scripts/import_markdown.py aktualisiert
Some checks failed
Deploy mindnet to llm-node / deploy (push) Failing after 2s

This commit is contained in:
Lars 2025-09-09 16:31:00 +02:00
parent 9362064ad1
commit 43a8c80508

View File

@ -2,19 +2,24 @@
# -*- coding: utf-8 -*- # -*- coding: utf-8 -*-
""" """
Script: scripts/import_markdown.py Markdown Qdrant (Notes, Chunks, Edges) Script: scripts/import_markdown.py Markdown Qdrant (Notes, Chunks, Edges)
Version: 3.4.1 Version: 3.5.0
Datum: 2025-09-09 Datum: 2025-09-09
Kurzbeschreibung Kurzbeschreibung
---------------- ----------------
- Liest Markdown-Dateien ein, erzeugt Notes/Chunks/Edges **idempotent**. - Liest Markdown-Dateien ein, erzeugt Notes/Chunks/Edges **idempotent**.
- Change-Detection (nur **Inhalte**, keine FS-Zeitstempel) konfigurierbar: - **Änderungserkennung** (nur Inhalte, keine FS-Zeitstempel) konfigurierbar:
* ``--hash-mode``: body | frontmatter | body+frontmatter | full (Alias) * ``--hash-mode``: body | frontmatter | body+frontmatter | full (Alias)
- Env: ``MINDNET_HASH_MODE`` **oder** ``MINDNET_HASH_COMPARE`` (Body/Frontmatter/Full) - Env: ``MINDNET_HASH_MODE`` **oder** ``MINDNET_HASH_COMPARE`` (Body/Frontmatter/Full)
* ``--hash-normalize``: canonical | none (Default: canonical) * ``--hash-normalize``: canonical | none (Default: canonical)
* ``--hash-source``: parsed (Default) | raw * ``--hash-source``: parsed (Default) | raw
- "raw" hasht den **ungeparsten** Body aus der Datei (Frontmatter vorher entfernt). - "raw" hasht den **ungeparsten** Body (Frontmatter via Regex entfernt).
- Optionales Diff-Logging: ``--debug-hash-diff`` zeigt bei Bedarf einen kompakten Diff. * **NEU**: ``--compare-text`` (oder ENV ``MINDNET_COMPARE_TEXT=true``)
- Zusätzlich zum Hash wird der *parsed* ``fulltext`` direkt verglichen.
- Erkennt Änderungen auch dann, wenn eine Normalisierung Unterschiede glättet.
* **NEU**: Signaturabgleich:
- Falls sich die Signatur (z. B. bodyfull, parsedraw, canonicalnone) zwischen Alt/Neu unterscheidet,
gilt die Note als **geändert** (Einmal-Update, um neue Signatur zu persistieren).
ENV / Qdrant ENV / Qdrant
------------ ------------
@ -22,15 +27,22 @@ ENV / Qdrant
- COLLECTION_PREFIX (Default: mindnet) - COLLECTION_PREFIX (Default: mindnet)
- VECTOR_DIM (Default: 384) - VECTOR_DIM (Default: 384)
- MINDNET_NOTE_SCOPE_REFS: true|false (Default: false) - MINDNET_NOTE_SCOPE_REFS: true|false (Default: false)
- MINDNET_COMPARE_TEXT: true|false (Default: false)
Aufruf Aufruf-Beispiele
------ ----------------
# Standard (nur Body, kanonisiert, parsed-Quelle)
python3 -m scripts.import_markdown --vault ./vault python3 -m scripts.import_markdown --vault ./vault
python3 -m scripts.import_markdown --vault ./vault --apply
python3 -m scripts.import_markdown --vault ./vault --apply --hash-source raw --hash-normalize none
MINDNET_HASH_COMPARE=Full python3 -m scripts.import_markdown --vault ./vault --apply
"""
# Sensibel (jede Kleinigkeit): raw-Quelle + keine Normalisierung
python3 -m scripts.import_markdown --vault ./vault --apply --hash-source raw --hash-normalize none
# Full-Vergleich (Body+Frontmatter)
MINDNET_HASH_COMPARE=Full python3 -m scripts.import_markdown --vault ./vault --apply
# Zusätzlich Body-Text direkt vergleichen (maximale Sicherheit)
python3 -m scripts.import_markdown --vault ./vault --apply --compare-text
"""
from __future__ import annotations from __future__ import annotations
import argparse import argparse
import difflib import difflib
@ -143,11 +155,14 @@ def main() -> None:
help="(Optional) erzeugt zusätzlich references:note (Default: aus)") help="(Optional) erzeugt zusätzlich references:note (Default: aus)")
ap.add_argument("--debug-hash-diff", action="store_true", ap.add_argument("--debug-hash-diff", action="store_true",
help="Zeigt bei Bedarf einen kurzen Diff zwischen altem und neuem Body") help="Zeigt bei Bedarf einen kurzen Diff zwischen altem und neuem Body")
ap.add_argument("--compare-text", action="store_true",
help="Parsed fulltext zusätzlich direkt vergleichen (über Hash hinaus)")
args = ap.parse_args() args = ap.parse_args()
note_scope_refs_env = (os.environ.get("MINDNET_NOTE_SCOPE_REFS", "false").strip().lower() == "true") note_scope_refs_env = (os.environ.get("MINDNET_NOTE_SCOPE_REFS", "false").strip().lower() == "true")
note_scope_refs = args.note_sCope_refs if hasattr(args, "note_sCope_refs") else args.note_scope_refs # defensive note_scope_refs = args.note_scope_refs or note_scope_refs_env
note_scope_refs = note_scope_refs or note_scope_refs_env
compare_text = args.compare_text or (os.environ.get("MINDNET_COMPARE_TEXT", "false").strip().lower() == "true")
cfg = QdrantConfig.from_env() cfg = QdrantConfig.from_env()
client = get_client(cfg) client = get_client(cfg)
@ -200,18 +215,36 @@ def main() -> None:
# Change-Detection (nur Inhalte, keine FS-Timestamps) # Change-Detection (nur Inhalte, keine FS-Timestamps)
old_payload = None if args.force_replace else fetch_existing_note_payload(client, cfg.prefix, note_id) old_payload = None if args.force_replace else fetch_existing_note_payload(client, cfg.prefix, note_id)
old_hash = None if not old_payload else old_payload.get("hash_fulltext") old_hash = None if not old_payload else old_payload.get("hash_fulltext")
old_sig = (old_payload or {}).get("hash_signature")
new_hash = note_pl.get("hash_fulltext") new_hash = note_pl.get("hash_fulltext")
changed = args.force_replace or (old_hash != new_hash) new_sig = note_pl.get("hash_signature")
# 1) Signaturwechsel → als geändert behandeln (Einmal-Update)
sig_changed = bool(old_sig) and bool(new_sig) and (old_sig.split(":")[:3] != new_sig.split(":")[:3])
# 2) Hash-Vergleich
hash_changed = (old_hash != new_hash)
# 3) Optional: Parsed-Text direkt vergleichen (zusätzlich)
text_changed = False
if compare_text:
old_text = (old_payload or {}).get("fulltext") or ""
new_text = note_pl.get("fulltext") or ""
text_changed = (old_text != new_text)
changed = args.force_replace or sig_changed or hash_changed or text_changed
# Optionales Debugging: kompakten Diff anzeigen # Optionales Debugging: kompakten Diff anzeigen
if args.debug_hash_diff: if args.debug_hash_diff:
old_text = (old_payload or {}).get("fulltext") or "" old_text = (old_payload or {}).get("fulltext") or ""
new_text = note_pl.get("fulltext") or "" new_text = note_pl.get("fulltext") or ""
# Wenn Hash gleich, aber Text verschieden → Hinweis auf Normalisierung/Quelle # Hinweis, wenn Hash gleich aber Text verschieden (oder Signaturwechsel)
if (old_hash == new_hash) and old_text != new_text: if (not hash_changed) and (old_text != new_text or sig_changed):
print(json.dumps({ print(json.dumps({
"debug": "hash_equal_but_text_differs", "debug": "hash_equal_but_text_or_signature_differs",
"note_id": note_id, "note_id": note_id,
"old_sig": old_sig,
"new_sig": new_sig,
"hash_mode": args.hash_mode or os.environ.get("MINDNET_HASH_MODE") or os.environ.get("MINDNET_HASH_COMPARE", "body"), "hash_mode": args.hash_mode or os.environ.get("MINDNET_HASH_MODE") or os.environ.get("MINDNET_HASH_COMPARE", "body"),
"hash_normalize": args.hash_normalize or os.environ.get("MINDNET_HASH_NORMALIZE", "canonical"), "hash_normalize": args.hash_normalize or os.environ.get("MINDNET_HASH_NORMALIZE", "canonical"),
"hash_source": args.hash_source or os.environ.get("MINDNET_HASH_SOURCE", "parsed"), "hash_source": args.hash_source or os.environ.get("MINDNET_HASH_SOURCE", "parsed"),
@ -257,6 +290,10 @@ def main() -> None:
"hash_mode": args.hash_mode or os.environ.get("MINDNET_HASH_MODE") or os.environ.get("MINDNET_HASH_COMPARE", "body"), "hash_mode": args.hash_mode or os.environ.get("MINDNET_HASH_MODE") or os.environ.get("MINDNET_HASH_COMPARE", "body"),
"hash_normalize": args.hash_normalize or os.environ.get("MINDNET_HASH_NORMALIZE", "canonical"), "hash_normalize": args.hash_normalize or os.environ.get("MINDNET_HASH_NORMALIZE", "canonical"),
"hash_source": args.hash_source or os.environ.get("MINDNET_HASH_SOURCE", "parsed"), "hash_source": args.hash_source or os.environ.get("MINDNET_HASH_SOURCE", "parsed"),
"hash_signature": note_pl.get("hash_signature"),
"sig_changed": sig_changed,
"hash_changed": hash_changed,
"text_changed": text_changed,
} }
print(json.dumps(summary, ensure_ascii=False)) print(json.dumps(summary, ensure_ascii=False))