From 23d06701266c761decbade97f25f5367e6ecb9b8 Mon Sep 17 00:00:00 2001 From: Lars Date: Mon, 17 Nov 2025 15:39:18 +0100 Subject: [PATCH] app/core/derive_edges.py aktualisiert --- app/core/derive_edges.py | 613 +++++++++++++-------------------------- 1 file changed, 208 insertions(+), 405 deletions(-) diff --git a/app/core/derive_edges.py b/app/core/derive_edges.py index 0465cb4..c1cbba9 100644 --- a/app/core/derive_edges.py +++ b/app/core/derive_edges.py @@ -1,438 +1,241 @@ #!/usr/bin/env python3 # -*- coding: utf-8 -*- -""" -Modul: app/core/derive_edges.py -Version: 2.2.0 (V2-superset mit "typed inline relations" + Obsidian-Callouts) - -Zweck ------ -Bewahrt die bestehende Edgelogik (belongs_to, prev/next, references, backlink) -und ergänzt: -- Typ-Default-Kanten gemäß config/types.yaml (edge_defaults je Notiztyp) -- **Explizite, getypte Inline-Relationen** direkt im Chunk-Text: - * [[rel:depends_on | Target Title]] - * [[rel:related_to Target Title]] -- **Obsidian-Callouts** zur Pflege von Kanten im Markdown: - * > [!edge] related_to: [[Vector DB Basics]] - * Mehrere Zeilen im Callout werden unterstützt (alle Zeilen beginnen mit '>'). - -Konfiguration -------------- -- ENV MINDNET_TYPES_FILE (Default: ./config/types.yaml) - -Hinweis -------- -Diese Implementierung ist rückwärtskompatibel zur bisherigen Signatur. -""" from __future__ import annotations -import os import re -from typing import Iterable, List, Optional, Tuple, Set +from typing import Dict, Iterable, List, Optional, Tuple -try: - import yaml # optional, nur für types.yaml -except Exception: # pragma: no cover - yaml = None -# ---------------------------- Utilities ------------------------------------ - -def _get(d: dict, *keys, default=None): - for k in keys: - if k in d and d[k] is not None: - return d[k] - return default - -def _chunk_text_for_refs(chunk: dict) -> str: - # bevorzugt 'window' → dann 'text' → 'content' → 'raw' - return ( - _get(chunk, "window") - or _get(chunk, "text") - or _get(chunk, "content") - or _get(chunk, "raw") - or "" - ) - -def _dedupe_seq(seq: Iterable[str]) -> List[str]: - seen: Set[str] = set() - out: List[str] = [] - for s in seq: - if s not in seen: - seen.add(s) - out.append(s) - return out - -def _edge(kind: str, scope: str, source_id: str, target_id: str, note_id: str, extra: Optional[dict] = None) -> dict: - pl = { +# ------------------------------ +# Edge payload helper +# ------------------------------ +def _edge_payload( + *, + note_id: str, + chunk_id: Optional[str], + kind: str, + source_id: str, + target_id: str, + rule_id: str, + scope: str = "chunk", + confidence: Optional[float] = None, +) -> Dict: + p = { + "note_id": note_id, + "chunk_id": chunk_id, "kind": kind, - "relation": kind, # v2 Feld (alias) - "scope": scope, # "chunk" | "note" + "scope": scope, "source_id": source_id, "target_id": target_id, - "note_id": note_id, # Träger/Quelle der Kante (aktuelle Note) + "rule_id": rule_id, } - if extra: - pl.update(extra) - return pl + if confidence is not None: + p["confidence"] = float(confidence) + return p -def _mk_edge_id(kind: str, s: str, t: str, scope: str, rule_id: Optional[str] = None) -> str: - base = f"{kind}:{s}->{t}#{scope}" - if rule_id: - base += f"|{rule_id}" - try: - import hashlib - return hashlib.blake2s(base.encode("utf-8"), digest_size=12).hexdigest() - except Exception: # pragma: no cover - return base -# ---------------------- Typen-Registry (types.yaml) ------------------------ +# ------------------------------ +# Inline [[wikilink]] parser +# ------------------------------ +_WIKILINK_RE = re.compile(r"\[\[([^\]]+)\]\]") -def _env(n: str, default: Optional[str] = None) -> str: - v = os.getenv(n) - return v if v is not None else (default or "" ) +def _iter_wikilinks(text: str) -> Iterable[str]: + for m in _WIKILINK_RE.finditer(text): + yield m.group(1).strip() -def _load_types_registry() -> dict: - """Lädt die YAML-Registry aus MINDNET_TYPES_FILE oder ./config/types.yaml""" - p = _env("MINDNET_TYPES_FILE", "./config/types.yaml") - if not os.path.isfile(p) or yaml is None: - return {} - try: - with open(p, "r", encoding="utf-8") as f: - data = yaml.safe_load(f) or {} - return data - except Exception: - return {} -def _get_types_map(reg: dict) -> dict: - if isinstance(reg, dict) and isinstance(reg.get("types"), dict): - return reg["types"] - return reg if isinstance(reg, dict) else {} +# ------------------------------ +# Callout parser +# Syntax: +# > [!edge] related_to: [[Vector DB Basics]] [[Embeddings 101]] +# Mehrere Ziele pro Zeile erlaubt. +# ------------------------------ +_CALLOUT_RE = re.compile( + r"^\s*>\s*\[!edge\]\s*([a-z_]+)\s*:\s*(.+)$", + flags=re.IGNORECASE, +) -def _edge_defaults_for(note_type: Optional[str], reg: dict) -> List[str]: - """ - Liefert die edge_defaults-Liste für den gegebenen Notiztyp. - Fallback-Reihenfolge: - 1) reg['types'][note_type]['edge_defaults'] - 2) reg['defaults']['edge_defaults'] (oder 'default'/'global') - 3) [] - """ - types_map = _get_types_map(reg) - # 1) exakter Typ - if note_type and isinstance(types_map, dict): - t = types_map.get(note_type) - if isinstance(t, dict) and isinstance(t.get("edge_defaults"), list): - return [str(x) for x in t["edge_defaults"] if isinstance(x, str)] - # 2) Fallback - for key in ("defaults", "default", "global"): - v = reg.get(key) - if isinstance(v, dict) and isinstance(v.get("edge_defaults"), list): - return [str(x) for x in v["edge_defaults"] if isinstance(x, str)] - # 3) leer - return [] +def _parse_callout_line(line: str) -> Optional[Tuple[str, List[str]]]: + m = _CALLOUT_RE.match(line) + if not m: + return None + relation = m.group(1).strip().lower() + rhs = m.group(2) + targets = [t.strip() for t in _WIKILINK_RE.findall(rhs) if t.strip()] + if not targets: + return None + return (relation, targets) -# ------------------------ Parser für Links --------------------------------- -# Normale Wikilinks (Fallback) -_WIKILINK_RE = re.compile(r"\[\[(?:[^\|\]]+\|)?([a-zA-Z0-9_\-#:. ]+)\]\]") +# ------------------------------ +# Defaults aus types.yaml anwenden (wenn konfiguriert) +# types_cfg Beispiel: +# { "types": { "project": { "edge_defaults": ["references","depends_on"] }, ... } } +# ------------------------------ +def _edge_defaults_for_type(types_cfg: Dict, note_type: str) -> List[str]: + tdef = (types_cfg or {}).get("types", {}).get(note_type, {}) + vals = tdef.get("edge_defaults") or [] + return [str(v).strip().lower() for v in vals if str(v).strip()] -# Getypte Inline-Relationen: -# [[rel:depends_on | Target]] -# [[rel:related_to Target]] -_REL_PIPE = re.compile(r"\[\[\s*rel:(?P[a-z_]+)\s*\|\s*(?P[^\]]+?)\s*\]\]", re.IGNORECASE) -_REL_SPACE = re.compile(r"\[\[\s*rel:(?P[a-z_]+)\s+(?P[^\]]+?)\s*\]\]", re.IGNORECASE) -def _extract_typed_relations(text: str) -> Tuple[List[Tuple[str,str]], str]: - """ - Gibt Liste (kind, target) zurück und den Text mit entfernten getypten Relation-Links, - damit die generische Wikilink-Erkennung sie nicht doppelt zählt. - """ - pairs: List[Tuple[str,str]] = [] - def _collect(m): - k = (m.group("kind") or "").strip().lower() - t = (m.group("target") or "").strip() - if k and t: - pairs.append((k, t)) - return "" # löschen - text = _REL_PIPE.sub(_collect, text) - text = _REL_SPACE.sub(_collect, text) - return pairs, text +# ------------------------------ +# Hauptfunktion: Edges ableiten +# Erwartete Inputs: +# note: { "note_id","title","type","text", ... } +# chunks: [ { "chunk_id","note_id","index","ord","text","window", ... }, ... ] +# types_cfg: geladene types.yaml als Dict +# ------------------------------ +def derive_edges( + note: Dict, + chunks: List[Dict], + types_cfg: Optional[Dict] = None, +) -> List[Dict]: + note_id = note.get("note_id") or note.get("id") + note_title = note.get("title") or "" + note_type = (note.get("type") or "").strip().lower() + text = note.get("text") or "" -# ---- Obsidian Callout Parser ---------------------------------------------- -# Callout-Start erkennt Zeilen wie: > [!edge] ... (case-insensitive) -_CALLOUT_START = re.compile(r"^\s*>\s*\[!edge\]\s*(.*)$", re.IGNORECASE) + edges: List[Dict] = [] -# Innerhalb von Callouts erwarten wir je Zeile Muster wie: -# related_to: [[Vector DB Basics]] -# depends_on: [[A]], [[B]] -# similar_to: Qdrant Vektordatenbank -_REL_LINE = re.compile(r"^(?P[a-z_]+)\s*:\s*(?P.+?)\s*$", re.IGNORECASE) + # 1) Sequenz-Edges je Note: belongs_to / next / prev + for i, ch in enumerate(chunks): + cid = ch.get("chunk_id") + # belongs_to + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=cid, + kind="belongs_to", + source_id=cid, + target_id=note_id, + rule_id="structure:v1:belongs_to", + scope="chunk", + ) + ) + # next/prev + if i + 1 < len(chunks): + nxt = chunks[i + 1]["chunk_id"] + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=cid, + kind="next", + source_id=cid, + target_id=nxt, + rule_id="structure:v1:next", + scope="chunk", + ) + ) + if i - 1 >= 0: + prv = chunks[i - 1]["chunk_id"] + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=cid, + kind="prev", + source_id=cid, + target_id=prv, + rule_id="structure:v1:prev", + scope="chunk", + ) + ) -_WIKILINKS_IN_LINE = re.compile(r"\[\[([^\]]+)\]\]") - -def _extract_callout_relations(text: str) -> Tuple[List[Tuple[str,str]], str]: - """ - Findet Obsidian-Callouts vom Typ [!edge] und extrahiert (kind, target). - Entfernt den gesamten Callout-Block aus dem Text, damit Wikilinks daraus - nicht zusätzlich als "references" gezählt werden. - """ - if not text: - return [], text - - lines = text.splitlines() - out_pairs: List[Tuple[str,str]] = [] - keep_lines: List[str] = [] - - i = 0 - while i < len(lines): - m = _CALLOUT_START.match(lines[i]) - if not m: - keep_lines.append(lines[i]) - i += 1 - continue - - # Wir sind in einem Callout-Block; erste Zeile nach dem Marker: - # Rest dieser Zeile nach [!edge] mitnehmen - block_lines: List[str] = [] - first_rest = m.group(1) or "" - if first_rest.strip(): - block_lines.append(first_rest) - - # Folgezeilen sind Teil des Callouts, solange sie weiterhin mit '>' beginnen - i += 1 - while i < len(lines) and lines[i].lstrip().startswith('>'): - # Entferne führendes '>' und evtl. Leerzeichen - block_lines.append(lines[i].lstrip()[1:].lstrip()) - i += 1 - - # Parse jede Blockzeile eigenständig - for bl in block_lines: - mrel = _REL_LINE.match(bl) - if not mrel: - continue - kind = (mrel.group("kind") or "").strip().lower() - targets = mrel.group("targets") or "" - # Wikilinks bevorzugt - found = _WIKILINKS_IN_LINE.findall(targets) - if found: - for t in found: - t = t.strip() - if t: - out_pairs.append((kind, t)) - else: - # Fallback: Split per ',' oder ';' - for raw in re.split(r"[,;]", targets): - t = raw.strip() - if t: - out_pairs.append((kind, t)) - # Wichtig: Callout wird NICHT in keep_lines übernommen (entfernt) - continue - - remainder = "\n".join(keep_lines) - return out_pairs, remainder - -def _extract_wikilinks(text: str) -> List[str]: - ids: List[str] = [] - for m in _WIKILINK_RE.finditer(text or ""): - ids.append(m.group(1).strip()) - return ids - -# --------------------------- Hauptfunktion --------------------------------- - -def build_edges_for_note( - note_id: str, - chunks: List[dict], - note_level_references: Optional[List[str]] = None, - include_note_scope_refs: bool = False, -) -> List[dict]: - """ - Erzeugt Kanten für eine Note. - - - belongs_to: für jeden Chunk (chunk -> note) - - next / prev: zwischen aufeinanderfolgenden Chunks - - references: pro Chunk aus window/text (via Wikilinks) - - typed inline relations: [[rel:KIND | Target]] oder [[rel:KIND Target]] - - Obsidian Callouts: > [!edge] KIND: [[Target]] - - optional note-scope references/backlinks: dedupliziert über alle Chunk-Funde + note_level_references - - typenbasierte Default-Kanten (edge_defaults) je gefundener Referenz - """ - edges: List[dict] = [] - - # --- 0) Note-Typ ermitteln (aus erstem Chunk erwartet) --- - note_type = None - if chunks: - note_type = _get(chunks[0], "type") - - # --- 1) belongs_to --- + # 2) Inline-Wikilinks ([[Title]]) => references (note-scope + chunk-scope) + # - chunk-scope: pro Chunk in dessen Text/Window + # - note-scope: Gesamttext der Note + # Hinweis: target_id wird hier als Titel gespeichert; später kann ein Resolver auf note_id mappen. + # chunk-scope for ch in chunks: - cid = _get(ch, "chunk_id", "id") - if not cid: - continue - edges.append(_edge("belongs_to", "chunk", cid, note_id, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id("belongs_to", cid, note_id, "chunk", "structure:belongs_to:v1"), - "provenance": "rule", - "rule_id": "structure:belongs_to:v1", - "confidence": 1.0, - })) + cid = ch.get("chunk_id") + body = (ch.get("window") or ch.get("text") or "") + touched = False + for tgt in _iter_wikilinks(body): + touched = True + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=cid, + kind="references", + source_id=cid, + target_id=tgt, # Titel + rule_id="inline:rel:v1:references", + scope="chunk", + confidence=0.8, + ) + ) + # Optional: wenn in einem Chunk Wikilinks vorkamen, kannst du (später) einen counter o. ä. setzen. + _ = touched - # --- 2) next/prev --- - for i in range(len(chunks) - 1): - a, b = chunks[i], chunks[i + 1] - a_id = _get(a, "chunk_id", "id") - b_id = _get(b, "chunk_id", "id") - if not a_id or not b_id: - continue - edges.append(_edge("next", "chunk", a_id, b_id, note_id, { - "chunk_id": a_id, - "edge_id": _mk_edge_id("next", a_id, b_id, "chunk", "structure:order:v1"), - "provenance": "rule", - "rule_id": "structure:order:v1", - "confidence": 0.95, - })) - edges.append(_edge("prev", "chunk", b_id, a_id, note_id, { - "chunk_id": b_id, - "edge_id": _mk_edge_id("prev", b_id, a_id, "chunk", "structure:order:v1"), - "provenance": "rule", - "rule_id": "structure:order:v1", - "confidence": 0.95, - })) - - # --- 3) references (chunk-scope) + inline relations + callouts + abgeleitete Relationen je Ref --- - reg = _load_types_registry() - defaults = _edge_defaults_for(note_type, reg) - refs_all: List[str] = [] + # note-scope (Gesamttext) + for tgt in _iter_wikilinks(text): + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=None, + kind="references", + source_id=note_id, + target_id=tgt, # Titel + rule_id="explicit:ref:v1:wikilink", + scope="note", + confidence=0.8, + ) + ) + # 3) Callouts: + # > [!edge] related_to: [[A]] [[B]] + # ⇒ pro Ziel A/B je ein Edge mit rule_id="callout:edge:v1:" for ch in chunks: - cid = _get(ch, "chunk_id", "id") - if not cid: - continue - raw = _chunk_text_for_refs(ch) - - # a) typed inline relations zuerst extrahieren - typed, remainder = _extract_typed_relations(raw) - for kind, target in typed: - edges.append(_edge(kind, "chunk", cid, target, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id(kind, cid, target, "chunk", "inline:rel:v1"), - "provenance": "explicit", - "rule_id": "inline:rel:v1", - "confidence": 0.95, - })) - # symmetrische Relationen zusätzlich rückwärts - if kind in {"related_to", "similar_to"}: - edges.append(_edge(kind, "chunk", target, cid, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id(kind, target, cid, "chunk", "inline:rel:v1"), - "provenance": "explicit", - "rule_id": "inline:rel:v1", - "confidence": 0.95, - })) - - # b) Obsidian Callouts extrahieren (und aus remainder entfernen) - call_pairs, remainder2 = _extract_callout_relations(remainder) - for kind, target in call_pairs: - k = (kind or "").strip().lower() - if not k or not target: + cid = ch.get("chunk_id") + body = (ch.get("window") or ch.get("text") or "") + for line in body.splitlines(): + parsed = _parse_callout_line(line) + if not parsed: continue - edges.append(_edge(k, "chunk", cid, target, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id(k, cid, target, "chunk", "callout:edge:v1"), - "provenance": "explicit", - "rule_id": "callout:edge:v1", - "confidence": 0.95, - })) - if k in {"related_to", "similar_to"}: - edges.append(_edge(k, "chunk", target, cid, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id(k, target, cid, "chunk", "callout:edge:v1"), - "provenance": "explicit", - "rule_id": "callout:edge:v1", - "confidence": 0.95, - })) + relation, targets = parsed + # normalize relation name + relation = relation.lower() + # einheitliches Rule-Tagging für Callouts: + rule_tag = f"callout:edge:v1:{relation}" + for tgt in targets: + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=cid, + kind=relation, + source_id=cid, + target_id=tgt, # Titel + rule_id=rule_tag, + scope="chunk", + confidence=0.7, + ) + ) - # c) generische Wikilinks (remainder2) → "references" - refs = _extract_wikilinks(remainder2) - for r in refs: - # reale Referenz (wie bisher) - edges.append(_edge("references", "chunk", cid, r, note_id, { - "chunk_id": cid, - "ref_text": r, - "edge_id": _mk_edge_id("references", cid, r, "chunk", "explicit:wikilink:v1"), - "provenance": "explicit", - "rule_id": "explicit:wikilink:v1", - "confidence": 1.0, - })) - # abgeleitete Kanten je default-Relation + # 4) Ableitungs-Edges (edge_defaults) aus types.yaml + # Beispiel: project -> ["references","depends_on"] + # Für jede Chunk-Einheit eine schwach gewichtete Default-Beziehung gegen den Note-Titel, + # damit es als Navigationskanten funktioniert, bis ein Resolver Titeleindeutigkeit herstellt. + defaults = _edge_defaults_for_type(types_cfg or {}, note_type) + if defaults: + rule_prefix = f"edge_defaults:{note_type}" + for ch in chunks: + cid = ch.get("chunk_id") for rel in defaults: - if rel == "references": - continue # doppelt vermeiden - edges.append(_edge(rel, "chunk", cid, r, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id(rel, cid, r, "chunk", f"edge_defaults:{note_type}:{rel}:v1"), - "provenance": "rule", - "rule_id": f"edge_defaults:{note_type}:{rel}:v1", - "confidence": 0.7, - })) - # symmetrisch? - if rel in {"related_to", "similar_to"}: - edges.append(_edge(rel, "chunk", r, cid, note_id, { - "chunk_id": cid, - "edge_id": _mk_edge_id(rel, r, cid, "chunk", f"edge_defaults:{note_type}:{rel}:v1"), - "provenance": "rule", - "rule_id": f"edge_defaults:{note_type}:{rel}:v1", - "confidence": 0.7, - })) - refs_all.extend(refs) + edges.append( + _edge_payload( + note_id=note_id, + chunk_id=cid, + kind=rel, + source_id=cid, + target_id=note_title or note_id, # weiche Zielmarke + rule_id=f"{rule_prefix}:{rel}", + scope="chunk", + confidence=0.7, + ) + ) - # --- 4) optional: note-scope references/backlinks (+ defaults) --- - if include_note_scope_refs: - refs_note = list(refs_all or []) - if note_level_references: - refs_note.extend([r for r in note_level_references if isinstance(r, str) and r]) - refs_note = _dedupe_seq(refs_note) - for r in refs_note: - # echte note-scope Referenz & Backlink (wie bisher) - edges.append(_edge("references", "note", note_id, r, note_id, { - "edge_id": _mk_edge_id("references", note_id, r, "note", "explicit:note_scope:v1"), - "provenance": "explicit", - "rule_id": "explicit:note_scope:v1", - "confidence": 1.0, - })) - edges.append(_edge("backlink", "note", r, note_id, note_id, { - "edge_id": _mk_edge_id("backlink", r, note_id, "note", "derived:backlink:v1"), - "provenance": "rule", - "rule_id": "derived:backlink:v1", - "confidence": 0.9, - })) - # und zusätzlich default-Relationen (note-scope) - for rel in defaults: - if rel == "references": - continue - edges.append(_edge(rel, "note", note_id, r, note_id, { - "edge_id": _mk_edge_id(rel, note_id, r, "note", f"edge_defaults:{note_type}:{rel}:v1"), - "provenance": "rule", - "rule_id": f"edge_defaults:{note_type}:{rel}:v1", - "confidence": 0.7, - })) - if rel in {"related_to", "similar_to"}: - edges.append(_edge(rel, "note", r, note_id, note_id, { - "edge_id": _mk_edge_id(rel, r, note_id, "note", f"edge_defaults:{note_type}:{rel}:v1"), - "provenance": "rule", - "rule_id": f"edge_defaults:{note_type}:{rel}:v1", - "confidence": 0.7, - })) - - # --- 5) Dedupe (Schlüssel: source_id, target_id, relation, rule_id) --- - seen: Set[Tuple[str,str,str,str]] = set() - out: List[dict] = [] + # 5) De-Duplizierung (idempotent): key = (source_id, target_id, kind, rule_id) + unique: Dict[Tuple[str, str, str, str], Dict] = {} for e in edges: - s = str(e.get("source_id") or "") - t = str(e.get("target_id") or "") - rel = str(e.get("relation") or e.get("kind") or "edge") - rule = str(e.get("rule_id") or "") - key = (s, t, rel, rule) - if key in seen: - continue - seen.add(key) - out.append(e) - return out + k = (e["source_id"], e["target_id"], e["kind"], e["rule_id"]) + unique[k] = e + return list(unique.values())