feat: index semantic records in qdrant

harness/tht/cli/vector_cmd.py

@@ -2,9 +2,15 @@ from pathlib import Path
 
 import typer
 
-from tht.cli._guards import has_vector_write_rest, require_server_profile, require_vector_write_allowed
+from tht.cli._guards import (
+    has_vector_write_rest,
+    require_server_profile,
+    require_vector_write_allowed,
+)
 from tht.cli.config_cmd import CONFIG_OPT
 from tht.cli.schema_cmd import _load_config_or_exit, annotations_path, physical_path
+from tht.ports.vector import VectorWriteRecord
+from tht.vectorstore.store import SyncStats, content_hash
 
 vector_app = typer.Typer(help="Indice semantico pgvector (derivato, rigenerabile)")
 
@@ -69,6 +75,31 @@ def open_searcher(cfg):
     return AdapterSearcher()
 
 
+def sync_canonical_records(collection, records, *, store, embedder):
+    kinds = sorted({record.kind for record in records})
+    existing = store.existing_hashes(collection, kinds)
+    pending = []
+    stats = SyncStats()
+    changed = []
+    for record in records:
+        hashed = content_hash(record.content)
+        current = existing.get(record.id)
+        if current == hashed:
+            stats.unchanged += 1
+            continue
+        changed.append((record, hashed, current is None))
+    if changed:
+        embeddings = embedder.embed_documents([record.content for record, *_ in changed])
+        for (record, hashed, is_added), embedding in zip(changed, embeddings, strict=True):
+            pending.append(VectorWriteRecord(record=record, embedding=embedding, content_hash=hashed))
+            if is_added:
+                stats.added += 1
+            else:
+                stats.updated += 1
+        store.upsert(collection, pending)
+    return stats
+
+
 def _print_stats(stats) -> None:
     typer.secho(
         f"OK: {stats.added} nuovi, {stats.updated} aggiornati, "
@@ -88,7 +119,6 @@ def init_cmd(
     from sqlalchemy.exc import OperationalError
 
     from tht.vectorstore.embeddings import EmbeddingsError
-
     from tht.vectorstore.reader import ALL_TABLES
 
     cfg = _load_config_or_exit(config)
@@ -131,8 +161,12 @@ def index_schema_cmd(config: Path = CONFIG_OPT) -> None:
     physical = PhysicalSchema.from_yaml(phys_file)
     annotations = Annotations.from_yaml(annotations_path(cfg))
     records = schema_records(physical, annotations)
-    store = open_store(cfg, "schema_records")
-    stats = store.sync(
-        records, make_embedder(cfg.embeddings), kinds={"schema_table", "schema_column"}
+    from tht.adapters.factory import build_vector_store
+
+    stats = sync_canonical_records(
+        "schema_records",
+        records,
+        store=build_vector_store(cfg, require_write=True),
+        embedder=make_embedder(cfg.embeddings),
     )
     _print_stats(stats)