JSONbored · loopover-orb · Jul 3, 2026 · Jul 3, 2026
@@ -7,16 +7,19 @@ function isTestFile(file) {
     /(^|\/)(test|tests|spec|__tests__)\//i.test(file) ||
     /(^|\/)src\/test\//i.test(file) ||
     /(^|\/)[^/]+_test\.(go|py|rb)$/i.test(file) ||
+    /(^|\/)test_[^/]*\.py$/i.test(file) || // pytest's default `test_*.py` prefix (the suffix rule above only catches `*_test.py`)
     /(^|\/)[^/]+_spec\.rb$/i.test(file) ||
-    /\.(test|spec)\.(ts|tsx|js|jsx|py|rb|rs)$/i.test(file) ||
+    /\.(test|spec)\.(ts|tsx|mts|cts|js|jsx|mjs|cjs|py|rb|rs)$/i.test(file) ||
+    /(^|\/)[^/]+\.(cy|e2e)\.(ts|tsx|mts|cts|js|jsx|mjs|cjs)$/i.test(file) ||
     // JVM/.NET/Swift PascalCase test-class suffix (case-sensitive, matching the
     // signal classifiers) so C#/Swift/Groovy tests aren't counted as source.
-    /(^|\/)\w*(Tests?|Spec)\.(java|kt|kts|scala|cs|swift|groovy)$/.test(file)
+    /(^|\/)\w*(Tests?|Spec)\.(java|kt|kts|scala|cs|swift|groovy)$/.test(file) ||
+    /(^|\/)__snapshots__\//i.test(file)
   );
 }
 
 function isCodeFile(file) {
-  return /\.(ts|tsx|js|jsx|py|rb|rs|kt|scala|java|go|sql|cs|swift|groovy)$/i.test(file) && !isTestFile(file);
+  return /\.(ts|tsx|mts|cts|js|jsx|mjs|cjs|py|rb|rs|kt|scala|java|go|sql|cs|swift|groovy)$/i.test(file) && !isTestFile(file);
 }
 
 function lineCount(file) {

@@ -12,43 +12,22 @@
 import sys
 from pathlib import Path
 
-# JVM/.NET/Swift PascalCase test-class suffix (case-sensitive, on the original
-# path) so C#/Swift/Groovy tests aren't counted as source once their extensions
-# are recognized as code. Matches the signal classifiers' isTestPath rule.
-_JVM_TEST_SUFFIX_RE = re.compile(r"(?:Tests?|Spec)\.(?:java|kt|kts|scala|cs|swift|groovy)$")
+# Every rule is an END/SEGMENT-anchored regex, a faithful mirror of the server isTestPath
+# (src/signals/test-evidence.ts). Plain substring tokens over-matched: "/__snapshots__/" missed a root-level
+# dir, and ".test.mjs" matched non-tests like `dist/widget.test.mjs.map` where the extension is not end-of-path.
+_TEST_PATH_RES = (
+    re.compile(r"(?:^|/)(?:tests?|spec|__tests__|__snapshots__|src/test)/", re.IGNORECASE),  # dir conventions
+    re.compile(r"(?:^|/)[^/]+_test\.(?:go|py|rb)$", re.IGNORECASE),  # go/py/rb *_test suffix
+    re.compile(r"(?:^|/)test_[^/]*\.py$", re.IGNORECASE),  # pytest test_*.py prefix
+    re.compile(r"(?:^|/)[^/]+_spec\.rb$", re.IGNORECASE),  # RSpec *_spec.rb suffix
+    re.compile(r"\.(?:test|spec)\.(?:ts|tsx|mts|cts|js|jsx|mjs|cjs|py|rb|rs)$", re.IGNORECASE),  # .test/.spec.<ext>
+    re.compile(r"(?:^|/)[^/]+\.(?:cy|e2e)\.(?:ts|tsx|mts|cts|js|jsx|mjs|cjs)$", re.IGNORECASE),  # Cypress/Playwright
+    re.compile(r"(?:^|/)\w*(?:Tests?|Spec)\.(?:java|kt|kts|scala|cs|swift|groovy)$"),  # JVM/.NET/Swift (case-sensitive)
+)
 
 
 def is_test_file(path: str) -> bool:
-    lowered = path.lower()
-    basename = lowered.rsplit("/", 1)[-1]
-    patterns = (
-        "/test/",
-        "/tests/",
-        "/spec/",
-        "/__tests__/",
-        "/src/test/",
-        "_test.go",
-        "_test.py",
-        "_test.rb",
-        "_spec.rb",
-        ".test.ts",
-        ".test.tsx",
-        ".test.js",
-        ".test.jsx",
-        ".test.py",
-        ".test.rb",
-        ".test.rs",
-        ".spec.ts",
-        ".spec.tsx",
-        ".spec.js",
-        ".spec.jsx",
-        ".spec.py",
-        ".spec.rb",
-        ".spec.rs",
-    )
-    if _JVM_TEST_SUFFIX_RE.search(path):
-        return True
-    return any(token in lowered for token in patterns) or any(basename.endswith(suffix) for suffix in ("_test.go", "_test.py", "_test.rb"))
+    return any(rx.search(path) for rx in _TEST_PATH_RES)
 
 
 def load_gittensor(gittensor_root: str):
@@ -160,7 +139,7 @@ def metadata_fallback(metadata: dict) -> dict:
         lines = max(int(entry.get("additions") or 0) + int(entry.get("deletions") or 0), 0)
         if is_test_file(path):
             tests += lines
-        elif path.endswith((".ts", ".tsx", ".js", ".jsx", ".py", ".rb", ".rs", ".go", ".java", ".kt", ".scala", ".sql", ".cs", ".swift", ".groovy")):
+        elif path.endswith((".ts", ".tsx", ".mts", ".cts", ".js", ".jsx", ".mjs", ".cjs", ".py", ".rb", ".rs", ".go", ".java", ".kt", ".scala", ".sql", ".cs", ".swift", ".groovy")):
             source += lines
         else:
             non_code += lines

@@ -0,0 +1,77 @@
+import { describe, expect, it } from "vitest";
+import { spawnSync } from "node:child_process";
+import { fileURLToPath } from "node:url";
+import { dirname, join } from "node:path";
+
+// The MCP's local score-preview script inlines its own isTestFile/isCodeFile (it ships in the standalone Node
+// bin package and can't import from src/). They must mirror the server's isTestPath/isCodeFile, or a miner's
+// LOCAL preview classifies files differently than the gate would. This spawns the real script and checks the
+// token classification, which is where the drift would surface.
+const scriptsDir = join(dirname(fileURLToPath(import.meta.url)), "../../packages/gittensory-mcp/scripts");
+const scriptMjs = join(scriptsDir, "gittensor-score-preview.mjs");
+const scriptPy = join(scriptsDir, "gittensor-score-preview.py");
+
+const SAMPLE = [
+  { path: "src/loader.mts", additions: 10, deletions: 0 }, // module-ext source — was non-code before the fix
+  { path: "e2e/login.cy.ts", additions: 5, deletions: 0 }, // Cypress test — was counted as source before
+  { path: "src/test_api.py", additions: 3, deletions: 0 }, // pytest test_*.py prefix — was counted as source before
+  { path: "__snapshots__/Card.tsx.snap", additions: 4, deletions: 0 }, // ROOT-LEVEL snapshot dir — segment-aware match
+];
+const SAMPLE_TEST_LINES = 5 + 3 + 4; // cy + pytest + root-level snapshot
+
+function runPreview(changedFiles: Array<{ path: string; additions: number; deletions: number }>): { sourceTokenScore: number; testTokenScore: number; nonCodeTokenScore: number } {
+  const res = spawnSync(process.execPath, [scriptMjs], { input: JSON.stringify({ changedFiles }), encoding: "utf8" });
+  expect(res.status, res.stderr).toBe(0);
+  return JSON.parse(res.stdout);
+}
+
+function findPython(): string | null {
+  for (const cmd of ["python3", "python"]) {
+    const r = spawnSync(cmd, ["--version"], { encoding: "utf8" });
+    if (r.status === 0) return cmd;
+  }
+  return null;
+}
+
+describe("gittensor-score-preview.mjs classifier parity with the server", () => {
+  it("counts module-ext source as code and pytest-prefix/Cypress files as tests, matching isTestPath/isCodeFile", () => {
+    const out = runPreview([...SAMPLE, { path: "app/FooTests.java", additions: 7, deletions: 0 }]); // + JVM test control
+    expect(out.sourceTokenScore).toBe(10); // only src/loader.mts is source now
+    expect(out.testTokenScore).toBe(SAMPLE_TEST_LINES + 7); // cy + pytest + root snapshot + JVM tests
+    expect(out.nonCodeTokenScore).toBe(0); // the .mts is no longer misfiled as non-code
+  });
+
+  it("the .py fallback classifier agrees with the .mjs (skipped when no python is available)", () => {
+    // metadata_fallback runs when GITTENSOR_ROOT is unset; its source-extension tuple must also carry the module
+    // extensions so a .mts is counted as source, not non-code.
+    const python = findPython();
+    if (!python) return; // environment has no python — the .mjs test above covers the shared intent
+    const env = { ...process.env };
+    delete env.GITTENSOR_ROOT;
+    const res = spawnSync(python, [scriptPy], { input: JSON.stringify({ changedFiles: SAMPLE }), encoding: "utf8", env });
+    expect(res.status, res.stderr).toBe(0);
+    const out = JSON.parse(res.stdout);
+    expect(out.sourceTokenScore).toBe(10);
+    expect(out.testTokenScore).toBe(SAMPLE_TEST_LINES);
+    expect(out.nonCodeTokenScore).toBe(0);
+  });
+
+  it("does not misclassify a *.test.mjs.map source-map as a test (extension anchored to end-of-path, matching the server)", () => {
+    // A substring match on ".test.mjs" wrongly flagged non-tests like dist/widget.test.mjs.map; the rule must be
+    // end-anchored like isTestPath. It's a source-map — neither test nor code — so it counts as non-code.
+    const files = [{ path: "dist/widget.test.mjs.map", additions: 2, deletions: 0 }];
+    const mjs = runPreview(files);
+    expect(mjs.testTokenScore).toBe(0);
+    expect(mjs.nonCodeTokenScore).toBe(2);
+
+    const python = findPython();
+    if (!python) return;
+    const env = { ...process.env };
+    delete env.GITTENSOR_ROOT;
+    const res = spawnSync(python, [scriptPy], { input: JSON.stringify({ changedFiles: files }), encoding: "utf8", env });
+    expect(res.status, res.stderr).toBe(0);
+    const py = JSON.parse(res.stdout);
+    expect(py.testTokenScore).toBe(0);
+    expect(py.nonCodeTokenScore).toBe(2);
+  });
+});