feat(web): support .gitattributes linguist-language overrides in file viewer (#1048)

Author: brendan-kellam

CHANGELOG.md

@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+- Added support for `.gitattributes` `linguist-language` overrides in the file viewer ([#1048](https://github.com/sourcebot-dev/sourcebot/pull/1048))
+
 ### Fixed
 - Fixed Ask GitHub landing page chat box placement to be centered on the page instead of at the bottom. [#1046](https://github.com/sourcebot-dev/sourcebot/pull/1046)
 

packages/web/src/features/git/getFileSourceApi.ts

@@ -2,6 +2,7 @@ import { sew } from '@/actions';
 import { getBrowsePath } from '@/app/[domain]/browse/hooks/utils';
 import { getAuditService } from '@/ee/features/audit/factory';
 import { SINGLE_TENANT_ORG_DOMAIN } from '@/lib/constants';
+import { parseGitAttributes, resolveLanguageFromGitAttributes } from '@/lib/gitattributes';
 import { detectLanguageFromFilename } from '@/lib/languageDetection';
 import { ServiceError, notFound, fileNotFound, invalidGitRef, unexpectedError } from '@/lib/serviceError';
 import { getCodeHostBrowseFileAtBranchUrl } from '@/lib/utils';
@@ -65,7 +66,17 @@ export const getFileSource = async ({ path: filePath, repo: repoName, ref }: Fil
         throw error;
     }
 
-    const language = detectLanguageFromFilename(filePath);
+    let gitattributesContent: string | undefined;
+    try {
+        gitattributesContent = await git.raw(['show', `${gitRef}:.gitattributes`]);
+    } catch {
+        // No .gitattributes in this repo/ref, that's fine
+    }
+
+    const language = gitattributesContent
+        ? (resolveLanguageFromGitAttributes(filePath, parseGitAttributes(gitattributesContent)) ?? detectLanguageFromFilename(filePath))
+        : detectLanguageFromFilename(filePath);
+
     const externalWebUrl = getCodeHostBrowseFileAtBranchUrl({
         webUrl: repo.webUrl,
         codeHostType: repo.external_codeHostType,

packages/web/src/lib/gitattributes.ts

@@ -0,0 +1,72 @@
+import micromatch from 'micromatch';
+
+// GitAttributes holds parsed .gitattributes rules for overriding language detection.
+export interface GitAttributes {
+    rules: GitAttributeRule[];
+}
+
+interface GitAttributeRule {
+    pattern: string;
+    attrs: Record<string, string>;
+}
+
+// parseGitAttributes parses the content of a .gitattributes file.
+// Each non-comment, non-empty line has the form: pattern attr1 attr2=value ...
+// Attributes can be:
+//   - "linguist-vendored" (set/true), "-linguist-vendored" (unset/false)
+//   - "linguist-language=Go"
+//   - etc.
+export function parseGitAttributes(content: string): GitAttributes {
+    const rules: GitAttributeRule[] = [];
+
+    for (const raw of content.split('\n')) {
+        const line = raw.trim();
+        if (line === '' || line.startsWith('#')) {
+            continue;
+        }
+
+        const fields = line.split(/\s+/);
+        if (fields.length < 2) {
+            continue;
+        }
+
+        const pattern = fields[0];
+        const attrs: Record<string, string> = {};
+
+        for (const field of fields.slice(1)) {
+            if (field.startsWith('!')) {
+                // !attr means unspecified (reset to default)
+                attrs[field.slice(1)] = 'unspecified';
+            } else if (field.startsWith('-')) {
+                // -attr means unset (false)
+                attrs[field.slice(1)] = 'false';
+            } else {
+                const eqIdx = field.indexOf('=');
+                if (eqIdx !== -1) {
+                    // attr=value
+                    attrs[field.slice(0, eqIdx)] = field.slice(eqIdx + 1);
+                } else {
+                    // attr alone means set (true)
+                    attrs[field] = 'true';
+                }
+            }
+        }
+
+        rules.push({ pattern, attrs });
+    }
+
+    return { rules };
+}
+
+// resolveLanguageFromGitAttributes returns the linguist-language override for
+// the given file path based on the parsed .gitattributes rules, or undefined
+// if no rule matches. Last matching rule wins, consistent with gitattributes semantics.
+export function resolveLanguageFromGitAttributes(filePath: string, gitAttributes: GitAttributes): string | undefined {
+    let language: string | undefined;
+    for (const rule of gitAttributes.rules) {
+        if (micromatch.isMatch(filePath, rule.pattern) && rule.attrs['linguist-language']) {
+            language = rule.attrs['linguist-language'];
+        }
+    }
+    return language;
+}