update to new search index schema

Author: madebygps

brain_cli/notes_commands.py

@@ -16,7 +16,6 @@
 PREVIEW_LENGTH = 80
 TABLE_COL_WIDTH_INDEX = 3
 TABLE_COL_WIDTH_CATEGORY = 12
-TABLE_COL_WIDTH_WORDS = 7
 TABLE_COL_WIDTH_SCORE = 7
 
 
@@ -59,17 +58,13 @@ def search(
     top: int = typer.Option(
         DEFAULT_TOP_RESULTS, "--top", "-n", help="Maximum number of results to return"
     ),
-    semantic: bool = typer.Option(
-        False, "--semantic", "-s", help="Use semantic search (slower but more relevant)"
-    ),
     detailed: bool = typer.Option(False, "--detailed", "-d", help="Show full content of results"),
 ):
     """Search book notes using Azure AI Search.
 
     Examples:
         brain notes search "discipline and habits"
         brain notes search "productivity tips" --top 5
-        brain notes search "notes on learning" --semantic
         brain notes search "deep work" --detailed
     """
     try:
@@ -82,21 +77,14 @@ def search(
 
         # Perform search
         with console.status(f"[bold cyan]Searching for '{query}'...", spinner="dots"):
-            if semantic:
-                results = search_client.semantic_search(query, top=top)
-                search_type = "Semantic"
-            else:
-                results = search_client.search(query, top=top)
-                search_type = "Text"
+            results = search_client.search(query, top=top)
 
         if not results:
             console.print(f"\n[yellow]No results found for '{query}'[/yellow]")
             return
 
         # Display results
-        console.print(
-            f"\n[bold cyan]{search_type} Search Results[/bold cyan] for: [bold]'{query}'[/bold]"
-        )
+        console.print(f"\n[bold cyan]Search Results[/bold cyan] for: [bold]'{query}'[/bold]")
         console.print(f"[dim]Found {len(results)} result(s)[/dim]\n")
 
         if detailed:
@@ -106,9 +94,8 @@ def search(
                 panel_content = f"**Book:** {result.title}\n"
                 panel_content += f"**Category:** {result.category}\n"
                 panel_content += f"**Source:** {result.source}\n"
-                panel_content += (
-                    f"**Score:** {result.score:.2f} | **Words:** {result.word_count}\n\n"
-                )
+                panel_content += f"**File:** {result.file_path}\n"
+                panel_content += f"**Score:** {result.score:.2f}\n\n"
                 panel_content += "---\n\n"
                 panel_content += result.content
 
@@ -126,7 +113,6 @@ def search(
             table.add_column("#", style="dim", width=TABLE_COL_WIDTH_INDEX, justify="right")
             table.add_column("Book", style="bold")
             table.add_column("Category", style="yellow", width=TABLE_COL_WIDTH_CATEGORY)
-            table.add_column("Words", justify="right", width=TABLE_COL_WIDTH_WORDS)
             table.add_column("Score", justify="right", width=TABLE_COL_WIDTH_SCORE)
             table.add_column("Preview", style="dim")
 
@@ -137,7 +123,6 @@ def search(
                     str(i),
                     result.title,
                     result.category,
-                    str(result.word_count),
                     f"{result.score:.2f}",
                     preview,
                 )

brain_core/constants.py

@@ -70,3 +70,7 @@
         "output": 0.002 / 1000,
     },
 }
+
+# Azure AI Search - Vector embeddings
+EMBEDDING_MODEL = "text-embedding-3-small"  # Azure OpenAI embedding model
+EMBEDDING_DIMENSIONS = 384  # Dimensions for text-embedding-3-small

brain_core/notes_search_client.py

@@ -4,6 +4,7 @@
 
 from azure.core.credentials import AzureKeyCredential
 from azure.search.documents import SearchClient
+from azure.search.documents.models import VectorizedQuery
 
 
 @dataclass
@@ -15,10 +16,8 @@ class SearchResult:
     content: str
     source: str
     category: str
-    file_name: str
-    word_count: int
+    file_path: str
     score: float
-    created_at: str
     metadata: dict
 
 
@@ -63,32 +62,35 @@ def search(self, query: str, top: int = 10, search_mode: str = "any") -> list[Se
                     content=result.get("content", ""),
                     source=result.get("source", "Unknown"),
                     category=result.get("category", ""),
-                    file_name=result.get("file_name", ""),
-                    word_count=result.get("word_count", 0),
+                    file_path=result.get("file_path", ""),
                     score=result.get("@search.score", 0.0),
-                    created_at=result.get("created_at", ""),
                     metadata={k: v for k, v in result.items() if not k.startswith("@")},
                 )
             )
 
         return search_results
 
-    def semantic_search(self, query: str, top: int = 10) -> list[SearchResult]:
-        """Perform a semantic search using Azure's semantic ranking.
+    def vector_search(
+        self, query_vector: list[float], top: int = 10, text_query: str | None = None
+    ) -> list[SearchResult]:
+        """Perform a vector search using embeddings.
 
         Args:
-            query: Search query text
+            query_vector: Query embedding vector (384 dimensions for text-embedding-3-small)
             top: Maximum number of results to return
+            text_query: Optional text query for hybrid search
 
         Returns:
-            List of SearchResult objects with semantic relevance
+            List of SearchResult objects with vector similarity
         """
+        vector_query = VectorizedQuery(
+            vector=query_vector, k_nearest_neighbors=top, fields="content_vector"
+        )
+
         results = self.client.search(
-            search_text=query,
+            search_text=text_query,
+            vector_queries=[vector_query],
             top=top,
-            query_type="semantic",
-            semantic_configuration_name="default",
-            include_total_count=True,
         )
 
         search_results = []
@@ -100,10 +102,8 @@ def semantic_search(self, query: str, top: int = 10) -> list[SearchResult]:
                     content=result.get("content", ""),
                     source=result.get("source", "Unknown"),
                     category=result.get("category", ""),
-                    file_name=result.get("file_name", ""),
-                    word_count=result.get("word_count", 0),
-                    score=result.get("@search.reranker_score", result.get("@search.score", 0.0)),
-                    created_at=result.get("created_at", ""),
+                    file_path=result.get("file_path", ""),
+                    score=result.get("@search.score", 0.0),
                     metadata={k: v for k, v in result.items() if not k.startswith("@")},
                 )
             )