graphifyy 0.9.76__tar.gz → 0.9.78__tar.gz

{graphifyy-0.9.76/graphifyy.egg-info → graphifyy-0.9.78}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: graphifyy
-Version: 0.9.76
+Version: 0.9.78
 Summary: AI coding assistant skill (Claude Code, CodeBuddy, Codex, OpenCode, Kilo Code, Cursor, Gemini CLI, Aider, OpenClaw, Factory Droid, Trae, Hermes, Kiro, Pi, Devin CLI, Google Antigravity) - turn any folder of code, docs, papers, images, or videos into a queryable knowledge graph
 License-Expression: Apache-2.0
 Project-URL: Homepage, https://graphify.com
@@ -219,6 +219,10 @@ graphify-out/
 └── graph.json       the full graph — query it anytime without re-reading your files
 ```
 
+The persisted graph includes `graph.schema_version` so integrations can detect
+incompatible format changes, plus `graph.graphify_version` identifying the
+Graphify release that produced it.
+
 **Works in** Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, and 15+ more — [pick your platform](#install).
 
 ---
@@ -277,6 +281,7 @@ What you get out of the box:
 | LOCOMO (n=300) | recall@10 | **0.497** | mem0 0.048, supermemory 0.149 |
 | LOCOMO (n=300) | QA accuracy | 45.3% | supermemory 49.7%, mem0 27.3% |
 | LongMemEval-S (n=50) | QA accuracy | **76%** | tied with dense RAG |
+| ERPNext cross-tool (n=6) | key-fact coverage | **82.0%** | grep/read baseline 70.8% |
 | Graph build | LLM credits | **0** | per-token for most systems |
 
 Every system ran on the same harness with the same model and budgets, scored by a judge blind-validated against a second judge (90.6% agreement, Cohen's kappa 0.81). Full per-system tables, the code-intelligence result, and reproduction commands: **[BENCHMARKS.md](./BENCHMARKS.md)**.
@@ -1011,6 +1016,12 @@ graphify label ./my-project                                    # (re)name commun
 graphify label ./my-project --backend=openai --model gpt-4o   # force a specific backend and model
 ```
 
+`--no-dedup` also skips coalescing distinct non-AST nodes solely because they
+share a file and label. The Python equivalents are `build(chunks, dedup=False)`,
+`build_merge(chunks, graph_path, dedup=False)`, and
+`build_from_json(extraction, dedup=False)`. AST/semantic twins still reconcile to
+the canonical AST node, and document-file twin reconciliation remains enabled.
+
 > **Community names:** inside an agent (Claude Code, Gemini CLI) the agent names communities itself. When you run the bare CLI, `cluster-only` auto-names them with the configured backend (built-in or custom OpenAI-compatible provider) — pass `--no-label` to keep `Community N`, or run `graphify label` to (re)generate names on demand.
 
 ---
@@ -1053,6 +1064,12 @@ Made with [contrib.rocks](https://contrib.rocks).
 
 ---
 
+## Translations
+
+The README is available in 32 languages. Use the language switcher at the top of this file to read it in yours, or browse [`docs/translations/`](docs/translations/). To improve a translation or add a new one, open a pull request against the matching file there.
+
+---
+
 ## Community and links
 
 <p align="center">

{graphifyy-0.9.76 → graphifyy-0.9.78}/README.md

@@ -68,6 +68,10 @@ graphify-out/
 └── graph.json       the full graph — query it anytime without re-reading your files
 ```
 
+The persisted graph includes `graph.schema_version` so integrations can detect
+incompatible format changes, plus `graph.graphify_version` identifying the
+Graphify release that produced it.
+
 **Works in** Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, and 15+ more — [pick your platform](#install).
 
 ---
@@ -126,6 +130,7 @@ What you get out of the box:
 | LOCOMO (n=300) | recall@10 | **0.497** | mem0 0.048, supermemory 0.149 |
 | LOCOMO (n=300) | QA accuracy | 45.3% | supermemory 49.7%, mem0 27.3% |
 | LongMemEval-S (n=50) | QA accuracy | **76%** | tied with dense RAG |
+| ERPNext cross-tool (n=6) | key-fact coverage | **82.0%** | grep/read baseline 70.8% |
 | Graph build | LLM credits | **0** | per-token for most systems |
 
 Every system ran on the same harness with the same model and budgets, scored by a judge blind-validated against a second judge (90.6% agreement, Cohen's kappa 0.81). Full per-system tables, the code-intelligence result, and reproduction commands: **[BENCHMARKS.md](./BENCHMARKS.md)**.
@@ -860,6 +865,12 @@ graphify label ./my-project                                    # (re)name commun
 graphify label ./my-project --backend=openai --model gpt-4o   # force a specific backend and model
 ```
 
+`--no-dedup` also skips coalescing distinct non-AST nodes solely because they
+share a file and label. The Python equivalents are `build(chunks, dedup=False)`,
+`build_merge(chunks, graph_path, dedup=False)`, and
+`build_from_json(extraction, dedup=False)`. AST/semantic twins still reconcile to
+the canonical AST node, and document-file twin reconciliation remains enabled.
+
 > **Community names:** inside an agent (Claude Code, Gemini CLI) the agent names communities itself. When you run the bare CLI, `cluster-only` auto-names them with the configured backend (built-in or custom OpenAI-compatible provider) — pass `--no-label` to keep `Community N`, or run `graphify label` to (re)generate names on demand.
 
 ---
@@ -902,6 +913,12 @@ Made with [contrib.rocks](https://contrib.rocks).
 
 ---
 
+## Translations
+
+The README is available in 32 languages. Use the language switcher at the top of this file to read it in yours, or browse [`docs/translations/`](docs/translations/). To improve a translation or add a new one, open a pull request against the matching file there.
+
+---
+
 ## Community and links
 
 <p align="center">

{graphifyy-0.9.76 → graphifyy-0.9.78}/graphify/analyze.py

@@ -630,24 +630,39 @@ def graph_diff(G_old: nx.Graph, G_new: nx.Graph) -> dict:
             return (u, v, data.get("relation", ""))
         return (min(u, v), max(u, v), data.get("relation", ""))
 
-    old_edge_keys = {
-        edge_key(G_old, u, v, d)
+    def edge_dir(data: dict) -> tuple | None:
+        # Stored direction of an undirected edge (build_from_json, load_node_link_graph).
+        if "_src" in data and "_tgt" in data:
+            return (data["_src"], data["_tgt"])
+        return None
+
+    old_edge_dirs = {
+        edge_key(G_old, u, v, d): edge_dir(d)
         for u, v, d in G_old.edges(data=True)
     }
-    new_edge_keys = {
-        edge_key(G_new, u, v, d)
+    new_edge_dirs = {
+        edge_key(G_new, u, v, d): edge_dir(d)
         for u, v, d in G_new.edges(data=True)
     }
-
-    added_edge_keys = new_edge_keys - old_edge_keys
-    removed_edge_keys = old_edge_keys - new_edge_keys
+    old_edge_keys = set(old_edge_dirs)
+    new_edge_keys = set(new_edge_dirs)
+
+    # A reversed edge (a->b became b->a) keeps its undirected key. Report it as
+    # removed + added only when both snapshots carry _src/_tgt and they differ,
+    # so graphs without the markers diff exactly as before (#4067).
+    reversed_edge_keys = {
+        k for k in old_edge_keys & new_edge_keys
+        if old_edge_dirs[k] and new_edge_dirs[k] and old_edge_dirs[k] != new_edge_dirs[k]
+    }
+    added_edge_keys = (new_edge_keys - old_edge_keys) | reversed_edge_keys
+    removed_edge_keys = (old_edge_keys - new_edge_keys) | reversed_edge_keys
 
     new_edges_list = []
     for u, v, d in G_new.edges(data=True):
         if edge_key(G_new, u, v, d) in added_edge_keys:
             new_edges_list.append({
-                "source": u,
-                "target": v,
+                "source": d.get("_src", u),
+                "target": d.get("_tgt", v),
                 "relation": d.get("relation", ""),
                 "confidence": d.get("confidence", ""),
             })
@@ -656,8 +671,8 @@ def graph_diff(G_old: nx.Graph, G_new: nx.Graph) -> dict:
     for u, v, d in G_old.edges(data=True):
         if edge_key(G_old, u, v, d) in removed_edge_keys:
             removed_edges_list.append({
-                "source": u,
-                "target": v,
+                "source": d.get("_src", u),
+                "target": d.get("_tgt", v),
                 "relation": d.get("relation", ""),
                 "confidence": d.get("confidence", ""),
             })

{graphifyy-0.9.76 → graphifyy-0.9.78}/graphify/build.py

@@ -874,13 +874,17 @@ def _doc_twin_remap(nodes: list) -> dict[str, str]:
     return remap
 
 
-def build_from_json(extraction: dict, *, directed: bool = False, root: str | Path | None = None) -> nx.Graph:
+def build_from_json(extraction: dict, *, directed: bool = False, root: str | Path | None = None,
+                    dedup: bool = True) -> nx.Graph:
     """Build a NetworkX graph from an extraction dict.
 
     directed=True produces a DiGraph that preserves edge direction (source→target).
     directed=False (default) produces an undirected Graph for backward compatibility.
     root: if given, absolute source_file paths from semantic subagents are made
         relative to root so all nodes share a consistent path key (#932).
+    dedup=False preserves distinct non-AST IDs rather than coalescing nodes by
+        file and label. AST/semantic and document-file twin reconciliation
+        remain enabled.
     """
     _root = str(Path(root).resolve()) if root else None
     # NetworkX <= 3.1 serialised edges as "links"; remap to "edges" for compatibility.
@@ -1158,6 +1162,8 @@ def build_from_json(extraction: dict, *, directed: bool = False, root: str | Pat
         if key in _loc_collisions:
             continue  # ambiguous key: no safe canonical winner, leave ghost intact
         if key in _loc_nodes and _loc_nodes[key] != nid:
+            if not dedup and G.nodes[_loc_nodes[key]].get("_origin") != "ast":
+                continue
             _noloc_nodes[key] = nid
         elif key not in _loc_nodes:
             # Spec-conformant method ghost omitting class segment / leading dot
@@ -1599,7 +1605,7 @@ def build(
             protected_ids=protected_ids,
         )
         _dedup_collapsed = _before_dedup - len(combined["nodes"])
-    G = build_from_json(combined, directed=directed, root=_root)
+    G = build_from_json(combined, directed=directed, root=_root, dedup=dedup)
     # CLI reads this to tell a dedup shrink from a file deletion (#3774).
     # Popped before to_json so it is not stored in graph.json.
     if _dedup_collapsed:

{graphifyy-0.9.76 → graphifyy-0.9.78}/graphify/cli.py

@@ -2684,6 +2684,12 @@ def dispatch_command(cmd: str) -> None:
             out_data = _jg.node_link_data(merged, edges="links")
         except TypeError:
             out_data = _jg.node_link_data(merged)
+        # Write direction back into arc order, as to_json does (#563): the
+        # loader stamped _src/_tgt, and compose() kept them on each edge.
+        for _link in out_data.get("links", []):
+            _ts, _tt = _link.pop("_src", None), _link.pop("_tgt", None)
+            if _ts is not None and _tt is not None:
+                _link["source"], _link["target"] = _ts, _tt
         from graphify.paths import write_json_atomic
         write_json_atomic(_current_path, out_data, indent=2)
         sys.exit(0)
@@ -3081,6 +3087,8 @@ def dispatch_command(cmd: str) -> None:
         _raw = json.loads(graph_path.read_text(encoding="utf-8"))
         if "links" not in _raw and "edges" in _raw:
             _raw = dict(_raw, links=_raw["edges"])
+        from graphify.paths import restore_arc_direction as _rad
+        _raw = _rad(_raw)
         try:
             G = _jg.node_link_graph(_raw, edges="links")
         except TypeError:

{graphifyy-0.9.76 → graphifyy-0.9.78}/graphify/detect.py

@@ -22,7 +22,7 @@ from graphify.google_workspace import (
     convert_google_workspace_file,
     google_workspace_enabled,
 )
-from graphify.paths import GRAPHIFY_OUT, out_path
+from graphify.paths import GRAPHIFY_OUT, GRAPHIFY_OUT_NAME, out_path
 
 
 class FileType(str, Enum):
@@ -2397,6 +2397,27 @@ def _collapse_manifest_duplicates(
     return result
 
 
+def _manifest_storage_anchor(manifest_path: str | Path | None, root: Path | None) -> Path | None:
+    """When a manifest lives under <repo>/graphify-out/manifest.json and root
+    is a subfolder of <repo>, stored relative keys are anchored to <repo> (#3785)."""
+    if root is None or not manifest_path:
+        return root
+    try:
+        mp = Path(manifest_path).resolve()
+        if mp.parent.name in ("graphify-out", GRAPHIFY_OUT_NAME):
+            manifest_base = mp.parent.parent
+            root_resolved = Path(root).resolve()
+            if root_resolved != manifest_base:
+                try:
+                    root_resolved.relative_to(manifest_base)
+                    return manifest_base
+                except ValueError:
+                    pass
+    except (OSError, RuntimeError):
+        pass
+    return root
+
+
 def load_manifest(
     manifest_path: str = _MANIFEST_PATH,
     *,
@@ -2421,8 +2442,9 @@ def load_manifest(
         return raw
     if root is None:
         return _collapse_manifest_duplicates(raw.items(), _nfc)
+    anchor = _manifest_storage_anchor(manifest_path, root) or root
     return _collapse_manifest_duplicates(
-        raw.items(), lambda k: _nfc(_to_absolute_from_storage(k, root))
+        raw.items(), lambda k: _nfc(_to_absolute_from_storage(k, anchor))
     )
 
 
@@ -2624,18 +2646,10 @@ def save_manifest(
             "semantic_hash": sem_h,
         }
         manifest[key] = entry
-    if root is not None:
-        # Persist in portable form: forward-slash relative paths. Keys outside
-        # ``root`` (out-of-tree symlinked corpora, --include sources) keep
-        # their absolute form so the manifest round-trips on the saving
-        # machine even when not every entry can be portably encoded.
-        # NFC after relativize so on-disk keys match what load_manifest
-        # re-anchors and compares against (#2221). A seeded absolute key and
-        # a seeded relative key for the same file collapse here too (#1964)
-        # -- _collapse_manifest_duplicates keeps whichever is more recently
-        # seen instead of whichever the dict comprehension iterates last.
+    storage_root = _manifest_storage_anchor(manifest_path, root) if root is not None else None
+    if storage_root is not None:
         manifest = _collapse_manifest_duplicates(
-            manifest.items(), lambda k: _nfc(_to_relative_for_storage(k, root))
+            manifest.items(), lambda k: _nfc(_to_relative_for_storage(k, storage_root))
         )
     else:
         manifest = _collapse_manifest_duplicates(manifest.items(), _nfc)
@@ -2813,10 +2827,32 @@ def detect_incremental(
     # current scan was EXCLUDED (ignore rules / --exclude changed) and must
     # not be reported as deleted. Mirrors the watch-side excluded-vs-deleted
     # distinction (#1795).
+    try:
+        root_res: Path | None = Path(root).resolve() if root is not None else None
+    except (OSError, RuntimeError):
+        root_res = Path(root) if root is not None else None
+
+    def _in_root(path_str: str) -> bool:
+        if root_res is None:
+            return True
+        p = Path(path_str)
+        try:
+            p.relative_to(root_res)
+            return True
+        except ValueError:
+            pass
+        try:
+            p.resolve().relative_to(root_res)
+            return True
+        except (ValueError, OSError, RuntimeError):
+            return False
+
     current_files = {_nfc(f) for flist in full["files"].values() for f in flist}
     deleted_files: list[str] = []
     excluded_files: list[str] = []
     for f in manifest:
+        if not _in_root(f):
+            continue
         if _nfc(f) in current_files:
             continue
         try:

{graphifyy-0.9.76 → graphifyy-0.9.78}/graphify/export.py

@@ -172,6 +172,20 @@ from graphify.exporters.base import COMMUNITY_COLORS  # noqa: E402,F401
 from graphify.exporters.html import to_html  # noqa: E402,F401
 
 
+# Increment when the persisted graph structure changes incompatibly for consumers.
+GRAPH_SCHEMA_VERSION = 1
+
+
+def _graphify_version() -> str | None:
+    """Return the installed graphify version for graph provenance."""
+    try:
+        from importlib.metadata import version
+
+        return version("graphifyy")
+    except Exception:
+        return None
+
+
 # Fallback scores for an edge that carries a confidence tier but no
 # confidence_score. The INFERRED default was 0.5, which references/extraction-spec.md
 # rules out in as many words — "never omit it, never use 0.5 as a default" — and
@@ -433,6 +447,11 @@ def to_json(G: nx.Graph, communities: dict[int, list[str]], output_path: str, *,
     if isinstance(data.get("graph"), dict) and "hyperedges" in data["graph"]:
         data["graph"]["hyperedges"] = hyperedges
     data["hyperedges"] = hyperedges
+    graph_metadata = data.setdefault("graph", {})
+    graph_metadata["schema_version"] = GRAPH_SCHEMA_VERSION
+    graphify_version = _graphify_version()
+    if graphify_version is not None:
+        graph_metadata["graphify_version"] = graphify_version
     # Fallback provenance comes from the repo the graph is being written INTO
     # (output_path lives in <target>/graphify-out/), never the shell's cwd —
     # the same cwd-anchoring mistake #2316 fixed for `update`.
@@ -517,6 +536,7 @@ def to_cypher(G: nx.Graph, output_path: str) -> None:
         lines.append(f"MERGE (n:{ftype} {{id: '{node_id_esc}', label: '{label}'}});")
     lines.append("")
     for u, v, data in G.edges(data=True):
+        u, v = data.get("_src", u), data.get("_tgt", v)
         rel = _cypher_label(
             (data.get("relation", "RELATES_TO") or "RELATES_TO").upper(),
             "RELATES_TO",
@@ -1223,6 +1243,7 @@ def to_canvas(
     # Generate edges - only between nodes both in canvas, cap at 200 highest-weight
     all_edges_weighted: list[tuple[float, str, str, str]] = []
     for u, v, edata in G.edges(data=True):
+        u, v = edata.get("_src", u), edata.get("_tgt", v)
         if u in all_canvas_nodes and v in all_canvas_nodes:
             weight = edata.get("weight", 1.0)
             relation = edata.get("relation", "")

{graphifyy-0.9.76 → graphifyy-0.9.78}/graphify/exporters/graphdb.py

@@ -91,8 +91,8 @@ def push_to_neo4j(
             session.run(
                 f"MATCH (a {{id: $src}}), (b {{id: $tgt}}) "
                 f"MERGE (a)-[r:{rel}]->(b) SET r += $props",
-                src=u,
-                tgt=v,
+                src=data.get("_src", u),
+                tgt=data.get("_tgt", v),
                 props=props,
             )
             edges_pushed += 1
@@ -203,7 +203,7 @@ def push_to_falkordb(
         graph.query(
             f"MATCH (a {{id: $src}}), (b {{id: $tgt}}) "
             f"MERGE (a)-[r:{rel}]->(b) SET r += $props",
-            {"src": u, "tgt": v, "props": props},
+            {"src": data.get("_src", u), "tgt": data.get("_tgt", v), "props": props},
         )
         edges_pushed += 1