chore: Dead-Code entfernen (90 Funde, 4-fach cross-model-verifiziert)

extractive/eval/extractive_eval.py

@@ -6,8 +6,6 @@
 from rapidfuzz import fuzz
 
 import re
-import sys
-from pathlib import Path
 
 from shared.db_schema import SCHEMA_SQL as _SCHEMA_MIGRATION
 

extractive/pipeline/pdf_chunker.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 import re
 from dataclasses import dataclass
-from pathlib import Path
 
 import pdfplumber
 

generative/agents/_litellm_backend.py

@@ -9,7 +9,6 @@
 Vollständige Provider-Liste: https://docs.litellm.ai/docs/providers
 """
 from __future__ import annotations
-import sys
 import time
 
 import litellm

generative/agents/base.py

@@ -11,14 +11,13 @@
 import threading
 import time
 from dataclasses import dataclass
-from pathlib import Path
 from typing import Optional
 
 from generative.config import MODEL_OPUS, CACHE_DIR, BACKEND
 
 # Re-Export für Backwards-Compat — Agenten importieren aus agents.base
 from generative.agents.tracing import trace_event, trace_run_start, set_tracing_backend, flush_tracing  # noqa: F401
-from generative.agents.tracing import _RUN_ID  # single source of truth for run ID
+from generative.agents.tracing import _RUN_ID  # noqa: F401  # re-export, extern genutzt (orchestrator, eval_quality)
 
 # OTel-Tracer für Phoenix. Bleibt None bis orchestrator._setup_phoenix_tracing()
 # ihn via set_llm_tracer() explizit setzt (nur bei ATOMIC_AGENT_TRACING=phoenix).

generative/agents/context_builder.py

@@ -8,7 +8,7 @@
 
 import yaml
 
-from generative.config import VAULT, WISSEN, SCRIPTS_DIR  # BA_DIR-Import entfernt (v31)
+from generative.config import VAULT, SCRIPTS_DIR  # BA_DIR-Import entfernt (v31)
 
 # Vault-Ordner die NICHT als existing_concepts zählen
 SKIP_DIRS = {"00-inbox", "98-system", "99-archive", "08-dashboards", ".obsidian", ".trash"}

generative/agents/extractor.py

@@ -5,7 +5,6 @@
 Seitenzahl, deutsche Sprache, Akronyme aufgelöst, aliases-Liste, kein Pass-Through.
 """
 from __future__ import annotations
-import json
 import re
 
 from generative.agents.base import call_claude_async

generative/agents/langfuse_backend.py

@@ -10,10 +10,8 @@
 import base64
 import json
 import os
-import time
 import uuid
 from datetime import datetime, timedelta
-from typing import Any
 
 
 class LangfuseBackend:

generative/agents/planner.py

@@ -1,7 +1,6 @@
 """Planner-Agent: TOC + Intro + Fazit + Relevanz-Profil → ConceptPlan."""
 from __future__ import annotations
 
-import re
 
 # Generika-Blacklist (portiert aus extractive/gliner_planner.py, angepasst für LLM-Output).
 # Fängt seltene LLM-"Ausrutscher" ab wenn trotz Prompt-Vorgabe abstrakte Einzel-Konzepte

generative/calibration/run.py

@@ -15,7 +15,6 @@
 from generative.eval_quality_v4 import (  # noqa: E402
     _QUALITY_HISTORY as HISTORY_PATH,
     eval_note,
-    print_summary,
     save_result,
 )
 from generative.config import AGENT_VERSION  # noqa: E402

generative/db.py

@@ -17,7 +17,6 @@
 
 import os
 import sqlite3
-import sys
 from contextlib import contextmanager
 from datetime import datetime
 from pathlib import Path

generative/eval_batch.py

@@ -12,7 +12,6 @@
 from __future__ import annotations
 import argparse
 import re
-import sys
 from pathlib import Path
 
 

generative/eval_dashboard_server.py

@@ -11,7 +11,6 @@
 
 import argparse
 import json
-import sys
 import webbrowser
 from datetime import datetime
 from http.server import BaseHTTPRequestHandler, HTTPServer
@@ -232,7 +231,7 @@ def _read_calibration_data(allowed_note_paths: set | None = None,
             "avg_agree": avg_agree,
             "has_labels": n_labeled > 0,
         }
-    except Exception as e:
+    except Exception:
         return {"rows": [], "n_eval": 0, "n_labeled": 0, "avg_agree": None, "has_labels": False}
 
 

generative/eval_progress.py

@@ -9,8 +9,6 @@
 from __future__ import annotations
 import argparse
 import json
-import sys
-from pathlib import Path
 
 from generative.config import CACHE_DIR
 

generative/eval_quality.py

@@ -128,8 +128,7 @@ def _page_num_from_str(page_str: str) -> int | None:
 def _semantic_score(quote: str, page_text: str) -> float:
     """Sentence-Transformers Cosine-Similarity. 0.0 wenn Modell nicht geladen."""
     try:
-        from generative.pipeline.embeddings import _model, cosine
-        import numpy as np
+        from generative.pipeline.embeddings import _model
         model = _model()
         if model is None:
             return 0.0

generative/eval_quality_v4.py

@@ -32,7 +32,7 @@
     sys.exit("rapidfuzz fehlt: pip install rapidfuzz")
 
 from generative.agents import base
-from generative.config import AGENT_VERSION, CACHE_DIR, EVAL_ADAPTIVE_K_HIGH, EVAL_ADAPTIVE_K_MID, MODEL_OPUS, MODEL_CONFIG, QUALITY_HISTORY
+from generative.config import AGENT_VERSION, MODEL_OPUS, MODEL_CONFIG, QUALITY_HISTORY
 from decision_engine import ClaimDecision, ClaimInput, DEFAULT_CONFIG, Label, determine_decision
 from decision_engine.aggregation import aggregate as aggregate_decisions
 from decision_engine.models import QualityFlag
@@ -558,7 +558,6 @@ def _aggregate(
     # Atomic-Agent-spezifische Felder wie Wilson-CI, PDF-Chunk-Zahl und LLM-Usage an.
     valid_claims = int(engine_metrics["valid_claims"])
     parse_error_count = counts[PARSE_ERROR]
-    confirmed = counts[SUPPORTED_EXACT] + counts[SUPPORTED_PARAPHRASE]
     hallucinated = counts[NOT_IN_CONTEXT] + counts[CONTRADICTED]
     with_evidence = sum(1 for score in claim_scores if score["evidence"])
     evidence_verified_count = sum(1 for score in claim_scores if score["evidence_verified"] is True)

generative/migrate_jsonl_to_sqlite.py

@@ -10,7 +10,6 @@
 
 import argparse
 import json
-import sys
 from pathlib import Path
 
 from generative import db

generative/orchestrator.py

@@ -57,7 +57,6 @@ def _span(name: str, **attrs):
     CRITIC_AUTO_THRESHOLD,
     ER_BODY_COSINE_THRESHOLD,
     ER_TITLE_COSINE_THRESHOLD,
-    ER_BLOCKING_JACCARD,
     ER_MAX_TOKEN_DIFF,
     ER_HUB_GENERIC_TOKENS,
     ENABLE_ENTITY_RESOLUTION,

generative/pipeline/acronym_fix.py

@@ -137,7 +137,6 @@ def extract_acronym_pairs(text: str) -> dict[str, str]:
     - „Short Form (Long Form)" — typisch in Lehrbüchern, Glossaren
     """
     pairs: dict[str, str] = {}
-    seen_orders: dict[str, str] = {}  # Prävention von späteren falschen Re-Definitionen
 
     for match in _PAT_LONG_PAREN_SHORT.finditer(text):
         before = match.group(1).strip()

generative/pipeline/embeddings.py

@@ -10,7 +10,6 @@
 from __future__ import annotations
 import re
 import threading
-from functools import lru_cache
 
 # Singleton-Modell — Lade-Zeit ~3s, danach pro Embedding ~10ms
 _MODEL = None

generative/pipeline/pdf_chunker.py

@@ -8,7 +8,7 @@
 import re
 import subprocess
 import sys
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from pathlib import Path
 
 from generative.config import CHUNK_WORDS, MIN_WORDS_PER_PAGE

generative/pipeline/vault_writer.py

@@ -8,7 +8,7 @@
 
 import yaml
 
-from generative.config import VAULT, WISSEN, INBOX, LITERATURE_DIR, CRITIC_AUTO_THRESHOLD
+from generative.config import VAULT, INBOX, LITERATURE_DIR, CRITIC_AUTO_THRESHOLD
 from generative.schemas.atomic_note import AtomicNoteDraft
 
 
@@ -565,7 +565,6 @@ def find_existing_in_inbox(source_file: str, title: str,
         if is_pristine_pipeline_note(text):
             return f
     return matches[0][0]
-    return None
 
 
 def _read_source_field(note_path: Path) -> str | None:

generative/tests/test_acronym_fix.py

@@ -1,7 +1,5 @@
 """Tests für Schwartz-Hearst-basierte Akronym-Erkennung + Body-Insertion."""
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 from generative.pipeline.acronym_fix import (

generative/tests/test_chapter_regex.py

@@ -8,8 +8,6 @@
 split_by_chapters() als auch in drop_frontmatter_pages().
 """
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 from generative.pipeline.pdf_chunker import (

generative/tests/test_chunk_recall.py

@@ -8,8 +8,6 @@
                          extract_overview()-Output behalten (Planner-Recall-Decke).
 """
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 import pytest

generative/tests/test_cli.py

@@ -1,7 +1,6 @@
 """Tests für den atomic-notes-Konsolen-Entry-Point (generative/cli.py)."""
 from __future__ import annotations
 
-import pytest
 
 from generative import cli
 

generative/tests/test_collect_agreement.py

@@ -6,8 +6,6 @@
 `note_claim_agreement` paart stattdessen Mensch- und Pipeline-Label pro claim_idx.
 """
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 from generative.calibration.collect import note_claim_agreement

generative/tests/test_collect_db_write.py

@@ -11,7 +11,6 @@
 from __future__ import annotations
 import json
 import sqlite3
-import sys
 from pathlib import Path
 
 

generative/tests/test_config.py

@@ -2,7 +2,6 @@
 
 import importlib
 import sys
-from pathlib import Path
 
 import pytest
 

generative/tests/test_cost.py

@@ -1,4 +1,3 @@
-import sys
 from unittest.mock import patch
 from generative.config import compute_cost_per_call
 

generative/tests/test_cross_draft_hub_suggest.py

@@ -7,8 +7,6 @@
 Seiteneffektfrei — schlägt nur vor, erzeugt keine Note.
 """
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 from generative.schemas.atomic_note import AtomicNoteDraft

generative/tests/test_dashboard_avg_agree.py

@@ -6,8 +6,6 @@
 (katastrophale Fehlkalibrierung) statt "–" (keine Daten).
 """
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 from generative.eval_dashboard_server import _avg_agreement

generative/tests/test_db.py

@@ -1,7 +1,6 @@
 import sqlite3
 import tempfile
 import os
-import sys
 
 
 def test_pipeline_run_has_cost_usd():

generative/tests/test_decision_engine.py

@@ -1,7 +1,5 @@
 from __future__ import annotations
 
-import sys
-from pathlib import Path
 
 
 from decision_engine import ClaimDecision, ClaimInput, DEFAULT_CONFIG, Label, determine_decision

generative/tests/test_doctor.py

@@ -1,7 +1,6 @@
 """Tests für atomic-notes doctor (generative/doctor.py) — Preflight-Checks als pure Funktionen."""
 from __future__ import annotations
 
-from pathlib import Path
 
 from generative import doctor
 

generative/tests/test_dry_run_eval_scope.py

@@ -1,6 +1,5 @@
 from __future__ import annotations
 
-import sys
 from pathlib import Path
 
 

generative/tests/test_er_stage1_filter.py

@@ -7,8 +7,6 @@
     (b) Hub-Generic-Blocklist: shorter darf nicht ⊆ ER_HUB_GENERIC_TOKENS sein
 """
 from __future__ import annotations
-import sys
-from pathlib import Path
 
 
 from generative.orchestrator import er_stage1_decision

generative/tests/test_figure_alt.py

@@ -8,7 +8,6 @@
 """
 from __future__ import annotations
 
-import sys
 from pathlib import Path
 
 ROOT = Path(__file__).resolve().parent.parent

generative/tests/test_figure_feasibility.py

@@ -6,8 +6,6 @@
 """
 from __future__ import annotations
 
-import sys
-from pathlib import Path
 
 
 from generative.pipeline.pdf_chunker import Chunk

generative/tests/test_inline_eval_toggle.py

@@ -1,7 +1,5 @@
 from __future__ import annotations
 
-import sys
-from pathlib import Path
 
 
 from generative.orchestrator import inline_eval_enabled

generative/tests/test_overwrite_guard.py

@@ -4,7 +4,6 @@
 überschreiben (Datenverlust). content-hash-Guard: pristine Pipeline-Notes dürfen
 idempotent überschrieben werden, editierte nicht.
 """
-from pathlib import Path
 
 from generative.schemas.atomic_note import AtomicNoteDraft
 

generative/tests/test_pdf_chunker.py

@@ -7,8 +7,6 @@
 """
 from __future__ import annotations
 import re
-import sys
-from pathlib import Path
 
 
 from generative.pipeline.pdf_chunker import drop_frontmatter_pages, page_range_of_text, concept_text_window

generative/tests/test_pdf_enrich.py

@@ -1,5 +1,4 @@
 import sys
-from pathlib import Path
 
 def test_extract_text_returns_string(tmp_path):
     """extract_text() gibt String zurück (auch bei leerer PDF)."""