fix(web): Handle ambiguous languages (#1026)

* feat(web): replace language param with path in symbol tools, make repo mandatory

- find_symbol_definitions and find_symbol_references now accept `path`
  (the file where the symbol was encountered) instead of `language`
- Language is derived internally via detectLanguageFromFilename
- `repo` is now required in both tool schemas
- Also fixes ambiguous extension overrides in languageDetection.ts

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* delete

* changelog

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: brendan-kellam

CHANGELOG.md

@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Changed
+- Changed language detection to resolve file extensions with multiple language resolutions (e.g., .md) to the most common resolution. [#1026](https://github.com/sourcebot-dev/sourcebot/pull/1026)
+
 ## [4.15.11] - 2026-03-20
 
 ### Changed

packages/web/src/lib/languageDetection.ts

@@ -1,6 +1,23 @@
 import * as linguistLanguages from 'linguist-languages';
 import path from 'path';
 
+// Override map for extensions that are ambiguous in linguist-languages.
+// These are extensions where linguist maps to multiple languages, but one
+// is overwhelmingly more common in practice.
+const ambiguousExtensionOverrides: Record<string, string> = {
+    '.cs': 'C#',        // Not Smalltalk
+    '.fs': 'F#',        // Not Forth, GLSL, or Filterscript
+    '.html': 'HTML',    // Not Ecmarkup
+    '.json': 'JSON',    // Not OASv2-json, OASv3-json
+    '.md': 'Markdown',  // Not GCC Machine Description
+    '.rs': 'Rust',      // Not RenderScript (deprecated)
+    '.tsx': 'TSX',      // Not XML
+    '.ts': 'TypeScript', // Not XML
+    '.txt': 'Text',     // Not Adblock Filter List, Vim Help File
+    '.yaml': 'YAML',    // Not MiniYAML, OASv2-yaml, OASv3-yaml
+    '.yml': 'YAML',
+};
+
 const extensionToLanguage = new Map<string, string>();
 
 for (const [languageName, languageData] of Object.entries(linguistLanguages)) {
@@ -31,6 +48,12 @@ export const detectLanguageFromFilename = (filename: string): string => {
 
     // Check for extension match
     const ext = path.extname(filename).toLowerCase();
+
+    // Check override map first for ambiguous extensions
+    if (ext && ext in ambiguousExtensionOverrides) {
+        return ambiguousExtensionOverrides[ext];
+    }
+
     if (ext && extensionToLanguage.has(ext)) {
         return extensionToLanguage.get(ext)!;
     }