 @@
 Grid Editor helper functions — barrel re-export module.
 
 This file re-exports all public symbols from the split sub-modules
-so that existing ``from grid_editor_helpers import ...`` statements
+so that existing ``from grid.editor.helpers import ...`` statements
 continue to work without changes.
 
 Sub-modules:
@@ -55,4 +55,4 @@ from .zones import (  # noqa: F401
 )
 
 # --- Re-export from cv_words_first (used by cv_box_layout.py) ---------------
-from cv_words_first import _cluster_rows  # noqa: F401
+from ocr.words_first import _cluster_rows  # noqa: F401

klausur-service/backend/grid/editor/zones.py

@@ -12,8 +12,8 @@ import logging
 import re
 from typing import Any, Dict, List, Optional
 
-from cv_vocab_types import PageZone
-from cv_words_first import _cluster_rows, _build_cells
+from ocr.types import PageZone
+from ocr.words_first import _cluster_rows, _build_cells
 
 from .columns import (
     _cluster_columns_by_alignment,

klausur-service/backend/korrektur/__init__.py

@@ -2,5 +2,5 @@
 korrektur package — exam correction, EH templates, PDF export.
 
 Backward-compatible re-exports: consumers can still use
-``from eh_pipeline import ...`` etc. via the shim files in backend/.
+``from korrektur.eh_pipeline import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/metrics/__init__.py

@@ -2,5 +2,5 @@
 metrics package — PostgreSQL metrics database operations.
 
 Backward-compatible re-exports: consumers can still use
-``from metrics_db import ...`` etc. via the shim files in backend/.
+``from metrics.db import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/ocr/__init__.py

@@ -2,7 +2,7 @@
 OCR package — restructured from cv_* flat modules.
 
 Backward-compatible re-exports: consumers can still use
-``from cv_layout import ...`` etc. via the shim files in backend/.
+``from ocr.layout.layout import ...`` etc. via the shim files in backend/.
 """
 
 from .types import *  # noqa: F401,F403

klausur-service/backend/ocr/detect/box_layout.py

@@ -149,7 +149,7 @@ def build_box_zone_grid(
 
     Returns the same format as _build_zone_grid (columns, rows, cells, header_rows).
     """
-    from grid_editor_helpers import _build_zone_grid
+    from grid.editor.helpers import _build_zone_grid
 
     if not zone_words:
         return {

klausur-service/backend/ocr/labeling/api.py

@@ -48,7 +48,7 @@ except ImportError:
     pass
 
 try:
-    from training_export_service import (  # noqa: F401
+    from training.export_service import (  # noqa: F401
         TrainingExportService,
         TrainingSample,
         get_training_export_service,

klausur-service/backend/ocr/labeling/helpers.py

@@ -57,7 +57,7 @@ except ImportError:
 
 # Try to import Training Export Service
 try:
-    from training_export_service import (
+    from training.export_service import (
         TrainingExportService,
         TrainingSample,
         get_training_export_service,

klausur-service/backend/ocr/labeling/routes.py

@@ -20,7 +20,7 @@ from typing import Optional, List
 from datetime import datetime
 import uuid
 
-from metrics_db import (
+from metrics.db import (
     create_ocr_labeling_session,
     get_ocr_labeling_sessions,
     get_ocr_labeling_session,

klausur-service/backend/ocr/labeling/upload_routes.py

@@ -17,7 +17,7 @@ from typing import Optional, List
 import uuid
 import os
 
-from metrics_db import (
+from metrics.db import (
     get_ocr_labeling_session,
     add_ocr_labeling_item,
     get_ocr_labeling_item,
@@ -42,7 +42,7 @@ except ImportError:
     pass
 
 try:
-    from training_export_service import TrainingSample, get_training_export_service
+    from training.export_service import TrainingSample, get_training_export_service
 except ImportError:
     pass
 
@@ -271,7 +271,7 @@ async def run_ocr_for_item(item_id: str):
     if ocr_text is None:
         raise HTTPException(status_code=500, detail="OCR failed")
 
-    from metrics_db import get_pool
+    from metrics.db import get_pool
     pool = await get_pool()
     if pool:
         async with pool.acquire() as conn:

klausur-service/backend/ocr/pipeline/auto.py

@@ -15,7 +15,7 @@ from .reprocess import router as _reprocess_router
 from .auto_steps import router as _steps_router
 
 # Combine both sub-routers into a single router for backwards compatibility.
-# The consumer imports `from ocr_pipeline_auto import router as _auto_router`.
+# The consumer imports `from ocr.pipeline.auto import router as _auto_router`.
 router = APIRouter(prefix="/api/v1/ocr-pipeline", tags=["ocr-pipeline"])
 router.include_router(_reprocess_router)
 router.include_router(_steps_router)

klausur-service/backend/ocr/pipeline/auto_steps.py

@@ -17,7 +17,7 @@ import numpy as np
 from fastapi import APIRouter, HTTPException, Request
 from fastapi.responses import StreamingResponse
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     OLLAMA_REVIEW_MODEL,
     PageRegion,
     RowGeometry,

klausur-service/backend/ocr/pipeline/columns.py

@@ -14,7 +14,7 @@ from typing import Dict, List
 import cv2
 from fastapi import APIRouter, HTTPException
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     _detect_header_footer_gaps,
     _detect_sub_columns,
     classify_column_types,

klausur-service/backend/ocr/pipeline/deskew.py

@@ -12,7 +12,7 @@ from datetime import datetime
 import cv2
 from fastapi import APIRouter, HTTPException
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     create_ocr_image,
     deskew_image,
     deskew_image_by_word_alignment,

klausur-service/backend/ocr/pipeline/dewarp.py

@@ -17,7 +17,7 @@ from typing import Any, Dict
 import cv2
 from fastapi import APIRouter, HTTPException, Query
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     _apply_shear,
     create_ocr_image,
     dewarp_image,

klausur-service/backend/ocr/pipeline/geometry.py

@@ -8,7 +8,7 @@ This module was split into:
   - ocr_pipeline_columns.py    (Column detection + ground truth)
 
 The `router` object is assembled here by including all sub-routers.
-Importers that did `from ocr_pipeline_geometry import router` continue to work.
+Importers that did `from ocr.pipeline.geometry import router` continue to work.
 """
 
 from fastapi import APIRouter

klausur-service/backend/ocr/pipeline/htr_api.py

@@ -243,7 +243,7 @@ async def recognize_from_session(req: HTRSessionRequest):
     Set use_clean=true to prefer the clean image (after handwriting removal step).
     This is useful when you want to do HTR on isolated handwriting regions.
     """
-    from ocr_pipeline_session_store import get_session_db, get_session_image
+    from ocr.pipeline.session_store import get_session_db, get_session_image
 
     session = await get_session_db(req.session_id)
     if not session:

klausur-service/backend/ocr/pipeline/llm_review.py

@@ -15,7 +15,7 @@ from typing import Dict, List
 from fastapi import APIRouter, HTTPException, Request
 from fastapi.responses import StreamingResponse
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     OLLAMA_REVIEW_MODEL,
     llm_review_entries,
     llm_review_entries_streaming,

klausur-service/backend/ocr/pipeline/ocr_merge.py

@@ -15,7 +15,7 @@ import cv2
 import numpy as np
 from fastapi import APIRouter, HTTPException
 
-from cv_words_first import build_grid_from_words
+from ocr.words_first import build_grid_from_words
 from .common import _cache, _append_pipeline_log
 from .session_store import get_session_image, update_session_db
 
@@ -136,7 +136,7 @@ async def paddle_kombi(session_id: str):
     img_png, img_bgr = await _load_session_image(session_id)
     img_h, img_w = img_bgr.shape[:2]
 
-    from cv_ocr_engines import ocr_region_paddle
+    from ocr.engines.engines import ocr_region_paddle
 
     t0 = time.time()
 
@@ -202,8 +202,8 @@ async def rapid_kombi(session_id: str):
     img_png, img_bgr = await _load_session_image(session_id)
     img_h, img_w = img_bgr.shape[:2]
 
-    from cv_ocr_engines import ocr_region_rapid
-    from cv_vocab_types import PageRegion
+    from ocr.engines.engines import ocr_region_rapid
+    from ocr.types import PageRegion
 
     t0 = time.time()
 

klausur-service/backend/ocr/pipeline/orientation_api.py

@@ -9,7 +9,7 @@ from typing import Any, Dict
 import cv2
 from fastapi import APIRouter, HTTPException
 
-from cv_vocab_pipeline import detect_and_fix_orientation
+from ocr.cv_pipeline import detect_and_fix_orientation
 from .page_crop import detect_page_splits
 from .session_store import update_session_db
 

klausur-service/backend/ocr/pipeline/overlay_structure.py

@@ -17,8 +17,8 @@ from fastapi.responses import Response
 
 from .common import _get_base_image_png
 from .session_store import get_session_db
-from cv_color_detect import _COLOR_HEX, _COLOR_RANGES
-from cv_box_detect import detect_boxes, split_page_into_zones
+from ocr.detect.color_detect import _COLOR_HEX, _COLOR_RANGES
+from ocr.detect.box_detect import detect_boxes, split_page_into_zones
 
 logger = logging.getLogger(__name__)
 

klausur-service/backend/ocr/pipeline/postprocess.py

@@ -19,7 +19,7 @@ from .reconstruction import router as _reconstruction_router
 from .validation import router as _validation_router
 
 # Composite router — drop-in replacement for the old monolithic router.
-# ocr_pipeline_api.py imports ``from ocr_pipeline_postprocess import router``.
+# ocr_pipeline_api.py imports ``from ocr.pipeline.postprocess import router``.
 router = APIRouter()
 router.include_router(_llm_review_router)
 router.include_router(_reconstruction_router)

klausur-service/backend/ocr/pipeline/regression_endpoints.py

@@ -14,7 +14,7 @@ from typing import Any, Dict, Optional
 
 from fastapi import APIRouter, HTTPException, Query
 
-from grid_editor_api import _build_grid_core
+from grid.editor.api import _build_grid_core
 from .session_store import (
     get_session_db,
     list_ground_truth_sessions_db,

klausur-service/backend/ocr/pipeline/rows.py

@@ -17,7 +17,7 @@ import cv2
 import numpy as np
 from fastapi import APIRouter, HTTPException
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     create_ocr_image,
     detect_column_geometry,
     detect_row_geometry,
@@ -120,7 +120,7 @@ async def detect_rows(session_id: str):
     # merges rows.  Word-grouping directly clusters words by Y proximity,
     # which is more robust for these cases.
     if is_sub_session and word_dicts:
-        from cv_layout import _build_rows_from_word_grouping
+        from ocr.layout.layout import _build_rows_from_word_grouping
         rows = _build_rows_from_word_grouping(
             word_dicts, left_x, right_x, top_y, bottom_y,
             right_x - left_x, bottom_y - top_y,

klausur-service/backend/ocr/pipeline/sessions_crud.py

@@ -15,7 +15,7 @@ import cv2
 import numpy as np
 from fastapi import APIRouter, File, Form, HTTPException, Query, UploadFile
 
-from cv_vocab_pipeline import render_image_high_res, render_pdf_high_res
+from ocr.cv_pipeline import render_image_high_res, render_pdf_high_res
 from .common import (
     VALID_DOCUMENT_CATEGORIES,
     UpdateSessionRequest,

klausur-service/backend/ocr/pipeline/sessions_images.py

@@ -17,7 +17,7 @@ import numpy as np
 from fastapi import APIRouter, HTTPException, Query
 from fastapi.responses import Response
 
-from cv_vocab_pipeline import create_ocr_image, detect_document_type
+from ocr.cv_pipeline import create_ocr_image, detect_document_type
 from .common import (
     VALID_DOCUMENT_CATEGORIES,
     _append_pipeline_log,

klausur-service/backend/ocr/pipeline/structure.py

@@ -15,9 +15,9 @@ import numpy as np
 from fastapi import APIRouter, HTTPException
 from pydantic import BaseModel
 
-from cv_box_detect import detect_boxes
-from cv_color_detect import _COLOR_RANGES, _COLOR_HEX
-from cv_graphic_detect import detect_graphic_elements
+from ocr.detect.box_detect import detect_boxes
+from ocr.detect.color_detect import _COLOR_RANGES, _COLOR_HEX
+from ocr.detect.graphic_detect import detect_graphic_elements
 from .session_store import (
     get_session_db,
     update_session_db,
@@ -100,7 +100,7 @@ async def detect_structure(session_id: str):
     )
 
     # --- Zone splitting ---
-    from cv_box_detect import split_page_into_zones as _split_zones
+    from ocr.detect.box_detect import split_page_into_zones as _split_zones
     zones = _split_zones(content_x, content_y, content_w_px, content_h_px, boxes)
 
     # --- Color region sampling ---
@@ -123,7 +123,7 @@ async def detect_structure(session_id: str):
             med_s = float(np.median(roi_hsv[:, :, 1]))
             med_v = float(np.median(roi_hsv[:, :, 2]))
             if med_s > 15:
-                from cv_color_detect import _hue_to_color_name
+                from ocr.detect.color_detect import _hue_to_color_name
                 bg_name = _hue_to_color_name(med_h)
                 bg_hex = _COLOR_HEX.get(bg_name, "#6b7280")
             else:

klausur-service/backend/ocr/pipeline/words.py

@@ -23,7 +23,7 @@ import numpy as np
 from fastapi import APIRouter, HTTPException
 from pydantic import BaseModel
 
-from cv_words_first import build_grid_from_words
+from ocr.words_first import build_grid_from_words
 from .session_store import (
     get_session_db,
     get_session_image,
@@ -72,7 +72,7 @@ async def paddle_direct(session_id: str):
 
     img_h, img_w = img_bgr.shape[:2]
 
-    from cv_ocr_engines import ocr_region_paddle
+    from ocr.engines.engines import ocr_region_paddle
 
     t0 = time.time()
     word_dicts = await ocr_region_paddle(img_bgr, region=None)

klausur-service/backend/ocr/pipeline/words_detect.py

@@ -17,7 +17,7 @@ import numpy as np
 from fastapi import APIRouter, HTTPException, Request
 from fastapi.responses import StreamingResponse
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     PageRegion,
     RowGeometry,
     _cells_to_vocab_entries,
@@ -27,7 +27,7 @@ from cv_vocab_pipeline import (
     create_ocr_image,
     detect_column_geometry,
 )
-from cv_words_first import build_grid_from_words
+from ocr.words_first import build_grid_from_words
 from .session_store import (
     get_session_db,
     update_session_db,
@@ -223,7 +223,7 @@ async def _words_first_path(
     img_h, img_w = dewarped_bgr.shape[:2]
 
     if engine == "paddle":
-        from cv_ocr_engines import ocr_region_paddle
+        from ocr.engines.engines import ocr_region_paddle
         wf_word_dicts = await ocr_region_paddle(dewarped_bgr, region=None)
         cached["_paddle_word_dicts"] = wf_word_dicts
     else:

klausur-service/backend/ocr/pipeline/words_stream.py

@@ -15,7 +15,7 @@ from typing import Any, Dict, List
 import numpy as np
 from fastapi import Request
 
-from cv_vocab_pipeline import (
+from ocr.cv_pipeline import (
     PageRegion,
     RowGeometry,
     _cells_to_vocab_entries,

klausur-service/backend/ocr/review/spell.py

@@ -219,7 +219,7 @@ def spell_review_entries_sync(entries: List[Dict]) -> Dict:
     # Use SmartSpellChecker if available
     _smart = None
     try:
-        from smart_spell import SmartSpellChecker
+        from ocr.spell.smart_spell import SmartSpellChecker
         _smart = SmartSpellChecker()
         logger.debug("spell_review: using SmartSpellChecker")
     except Exception:

klausur-service/backend/ocr/spell/core.py

@@ -159,7 +159,7 @@ class _SmartSpellCoreBase:
             return True
         # Also accept known abbreviations (sth, sb, adj, etc.)
         try:
-            from cv_ocr_engines import _KNOWN_ABBREVIATIONS
+            from ocr.engines.engines import _KNOWN_ABBREVIATIONS
             if w in _KNOWN_ABBREVIATIONS:
                 return True
         except ImportError:

klausur-service/backend/ocr/spell/text.py

@@ -35,7 +35,7 @@ class SmartSpellChecker(_SmartSpellCoreBase):
         """
         # Import known abbreviations for vocabulary context
         try:
-            from cv_ocr_engines import _KNOWN_ABBREVIATIONS
+            from ocr.engines.engines import _KNOWN_ABBREVIATIONS
         except ImportError:
             _KNOWN_ABBREVIATIONS = set()
 
@@ -149,7 +149,7 @@ class SmartSpellChecker(_SmartSpellCoreBase):
         # --- Pass 1: Boundary repair between adjacent unknown words ---
         # Import abbreviations for the heuristic below
         try:
-            from cv_ocr_engines import _KNOWN_ABBREVIATIONS as _ABBREVS
+            from ocr.engines.engines import _KNOWN_ABBREVIATIONS as _ABBREVS
         except ImportError:
             _ABBREVS = set()
 

klausur-service/backend/training/__init__.py

@@ -2,5 +2,5 @@
 training package — training API, simulation, export, TrOCR.
 
 Backward-compatible re-exports: consumers can still use
-``from training_api import ...`` etc. via the shim files in backend/.
+``from training.api import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/training/routes.py

@@ -211,7 +211,7 @@ async def delete_model_version(version_id: str):
 @router.get("/dataset/stats", response_model=dict)
 async def get_dataset_stats():
     """Get statistics about the training dataset."""
-    from metrics_db import get_zeugnis_stats
+    from metrics.db import get_zeugnis_stats
 
     zeugnis_stats = await get_zeugnis_stats()
 

klausur-service/backend/vocab/__init__.py

@@ -2,5 +2,5 @@
 Vocab package — restructured from vocab_* flat modules.
 
 Backward-compatible re-exports: consumers can still use
-``from vocab_worksheet_api import ...`` etc. via the shim files in backend/.
+``from vocab.worksheet.api import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/vocab/worksheet/analysis_api.py

@@ -27,7 +27,7 @@ from .generation import convert_pdf_page_to_image
 
 # Try to import Tesseract extractor
 try:
-    from tesseract_vocab_extractor import (
+    from ocr.engines.tesseract_extractor import (
         extract_bounding_boxes, TESSERACT_AVAILABLE,
     )
 except ImportError:
@@ -264,7 +264,7 @@ async def extract_with_boxes(session_id: str, page_number: int):
     # Deskew image before OCR
     deskew_angle = 0.0
     try:
-        from cv_vocab_pipeline import deskew_image_by_word_alignment, CV2_AVAILABLE
+        from ocr.cv_pipeline import deskew_image_by_word_alignment, CV2_AVAILABLE
         if CV2_AVAILABLE:
             image_data, deskew_angle = deskew_image_by_word_alignment(image_data)
             logger.info(f"Deskew: {deskew_angle:.2f}° for page {page_number}")

klausur-service/backend/vocab/worksheet/compare_api.py

@@ -23,7 +23,7 @@ from .generation import convert_pdf_page_to_image
 
 # Try to import Tesseract extractor
 try:
-    from tesseract_vocab_extractor import (
+    from ocr.engines.tesseract_extractor import (
         run_tesseract_pipeline,
         match_positions_to_vocab, TESSERACT_AVAILABLE,
     )
@@ -32,7 +32,7 @@ except ImportError:
 
 # Try to import CV Pipeline
 try:
-    from cv_vocab_pipeline import run_cv_pipeline, CV_PIPELINE_AVAILABLE
+    from ocr.cv_pipeline import run_cv_pipeline, CV_PIPELINE_AVAILABLE
 except ImportError:
     CV_PIPELINE_AVAILABLE = False
 
@@ -328,7 +328,7 @@ async def analyze_grid(session_id: str, page_number: int, use_tesseract: bool =
             # Run Tesseract if not already cached
             if not tess_page_data:
                 logger.info("Running Tesseract for grid analysis (not cached)")
-                from tesseract_vocab_extractor import run_tesseract_pipeline as _run_tess
+                from ocr.engines.tesseract_extractor import run_tesseract_pipeline as _run_tess
                 tess_page_data = await _run_tess(image_data, lang="eng+deu")
                 session[f"tesseract_page_{page_number}"] = tess_page_data
                 session["tesseract_words"] = tess_page_data.get("words", [])

klausur-service/backend/vocab/worksheet/ocr.py

@@ -39,7 +39,7 @@ except ImportError:
 
 # CV pipeline helpers
 try:
-    from cv_vocab_pipeline import (
+    from ocr.cv_pipeline import (
         deskew_two_pass,
         dewarp_image,
         detect_and_fix_orientation,
@@ -54,7 +54,7 @@ except ImportError:
     _fix_phonetic_brackets = None  # type: ignore[assignment]
 
 try:
-    from cv_cell_grid import (
+    from ocr.cell_grid.cell_grid import (
         _merge_wrapped_rows,
         _merge_phonetic_continuation_rows,
         _merge_continuation_rows,
@@ -65,17 +65,17 @@ except ImportError:
     _merge_continuation_rows = None  # type: ignore[assignment]
 
 try:
-    from cv_ocr_engines import ocr_region_rapid
+    from ocr.engines.engines import ocr_region_rapid
 except ImportError:
     ocr_region_rapid = None  # type: ignore[assignment]
 
 try:
-    from cv_vocab_types import PageRegion
+    from ocr.types import PageRegion
 except ImportError:
     PageRegion = None  # type: ignore[assignment]
 
 try:
-    from ocr_pipeline_ocr_merge import (
+    from ocr.pipeline.ocr_merge import (
         _split_paddle_multi_words,
         _merge_paddle_tesseract,
         _deduplicate_words,
@@ -86,12 +86,12 @@ except ImportError:
     _deduplicate_words = None  # type: ignore[assignment]
 
 try:
-    from cv_words_first import build_grid_from_words
+    from ocr.words_first import build_grid_from_words
 except ImportError:
     build_grid_from_words = None  # type: ignore[assignment]
 
 try:
-    from ocr_pipeline_session_store import (
+    from ocr.pipeline.session_store import (
         create_session_db as create_pipeline_session_db,
         update_session_db as update_pipeline_session_db,
     )
@@ -173,7 +173,7 @@ async def _run_ocr_pipeline_for_page(
     # 5. Content crop (removes scanner borders, gutter shadows)
     t0 = _time.time()
     try:
-        from page_crop import detect_and_crop_page
+        from ocr.pipeline.page_crop import detect_and_crop_page
         cropped_bgr, crop_result = detect_and_crop_page(dewarped_bgr)
         if crop_result.get("crop_applied"):
             dewarped_bgr = cropped_bgr
@@ -186,7 +186,7 @@ async def _run_ocr_pipeline_for_page(
     # 5b. Scan quality assessment
     scan_quality_report = None
     try:
-        from scan_quality import score_scan_quality
+        from ocr.pipeline.scan_quality import score_scan_quality
         scan_quality_report = score_scan_quality(dewarped_bgr)
     except Exception as e:
         logger.warning(f"  scan quality: failed ({e})")
@@ -200,7 +200,7 @@ async def _run_ocr_pipeline_for_page(
     is_degraded = scan_quality_report.is_degraded if scan_quality_report else False
     if is_degraded and enable_enhance:
         try:
-            from ocr_image_enhance import enhance_for_ocr
+            from ocr.image_enhance import enhance_for_ocr
             dewarped_bgr = enhance_for_ocr(dewarped_bgr, is_degraded=True)
             logger.info("  enhancement: applied (degraded scan)")
         except Exception as e:
@@ -212,8 +212,8 @@ async def _run_ocr_pipeline_for_page(
 
     # RapidOCR (local ONNX)
     try:
-        from cv_ocr_engines import ocr_region_rapid
-        from cv_vocab_types import PageRegion
+        from ocr.engines.engines import ocr_region_rapid
+        from ocr.types import PageRegion
         full_region = PageRegion(type="full_page", x=0, y=0, width=img_w, height=img_h)
         rapid_words = ocr_region_rapid(dewarped_bgr, full_region) or []
     except Exception as e:
@@ -243,8 +243,8 @@ async def _run_ocr_pipeline_for_page(
         })
 
     # Merge dual-engine results
-    from ocr_pipeline_ocr_merge import _split_paddle_multi_words, _merge_paddle_tesseract, _deduplicate_words
-    from cv_words_first import build_grid_from_words
+    from ocr.pipeline.ocr_merge import _split_paddle_multi_words, _merge_paddle_tesseract, _deduplicate_words
+    from ocr.words_first import build_grid_from_words
 
     rapid_split = _split_paddle_multi_words(rapid_words) if rapid_words else []
     if rapid_split or tess_words:
@@ -300,7 +300,7 @@ async def _run_ocr_pipeline_for_page(
     # 8. Run full grid-build (with pipe-autocorrect, word-gap merge, etc.)
     t0 = _time.time()
     try:
-        from grid_editor_api import _build_grid_core
+        from grid.editor.api import _build_grid_core
         session_data = {
             "word_result": word_result,
         }

klausur-service/backend/vocab/worksheet/upload_api.py

@@ -36,7 +36,7 @@ LOCAL_STORAGE_PATH = os.getenv("VOCAB_STORAGE_PATH", "/app/vocab-worksheets")
 
 try:
     import numpy as np
-    from cv_preprocessing import render_pdf_high_res, detect_and_fix_orientation
+    from ocr.preprocessing.preprocessing import render_pdf_high_res, detect_and_fix_orientation
     OCR_PIPELINE_AVAILABLE = True
 except ImportError:
     np = None  # type: ignore[assignment]

klausur-service/backend/worksheet/__init__.py

@@ -2,5 +2,5 @@
 worksheet package — worksheet editor, NRU generator, cleanup.
 
 Backward-compatible re-exports: consumers can still use
-``from worksheet_editor_api import ...`` etc. via the shim files in backend/.
+``from worksheet.editor_api import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/worksheet/editor_api.py

@@ -366,7 +366,7 @@ async def reconstruct_document_from_session(request: ReconstructRequest):
 async def get_available_sessions():
     """Get list of available vocab sessions that can be reconstructed."""
     try:
-        from vocab_worksheet_api import _sessions
+        from vocab.worksheet.api import _sessions
 
         available = []
         for session_id, session in _sessions.items():

klausur-service/backend/worksheet/editor_reconstruct.py

@@ -31,7 +31,7 @@ async def reconstruct_document_logic(request: ReconstructRequest) -> Reconstruct
     Returns ReconstructResponse ready to send to the client.
     """
     from fastapi import HTTPException
-    from vocab_worksheet_api import _sessions, convert_pdf_page_to_image
+    from vocab.worksheet.api import _sessions, convert_pdf_page_to_image
 
     # Check if session exists
     if request.session_id not in _sessions:

klausur-service/backend/zeugnis/__init__.py

@@ -2,5 +2,5 @@
 zeugnis package — certificate crawler, models, storage.
 
 Backward-compatible re-exports: consumers can still use
-``from zeugnis_api import ...`` etc. via the shim files in backend/.
+``from zeugnis.api import ...`` etc. via the shim files in backend/.
 """

klausur-service/backend/zeugnis/api_docs.py

@@ -16,7 +16,7 @@ from .models import (
 from .crawler import (
     start_crawler, stop_crawler, get_crawler_status,
 )
-from metrics_db import (
+from metrics.db import (
     get_zeugnis_documents, get_zeugnis_stats,
     log_zeugnis_event, get_pool,
 )

klausur-service/backend/zeugnis/api_sources.py

@@ -15,7 +15,7 @@ from .models import (
     BUNDESLAENDER,
     generate_id, get_training_allowed, get_bundesland_name, get_license_for_bundesland,
 )
-from metrics_db import (
+from metrics.db import (
     get_zeugnis_sources, upsert_zeugnis_source, get_pool,
 )
 

klausur-service/backend/zeugnis/control.py

@@ -31,7 +31,7 @@ async def start_crawler(bundesland: Optional[str] = None, source_id: Optional[st
 
     async def run_crawler():
         try:
-            from metrics_db import get_pool
+            from metrics.db import get_pool
             pool = await get_pool()
 
             if pool:

klausur-service/backend/zeugnis/seed_data.py

@@ -322,8 +322,8 @@ SEED_DATA: Dict[str, Dict[str, Any]] = {
 
 async def populate_seed_data():
     """Populate database with seed data."""
-    from metrics_db import get_pool, upsert_zeugnis_source
-    from zeugnis_models import generate_id
+    from metrics.db import get_pool, upsert_zeugnis_source
+    from zeugnis.models import generate_id
 
     pool = await get_pool()
     if not pool:
@@ -412,4 +412,4 @@ if __name__ == "__main__":
 
     print("\n" + "=" * 60)
     print("To populate database, run:")
-    print("  python -c 'import asyncio; from zeugnis_seed_data import populate_seed_data; asyncio.run(populate_seed_data())'")
+    print("  python -c 'import asyncio; from zeugnis.seed_data import populate_seed_data; asyncio.run(populate_seed_data())'")

klausur-service/backend/zeugnis/worker.py

@@ -83,7 +83,7 @@ class ZeugnisCrawler:
 
         # Initialize database connection
         try:
-            from metrics_db import get_pool
+            from metrics.db import get_pool
             self.db_pool = await get_pool()
         except Exception as e:
             print(f"Failed to get database pool: {e}")