descry-codegraph 0.3.2__py3-none-any.whl → 0.3.4__py3-none-any.whl

descry/__init__.py

@@ -1,4 +1,4 @@
 """Descry — Codebase knowledge graph: polyglot indexer, semantic search, call-graph analysis."""
 
-__version__ = "0.3.2"
+__version__ = "0.3.4"
 __all__ = ["__version__"]

descry/cli.py

@@ -308,7 +308,9 @@ def main():
         default=3,
         help="Levels of calls below the start (default: 3)",
     )
-    sub.add_argument("--target", default=None, help="Target symbol to reach")
+    sub.add_argument(
+        "--target", default=None, help="Symbol to reach, by name or node ID"
+    )
     sub.add_argument(
         "--inline-threshold",
         type=int,
@@ -329,7 +331,13 @@ def main():
     sub.add_argument("--limit", type=int, default=10, help="Max results (default: 10)")
     sub.add_argument("--lang", default=None, help="Filter by language")
     sub.add_argument("--crate", default=None, help="Filter by crate")
-    sub.add_argument("--type", default=None, help="Filter by symbol type")
+    sub.add_argument(
+        "--type",
+        default=None,
+        type=str.lower,
+        choices=["function", "class", "method", "constant", "file", "all"],
+        help="Filter by symbol type",
+    )
     sub.add_argument(
         "--exclude-tests", action="store_true", help="Exclude test symbols"
     )

descry/embeddings.py

@@ -774,7 +774,9 @@ class SemanticSearcher:
         self.embeddings = embeddings / np.where(norms == 0, 1, norms)
         logger.info("Embeddings generated successfully")
 
-    def search(self, query: str, limit: int = 10, min_score: float = 0.3) -> list:
+    def search(
+        self, query: str, limit: int = 10, min_score: float = 0.3, keep=None
+    ) -> list:
         """Search for nodes semantically similar to the query.
 
         Args:
@@ -782,6 +784,7 @@ class SemanticSearcher:
             limit: Maximum number of results
             min_score: Minimum final score, i.e. cosine similarity after
                 the in-degree and type boosts are added (may exceed 1).
+            keep: The node IDs results may have, None for any.
 
         Returns:
             List of (node, score) tuples sorted by relevance
@@ -826,10 +829,19 @@ class SemanticSearcher:
             np.linalg.norm(self.embeddings, axis=1) * np.linalg.norm(query_embedding)
         )
 
+        if keep is not None:
+            # Nodes outside `keep` never enter the pool
+            kept = np.fromiter(
+                (n["id"] in keep for n in self.nodes), bool, len(self.nodes)
+            )
+            similarities = np.where(kept, similarities, -np.inf)
+            if not kept.any():
+                return []
+
         # Take a candidate pool by base score, then apply min_score to the
         # final score, so a node whose boosts (worth up to +0.17) lift it over
         # the threshold is kept.
-        pool = min(len(similarities), limit * 3)
+        pool = min(len(similarities) if keep is None else int(kept.sum()), limit * 3)
         top_indices = np.argpartition(-similarities, pool - 1)[:pool]
         top_indices = top_indices[np.argsort(-similarities[top_indices])]
 

descry/generate.py

@@ -18650,6 +18650,15 @@ class CallResolver:
             if self.standard[lang].key(callee, source_id) == callee:
                 held = self._returned_type(constructed, source_id)
                 self._result = held or self._result
+        elif verdict[1] == _EXTERNAL and not origin and lang == "dart":
+            # Dart constructs without `new`: a call of a standard type's
+            # name gives an instance of it.
+            library = self.standard[lang]
+            if (
+                callee in library.data["types"]
+                and library.key(callee, source_id) == callee
+            ):
+                self._result = self._returned_type(callee, source_id) or self._result
         elif verdict[1] == _EXTERNAL and not origin and lang == "typescript":
             known, value = self._js_global(callee, source_id)
             if known:
@@ -24878,12 +24887,19 @@ class CodeGraphBuilder:
         for root, dirs, files in os.walk(self.root_dir, followlinks=False):
             dirs.sort()
             files.sort()
-            # Filter by name first (excluded dirs, dotfiles), and by what git
-            # ignores.
+            # Filter by name first (excluded dirs, dotfiles, the build output
+            # of a .NET or an Eclipse project), and by what git ignores.
+            # `bin` holds sources in Dart and Rust.
+            built = any(
+                f.endswith((".csproj", ".vbproj", ".fsproj")) or f == ".classpath"
+                for f in files
+            )
             dirs[:] = [
                 d
                 for d in kept(root, dirs)
-                if not d.startswith(".") and d not in self.excluded_dirs
+                if not d.startswith(".")
+                and d not in self.excluded_dirs
+                and not (built and d == "bin")
             ]
             # Then filter by realpath containment — drop directories that are
             # symlinks pointing outside the project root.

descry/handlers.py

@@ -92,8 +92,8 @@ _DEFAULT_EXCLUDED_DIRS = {
     "venv",
     # Go
     "vendor",
-    # .NET
-    "bin",
+    # .NET (its `bin`, and an Eclipse project's, is skipped beside the
+    # project file)
     "obj",
     # Dart / Flutter
     ".dart_tool",
@@ -2086,23 +2086,23 @@ class DescryService:
                 # and respects the cache lock).
                 searcher = await self._get_semantic_searcher()
                 if searcher is not None:
+                    # Under the filters the keyword side has
                     semantic_results = await asyncio.to_thread(
                         searcher.search,
                         " ".join(terms),
                         limit=limit * 2,
                         min_score=0.25,
+                        keep=q.filtered_ids(lang, crate, symbol_type, exclude_tests),
                     )
             except Exception as e:  # noqa: BLE001 — boundary over the embedding backend; search degrades to keyword-only
                 logger.warning(f"Semantic search failed, using keyword only: {e}")
 
-        if semantic_results and tfidf_results:
+        if semantic_results:
             combined = reciprocal_rank_fusion(tfidf_results, semantic_results)
             results = [node for node, _ in combined[:limit]]
             method = "hybrid"
-        elif tfidf_results:
-            results = tfidf_results[:limit]
         else:
-            results = []
+            results = tfidf_results[:limit]
 
         return {"ok": True, "error": None, "results": results, "method": method}
 

descry/query.py

@@ -1318,7 +1318,8 @@ class GraphQuerier:
         once, at the first level that reaches it, and met again as `seen`,
         the level of that entry (the start is level 0).
         Returns ``(root, path, cut)``: the tree, the node IDs from the start
-        to the first node named ``target``, and why it stopped early
+        to the first node ``target`` names, by name or node ID, and why it
+        stopped early
         ("timeout", "node limit") or None."""
         start_time = time.time()
 
@@ -1353,7 +1354,7 @@ class GraphQuerier:
 
         root = tree_node(start_id, None)
         paths = {start_id: [start_id]}
-        if target and root["name"] == target:
+        if target in (root["name"], root["id"]):
             return root, paths[start_id], None
         level = [root]
         for _ in range(depth):
@@ -1378,7 +1379,7 @@ class GraphQuerier:
                     child = tree_node(child_id, marks.get(child_id))
                     paths[child_id] = [*paths[parent["id"]], child_id]
                     parent["children"].append(child)
-                    if target and child["name"] == target:
+                    if target in (child["name"], child["id"]):
                         return root, paths[child_id], None
                     next_level.append(child)
             level = next_level
@@ -1500,7 +1501,7 @@ class GraphQuerier:
         cut (why the trace stopped early, when it did). Each tree node has: id,
         name, type, tokens, file, line, code, lang, candidates, children,
         more, more_possible, and `seen`, the level its entry is at, where it
-        is placed there already. The trace stops at the first node named `target`.
+        is placed there already. The trace stops at the first node `target` names, by name or node ID.
         """
         if timeout_ms is None:
             timeout_ms = self._timeout_ms
@@ -1666,6 +1667,33 @@ class GraphQuerier:
             idf[term] = math.log((total_docs + 1) / (df + 1)) + 1
         return idf
 
+    def filtered_ids(
+        self, lang=None, crate=None, symbol_type=None, exclude_tests=False
+    ):
+        """The IDs of the nodes `search_docs`'s filters keep, None when none
+        is given: the language, the crate or module, the type, and no tests."""
+        if not (lang or crate or symbol_type or exclude_tests):
+            return None
+        self._ensure_filter_indices()
+        kept = set(self.nodes.keys())
+        if exclude_tests and self._index_is_test:
+            kept -= self._index_is_test
+        if crate:
+            kept &= self._index_by_crate.get(crate, set())
+        if lang and lang != "all":
+            lang = _LANG_ALIASES.get(lang.lower(), lang.lower())
+            kept &= self._index_by_lang.get(lang, set())
+        if symbol_type and symbol_type != "all":
+            type_map = {
+                "function": "Function",
+                "class": "Class",
+                "method": "Method",
+                "constant": "Constant",
+                "file": "File",
+            }
+            kept &= self._index_by_type.get(type_map.get(symbol_type.lower()), set())
+        return kept
+
     def search_docs(
         self, query_terms, lang=None, crate=None, symbol_type=None, exclude_tests=False
     ):
@@ -1695,38 +1723,10 @@ class GraphQuerier:
         if num_terms == 0:
             return []
 
-        # Apply filters if any specified
-        if lang or crate or symbol_type or exclude_tests:
-            self._ensure_filter_indices()
-            candidate_ids = set(self.nodes.keys())
-
-            # Filter out test files and functions if requested
-            if exclude_tests and self._index_is_test:
-                candidate_ids -= self._index_is_test
-
-            if crate:
-                # If crate filter specified but doesn't exist, return empty results
-                if crate not in self._index_by_crate:
-                    return []
-                candidate_ids &= self._index_by_crate[crate]
-            if lang and lang != "all":
-                lang = _LANG_ALIASES.get(lang.lower(), lang.lower())
-                candidate_ids &= self._index_by_lang.get(lang, set())
-            if symbol_type and symbol_type != "all":
-                type_map = {
-                    "function": "Function",
-                    "class": "Class",
-                    "method": "Method",
-                    "constant": "Constant",
-                    "file": "File",
-                }
-                type_key = type_map.get(symbol_type.lower())
-                if type_key and type_key in self._index_by_type:
-                    candidate_ids &= self._index_by_type[type_key]
-
-            candidates = [self.nodes[nid] for nid in candidate_ids]
-        else:
-            candidates = self.data["nodes"]
+        kept = self.filtered_ids(lang, crate, symbol_type, exclude_tests)
+        candidates = (
+            self.data["nodes"] if kept is None else [self.nodes[nid] for nid in kept]
+        )
 
         for node in candidates:
             score = 0.0

{descry_codegraph-0.3.2.dist-info → descry_codegraph-0.3.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: descry-codegraph
-Version: 0.3.2
+Version: 0.3.4
 Summary: Codebase knowledge graph: polyglot indexer, semantic search, call-graph analysis
 Project-URL: Homepage, https://github.com/06chaynes/descry
 Project-URL: Repository, https://github.com/06chaynes/descry
@@ -267,7 +267,7 @@ For customization, add a `.descry.toml` to your project root.
 
 ```toml
 [project]
-# OPTIONAL — REPLACES the 25-element default. Leave unset to keep defaults
+# OPTIONAL — REPLACES the 24-element default. Leave unset to keep defaults
 # (.git, .gradle, .next, .svelte-kit, .venv, node_modules, target, build, …).
 # excluded_dirs = ["target", "node_modules", "dist", "build", "vendor"]
 # OPTIONAL — in a git repository, files git ignores are not indexed.
@@ -393,7 +393,7 @@ Share of project calls resolved without SCIP, measured on public repositories at
 | C# | serilog `bebc771`, Newtonsoft.Json `52fa3ae` (`Src/Newtonsoft.Json`) | 94%, 91% |
 | C | jq `a19efdd`, libuv `954c1d1` | 99%, 79% |
 | C++ | leveldb `7ee830d`, fmt `c4927e6` | 97%, 86% |
-| Dart | shelf `b2cb503`, riverpod `1efba1c` (`packages/riverpod`) | 93%, 91% |
+| Dart | shelf `b2cb503`, riverpod `1efba1c` (`packages/riverpod`) | 95%, 91% |
 | Ruby | rack `a9833c8`, sinatra `cb22afd` | 70%, 39% |
 
 The remainder is unresolved, not wrong: of 450 resolved edges drawn at random from these repositories and read against the source, every one reaches the definition the call reaches. Ruby is low because the language declares no types and calls inside DSL blocks run with another `self`; libuv because most of its functions exist once for Unix and once for Windows, and the build files, not the source, choose between them.

{descry_codegraph-0.3.2.dist-info → descry_codegraph-0.3.4.dist-info}/RECORD

@@ -1,23 +1,23 @@
-descry/__init__.py,sha256=ZxLltjohUUqYcFkvFruQKgcOkihW-dwXAZEH0zxY9t4,148
+descry/__init__.py,sha256=wF1DadE3KUESLtSaENhYMj5JL-jXyNJP8GSF_l-S7NE,148
 descry/_comments.py,sha256=CCL0ywnHTWDZE2zKRpQwgbU2Re7B979XdbLyLIZooHc,2499
 descry/_env.py,sha256=xme87ibczJ0wBvFUlZGXeyqO6lFs0Xk2hg6Uxi1mVJU,251
 descry/_files.py,sha256=1fcXCw-tLXs9JOq67H0qHcKa0j538f65ZysH9op1HXM,1826
 descry/_graph.py,sha256=y3PExZ2XddHo__y-X9_uTtLD9rpGuHxcE7VtJUUvkoQ,5237
 descry/_stdlib.py,sha256=xA0hiCCspqeqD79znB6ADSMd3TjZvGkgsZlGUPqhrjI,3141
 descry/clang_parser.py,sha256=SImFWXPRIN81mZzHOwkJkWH8JV68ihJ75Wo9kyhV35I,76089
-descry/cli.py,sha256=4IUlBKdW8kcQuVPCjrzX1RK3umWkhugmJVnzqFgZwAE,14418
+descry/cli.py,sha256=IY-2aDksq7Yw-mRWGYVoPG7Njqdj6Z0D6X8_lKvjDm4,14576
 descry/cross_lang.py,sha256=tGBhUrkt3EPu0A2J8Fwi3RgneLtTbZjWiIaPlqk4tuI,14906
 descry/dart_parser.py,sha256=yOFLNqUdSSF1tS4YPuYGZLyY5o3xKUIh5QwFQDhcI30,77634
 descry/dotnet_parser.py,sha256=kzH28pmY6XXpOnFcghta_srarojrMmz7_I3887v0Yfo,57094
-descry/embeddings.py,sha256=ahDwLXDIB6-mYAK1Ic8QpwPhTzVaULiojpyaIl6d1q0,37221
-descry/generate.py,sha256=InbUpzYKkAYHy-ytCjT_6BWOCDeVk2cLAHAYryadh1w,1072736
+descry/embeddings.py,sha256=VyXynjY54soASnDcSehzvVgGW4q5ucSUOSVpmCJ1WsI,37677
+descry/generate.py,sha256=lQLjvQoaIOk3WUfnU5LnaABbA1Lq25W_VAVIrEapyD4,1073512
 descry/git_history.py,sha256=EGBzjorPBT3ZyUdYuCBqvb2aoueoH2X5vOd83P67Mb0,67277
 descry/go_parser.py,sha256=j2rMjgERA1A7Ma8sS1ha-QtoD4_EipUPVqs3XbsQRrE,40802
-descry/handlers.py,sha256=q7LQXIoUWlieYHeHmnGajjPZET3idaWYwvJeLQjmPAo,113990
+descry/handlers.py,sha256=LVeMuxPodte7bwh2qGvBPl2BKTDL0sIQTHTWUpnGpJ4,114135
 descry/java_parser.py,sha256=SK78X_Ykwkb9iLVpq9IemWLURZRF_l1XX6RFaHNkKuA,41424
 descry/mcp_server.py,sha256=W6dLKBBB5FJ8wcuzXc_FbjOcvcvIs1ZEABoYAUDsW9I,14216
 descry/php_parser.py,sha256=CUrfMO4lnYCmUGXR-FqkQ9kaoPsH8Gu13q3MwIMjsw4,56739
-descry/query.py,sha256=LqrSWuEbvQDqjfXo-dUZEX8aIo8S3hufTSiFI-FRfSk,96236
+descry/query.py,sha256=Q3nuaGaW9Lr3IUCHs85Rzyf8Jjmfu2iaTM-E6GmQF8A,96195
 descry/ruby_parser.py,sha256=LGpPaQX6QWKnPiRy9oUrDTg41QGmCWNvZ-m5tg6gF58,66233
 descry/scip/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 descry/scip/adapter.py,sha256=vLEQrO7MDqZipDlp6xsB_-86eAKwJhUk9aVKZfXceFI,7762
@@ -54,9 +54,9 @@ descry/web/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 descry/web/server.py,sha256=ESWuPSwXA0LtCo1bwWqiQN1wjvMDbXhJ3byqlPg99FU,38475
 descry/web/web/index.html,sha256=WvcTTkJSpdgDzYLk79YWlx2PG8d7CqwFwi2xgBH3-oc,151183
 descry/web/web/style.css,sha256=VTQ1mC-n0BYuDlI1ghozfDG-azLqhJcpWnA0eZtts1o,35816
-descry_codegraph-0.3.2.dist-info/METADATA,sha256=81TaY38igniVecGdM2zFuc1Uq61inpAr9jt8_DKrRCY,24021
-descry_codegraph-0.3.2.dist-info/WHEEL,sha256=W3fkpkm7-wf9vBI5Z-7s0eWkeM-spu78I8Neb98DeEg,87
-descry_codegraph-0.3.2.dist-info/entry_points.txt,sha256=qz8s4CwO8y8SAryD9PibrMPDgAbFV2L3sIafvl91_HQ,154
-descry_codegraph-0.3.2.dist-info/licenses/LICENSE,sha256=V8lsocOdIekq2MluAKSu-wGZBsWPBNHCgG7OBfUMlt0,1073
-descry_codegraph-0.3.2.dist-info/licenses/THIRD-PARTY-NOTICES.md,sha256=igEI8prnwdkwsNBWpIZGKDoTs-wuGD6ijhpDLI6ks2g,1944
-descry_codegraph-0.3.2.dist-info/RECORD,,
+descry_codegraph-0.3.4.dist-info/METADATA,sha256=DDgtb-9GJ_ke43N709Q6EO_qstv411baeBP1ZRsmmlI,24021
+descry_codegraph-0.3.4.dist-info/WHEEL,sha256=W3fkpkm7-wf9vBI5Z-7s0eWkeM-spu78I8Neb98DeEg,87
+descry_codegraph-0.3.4.dist-info/entry_points.txt,sha256=qz8s4CwO8y8SAryD9PibrMPDgAbFV2L3sIafvl91_HQ,154
+descry_codegraph-0.3.4.dist-info/licenses/LICENSE,sha256=V8lsocOdIekq2MluAKSu-wGZBsWPBNHCgG7OBfUMlt0,1073
+descry_codegraph-0.3.4.dist-info/licenses/THIRD-PARTY-NOTICES.md,sha256=igEI8prnwdkwsNBWpIZGKDoTs-wuGD6ijhpDLI6ks2g,1944
+descry_codegraph-0.3.4.dist-info/RECORD,,