python: resolve calls through an imported module name (M.f() / from a import b as M) - #334
Open
DavinchiPes wants to merge 1 commit into
Open
python: resolve calls through an imported module name (M.f() / from a import b as M)#334DavinchiPes wants to merge 1 commit into
DavinchiPes wants to merge 1 commit into
Conversation
`from analysis import pump_signature as P` followed by `P.find_pumps()` is the dominant call style in numeric Python (numpy's `import x as np`, generalised to a project's own modules). The extractor saw receiver `P`, found no type binding for it and dropped the edge — the right rule for an object of unknown class (trailhq#35), the wrong one for a module: `graft callers find_pumps` reported "no indexed callers" for a function with thirteen call sites. A module receiver is not an unknown object: the import statement names the exact file. So: - extract.ts collects, per Python file, every import that binds a MODULE to a local name (`import a.b as M`, `from a import b as M`, `from a import b`, `import a.b`, relative `from . import b`). A member call whose receiver did not resolve to a type but IS one of those names is emitted as a plain call carrying the module's dotted path as `specifier`. A receiver that resolved to a type keeps the typed-member path (a class alias is not a module). - resolve.ts resolves that specifier to a file (`a/b.py`, then `a/b/__init__.py`, under each ancestor directory of the caller nearest-first, so a `src/` layout needs no build-file parsing; relative imports climb from the caller's directory) and looks the name up inside that file only — a same-named function elsewhere in the repo can never become a false edge, and an ambiguous or missing name drops, as before. - pyReceiver also spells a pure dotted chain (`pkg.mod.f()`) so the `import pkg.mod` form matches. test/graph-python-module-alias.test.ts pins the four spellings plus the package-`__init__` case, and that neither a same-named decoy nor an untyped receiver gains an edge. Full suite: unchanged apart from the two new tests (claude-shim-resolve's four machine-dependent failures are pre-existing on main and untouched).
🌱 graft blast radius1 area changed → 16 areas can be affected. 226 dependent symbols, depth 2. flowchart TB
A0(("Graph Building<br/>65 symbols"))
A1(("Claude Integration<br/>24 symbols"))
A2(("Host Configuration<br/>24 symbols"))
A3(("CLI Orchestration<br/>17 symbols"))
A4(("Context Caching<br/>16 symbols"))
AX(("11 smaller areas<br/>80 symbols"))
classDef reached fill:#D9EDF3,stroke:#3AA7C9,stroke-width:1.5px,color:#0E313C;
class A0,A1,A2,A3,A4 reached;
classDef tail fill:#EEF2F3,stroke:#9AA4A9,stroke-width:1px,color:#3A4247;
class AX tail;
Who knows this code — 5 people across 17 areas
Ownership is git history over each area's own files, weighted towards recent work (120-day half-life). Merge commits and bots are dropped, and you are dropped from your own PR. A name with no All 226 dependent symbols, grouped by areaGraph Building — 65 symbols in 17 files
…165 further symbols not listed. Test signal per changed area — 1 ✓Reached = a node under a test path has a resolved edge into the changed symbol. It undercounts anything called indirectly — through a CLI, a spawned process or a dynamic import — so read a low ratio as “look here”, never as a coverage gate.
99 test suites also reference this code249 symbols, kept out of the diagram and the table so they cannot crowd out the areas a reviewer has to look at.
Open the interactive graph → — click an area to see its dependent symbols at file:line. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
from analysis import pump_signature as Pfollowed byP.find_pumps()is the dominant call style in numeric Python codebases (numpy'simport x as np, generalised to a project's own modules). The extractor saw receiverP, found no type binding for it and dropped the edge — the right rule for an object of unknown class (#35), the wrong one for a module. On our repograft callers find_pumpsreported "no indexed callers" for a function with thirteen call sites across 32 files that import this way;graft grepfound them all.Fix
A module receiver is not an unknown object: the import statement names the exact file.
import a.b as M,from a import b as M,from a import b,import a.b, relativefrom . import b). A member call whose receiver did not resolve to a type but is one of those names is emitted as a plain call carrying the module's dotted path asspecifier. A receiver that resolved to a type keeps the typed-member path (a class alias is not a module).a/b.py, thena/b/__init__.py, under each ancestor directory of the caller nearest-first, so asrc/layout needs no build-file parsing; relative imports climb from the caller's directory) and look the name up inside that file only. A same-named function elsewhere in the repo can never become a false edge; an ambiguous or missing name drops, as before — precision rule frommapclusters one entry per file instead of per directory #35 preserved.pyReceiveralso spells a pure dotted chain (pkg.mod.f()) so theimport pkg.modform matches.Tests
test/graph-python-module-alias.test.ts(TDD: written first, failed on main, passes now):__init__.pycase all produce anextractedcall edge into the module's file;obj.find_pumps()) still drops.Full suite: 1218/1222 — the four
claude-shim-resolvefailures are pre-existing onmain(machine-dependent global-install lookup) and untouched.Result on a real repo
Rebuilt the graph on a ~1 900-file Python repo:
graft callers find_pumps0 → 6 calling functions (13 sites),graft callers compute_all1 → 2; call counts in modules that use direct imports unchanged.🤖 Generated with Claude Code