189 files

custom_components/maintenance_supporter/helpers/documents.py

@@ -18,7 +18,7 @@ import hashlib
 import logging
 import os
 from pathlib import Path
-from typing import Any
+from typing import TYPE_CHECKING, Any
 from uuid import uuid4
 
 from homeassistant.core import HomeAssistant
@@ -28,6 +28,9 @@ from homeassistant.util import dt as dt_util
 
 from ..const import DOMAIN, MAX_DOCS_PER_OBJECT, SIGNAL_DOCUMENTS_UPDATED
 
+if TYPE_CHECKING:
+    from .document_text import DocumentTextIndex
+
 _LOGGER = logging.getLogger(__name__)
 
 DOC_STORE_VERSION = 1
@@ -60,6 +63,10 @@ class DocumentStore:
         self.hass = hass
         self._store: Store[dict[str, Any]] = Store(hass, DOC_STORE_VERSION, DOC_STORE_KEY)
         self._data: dict[str, Any] = {"documents": {}, "blobs": {}}
+        # The full-text index (#171) follows the blob lifecycle: told about
+        # every new blob, told when the last reference goes. Optional so the
+        # store stays usable on its own (tests, tooling).
+        self.text_index: DocumentTextIndex | None = None
 
     # ------------------------------------------------------------------
     # Paths
@@ -214,6 +221,7 @@ class DocumentStore:
         await self._async_save()
         if not wrote_new and not deduped:
             _LOGGER.debug("Adopted pre-existing blob %s into registry", digest[:12])
+        self.notify_blob_added(digest)
         return {
             "id": doc_id,
             "deduped": deduped,
@@ -221,6 +229,12 @@ class DocumentStore:
             **doc,
         }
 
+    def notify_blob_added(self, digest: str) -> None:
+        """A blob is (now) referenced — let the text index extract it in the
+        background. Idempotent: an already-extracted blob is a no-op."""
+        if self.text_index is not None:
+            self.text_index.schedule(digest)
+
     def _store_blob_sync(self, content: bytes) -> tuple[str, bool]:
         """Hash content and write the blob if absent. Returns (digest, wrote_new)."""
         digest = hashlib.sha256(content).hexdigest()
@@ -263,6 +277,7 @@ class DocumentStore:
         docs: list[dict[str, Any]],
         task_id_map: dict[str, str] | None = None,
         part_id_map: dict[str, str] | None = None,
+        id_map: dict[str, str] | None = None,
     ) -> int:
         """Recreate document metadata for an imported object (P6).
 
@@ -272,9 +287,17 @@ class DocumentStore:
         absent and the hygiene scan flags the doc as dangling. ``task_ids`` /
         ``part_ids`` are remapped through their old→new id maps so a doc's
         task and spare-part links survive the import; ids with no mapping are
-        dropped. Returns the number created.
+        dropped. ``id_map`` (if given) is filled with old→new document ids
+        for every doc whose export carried an ``id`` — the importer uses it
+        to re-point completion photos and part ``doc_id`` links. Returns the
+        number created.
         """
 
+        def _remember(meta: dict[str, Any], new_id: str) -> None:
+            old_id = meta.get("id")
+            if id_map is not None and isinstance(old_id, str) and old_id:
+                id_map[old_id] = new_id
+
         def _remap(meta: dict[str, Any]) -> list[str]:
             if not task_id_map:
                 return []
@@ -298,7 +321,9 @@ class DocumentStore:
                 # frontend would later window.open (matches ws_documents_add_link).
                 if not isinstance(url, str) or not url.lower().startswith(("http://", "https://")):
                     continue
-                self.documents[uuid4().hex] = {
+                new_id = uuid4().hex
+                _remember(meta, new_id)
+                self.documents[new_id] = {
                     "object_id": object_id,
                     "kind": KIND_WEBLINK,
                     "url": url,
@@ -320,7 +345,9 @@ class DocumentStore:
                     blob = {"size": size, "mime": mime, "refcount": 0}
                     self.blobs[digest] = blob
                 blob["refcount"] += 1
-                self.documents[uuid4().hex] = {
+                new_id = uuid4().hex
+                _remember(meta, new_id)
+                self.documents[new_id] = {
                     "object_id": object_id,
                     "kind": KIND_FILE,
                     "hash": digest,
@@ -464,15 +491,24 @@ class DocumentStore:
         if blob["refcount"] <= 0:
             size = int(blob.get("size", 0))
             self.blobs.pop(digest, None)
+            if self.text_index is not None:
+                await self.text_index.async_forget(digest)
             await self.hass.async_add_executor_job(self._delete_blob_sync, digest)
             return size
         return 0
 
     def _delete_blob_sync(self, digest: str) -> None:
+        """Delete a blob file and its extracted-text sidecar (executor)."""
+        from .document_text import text_sidecar_path
+
         try:
             self.blob_path(digest).unlink(missing_ok=True)
         except OSError:
             _LOGGER.warning("Could not delete blob %s", digest[:12])
+        try:
+            text_sidecar_path(self.hass, digest).unlink(missing_ok=True)
+        except (OSError, ValueError):
+            _LOGGER.debug("Could not delete text sidecar %s", digest[:12])
 
     # ------------------------------------------------------------------
     # Storage summary (backs the sensor + the panel overview)