mindnet/scripts/import_markdown.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Script: import_markdown.py — Markdown → Qdrant (Notes, Chunks, Edges)
Version: 3.1.0
Datum: 2025-09-09

Kurzbeschreibung
----------------
Liest Markdown-Dateien aus einem Vault ein und schreibt Notes, Chunks und Edges
idempotent nach Qdrant. Die Change-Detection basiert standardmäßig ausschließlich
auf dem **Body-Hash** (ENV: MINDNET_HASH_MODE), sodass reine Frontmatter-Änderungen
nicht unnötig Re-Imports auslösen (abwärtskompatibel).

Wichtige Kompatibilitäts-Erweiterung
------------------------------------
- Persistiert nun den **Volltext** der Note im Notes-Payload unter ``fulltext``.
  So kann der Export **verlustfrei** rekonstruiert werden, auch wenn die Chunk-
  Payloads (noch) kein ``text``-Feld enthalten.
- Speichert den ``path`` **relativ** zum Vault (führt absolute Pfade auf relative
  Pfade zurück), sodass Exporte zuverlässig unter ``--out`` landen.

Features
--------
- Deterministische IDs (UUIDv5) über qdrant_points.* (beibehalten)
- Idempotente Upserts für Notes/Chunks/Edges
- Optionale Vorreinigung pro geänderter Note: ``--purge-before-upsert``
- Robuste Edge-Erzeugung:
  * Chunk-Scope: belongs_to / prev / next / references (+ backlink)
  * Fallback-Mode: Note-Scope references, falls chunk_payloads keine references liefern
- Optionaler Note-Embedding-Vektor (``--embed-note``)

Qdrant / ENV
------------
- QDRANT_URL (oder QDRANT_HOST/QDRANT_PORT)
- QDRANT_API_KEY (optional)
- COLLECTION_PREFIX (Default: mindnet)
- VECTOR_DIM (Default: 384)
- MINDNET_HASH_MODE: "body" (Default) | "frontmatter" | "body+frontmatter"

Aufruf
------
    python3 -m scripts.import_markdown --vault ./vault
    python3 -m scripts.import_markdown --vault ./vault --apply
    python3 -m scripts.import_markdown --vault ./vault --apply --purge-before-upsert
    python3 -m scripts.import_markdown --vault ./vault --note-id 20250821-foo --apply
    # optionaler Note-Vektor
    python3 -m scripts.import_markdown --vault ./vault --apply --embed-note

Beispiele
---------
COLLECTION_PREFIX=mindnet QDRANT_URL=http://127.0.0.1:6333 \\
  python3 -m scripts.import_markdown --vault ./vault --apply --purge-before-upsert

Hinweise
--------
- Dry-Run (ohne --apply) zeigt Entscheidungen je Note als JSON (keine Schreibvorgänge).
- Backfill-Tools existieren weiterhin, werden aber nur als Reparatur eingesetzt.
"""

from __future__ import annotations

import argparse
import json
import os
import re
import sys
from typing import Dict, Iterable, List, Optional, Tuple

from dotenv import load_dotenv
import yaml
from qdrant_client.http import models as rest

# Core/Projekt-Module
from app.core.parser import (
    read_markdown,
    normalize_frontmatter,
    validate_required_frontmatter,
)
from app.core.note_payload import make_note_payload
from app.core.validate_note import validate_note_payload
from app.core.chunker import assemble_chunks
from app.core.chunk_payload import make_chunk_payloads
from app.core.embed import embed_texts, embed_one
from app.core.qdrant import QdrantConfig, get_client, ensure_collections
from app.core.qdrant_points import (
    points_for_chunks,
    points_for_note,
    points_for_edges,
    upsert_batch,
)
from app.core.edges import deriv_edges_for_note


# -----------------------------------------------------------------------------
# Hilfsfunktionen
# -----------------------------------------------------------------------------

def iter_md(root: str) -> List[str]:
    """Sammelt alle .md-Dateien unterhalb von root, filtert systemische Ordner."""
    out: List[str] = []
    for dirpath, _, filenames in os.walk(root):
        for fn in filenames:
            if not fn.lower().endswith(".md"):
                continue
            p = os.path.join(dirpath, fn)
            pn = p.replace("\\", "/")
            if any(ex in pn for ex in ["/.obsidian/", "/_backup_frontmatter/", "/_imported/"]):
                continue
            out.append(p)
    return sorted(out)


def collections(prefix: str) -> Tuple[str, str, str]:
    return f"{prefix}_notes", f"{prefix}_chunks", f"{prefix}_edges"


def fetch_existing_note_hash(client, prefix: str, note_id: str) -> Optional[str]:
    """Liest den bisherigen hash_fulltext aus der Notes-Collection (falls vorhanden)."""
    notes_col, _, _ = collections(prefix)
    f = rest.Filter(must=[rest.FieldCondition(
        key="note_id",
        match=rest.MatchValue(value=note_id),
    )])
    points, _ = client.scroll(
        collection_name=notes_col,
        scroll_filter=f,
        with_payload=True,
        with_vectors=False,
        limit=1,
    )
    if not points:
        return None
    pl = points[0].payload or {}
    return pl.get("hash_fulltext")


def purge_note_artifacts(client, prefix: str, note_id: str) -> None:
    """
    Löscht alle Chunks+Edges zu einer Note. Collections bleiben bestehen.
    """
    _, chunks_col, edges_col = collections(prefix)

    # Chunks der Note löschen
    f_chunks = rest.Filter(
        must=[rest.FieldCondition(key="note_id", match=rest.MatchValue(value=note_id))]
    )
    client.delete(collection_name=chunks_col, points_selector=f_chunks, wait=True)

    # Edges der Note löschen (Chunk- und Note-Scope)
    should = [
        # Chunk-IDs (note_id#c...)
        rest.FieldCondition(key="source_id", match=rest.MatchText(text=f"{note_id}#")),
        rest.FieldCondition(key="target_id", match=rest.MatchText(text=f"{note_id}#")),
        # Note-Scope (source/target == note_id)
        rest.FieldCondition(key="source_id", match=rest.MatchValue(value=note_id)),
        rest.FieldCondition(key="target_id", match=rest.MatchValue(value=note_id)),
    ]
    f_edges = rest.Filter(should=should)
    client.delete(collection_name=edges_col, points_selector=f_edges, wait=True)


def has_chunk_level_refs(chunk_payloads: List[Dict]) -> bool:
    return any(isinstance(ch.get("references"), list) and ch["references"] for ch in chunk_payloads)


def fallback_note_level_edges(note_meta: Dict, chunk_payloads: List[Dict]) -> List[Dict]:
    """
    Falls chunk_payloads keine 'references' enthalten, erzeugen wir:
      - belongs_to + prev/next aus chunk_payloads
      - references/backlink auf Note-Scope aus note_meta['references'] (falls vorhanden)
    """
    edges: List[Dict] = []

    # belongs_to + prev/next (Chunk-Scope)
    for ch in chunk_payloads:
        src = ch["id"]
        edges.append({"src_id": src, "dst_id": note_meta["id"], "edge_type": "belongs_to", "scope": "chunk"})
        nb = ch.get("neighbors") or {}
        prev_id = nb.get("prev")
        next_id = nb.get("next")
        if prev_id:
            edges.append({"src_id": src, "dst_id": prev_id, "edge_type": "next", "scope": "chunk"})
            edges.append({"src_id": prev_id, "dst_id": src, "edge_type": "prev", "scope": "chunk"})
        if next_id:
            edges.append({"src_id": src, "dst_id": next_id, "edge_type": "next", "scope": "chunk"})
            edges.append({"src_id": next_id, "dst_id": src, "edge_type": "prev", "scope": "chunk"})

    # Note-Scope references/backlink
    for tid in (note_meta.get("references") or []):
        edges.append({"src_id": note_meta["id"], "dst_id": tid, "edge_type": "references", "scope": "note"})
        edges.append({"src_id": tid, "dst_id": note_meta["id"], "edge_type": "backlink", "scope": "note"})

    # Dedupe
    uniq = {}
    for e in edges:
        key = (e["src_id"], e["dst_id"], e["edge_type"], e.get("scope", ""))
        uniq[key] = e
    return list(uniq.values())


def _normalize_rel_path(abs_path: str, vault_root: str) -> str:
    """Gibt einen **relativen** Pfad zurück, normalisiert auf forward slashes."""
    try:
        rel = os.path.relpath(abs_path, vault_root)
    except Exception:
        rel = abs_path  # Fallback
    return rel.replace("\\", "/").lstrip("/")


# -----------------------------------------------------------------------------
# Main
# -----------------------------------------------------------------------------

def main() -> None:
    load_dotenv()
    ap = argparse.ArgumentParser()
    ap.add_argument("--vault", required=True, help="Pfad zum Obsidian-Vault (Root-Ordner)")
    ap.add_argument("--apply", action="store_true", help="Schreibt in Qdrant; ohne Flag nur Dry-Run")
    ap.add_argument("--purge-before-upsert", action="store_true",
                    help="Vor Upsert Chunks & Edges der GEÄNDERTEN Note löschen")
    ap.add_argument("--note-id", help="Nur eine bestimmte Note-ID verarbeiten")
    ap.add_argument("--embed-note", action="store_true", help="Optional: Note-Volltext einbetten")
    ap.add_argument("--force-replace", action="store_true",
                    help="Änderungserkennung ignorieren und immer upserten (+ optional Purge)")
    args = ap.parse_args()

    # Qdrant
    cfg = QdrantConfig.from_env()
    client = get_client(cfg)
    ensure_collections(client, cfg.prefix, cfg.dim)
    notes_col, chunks_col, edges_col = collections(cfg.prefix)

    root = os.path.abspath(args.vault)
    files = iter_md(root)
    if not files:
        print("Keine Markdown-Dateien gefunden.", file=sys.stderr)
        sys.exit(2)

    processed = 0
    for path in files:
        parsed = read_markdown(path)
        fm = normalize_frontmatter(parsed.frontmatter)

        # Pflichtfelder prüfen (z. B. id, title) – bei Fehler: Note überspringen
        try:
            validate_required_frontmatter(fm)
        except Exception:
            continue

        if args.note_id and fm.get("id") != args.note_id:
            continue

        processed += 1

        # Note-Payload & Validierung
        note_pl = make_note_payload(parsed, vault_root=root)
        # **Kompat-Erweiterung**: Volltext & relativer Pfad sicherstellen
        if "fulltext" not in (note_pl or {}):
            note_pl["fulltext"] = parsed.body or ""
        # Pfad ggf. relativieren (falls make_note_payload es noch nicht tut)
        if note_pl.get("path"):
            note_pl["path"] = _normalize_rel_path(os.path.join(root, note_pl["path"])
                                                  if not os.path.isabs(note_pl["path"]) else note_pl["path"],
                                                  root)
        else:
            # fallback: relativer Pfad aus parsed.path
            note_pl["path"] = _normalize_rel_path(parsed.path, root)

        validate_note_payload(note_pl)
        note_id = note_pl["note_id"]

        # Change-Detection (Body-basiert per hash_fulltext)
        new_hash = note_pl.get("hash_fulltext")
        old_hash = None
        if not args.force_replace:
            old_hash = fetch_existing_note_hash(client, cfg.prefix, note_id)
        changed = args.force_replace or (old_hash != new_hash)

        # Chunks + Embeddings
        chunks = assemble_chunks(fm["id"], parsed.body, fm.get("type", "concept"))
        chunk_pls = make_chunk_payloads(fm, note_pl["path"], chunks)
        vecs = embed_texts([c.text for c in chunks])

        # Optional: Note-Vektor
        note_vec = embed_one(parsed.body) if args.embed_note else None

        # Edges
        edges = deriv_edges_for_note(fm, chunk_pls)
        if not edges or (not has_chunk_level_refs(chunk_pls) and (fm.get("references") or note_pl.get("references"))):
            # Ergänze fm['references'] falls noch nicht im fm vorhanden
            if "references" not in fm and "references" in note_pl:
                fm["references"] = note_pl["references"]
            edges = fallback_note_level_edges(fm, chunk_pls)

        # Zusammenfassung
        summary = {
            "note_id": note_id,
            "title": fm.get("title"),
            "chunks": len(chunk_pls),
            "edges": len(edges),
            "changed": changed,
            "decision": ("apply" if args.apply and changed else
                         "apply-skip-unchanged" if args.apply and not changed else
                         "dry-run"),
            "path": note_pl["path"],
        }
        print(json.dumps(summary, ensure_ascii=False))

        # Dry-Run?
        if not args.apply:
            continue

        # Optionaler Purge NUR für geänderte Notes
        if changed and args.purge_before_upsert:
            purge_note_artifacts(client, cfg.prefix, note_id)

        # Upserts: Notes / Chunks / Edges
        notes_name, note_pts = points_for_note(cfg.prefix, note_pl, note_vec, cfg.dim)
        upsert_batch(client, notes_name, note_pts)

        chunks_name, chunk_pts = points_for_chunks(cfg.prefix, chunk_pls, vecs)
        upsert_batch(client, chunks_name, chunk_pts)

        edges_name, edge_pts = points_for_edges(cfg.prefix, edges)
        upsert_batch(client, edges_name, edge_pts)

    print(f"Done. Processed notes: {processed}")


if __name__ == "__main__":
    main()