codebeacon 0.6.1__tar.gz → 0.6.3__tar.gz

{codebeacon-0.6.1 → codebeacon-0.6.3}/.cursorrules

@@ -27,7 +27,7 @@ Look up by class name — replace `{project}` with the relevant folder:
 
 | Project | Notes | Example |
 | --- | --- | --- |
-| codebeacon | 398 services, 3 entities | `affected_from_paths.py.md` |
+| codebeacon | 409 services, 3 entities | `affected_from_paths.py.md` |
 
 ### Step 3 — source file
 Open the paths surfaced by Steps 1–2.
@@ -38,7 +38,7 @@ Open the paths surfaced by Steps 1–2.
 
 | Project | Framework | Routes | Services | Entities | Components |
 | --- | --- | --- | --- | --- | --- |
-| codebeacon | python | 5 | 398 | 3 | 0 |
+| codebeacon | python | 5 | 409 | 3 | 0 |
 
 ---
 
@@ -56,19 +56,19 @@ Open the paths surfaced by Steps 1–2.
 ## Architecture
 
 **codebeacon**: **python** · python
-  Routes: 5 | Services: 398 | Entities: 3 | Components: 0
+  Routes: 5 | Services: 409 | Entities: 3 | Components: 0
 
 ## High-Impact Files
 
 Changes here affect many other files:
 
-- `codebeacon/semantic_pipeline.py` (imported by 56 files)
-- `codebeacon/pipeline.py` (imported by 12 files)
-- `tests/test_graph.py` (imported by 6 files)
-- `tests/integration/test_full_pipeline.py` (imported by 3 files)
-- `tests/test_known_bugs.py` (imported by 2 files)
+- `codebeacon/common/safety.py` (imported by 9 files)
+- `codebeacon/discover/detector.py` (imported by 6 files)
+- `codebeacon/wiki/generator.py` (imported by 6 files)
+- `codebeacon/affected.py` (imported by 5 files)
+- `codebeacon/common/symbols.py` (imported by 5 files)
 
 ---
 
-_Generated by [codebeacon](https://github.com/codebeacon/codebeacon) · 2026-05-31_
+_Generated by [codebeacon](https://github.com/codebeacon/codebeacon) · 2026-06-10_
 <!-- codebeacon:end -->

{codebeacon-0.6.1 → codebeacon-0.6.3}/AGENTS.md

@@ -27,7 +27,7 @@ Look up by class name — replace `{project}` with the relevant folder:
 
 | Project | Notes | Example |
 | --- | --- | --- |
-| codebeacon | 398 services, 3 entities | `affected_from_paths.py.md` |
+| codebeacon | 409 services, 3 entities | `affected_from_paths.py.md` |
 
 ### Step 3 — source file
 Open the paths surfaced by Steps 1–2.
@@ -38,7 +38,7 @@ Open the paths surfaced by Steps 1–2.
 
 | Project | Framework | Routes | Services | Entities | Components |
 | --- | --- | --- | --- | --- | --- |
-| codebeacon | python | 5 | 398 | 3 | 0 |
+| codebeacon | python | 5 | 409 | 3 | 0 |
 
 ---
 
@@ -56,19 +56,19 @@ Open the paths surfaced by Steps 1–2.
 ## Architecture
 
 **codebeacon**: **python** · python
-  Routes: 5 | Services: 398 | Entities: 3 | Components: 0
+  Routes: 5 | Services: 409 | Entities: 3 | Components: 0
 
 ## High-Impact Files
 
 Changes here affect many other files:
 
-- `codebeacon/semantic_pipeline.py` (imported by 56 files)
-- `codebeacon/pipeline.py` (imported by 12 files)
-- `tests/test_graph.py` (imported by 6 files)
-- `tests/integration/test_full_pipeline.py` (imported by 3 files)
-- `tests/test_known_bugs.py` (imported by 2 files)
+- `codebeacon/common/safety.py` (imported by 9 files)
+- `codebeacon/discover/detector.py` (imported by 6 files)
+- `codebeacon/wiki/generator.py` (imported by 6 files)
+- `codebeacon/affected.py` (imported by 5 files)
+- `codebeacon/common/symbols.py` (imported by 5 files)
 
 ---
 
-_Generated by [codebeacon](https://github.com/codebeacon/codebeacon) · 2026-05-31_
+_Generated by [codebeacon](https://github.com/codebeacon/codebeacon) · 2026-06-10_
 <!-- codebeacon:end -->

{codebeacon-0.6.1 → codebeacon-0.6.3}/CLAUDE.md

@@ -27,7 +27,7 @@ Look up by class name — replace `{project}` with the relevant folder:
 
 | Project | Notes | Example |
 | --- | --- | --- |
-| codebeacon | 398 services, 3 entities | `affected_from_paths.py.md` |
+| codebeacon | 409 services, 3 entities | `affected_from_paths.py.md` |
 
 ### Step 3 — source file
 Open the paths surfaced by Steps 1–2.
@@ -43,7 +43,7 @@ searches that span projects.
 
 | Project | Framework | Routes | Services | Entities | Components |
 | --- | --- | --- | --- | --- | --- |
-| codebeacon | python | 5 | 398 | 3 | 0 |
+| codebeacon | python | 5 | 409 | 3 | 0 |
 
 ---
 
@@ -61,19 +61,19 @@ searches that span projects.
 ## Architecture
 
 **codebeacon**: **python** · python
-  Routes: 5 | Services: 398 | Entities: 3 | Components: 0
+  Routes: 5 | Services: 409 | Entities: 3 | Components: 0
 
 ## High-Impact Files
 
 Changes here affect many other files:
 
-- `codebeacon/semantic_pipeline.py` (imported by 56 files)
-- `codebeacon/pipeline.py` (imported by 12 files)
-- `tests/test_graph.py` (imported by 6 files)
-- `tests/integration/test_full_pipeline.py` (imported by 3 files)
-- `tests/test_known_bugs.py` (imported by 2 files)
+- `codebeacon/common/safety.py` (imported by 9 files)
+- `codebeacon/discover/detector.py` (imported by 6 files)
+- `codebeacon/wiki/generator.py` (imported by 6 files)
+- `codebeacon/affected.py` (imported by 5 files)
+- `codebeacon/common/symbols.py` (imported by 5 files)
 
 ---
 
-_Generated by [codebeacon](https://github.com/codebeacon/codebeacon) · 2026-05-31_
+_Generated by [codebeacon](https://github.com/codebeacon/codebeacon) · 2026-06-10_
 <!-- codebeacon:end -->

{codebeacon-0.6.1 → codebeacon-0.6.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codebeacon
-Version: 0.6.1
+Version: 0.6.3
 Summary: Source code AST analysis tool for AI context generation — unified multi-framework knowledge graph
 Project-URL: Homepage, https://github.com/codebeacon/codebeacon
 Project-URL: Repository, https://github.com/codebeacon/codebeacon

codebeacon-0.6.3/codebeacon/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.6.3"

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/cache.py

@@ -33,6 +33,7 @@ from __future__ import annotations
 import hashlib
 import json
 import os
+import threading
 import time
 from pathlib import Path
 from typing import Any, Optional
@@ -63,6 +64,9 @@ class Cache:
         self._hash_memo: dict[str, str] = {}
         # Memoize stat() results within a single run as well
         self._stat_memo: dict[str, tuple[int, int]] = {}
+        # One Cache instance is shared by the wave ThreadPoolExecutor workers;
+        # the get()/put() check-then-act sequences need a lock to stay atomic.
+        self._lock = threading.RLock()
 
     def _key(self, file_path: str) -> str:
         """Map a (possibly absolute) ``file_path`` to its on-disk cache key.
@@ -117,8 +121,9 @@ class Cache:
 
     def file_hash(self, file_path: str) -> str:
         """Compute (and memoize) the SHA-256 hex digest of a file's contents."""
-        if file_path in self._hash_memo:
-            return self._hash_memo[file_path]
+        with self._lock:
+            if file_path in self._hash_memo:
+                return self._hash_memo[file_path]
 
         h = hashlib.sha256()
         try:
@@ -129,7 +134,8 @@ class Cache:
         except OSError:
             digest = ""
 
-        self._hash_memo[file_path] = digest
+        with self._lock:
+            self._hash_memo[file_path] = digest
         return digest
 
     def _file_stat(self, file_path: str) -> tuple[int, int]:
@@ -167,21 +173,22 @@ class Cache:
 
     def get(self, file_path: str) -> Optional[dict]:
         """Return the cached extraction result dict, or None if stale/missing."""
-        key = self._key(file_path)
-        entry = self._data.get(key)
-        if not entry:
-            return None
-        mtime_ns, size = self._file_stat(file_path)
-        if entry.get("mtime_ns") == mtime_ns and entry.get("size") == size:
+        with self._lock:
+            key = self._key(file_path)
+            entry = self._data.get(key)
+            if not entry:
+                return None
+            mtime_ns, size = self._file_stat(file_path)
+            if entry.get("mtime_ns") == mtime_ns and entry.get("size") == size:
+                return entry.get("result")
+            if entry.get("hash") != self.file_hash(file_path):
+                return None
+            # Content unchanged despite mtime bump — refresh the stat fields so
+            # the next run skips hashing.
+            entry["mtime_ns"] = mtime_ns
+            entry["size"] = size
+            self._dirty = True
             return entry.get("result")
-        if entry.get("hash") != self.file_hash(file_path):
-            return None
-        # Content unchanged despite mtime bump — refresh the stat fields so the
-        # next run skips hashing.
-        entry["mtime_ns"] = mtime_ns
-        entry["size"] = size
-        self._dirty = True
-        return entry.get("result")
 
     def put(self, file_path: str, result: Any, file_hash: Optional[str] = None) -> None:
         """Store an extraction result for a file.
@@ -199,16 +206,17 @@ class Cache:
             except (TypeError, ImportError):
                 result = {"_raw": str(result)}
 
-        mtime_ns, size = self._file_stat(file_path)
-        self._data[self._key(file_path)] = {
-            "hash": h,
-            "result": result,
-            "ts": time.time(),
-            "mtime_ns": mtime_ns,
-            "size": size,
-        }
-        self._hash_memo[file_path] = h
-        self._dirty = True
+        with self._lock:
+            mtime_ns, size = self._file_stat(file_path)
+            self._data[self._key(file_path)] = {
+                "hash": h,
+                "result": result,
+                "ts": time.time(),
+                "mtime_ns": mtime_ns,
+                "size": size,
+            }
+            self._hash_memo[file_path] = h
+            self._dirty = True
 
     def invalidate(self, file_path: str) -> None:
         """Remove a specific file's cache entry."""

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/common/safety.py

@@ -18,6 +18,7 @@ Three concerns:
 
 from __future__ import annotations
 
+import hashlib
 import re
 import subprocess
 from pathlib import Path
@@ -69,6 +70,52 @@ def escape_frontmatter_value(value: object) -> str:
     return text
 
 
+def cap_filename(name: str, limit: int = 200) -> str:
+    """Cap a filename stem to ``limit`` UTF-8 *bytes*, collision-safely.
+
+    A single path component is capped at 255 bytes by every mainstream
+    filesystem, so a node label long enough to overflow that — roughly 85 CJK
+    characters at 3 bytes each, or 255 ASCII — makes ``Path.write_text`` raise
+    ``OSError`` (ENAMETOOLONG) and aborts the *entire* obsidian / wiki export,
+    not just the one note. We cap to 200 bytes to leave headroom for the
+    trailing ``.md`` and any ``_N`` dedup suffix the caller appends.
+
+    Two properties matter:
+
+    * The budget is counted in **bytes**, not characters, so multi-byte scripts
+      (CJK, emoji) cannot slip past a character-count guard.
+    * When truncation actually happens we append ``_<sha1[:8]>`` of the original
+      name, so two labels sharing a long common prefix
+      (``"z"*250 + "_ALPHA"`` vs ``"z"*250 + "_BETA"``) still resolve to
+      distinct files instead of silently colliding.
+
+    Truncation slices the UTF-8 byte string and decodes with ``"ignore"`` so a
+    multi-byte character straddling the cut is dropped rather than producing
+    mojibake. Short names are returned untouched. Mirrors graphify 690b4e5.
+    """
+    encoded = name.encode("utf-8")
+    if len(encoded) <= limit:
+        return name
+    digest = hashlib.sha1(name.encode("utf-8")).hexdigest()[:8]
+    keep = max(limit - 9, 0)  # room for "_" + 8 hex chars
+    truncated = encoded[:keep].decode("utf-8", "ignore")
+    return f"{truncated}_{digest}"
+
+
+def safe_wiki_filename(label: str) -> str:
+    """Filename stem for a wiki article: filesystem-safe and byte-capped.
+
+    Lives here (not in wiki/generator.py) because BOTH sides of every wiki
+    link must agree on it: the generator names the file with it, and the
+    templates build `./<stem>.md` links with it. When the two used different
+    transforms, any label with a character outside [-_.\\w] (spaces, `#`,
+    parentheses, `<>` from generics) produced a file at one path and links
+    pointing at another — every such link was dead on arrival.
+    """
+    cleaned = "".join(c if c.isalnum() or c in "-_." else "_" for c in label)
+    return cap_filename(cleaned)
+
+
 def git_head(repo_path: str | Path) -> str:
     """Return the full HEAD commit SHA for ``repo_path``, or "" if not a repo.
 

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/common/symbols.py

@@ -27,6 +27,8 @@ class SymbolTable:
         self._implements_map: dict[str, list[str]] = {}
         # All known node IDs
         self._node_ids: set[str] = set()
+        # node_id → source_file, so resolved edges carry a real file path
+        self._node_files: dict[str, str] = {}
 
     def build(self, nodes: list[Node]) -> None:
         """Build symbol maps from a flat list of all extracted nodes.
@@ -35,6 +37,7 @@ class SymbolTable:
         """
         for node in nodes:
             self._node_ids.add(node.id)
+            self._node_files[node.id] = node.source_file or ""
 
             label = node.label
             if label not in self._class_map:
@@ -91,7 +94,10 @@ class SymbolTable:
             relation="injects",
             confidence="INFERRED" if is_interface_resolved else "EXTRACTED",
             confidence_score=0.8 if is_interface_resolved else 1.0,
-            source_file=ref.source_node_id,
+            # The source node's actual file — Edge.source_file is a file path
+            # everywhere else; stamping the node ID here leaked "proj::Name"
+            # strings into beacon.json's source_file fields.
+            source_file=self._node_files.get(ref.source_node_id, ""),
         )
 
     def resolve_all(

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/common/types.py

@@ -62,6 +62,8 @@ class ServiceInfo:
     methods: list = field(default_factory=list)    # list[str] - method names
     dependencies: list = field(default_factory=list)  # list[str] - injected type names (unresolved)
     annotations: list = field(default_factory=list)   # list[str] - @Service, @Injectable, etc.
+    implements: list = field(default_factory=list)    # list[str] - implemented interface names
+    extends: list = field(default_factory=list)       # list[str] - superclass names
 
 
 @dataclass(slots=True)

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/config.py

@@ -70,6 +70,10 @@ def load_config(path: str | Path) -> CodebeaconConfig:
 
     projects = []
     for p in projects_raw:
+        # A bare `-` in the YAML list arrives as None; a scalar entry as str.
+        # Both must surface as a config error, not a TypeError traceback.
+        if not isinstance(p, dict):
+            raise ValueError(f"Invalid project entry (expected mapping): {p!r}")
         if "name" not in p or "path" not in p:
             raise ValueError(f"Project entry missing 'name' or 'path': {p}")
         # Resolve path relative to config file location
@@ -82,8 +86,10 @@ def load_config(path: str | Path) -> CodebeaconConfig:
             type=p.get("type", "auto"),
         ))
 
-    output_raw = raw.get("output", {})
-    context_map = output_raw.get("context_map", {})
+    # `.get(key, {})` is not enough: a section present-but-empty in the YAML
+    # (`output:` on its own line) loads as None, and None.get(...) crashes.
+    output_raw = raw.get("output") or {}
+    context_map = output_raw.get("context_map") or {}
     output = OutputConfig(
         dir=output_raw.get("dir", ".codebeacon"),
         wiki=output_raw.get("wiki", True),
@@ -91,14 +97,14 @@ def load_config(path: str | Path) -> CodebeaconConfig:
         context_map_targets=context_map.get("targets", ["CLAUDE.md", ".cursorrules", "AGENTS.md"]),
     )
 
-    wave_raw = raw.get("wave", {})
+    wave_raw = raw.get("wave") or {}
     wave = WaveConfig(
         auto=wave_raw.get("auto", True),
         chunk_size=wave_raw.get("chunk_size", 300),
         max_parallel=wave_raw.get("max_parallel", 5),
     )
 
-    semantic_raw = raw.get("semantic", {})
+    semantic_raw = raw.get("semantic") or {}
     semantic = SemanticConfig(
         enabled=semantic_raw.get("enabled", False),
     )

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/contextmap/generator.py

@@ -247,14 +247,26 @@ def _collect_stats(G: nx.DiGraph) -> dict[str, dict[str, int]]:
 
 
 def _hub_files(G: nx.DiGraph, top_n: int = 5) -> list[tuple[str, int]]:
-    """Return (file_path, import_count) for the most-imported source files."""
-    counter: dict[str, int] = defaultdict(int)
-    for _, _, data in G.edges(data=True):
-        if data.get("relation") in ("imports", "imports_from"):
-            sf = data.get("source_file", "")
-            if sf:
-                counter[sf] += 1
-    ranked = sorted(counter.items(), key=lambda x: x[1], reverse=True)
+    """Return (file_path, importer_count) for the most-imported source files.
+
+    Counts DISTINCT importing files per edge-target file, mirroring
+    analyze.hub_files. The old version counted `edge["source_file"]` — the
+    importer's own file — so CLAUDE.md's "High-Impact Files" actually listed
+    the files that *import the most* (entry points), with counts inflated by
+    per-node edge remapping (e.g. "imported by 392 files" in a 60-file repo).
+    """
+    importers: dict[str, set[str]] = defaultdict(set)
+    for src, tgt, data in G.edges(data=True):
+        if data.get("relation") not in ("imports", "imports_from"):
+            continue
+        tgt_file = G.nodes[tgt].get("source_file", "")
+        src_file = G.nodes[src].get("source_file", "")
+        if tgt_file and src_file and tgt_file != src_file:
+            importers[tgt_file].add(src_file)
+    ranked = sorted(
+        ((fp, len(srcs)) for fp, srcs in importers.items()),
+        key=lambda x: (-x[1], x[0]),
+    )
     return ranked[:top_n]
 
 

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/discover/detector.py

@@ -218,10 +218,19 @@ def _detect_language_from_files(directory: Path) -> str:
         ".swift": "swift",
     }
     try:
-        for entry in directory.rglob("*"):
-            if entry.is_file() and entry.suffix in ext_to_lang:
-                lang = ext_to_lang[entry.suffix]
-                counts[lang] = counts.get(lang, 0) + 1
+        # os.walk with pruning, not rglob: an unbounded rglob descends into
+        # node_modules / .git / dist, which on a Node repo means enumerating
+        # tens of thousands of vendored files (and skews the language vote
+        # toward vendored JS) just to answer "what language is this project".
+        for dirpath, dirnames, filenames in os.walk(directory):
+            dirnames[:] = [
+                d for d in dirnames if d not in _SKIP_DIRS and not d.startswith(".")
+            ]
+            for fname in filenames:
+                ext = os.path.splitext(fname)[1]
+                if ext in ext_to_lang:
+                    lang = ext_to_lang[ext]
+                    counts[lang] = counts.get(lang, 0) + 1
     except (PermissionError, OSError):
         pass
 

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/export/hooks.py

@@ -25,7 +25,7 @@ import sys
 from pathlib import Path
 
 
-_POST_COMMIT_HOOK = r"""#!/usr/bin/env bash
+_POST_COMMIT_HOOK_TEMPLATE = r"""#!/usr/bin/env bash
 # codebeacon: incremental rebuild after each commit.
 # Detaches so `git commit` returns immediately; output goes to the log below.
 #
@@ -54,14 +54,51 @@ if [ -z "$CHANGED" ] || ! echo "$CHANGED" | grep -v '^\.codebeacon/' | grep -E "
   exit 0
 fi
 
-(
-  nohup codebeacon scan . --update >>"$LOG" 2>&1 &
-) >/dev/null 2>&1 &
-disown 2>/dev/null || true
+# Interpreter pinned at install time so the hook still works where the
+# `codebeacon` launcher is not on PATH — GUI git clients and CI runners
+# don't inherit a login shell's PATH, which made the old `nohup codebeacon`
+# form silently no-op. The heredoc detaches via subprocess so no `nohup`
+# is needed (it doesn't exist on Windows). Mirrors graphify #1127/#1161.
+CODEBEACON_PYTHON=__CODEBEACON_PYTHON__
+
+if [ -x "$CODEBEACON_PYTHON" ]; then
+  "$CODEBEACON_PYTHON" - "$LOG" <<'CODEBEACON_PY' >>"$LOG" 2>&1
+import subprocess, sys
+kwargs = {"stdin": subprocess.DEVNULL}
+if sys.platform == "win32":
+    kwargs["creationflags"] = 0x00000208  # DETACHED_PROCESS | CREATE_NEW_PROCESS_GROUP
+else:
+    kwargs["start_new_session"] = True
+log = open(sys.argv[1], "ab")
+subprocess.Popen(
+    [sys.executable, "-m", "codebeacon", "scan", ".", "--update"],
+    stdout=log, stderr=log, **kwargs,
+)
+CODEBEACON_PY
+else
+  # Pinned interpreter vanished (reinstall, deleted venv) — fall back to PATH.
+  (
+    nohup codebeacon scan . --update >>"$LOG" 2>&1 &
+  ) >/dev/null 2>&1 &
+  disown 2>/dev/null || true
+fi
 exit 0
 """
 
 
+def _render_post_commit_hook() -> str:
+    """Render the hook with the current interpreter pinned.
+
+    ``as_posix()`` keeps the path usable from git's bundled sh on Windows
+    (``C:/Python311/python.exe``); single quotes stop spaces or shell
+    metacharacters in the path from being interpreted.
+    """
+    python_path = "'" + Path(sys.executable).as_posix().replace("'", "'\\''") + "'"
+    # Plain token replacement — str.format would choke on the template's own
+    # shell braces (`${HOME}`, the heredoc's dict literal).
+    return _POST_COMMIT_HOOK_TEMPLATE.replace("__CODEBEACON_PYTHON__", python_path)
+
+
 def install_hooks(repo_path: str | Path) -> int:
     """Install the merge driver, ``.gitattributes`` entry, and post-commit hook.
 
@@ -105,8 +142,13 @@ def _is_git_repo(repo: Path) -> bool:
 
 def _register_merge_driver(repo: Path) -> None:
     """Register the ``codebeacon`` merge driver in ``.git/config``."""
-    codebeacon = shutil.which("codebeacon") or "codebeacon"
-    cmd = f'{codebeacon} merge-driver "%O" "%A" "%B"'
+    codebeacon = shutil.which("codebeacon")
+    if codebeacon:
+        cmd = f'"{Path(codebeacon).as_posix()}" merge-driver "%O" "%A" "%B"'
+    else:
+        # Launcher not on PATH (uv tool / pipx install, GUI client) — invoke
+        # the module through the pinned interpreter instead.
+        cmd = f'"{Path(sys.executable).as_posix()}" -m codebeacon merge-driver "%O" "%A" "%B"'
     subprocess.run(
         ["git", "config", "--local", "merge.codebeacon.name", "codebeacon union merge"],
         cwd=str(repo),
@@ -134,9 +176,17 @@ def _install_post_commit(repo: Path) -> None:
     hooks_dir = _hooks_dir(repo)
     hooks_dir.mkdir(parents=True, exist_ok=True)
     hook = hooks_dir / "post-commit"
-    if hook.exists() and "codebeacon" in hook.read_text(encoding="utf-8"):
-        return
-    hook.write_text(_POST_COMMIT_HOOK, encoding="utf-8")
+    content = _render_post_commit_hook()
+    if hook.exists():
+        existing = hook.read_text(encoding="utf-8")
+        if existing == content:
+            return
+        # A codebeacon-authored hook (our marker header) gets refreshed so
+        # re-running `hook install` picks up template fixes; anything else
+        # that merely mentions codebeacon is the user's — leave it alone.
+        if "codebeacon: incremental rebuild" not in existing and "codebeacon" in existing:
+            return
+    hook.write_text(content, encoding="utf-8")
     hook.chmod(hook.stat().st_mode | stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH)
 
 

{codebeacon-0.6.1 → codebeacon-0.6.3}/codebeacon/export/obsidian.py

@@ -33,7 +33,7 @@ from typing import Any
 
 import networkx as nx
 
-from codebeacon.common.safety import escape_frontmatter_value, sanitize_label
+from codebeacon.common.safety import cap_filename, escape_frontmatter_value, sanitize_label
 
 
 # ── Regexes ────────────────────────────────────────────────────────────────────
@@ -402,9 +402,18 @@ def _reverse_relation(relation: str) -> str:
 
 
 def _safe_note_name(label: str) -> str:
-    """Convert node label to a safe filename stem (no path separators)."""
+    """Convert node label to a safe filename stem (no path separators).
+
+    The byte cap (via ``cap_filename``) prevents a very long label — or an 85+
+    character CJK class name — from overflowing the 255-byte filesystem limit
+    and crashing ``_step1_generate_notes`` with ENAMETOOLONG. Because every
+    wikilink target is generated through this same function, the note's
+    filename and the links pointing at it stay byte-for-byte consistent even
+    after truncation.
+    """
     # Replace characters that confuse Obsidian wikilinks
-    return re.sub(r'[/\\#^|[\]]', "_", label).strip()
+    cleaned = re.sub(r'[/\\#^|[\]]', "_", label).strip()
+    return cap_filename(cleaned)
 
 
 # ── Step 2: Fix broken wikilinks ──────────────────────────────────────────────
@@ -632,7 +641,11 @@ def _step8_move_remaining(vault: Path) -> None:
         content = md.read_text(errors="ignore")
         m = _PROJECT_RE.search(content)
         if m and m.group(1):
-            svc = m.group(1)
+            # Cap here so the folder, the Step-9 hub note (`<svc>/<svc>.md`),
+            # and every backlink derived from the folder name stay consistent
+            # and under the 255-byte filesystem limit — note filenames are
+            # capped in _safe_note_name but this path wasn't.
+            svc = cap_filename(m.group(1))
             dest_dir = vault / svc
             dest_dir.mkdir(exist_ok=True)
             dest = dest_dir / md.name