A type-qualified Swift call (`Type.staticMethod()`, `Singleton.shared.method()`)
names the receiver type explicitly in source, so the resolved edge is an exact
reference — now emitted as EXTRACTED (1.0), matching the Python
qualified-class-method pass (_resolve_python_member_calls). Instance calls whose
receiver type comes from local inference (`obj.method()`) stay INFERRED (0.8).
Resolution and the single-definition god-node guard are unchanged.
This addresses the actionable part of #1533's "static calls" report: the edge
was always produced (graphify models calls as method->method), it was just
under-confident. Updated the confidence test to assert the instance/type-qualified
split.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Three residual ObjC extractor bugs from the #1475 thread, each reproduced
against the real tree-sitter-objc grammar:
1. NS_ASSUME_NONNULL_BEGIN before @interface made the parser fail to emit
a class_interface node at all (the whole interface was swallowed into
ERROR nodes), so headers using the macro produced no class node. Blank
the two argument-less annotation macros to equal-length spaces before
parsing (offset-preserving; macro-free files are byte-identical). The
reporter's "@class breaks it" hypothesis was wrong — only the macro does.
2. Quoted `#import "X.h"` edges dangled once a `.h`/`.m` pair existed: the
target used the bare stem, which the post-pass canonicalizes and then
_disambiguate_colliding_node_ids salts apart by path, so the import
target no longer matched. Resolve the include to a real file (mirroring
_import_c), and repoint imports/imports_from edges to the header variant
in _disambiguate_colliding_node_ids — taking precedence over the
same-source-file salt so a `.m` importing its own `.h` resolves to the
header instead of self-looping. Also repairs the equivalent latent
C-include dangling bug.
3. `[[Foo alloc] init]` produced no edge — walk_calls only reconstructed
selectors and skipped the receiver. Emit a `references` edge from the
allocating method to the class, resolved via the unique-class stub guard
(ensure_named_node + _rewire_unique_stub_nodes) so unknown/ambiguous
names produce no false edge. The calls-to-init edge is deliberately
deferred (init selectors are ambiguous across classes).
Reported by JabberYQ with a precise repro and test repo. Adds regression
tests incl. a self-loop guard on the import edges. Still open on #1475:
dot-syntax property accesses (Bug 5) and @selector target-action (Bug 6b).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Workspace imports with subpath exports (e.g.
`import { x } from "@scope/pkg/browser"`) now resolve through the
package's `exports` map instead of falling back to a bare path. Supports
string values, condition objects, nested conditions, and single-`*`
wildcard patterns (`"./*": "./src/*.js"`), falling back to the existing
bare-path/index resolution when there is no exports map or no match.
Adapted from #1541, taking only the exports-map resolver and not that
PR's competing import-node-ID normalization (current v8 already resolves
the node-ID mismatch via the #1529 id-remap post-pass, and the PR's
_file_stem approach regressed the relative-input alias case). Two
hardening changes over the original:
- `default` is consulted LAST in the condition priority (it is Node's
catch-all), so a matching `import`/`module`/`svelte` condition wins.
- Export targets that escape the package directory are rejected
(`_contained_in_package`), so a malicious `exports` value like
`"./x": "../../../etc/..."` cannot resolve to a file outside the package.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The AST cache is version-swept but the semantic/LLM cache had no pruning,
so it grew unbounded: it is content-hash-keyed, so every content change
or file deletion leaves a permanent orphan entry (reporter saw 152
entries for 124 live docs). This matters for the committed-cache workflow
where the semantic cache is published for warm CI rebuilds.
Adds prune_semantic_cache(root, live_hashes) and wires it into the end of
the extract path, sweeping cache/semantic/*.json entries whose hash is not
in the live set. The live set is computed from the FULL detected document
set (not the incremental changed-subset, which would delete valid
entries), using the same file_hash recipe save_semantic_cache uses.
Best-effort (unlink guarded), only touches cache/semantic/ (.tmp and
cache/ast/** untouched), and keeps the semantic cache unversioned so
releases never re-bill LLM extraction.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
#1529 (regression from the 0.9.0 full-repo-relative node-ID migration):
relative JS/TS imports resolve to repo-relative paths and ride the
extract() id-remap to canonical node IDs, but tsconfig path-alias and
workspace-package imports resolve to ABSOLUTE paths (their bases are
.resolve()'d), so the import-target ID baked in the on-disk prefix and
never matched the repo-relative definition node — the edge was dropped at
build (common on Next.js/SvelteKit `@/`-alias codebases). The id-remap
post-pass now also registers the absolute-resolved form of each input
path (file-level edges) and both the input-form and absolute-form symbol
prefixes (named-import edges), so alias/workspace import targets remap to
the canonical ID. Verified the built graph has no orphan nodes or
dangling edges.
#1531: tsconfig `paths` values are ordered fallback lists (tsc tries each
target until one resolves), but only targets[0] was kept. The alias map
now stores all targets in order, and a single _resolve_tsconfig_alias
helper (replacing six duplicated inline loops) returns the first target
whose candidate exists on disk, falling back to the first candidate when
none exist (no false edge). Wildcards, baseUrl, and array `extends` are
preserved.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Generated install/skill guidance told agents to invoke a literal `skill`
tool with `skill: "graphify"`, which is host-specific and not valid in
every environment. The always-on AGENTS fragment, packaged artifact,
expected snapshot, and _skill_registration() output now use host-generic
wording: "use the installed graphify skill or instructions". Also decodes
skillgen git blob reads as UTF-8 for Windows and replaces stale English
code-block examples in the translated READMEs.
The always-on roundtrip guard deliberately freezes the v8 baseline, so an
intentional wording change would otherwise fail it. Rather than only
patching the pytest mirror (which left the blocking CLI guard
--always-on-roundtrip red, as the original PR did), this adds an explicit,
reviewable ALWAYS_ON_SANCTIONED_EDITS registry: the guard applies the
approved old->new substitution to the baseline before the byte-for-byte
compare, so this exact sentence is allowed while any other drift still
fails. CLI guard and pytest test now agree and CI passes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two cross-platform fixes salvaged from #1502:
- to_graphml: nx.write_graphml raises ValueError on None attribute
values, so a node/edge carrying a null field crashed the export.
Coerce None -> "" for node and edge attributes before writing.
- save-result: add --answer-file as an alternative to --answer so long
or multiline answers can be passed via a file instead of a fragile
inline shell arg (notably Windows/PowerShell quoting). Exactly one of
--answer / --answer-file is required.
The rest of #1502 (a version downgrade and a hand-edited generated
skill-windows.md that fails skillgen --check, plus duplicated
windows-scripts) is left for rework on the PR.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_call_llm (used by the dedup LLM tiebreaker) built its Anthropic and
OpenAI-compatible clients with max_retries but no timeout, so requests
on this path silently ignored GRAPHIFY_API_TIMEOUT — unlike the primary
extraction paths (_call_openai_compat / _call_claude) which already pass
both. Add timeout=_resolve_api_timeout() to both constructors.
The PR branch self-neutralized: a v8 merge resolved the conflict in
favor of the max_retries-bearing line and dropped the original one-line
fix, so it is re-applied here on top of current v8 with max_retries
preserved. Adds regression coverage for both _call_llm branches, which
were previously untested.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Resolve Ruby `obj.method()` calls by the inferred type of the receiver
instead of by globally-unique method name. `p = Processor.new; p.run`
now emits a `calls` edge to `Processor#run` and survives name collisions
with unrelated `Worker#run` definitions, where the old name-based match
either resolved by luck or dropped the edge as ambiguous.
Introduces graphify/resolver_registry.py, a behavior-identical
formalization of the existing tail-of-extract() language resolution
passes (Swift #1356, Python #1446 become registered entries), and
graphify/ruby_resolution.py, its first new consumer. Receiver type is
inferred only from unambiguous local `var = ClassName.new` bindings;
ambiguous or unknown receivers resolve to nothing (no false positives).
Note: Ruby member calls are now excluded from name-based cross-file
resolution and resolved by inferred type only. This is an intentional
precision-over-recall change scoped to Ruby: a cross-file `var.method`
whose receiver type cannot be proven from a local `X.new` binding no
longer resolves by name-luck (it produces no edge rather than a possibly
wrong one), matching the project's confidence model.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
These were committed before .gitignore included the .DS_Store rule, so
gitignore never removed them from tracking. Untrack them (they remain
on local disk, just leave git) — the existing .gitignore rule keeps
them out going forward.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
On strict per-org concurrency/RPM caps (notably Moonshot/kimi), a parallel
`graphify extract --force` 429'd, and because the provider SDKs default to only
max_retries=2, the chunk gave up after two attempts, logged `chunk N failed`, and
was silently dropped — an incomplete graph plus the screen full of
rate_limit_reached_error spam in the report. The SDKs already back off
exponentially and honor Retry-After; they just need more attempts to outlast the
rate window.
The OpenAI-compatible, Azure, and Anthropic clients are now built with
max_retries=_resolve_max_retries() (default 6, override via GRAPHIFY_MAX_RETRIES;
0 disables). For very tight accounts, --max-concurrency 1 further cuts the
concurrency that trips org-level limits.
Reported by @bercedev (#1523). Tests: _resolve_max_retries default/env/invalid,
and the OpenAI-compatible client is constructed with retries.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A Java record's component types produced no edges, so a record's data dependencies
were invisible. The record_declaration handler now emits `references` edges from
each component type (field context; generic_arg for type arguments), skipping
primitives and — via #1518's in-scope type-parameter filtering — the record's own
type parameters. Handles standard, generic, primitive, and varargs components.
Ported from PR #1520 by @oleksii-tumanov (clean 3-way onto v8; builds on #1518).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
0.9.0 made the node-ID stem the full repo-relative path, but normalize_id collapses
every non-word run to "_", so the path separator is indistinguishable from inner
punctuation: foo/bar_baz.py and foo_bar/baz.py both normalized to foo_bar_baz and
still silently merged (the residual of #1504). The existing _disambiguate_colliding
_node_ids salt didn't help — it salted with _make_id(source_key, old_id), which
re-normalizes the path with the same lossy recipe, so the two colliders produced an
identical salted id.
When two distinct source paths' naive salts still collide, append a short stable
sha1(source_key)[:6] — injective over distinct paths — so they separate. Computed in
code from source_file (never trusted from the LLM), so AST<->semantic parity holds.
Blast radius: minimal/non-breaking — only the actual residual colliders get a hash
suffix. Non-colliding ids (the 99%, incl. the common #1504 case like two README.md
in different dirs) are byte-identical to 0.9.0 (verified: src/auth/session.py ->
src_auth_session, docs/v1/api/README.md -> docs_v1_api_readme unchanged). This is a
0.9.1 patch, not another migration.
Reported by @sub4biz (#1522). Regression tests cover both the collider-separation
and the non-collider-unchanged cases.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
`graphify update` preserved old edges keyed only on endpoint-node membership, so a
removed import's edge survived forever whenever both endpoint nodes still existed
(e.g. a.py no longer imports b, but both a and b are still present). The stale edge
drove phantom circular-dependency findings; `--force` didn't help (it only gates
the node-count shrink guard), only a clean rebuild did.
_rebuild_code (watch.py) now also drops preserved edges whose source_file was
re-extracted this run — an edge is owned by the file it was extracted from, and the
fresh extraction re-emits whichever ones still exist. Scoped to source_file
ownership (and the full-rebuild case, where evict_sources lists only deleted
files), so edges with no source_file or owned by a non-re-extracted file are kept —
cross-file edges that merely point at a re-extracted file (#1402 sourceless
stubs/rewire) are not over-pruned.
Reported by @UltronOfSpace (#1521). Regression test: a removed import's edge is
gone after update; verified no over-prune (still-present imports survive) and no
regression to deleted-file pruning.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The generic-parent support (#1511) emitted a spurious references/generic_arg edge
(and a sourceless stub node) for a bare type parameter — the T in
`class Box<T> extends Container<T>` — because the Java extractor had no type-
parameter tracking. _java_collect_type_refs now collects the in-scope type-
parameter names (walking class/interface/record/method/constructor declarations,
including bounded and multiple params) and skips references to them, while keeping
every real type. Scoped to the declaring node, so a real class that shares a name
with a type parameter elsewhere is still referenced.
Ported from PR #1518 by @oleksii-tumanov. Closes the nit flagged when #1511 landed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The #1462 same-label cross-file stub disambiguation (the origin_file key) only
existed in the generic extractor, so the six dedicated extractors — Julia,
Fortran, Go, Rust, PowerShell, ObjC — still collapsed same-named imported-type
stubs from different files into one conflated bare-id node (a false cross-package
link). Each now sets origin_file on its sourceless stub, identical to the generic
extractor; the generic _node_disambiguation_source_key consumes it, so two files
importing the same type stay distinct while source_file stays empty (the #1402
rewire onto a real definition is unchanged).
Ported from PR #1515 by @TPAteeq. Must ship with #1516: this widens origin_file to
six more languages, and #1516 is what strips it from graph.json. Verified: a 2-file
Go corpus now yields 2 distinct Widget stubs AND graph.json carries no origin_file.
Resolved a test-insertion conflict with #1516 by keeping both tests.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The origin_file disambiguation field (#1462) is internal — it's consumed by the
colliding-id pass and must not be persisted. In 0.9.0 it shipped into graph.json
as an absolute, machine-specific path (the same portability bug class as #555,
#932), breaking on clone/move. It's now popped at the single extract() output
chokepoint, AFTER _disambiguate_colliding_node_ids has consumed it, so all persist
paths (clustered, --no-cluster, cluster-only) emit clean output. _origin is kept
(the incremental watcher relies on it, #1116).
Ported from PR #1516 by @TPAteeq. Verified: fresh extract (clustered and
--no-cluster) produces graph.json with zero origin_file fields; the #1462 same-
label disambiguation still works (it runs before the strip).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The graph_has_legacy_ids nudge flagged fresh v9 graphs that contain Go code,
because the Go extractor scopes a type node by package directory
(_make_id(pkg_dir, name) -> "sub_thing"), which coincides with the OLD one-parent
file-stem form of pkg/sub/thing.go even though it's a current, by-design
package-scoped id (methods on a type across files in a package share one node).
The detector now inspects only file-level nodes (source_location "L1"), whose id is
unambiguously the file stem and which the migration actually re-keys. This still
catches a genuine pre-#1504 graph (its file nodes carry the old stem) while leaving
package/dir-scoped symbol ids alone. Verified end-to-end: a fresh Go graph no longer
triggers the rebuild nudge; a real legacy graph still does.
The migration itself was already correct — file nodes and Python/AST symbols
migrate to full-path ids; Go's package-scoped type ids are intentional and
unaffected. This only fixes the over-eager warning.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two pre-0.9.0 safety to-dos from the migration spike:
1. Auto-detect-and-warn: graph_has_legacy_ids() samples a loaded graph's node ids
and detects the pre-#1504 (parent-dir/filename) scheme. Read-only consumers that
don't re-extract — `graphify query` and the MCP serve loader — now print a
one-line nudge to rebuild with `extract --force` when they load an old graph.
Fires only on legacy graphs (verified: canonical graphs stay silent).
2. Re-key source_file contract: pinned with tests that a relative source_file is
migrated to the full-path id while an absolute/unrelativizable one is skipped
(its on-disk path must never leak into a persisted id). The is_absolute() guard
already enforced this; the tests make the contract explicit.
Full suite 2507 passed. Lands on v9.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
BREAKING node-ID format change. The stem that prefixes every node id was the
immediate parent dir + filename, so same-named files in different directories
collided into one last-writer-wins node and silently dropped graph content
(docs/v1/api/README.md and docs/v2/api/README.md both -> api_readme). The stem is
now the full repo-relative path (docs_v1_api_readme vs docs_v2_api_readme);
top-level files are unchanged (setup.py -> setup).
- extractors/base.py::_file_stem -> full path (as_posix; make_id collapses
separators). The two hand-copied stems (symbol_resolution, mcp_ingest) now
import the canonical one, so they can't drift again.
- llm.py system prompt + extraction-spec fragments aligned to the same rule,
fixing the #1509 AST<->LLM divergence (prompt used zero parent dirs -> ghosts).
Regenerated + blessed the per-host specs.
- build_from_json deterministically re-keys every non-AST node id from its
source_file via the new stem (and registers old-stem aliases), so a cached or
pre-migration semantic fragment carrying an old short id reconciles with the
AST node instead of spawning a ghost. The semantic cache is unversioned, so
this code-side re-key (not LLM prose) is what makes it survive the format change
with no re-bill. AST nodes are already canonical and skipped.
- Migration: existing graphs migrate automatically on the next build/update (the
re-key runs in build_from_json, including the whole graph fed back through
build_merge); `graphify extract --force` for a clean rebuild. Neo4j persisted
stores need a re-import; GraphML layouts go stale.
Full suite green (2505 passed); #1504 collision fixed and old-id fragments
re-keyed verified by smoke. Lands on v9 for review before a 0.9.0 release.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds an opt-in --timing flag that prints per-stage wall-clock timings to stderr so
slow stages are visible on large corpora. extract reports detect / AST extract /
semantic extract / build / cluster / analyze / export (and write on --no-cluster);
cluster-only reports load / cluster / analyze / label / report / export; both end
with a total. A small _StageTimer helper uses monotonic perf_counter.
Off by default and stderr-only, so default output and machine-read stdout/graph.json
are byte-identical (the extract arg loop already swallows unknown flags). Added a
regression test asserting the lines appear with --timing and are absent without it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
to_obsidian wrote one note per node straight into the target directory and
unconditionally replaced .obsidian/graph.json. Pointing --obsidian-dir at a real
vault could therefore clobber a user note whose name matched a graph node
(Database.md) and destroy the user's graph-view settings — silently, no backup,
irreversible.
graphify now records the files it owns in .graphify_obsidian_manifest.json and
refuses to overwrite any pre-existing file it didn't create: such a file is skipped
and reported in a single aggregated warning. A re-run still updates graphify's own
notes (they're in the manifest), and .obsidian/graph.json is only written when it
doesn't already exist or graphify owns it. The default graphify-out/obsidian output
and the flat note layout are unchanged.
Added regression tests: existing-vault preserves user note + .obsidian settings,
empty dir still gets the full vault, and a re-run updates own notes but not a
user-added file. The CHANGELOG also records the @oleksii-tumanov Java fixes
(#1512/#1510) and the @nuthalapativarun Windows GBK fix (#1505) committed just prior.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
On Windows where claude.cmd emits GBK/cp936 bytes, the claude-cli subprocess
decoding raised UnicodeDecodeError and crashed extraction. Both claude-cli
subprocess.run sites (_call_claude_cli and the claude-cli branch of _call_llm) now
pass errors="replace", so incidental non-UTF8 console chatter decodes to
replacement chars instead of crashing — the structured JSON payload (ASCII/UTF-8
on stdout) is unaffected. capture_output=True means the single errors= covers both
stdout and stderr.
Ported from PR #1507 by @nuthalapativarun (dropped an unrelated .gitignore change).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A generic parent — `class Foo extends Bar<T>` or `implements List<T>` — is a
`generic_type` node in the tree, but the inheritance extractor only fired on a
bare `type_identifier`, so those inherits/implements edges were silently dropped.
The parent type is now unwrapped to its base (`Bar<T>` -> `Bar`) for the
inherits/implements edge, and the type arguments are emitted as `generic_arg`
references.
Ported from PR #1511 by @oleksii-tumanov. Known pre-existing limitation (not
introduced here, worth a separate follow-up): the extractor has no type-parameter
tracking, so a bare parameter like `T` in `extends Container<T>` still produces a
`generic_arg` reference to `T`; the inherits/implements edge itself always targets
the real base type, never `T`.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Java `enum` and `@interface` (annotation) declarations weren't in _JAVA_CONFIG's
class_types, so they were never emitted as type nodes — a field typed as an enum or
a class annotated with a project annotation referenced a node that didn't exist
(dangling). enum_declaration and annotation_type_declaration are now extracted as
first-class (sourced) type nodes, so those references resolve onto them.
Ported from PR #1513 by @oleksii-tumanov.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A path query like `explain "app/api/route.ts"` tokenized to terms that matched no
node, so explain/affected returned "No node matching". Source-file paths are now
part of the search index and matched exactly (serve._find_node gains a leading
source-exact tier; affected.resolve_seed gains a source-file match). When several
nodes share a source_file (e.g. a file-level node plus a function node), the lookup
prefers the file-level node — the L1 node whose label basename matches the queried
filename, falling back to the unique L1 or unique basename match, else None.
Ported from PR #1503 by @behavio1. Maintainer fixes on top: aligned trailing-
separator handling between resolve_seed and _find_node (affected previously
returned None for a trailing-slash path that explain resolved), corrected the
stale "three-tier" _find_node docstring, and added regression tests for the
trailing-slash parity and the ambiguous-no-file-node -> None case.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two files that both import and use the same type as a bare name (e.g.
`from pathlib import Path` used as a type hint in a.py and b.py) collapsed into one
node, even with no project definition to anchor them. The referencing file is now
recorded in an internal `origin_file` field and used as the disambiguation key when
_disambiguate_colliding_node_ids splits same-id nodes — while source_file stays
empty, so the corpus-level rewire still collapses these stubs onto a real project
definition when one exists (the #1402 path is untouched). origin_file is read only
inside disambiguation; the rewire and the Java/C# type resolvers key off
source_file as before.
Ported from PR #1479 by @jiangyq9 (squash-merged onto current v8; the branch's
stale base made the raw diff appear to revert later features — the 3-way merge
applies only the surgical origin_file change). Resolved a test-adjacency conflict
with #1500 by keeping both cross-file tests.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds _resolve_csharp_type_references (graphify/extractors/csharp.py), the C#
counterpart to _resolve_java_type_references. It re-points dangling
inherits/implements/references edges from no-source shadow stubs onto the real
definitions, disambiguating same-named types across namespaces using the
referencing file's `using` directives + enclosing namespace; ambiguous matches
are refused (unique-hit guardrail), not guessed. Runs after id-disambiguation and
the sourceless-stub rewire, on the ambiguous remainder, behind a log-and-skip
try/except. _CSHARP_CONFIG is broadened to extract enum/struct/record as type
definitions so references to them resolve too.
Ported from PR #1466 by @TheFedaikin. Known follow-up: types declared in
nested/multiple namespaces in one file aren't registered as targets yet (fails
safe — under-resolves to a stub, never mis-resolves). Advances #1318 for C#.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Go's ensure_named_node emitted a SOURCED stub for a cross-file type reference,
unlike the generic/ObjC/Java extractors which emit sourceless ones. A sourced stub
bakes the referencing file's path into the node id at disambiguation time, which
blocks the corpus-level rewire from collapsing it onto the real definition — so a
type defined in one Go file and referenced from two others produced three nodes
(the real def plus two phantom per-file duplicates). Go now emits a sourceless
stub like the other languages, so the rewire collapses them to a single node.
Ported from PR #1501 by @TPAteeq. Closes the #1402 gap for Go.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
`graphify label --missing-only` restricts LLM labeling to communities that are
unnamed or still hold a `Community N` placeholder, preserving existing
non-placeholder labels read from .graphify_labels.json and merging new labels over
them. Lets a large graph be relabeled incrementally without re-naming (and paying
for) communities that already have good names.
Ported from PR #1481 by @jiangyq9. This supersedes the earlier #1421 by
@matiasduartee, which proposed the same flag — credit to @matiasduartee for the
original; #1481 is written against the current label signature (post-#1390
max-concurrency/batch-size) and merges clean, where #1421 had drifted.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Metal Shading Language is C++14, so .metal files are classified as code and routed
through the existing C++ extractor, mirroring the CUDA .cu/.cuh reuse. Also adds
.cu/.cuh/.metal to build.py's _LANG_FAMILY map (they were all missing), so the
cross-language phantom-`calls`-edge filter treats them as C++. README language
table updated.
Ported from PR #1480 by @jiangyq9. This supersedes the earlier #1450 by
@GoodOlClint (same .metal -> C++ approach and fixture) — credit to @GoodOlClint
for the original; #1480 additionally closes the CUDA family-map gap and updates
the docs, and merges clean against current v8.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
to_wiki emitted Obsidian `[[Title]]` wikilinks. Outside Obsidian `[[Domain Data
Models]]` resolves to a literal "Domain Data Models.md", but the article file is
the slugged "Domain_Data_Models.md", so nearly every community/god-node link
opened an empty page; god-node articles also linked node-level neighbors that
never get an article file, so those were dead even inside Obsidian.
Links are now standard `[display](slug.md)` with the target URL-encoded (spaces,
&, parentheses, # survive in CommonMark and Obsidian alike); a link whose target
has no article is rendered as plain text instead of left dangling. A label->slug
resolver is built up front so a link points at the real on-disk filename incl. the
case-fold collision suffix.
Ported from PR #1465 by @TPAteeq. Maintainer edit: trimmed the CHANGELOG wording
that overstated the guarantee ("always matches ... collision suffix and all") —
two articles sharing a byte-identical label still keep the first slug (pre-existing
behavior, same as the old [[label]]), so the claim is scoped to the case-fold case.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Some OpenAI-compatible gateways default to SSE streaming when `stream` is omitted,
but graphify always reads the result as a single resp.choices[0]. The call would
then fail against those gateways. Pass `stream: False` explicitly.
Ported from PR #1482 by @jiangyq9 (covers the extraction dispatch path,
_call_openai_compat). Maintainer fix on top: applied the same `stream: False` to
the second OpenAI-compatible call site, _call_llm, which feeds the --dedup-llm
tiebreaker and had the identical bug.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Annotations on Java classes/interfaces/records (e.g. @Service, @Entity) produced
no edge, so type-level framework wiring was invisible even though method-level
annotations were already captured. The class/interface/record handler now emits
`references` edges with the existing `attribute` context, reusing the renamed
`_java_annotation_names` helper (was `_java_method_annotation_names`; logic
unchanged) at both the type and method call sites.
Ported from PR #1487 by @oleksii-tumanov. Resolved an additive test conflict with
#1485 by keeping both new test functions.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Java field declarations produced no `references` edge for their type, so a class's
data dependencies (its field types) were missing from the graph even though
parameter and return types were already captured. The field handler now collects
the declared type via the same `_java_collect_type_refs` helper used elsewhere,
preserving the `field` and `generic_arg` contexts and skipping primitives
(int/boolean/etc.), matching the existing C#/PHP/Kotlin field handlers.
Ported from PR #1485 by @oleksii-tumanov.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A detailed report (#1475) showed the ObjC extractor silently dropping ~60% of
code-level relationships. Five fixes:
1. Dispatch: ObjC `.h` headers were parsed by extract_c (1 node, 0 edges, losing
every @interface/@protocol/@property/method). _get_extractor now routes a `.h`
to extract_objc when it contains an ObjC-only directive
(@interface/@protocol/@implementation/@import) — these are illegal in C/C++, so
the sniff never hijacks a genuine C/C++ header (verified: a plain C/C++ `.h`
stays on its existing extractor).
2. Calls: the method-body pass produced zero `calls` edges because it matched
child types `selector`/`keyword_argument_list`, but tree-sitter-objc tags
selector parts with the field name `method` (type `identifier`). The selector
is now reconstructed from every `method`-field child, explicitly skipping the
`receiver` field — so self/super/ClassName receivers are never mistaken for a
selector, and compound sends ([self a:x b:y]) resolve too. (Avoids the report's
suggested `"identifier"` fix, which would have matched receivers as selectors.)
3. Generic property types: NSArray<Product *> * wraps the type in a
generic_specifier, so the old direct-type_identifier scan saw nothing. The
element (Product) and container (NSArray) are now both referenced.
4. Class methods: `+ (…)shared` was labeled -shared; the +/- sigil is now read
from the method node's first child.
5. @import: `@import Foundation;` (a module_import node) now emits an imports edge.
Dot-syntax property `accesses` (Bug 5) and @selector(...) target-action edges
(Bug 6b) need type/name resolution policy and are left as follow-ups. Added six
regression tests; the reporter's claim that compound messages lack a
message_expression wrapper (Bug 6a) was checked against the real AST and refuted —
they share Bug 2's root cause, fixed by the same field-name change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Builds on the initial XAML support (#1460). Resolves a view to its ViewModel from
an explicit <Window.DataContext><vm:MainViewModel/>, a design-time
d:DataContext="{d:DesignInstance Type=...}", the View->ViewModel naming
convention, or Prism ViewModelLocator.AutoWireViewModel="True". Resolution is
always against an actually-extracted C# class node, so a name matching no class
(or an ambiguous 2+) emits no edge -- explicit DataContext is EXTRACTED,
convention/Prism are INFERRED. Also extracts binding paths ({Binding User.Name},
Path=Order.Total), commands (Command="{Binding SaveCommand}"), converters, and
CommunityToolkit [ObservableProperty]/[RelayCommand] generated members.
The #1460 event-handler hardening is preserved unchanged: events still resolve
only to methods with a .NET (object sender, ...EventArgs e) signature, and the
free-form-attribute denylist still prevents values like Content="Save" from
fabricating event edges (both regression tests still pass). ViewModel discovery is
bounded to the active extraction root.
Ported from PR #1473 by @MikeKatsoulakis (clean 3-way merge onto current v8).
Maintainer fix on top: the CommunityToolkit member reader now reads the
code-behind with errors="replace", so a non-UTF8 ViewModel .cs can't raise
UnicodeDecodeError and abort extract_xaml (matches every other reader in the
module). Added a regression test for that case.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
.vue files were dispatched to extract_js, which picks a tree-sitter grammar by
suffix. .vue is neither .ts nor .tsx, so the whole SFC -- <template> markup,
<script>, and <style> -- was fed to the JavaScript grammar, producing a top-level
ERROR node and recovering no imports, symbols, or type references.
A dedicated extract_vue masks everything outside <script> (replacing it with
spaces so symbol line numbers stay accurate) and parses just the script with the
grammar named by `lang` (ts default; tsx/js/jsx honored). .vue also joins the
cross-file symbol-resolution pass now that it parses cleanly.
Ported from PR #1468 by @papinto. Maintainer fix on top: the <script> open-tag
scan now skips over quoted attribute values, so a `>` inside one (Vue 3.3+ generic
components, e.g. generic="T extends Record<string, unknown>") no longer ends the
tag early and swallow the body. Added a regression test for that case.
(CHANGELOG also records #1470, committed just prior.)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
graphify reflect renders LESSONS.md from the memory docs and the
.graphify_analysis.json / .graphify_labels.json sidecars, but reflect --if-stale
only stat'd the memory docs and graph.json. So after community analysis or labels
changed (without the graph changing), --if-stale wrongly reported lessons fresh and
skipped the regen, leaving LESSONS.md stale. The freshness check now also stats the
analysis and labels sidecars, using the same custom --analysis / --labels paths the
run itself uses (so the check and the run can't disagree about which files matter),
and treats a missing sidecar as not-an-input. This makes the documented "no-op when
LESSONS.md is newer than every input" contract actually true.
Ported from PR #1470 by @oleksii-tumanov.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The Read|Glob PreToolUse hook (the "run graphify first" nudge, shared by the
Claude Code and CodeBuddy installers via _READ_SETTINGS_HOOK) decided whether to
nudge by substring-scanning the joined file_path/pattern/path for known
extensions. That had two opposite failures: '.js' is a substring of '.json' so
package.json / tsconfig.json spuriously fired, and .astro/.vue/.svelte weren't in
the set so Astro/Vue/Svelte projects never nudged on their primary source type.
The hook now compares each value's real trailing extension (segment after the
last '/', then after the last '.') against the set, and adds .astro/.vue/.svelte.
package.json -> tail .json (silent); **/*.astro -> tail .astro (fires); an
extension on a directory component (my.ts/file) correctly stays silent. The
graphify-out/ suppression and fail-open behavior are unchanged.
Ported from PR #1464 by @marketechniks onto current v8. Added three regression
tests on top of the PR's (multi-dot a.test.tsx / foo.min.js, a Windows backslash
path, and the directory-extension trap) to pin the trickiest parts of the new
segment-split logic.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Hermes and the other AGENTS.md hosts (Codex, Aider, OpenClaw, Droid, Trae, ...)
run the graphify CLI directly and do not dispatch subagents. The Step 3 extraction
guidance only described the no-API-key path as "fall straight through to subagent
dispatch", so on `/graphify .` those agents had no path they could take, fixated on
the GEMINI/ANTHROPIC key language, and looped for minutes insisting on a missing key
before eventually proceeding (a pure-code corpus is AST-only and needs no key at all).
Step 3 now opens with a hoisted, host-agnostic statement -- graphify needs no API
key, never prompt for one, never block on one; code is AST-only; a code-only corpus
skips semantic extraction entirely -- and the no-key fallback now spells out a
terminal-only path (write the empty semantic file and continue, or extract content
inline) instead of assuming subagent dispatch.
Applied in the shared core fragment and in the aider/devin core variants, so all 16
skill bodies carry it; the aider/devin change is registered as a sanctioned
monolith-roundtrip diff. Regression test (test_extraction_states_no_api_key_required_
for_every_host) renders every host and pins the wording, ordering, and the
non-subagent fallback so this can't silently regress.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Makes .xaml a first-class code input. extract_xaml() uses stdlib XML (no new
parser dependency) behind the same DOCTYPE/ENTITY and size guards as the .csproj
extractor, and captures: the root element, named controls (x:Name/Name) and their
control types, {Binding ...} references, x:Class, and -- the useful part -- a
bridge from the view markup to its .xaml.cs code-behind by resolving event-handler
attributes to the matching methods on the partial class.
Ported from PR #1460 by @MikeKatsoulakis onto current v8.
Maintainer hardening on top of the original PR: event resolution is now gated so
it can't fabricate edges. The original matched any attribute value against
code-behind method names, so Content="Save" next to a business method Save(), or
Tag="<a-handler-name>", produced spurious "event" edges. Resolution now requires
(a) the attribute is not a known free-form/identity property (Content, Text, Tag,
Title, ToolTip, Header, ...), (b) the value is a bare identifier, and (c) the
matched method actually has the .NET event-handler signature
(object sender, <T>EventArgs e) -- read from the code-behind source since the C#
extractor does not record parameter lists on method nodes. Added regression tests
for both false-positive cases.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The kimi, gemini, and deepseek backends hardcoded their base_url, so users
behind an OpenAI-compatible proxy/gateway or running a self-hosted relay had no
way to redirect them (unlike ollama/openai, which already read *_BASE_URL). Each
backend now reads KIMI_BASE_URL / GEMINI_BASE_URL / DEEPSEEK_BASE_URL and falls
back to its official default when unset, so behavior is unchanged for anyone who
doesn't set the variable.
Ported from PR #1458 by @jc2shile onto current v8. The PR branch carried 624
unrelated files from a stale base; this lands just the clean 16-line llm.py
change. Added subprocess-based tests covering both the override and the default
for all three backends (BACKENDS reads the env at import time, so each case runs
in a fresh interpreter).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
to_canvas sized each community group box for a ceil(sqrt(n))-column grid but the
placement loop hardcoded 3 columns, so any community bigger than ~9 members
rendered as a cramped 3-wide strip in an over-wide, mostly-empty box (and the box
width/height didn't even agree — w used sqrt(n), h used /3). The column count is
now computed once per community (inner_cols) and reused for box width, box height,
and card placement, so the cards fill the box. Cosmetic, no data change.
Ported from PR #1459 by @TPAteeq onto current v8 (clean: only the grid math
changed, the #1457 dedup helper is untouched). Verified the geometry on a real
canvas: n=25 -> 5x5 grid with every card inside its box; n=10 -> 4 columns.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
to_obsidian / to_canvas / to_wiki keyed filename dedup on the exact-case name,
so two labels differing only by case (e.g. `References` vs `references`) counted
as non-colliding and the second write clobbered the first on case-insensitive
filesystems (macOS/APFS, Windows/NTFS) — silently, no suffix, no warning.
Dedup now folds case (keyed on the lowercased name) while emitting the
original-case filename, so any pair that would collide on disk gets a numeric
suffix. The obsidian/canvas dedup is one shared helper (`_dedup_node_filenames`)
so they can't drift; wiki's slug dedup gets the matching fix; the `_COMMUNITY_*`
overview notes (which had no dedup at all) are covered; and a generated `base_1`
is re-checked so it can't overwrite a node literally labelled `base_1`.
Ported from PR #1457 by @TPAteeq onto current v8. Verified with a rigorous
edge-case battery (case-only collision, base_1 literal re-check -> base_1_1,
community-label case fold, determinism) plus the PR's tests; full suite 2404
passed, ruff + skillgen clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
get_community was the only graph tool still returning a bare numeric id, while
get_node and the query-traversal output already render the community_name
attribute to_json writes onto every node. Read the name from the community's
member nodes and put it in the header ("Community 12 — Auth & Sessions"),
sanitised like every other LLM-derived field.
Ported from PR #1448 by @rmart1308 onto current v8, with two additions: the name
is skipped when it is just the "Community N" placeholder (written for unnamed
communities) so the header never doubles to "Community 12 — Community 12", and
the formatting is extracted to a module-level _community_header() with focused
tests (named / placeholder / empty / sanitised). Full suite 2397 passed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
label_communities ran batches one LLM call at a time, so a large graph needed
hundreds of sequential calls even on backends that allow heavy concurrency. It
now fans batches out across a thread pool, mirroring extract_corpus_parallel:
results are returned per batch and merged on the main thread (labels dict is
never mutated concurrently, no lock), and workers==1 keeps the original
sequential path verbatim. ollama and claude-cli are forced serial unless the
matching GRAPHIFY_*_PARALLEL env opt-in is set (same guard as extract).
generate_community_labels threads max_concurrency + batch_size through, and the
cluster-only/label CLI parses --max-concurrency and --batch-size (both `--flag N`
and `--flag=N` forms; the space form is parsed explicitly so the value is not
mistaken for the positional scan path by the arg-walk's catch-all).
Output is deterministic regardless of concurrency (keyed by community id). Tests:
parallel == sequential result, batch-size controls batch count, batches actually
run concurrently, ollama forced serial, and the CLI parses both new flags. Full
suite 2393 passed; ruff clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Saving the same Q&A more than once duplicated lines in the "known dead ends" and
"corrections" sections: both lists were appended per memory doc with no key, while
node scoring already dedups by node. They now collapse by question, keeping the
most recent entry (docs are processed oldest-first, so a re-corrected question
shows its latest correction). Output stays deterministic, ordered by (date,
question). Applied to both the flat lists and the per-community buckets.
Found by a user running it on a 104-file Go codebase. Added a regression test
covering the dedupe and the recency-wins correction. Full suite 2389 passed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Both the agent (session start) and the post-commit hook can run reflect; the runs
are deterministic and idempotent, but back-to-back ones are wasted work. Add
`graphify reflect --if-stale`, which no-ops when LESSONS.md is already at least as
new as every input (the memory docs and the graph). The skill's session-start
guidance now uses `--if-stale`, so when the hook just refreshed the file the
agent's run costs almost nothing, while a skill-only install still refreshes
on demand.
New lessons_fresh() helper + 5 tests (mtime freshness in each direction, and the
CLI skip/run behavior). Regenerated per-host references + re-blessed expected/;
all five skillgen guards pass; full suite 2388 passed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds the deterministic work-memory loop: `save-result --outcome
useful|dead_end|corrected [--correction]` records how a saved Q&A turned out, and
`graphify reflect` aggregates graphify-out/memory/ into a deterministic
reflections/LESSONS.md an agent loads next session.
Source nodes are scored, not counted: signed, recency-decayed (useful +,
dead_end/corrected -, configurable --half-life-days, default 30), so a fresh dead
end outweighs a stale useful. A node is "preferred" only once corroborated by
>=--min-corroboration distinct results (default 2); others are "tentative", and
mixed-signal nodes render once as "contested" (recency-wins). Source nodes are
matched to the graph by label OR id, and citations whose node no longer exists are
dropped, so a plain `graphify update` after deleting code clears stale lessons.
Deterministic, no LLM; bare save-result and existing behavior unchanged.
Rigorously verified end-to-end on real data: corroboration boundary, recency flip,
contested verdict, foreign/malformed memory docs, cold start, 300-doc scale +
byte-stable output, and the node-gate dropping deleted-code lessons after update.
Full suite 2383 passed; skillgen --check clean; ruff clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
safishamsi