linguist-js 3.0.0 → 3.0.2

package/dist/analyser/pipeline/filter.js

@@ -1,3 +1,4 @@
+import ignore from 'ignore';
 export function filterFiles(files, opts) {
     return files.filter((file) => {
         // Skip binary files
@@ -9,6 +10,10 @@ export function filterFiles(files, opts) {
             // Skip vendored, generated, or documentation files
             return false;
         }
+        // Skip manually ignored files
+        if (opts.ignoredFiles?.length && ignore().add(opts.ignoredFiles).ignores((file.path))) {
+            return false;
+        }
         return true;
     });
 }

package/dist/input/fromFilesystem.js

@@ -1,3 +1,4 @@
+import binaryData from 'binary-extensions';
 import commonPrefix from 'common-path-prefix';
 import ignore from 'ignore';
 import FS from 'node:fs';
@@ -6,7 +7,6 @@ import Attributes from '../program/classes/attributes.js';
 import { getFileExtension, normPath } from '../program/fs/normalisedPath.js';
 import walkTree from '../program/fs/walkTree.js';
 import parseGitattributes from '../program/parsing/parseGitattributes.js';
-const binaryData = JSON.parse(FS.readFileSync(new URL('../../node_modules/binary-extensions/binary-extensions.json', import.meta.url), 'utf-8'));
 export default async function fromFilesystem(input, opts, vendorPaths) {
     const resolvedInput = input.map((path) => normPath(Path.resolve(path)));
     const commonRoot = (resolvedInput.length > 1 ? commonPrefix(resolvedInput) : resolvedInput[0]).replace(/\/?$/, '');

package/dist/program/data/loadDataFiles.js

@@ -1,3 +1,4 @@
+import YAML from 'js-yaml';
 import Cache from 'node-cache';
 import FS from 'node:fs';
 import Path from 'node:path';
@@ -14,8 +15,15 @@ async function loadWebFile(file) {
     // Load file content, falling back to the local file if the request fails
     const fileContent = await fetch(dataUrl(file))
         .then((data) => data.text())
-        .catch(async () => await loadLocalFile(file));
+        .catch(async (x) => void x);
+    if (!fileContent) {
+        return await loadLocalFile(file);
+    }
     cache.set(file, fileContent);
+    // Clean up lengthy files
+    if (file === 'generated.rb') {
+        return YAML.dump(parseGeneratedDataFile(fileContent));
+    }
     return fileContent;
 }
 async function loadLocalFile(file) {

package/dist/program/data/retrieveData.d.ts

@@ -1,10 +1,7 @@
-import { HeuristicsSchema, LanguagesScema, VendorSchema } from '../../types/schema.js';
+import { HeuristicsSchema, LanguagesScema } from '../../types/schema.js';
 type LoadedData = {
     langData: LanguagesScema;
-    vendorData: VendorSchema;
-    docData: VendorSchema;
     heuristicsData: HeuristicsSchema;
-    generatedData: string[];
     vendorPaths: string[];
 };
 /** Load data from github-linguist web repo or cached local file. */

package/dist/program/data/retrieveData.js

@@ -1,22 +1,19 @@
 import YAML from 'js-yaml';
-import { loadFile, parseGeneratedDataFile } from './loadDataFiles.js';
+import { loadFile } from './loadDataFiles.js';
 let data = null;
 async function initRetrieveData(offline) {
-    // Only load the data on mont
+    // Only load the data on mount
     if (data)
         return;
     const langData = (await loadFile('languages.yml', offline).then(YAML.load));
     const vendorData = (await loadFile('vendor.yml', offline).then(YAML.load));
     const docData = (await loadFile('documentation.yml', offline).then(YAML.load));
     const heuristicsData = (await loadFile('heuristics.yml', offline).then(YAML.load));
-    const generatedData = (await loadFile('generated.rb', offline).then(parseGeneratedDataFile));
+    const generatedData = (await loadFile('generated.rb', offline).then(YAML.load));
     const vendorPaths = [...vendorData, ...docData, ...generatedData];
     data = {
         langData,
-        vendorData,
-        docData,
         heuristicsData,
-        generatedData,
         vendorPaths,
     };
 }

package/dist/program/fs/walkTree.js

@@ -31,7 +31,8 @@ export default function walkTree(data) {
             const ignoredPaths = parseGitignore(gitignoreContents);
             const rootRelIgnoredPaths = ignoredPaths.map((ignorePath) => 
             // get absolute path of the ignore glob
-            normPath(folder, ignorePath)
+            // use '/**/' to ensure matches in subfolders are also ignored
+            normPath(folder, '**', ignorePath)
                 // convert abs ignore glob to be relative to the root folder
                 .replace(commonRoot + '/', ''));
             ignored.add(rootRelIgnoredPaths);

package/dist/program/utils/pcre.js

@@ -14,6 +14,7 @@ export default function pcre(regex) {
     // Remove PCRE-only syntax
     replace(/([*+]){2}/g, '$1');
     replace(/\(\?>/g, '(?:');
+    replace(/\\g\<[^>]+\>/g, '.*?'); // referenced named groups not supported
     // Remove start/end-of-file markers
     if (/\\[AZ]/.test(finalRegex)) {
         replace(/\\A/g, '^');

package/ext/heuristics.yml

@@ -176,6 +176,13 @@ disambiguations:
   rules:
   - language: Gerber Image
     pattern: '^[DGMT][0-9]{2}\*(?:\r?\n|\r)'
+- extensions: ['.container', '.mount', '.network', '.service', '.socket', '.target', '.timer']
+  rules:
+  - language: INI
+    and:
+    - pattern: '^[#\s]*\[[^\]]+\]\s*$'
+    - named_pattern: key_equals_value
+    - negative_pattern: '^[^#;=\[]+[^ \r\n=]$'
 - extensions: ['.cs']
   rules:
   - language: Smalltalk
@@ -597,6 +604,15 @@ disambiguations:
     pattern:
     - '(?i)\bDEFINE\s+(?:VARIABLE|TEMP-TABLE|BUFFER|QUERY|INPUT\s+PARAMETER|OUTPUT\s+PARAMETER)\b'
     - '(?i)\bEND(?:\s+(?:PROCEDURE|FUNCTION|DO|FOR\s+EACH))?\.'
+- extensions: ['.pc']
+  rules:
+  - language: Pro*C
+    pattern: '^\s*EXEC\s+SQL\s'
+  - language: pkg-config
+    and:
+      - pattern: '^Name:[ \t]+\S'
+      - pattern: '^Description:[ \t]+\S'
+      - pattern: '^Version:[ \t]+\S'
 - extensions: ['.php']
   rules:
   - language: Hack

package/ext/languages.yml

@@ -513,6 +513,16 @@ B4X:
   codemirror_mode: vb
   codemirror_mime_type: text/x-vb
   language_id: 96642275
+BAML:
+  type: programming
+  color: "#a855f7"
+  extensions:
+  - ".baml"
+  tm_scope: source.baml
+  ace_mode: io
+  codemirror_mode: stylus
+  codemirror_mime_type: text/x-styl
+  language_id: 502521509
 BASIC:
   type: programming
   extensions:
@@ -673,6 +683,16 @@ BlitzMax:
   tm_scope: source.blitzmax
   ace_mode: text
   language_id: 35
+Blueprint:
+  type: markup
+  color: "#3584E4"
+  extensions:
+  - ".blp"
+  aliases:
+  - blp
+  tm_scope: source.blueprint
+  ace_mode: text
+  language_id: 765545512
 Bluespec:
   type: programming
   color: "#12223c"
@@ -792,7 +812,7 @@ C#:
   codemirror_mode: clike
   codemirror_mime_type: text/x-csharp
   tm_scope: source.cs
-  color: "#178600"
+  color: "#7355dd"
   aliases:
   - csharp
   - cake
@@ -2618,6 +2638,8 @@ Git Config:
   filenames:
   - ".gitconfig"
   - ".gitmodules"
+  - ".tgitconfig"
+  - config.worktree
   ace_mode: ini
   codemirror_mode: properties
   codemirror_mime_type: text/x-properties
@@ -2885,6 +2907,23 @@ Groovy Server Pages:
   codemirror_mode: htmlembedded
   codemirror_mime_type: application/x-jsp
   language_id: 143
+GtkRC:
+  type: data
+  color: "#7fe719"
+  aliases:
+  - gtk
+  - gtk 1
+  - gtk 2
+  extensions:
+  - ".gtkrc"
+  filenames:
+  - gtkrc
+  - gtkrc-2.0
+  tm_scope: source.gtkrc
+  ace_mode: ruby
+  codemirror_mode: ruby
+  codemirror_mime_type: text/x-ruby
+  language_id: 876401352
 HAProxy:
   type: data
   color: "#106da9"
@@ -3234,12 +3273,19 @@ INI:
   - ".ini"
   - ".cfg"
   - ".cnf"
+  - ".container"
   - ".dof"
   - ".frm"
   - ".lektorproject"
+  - ".mount"
+  - ".network"
   - ".prefs"
   - ".pro"
   - ".properties"
+  - ".service"
+  - ".socket"
+  - ".target"
+  - ".timer"
   - ".url"
   filenames:
   - ".buckconfig"
@@ -3253,6 +3299,7 @@ INI:
   - vlcrc
   tm_scope: source.ini
   aliases:
+  - conf
   - dosini
   ace_mode: ini
   codemirror_mode: properties
@@ -3303,11 +3350,13 @@ Ignore List:
   - ".atomignore"
   - ".babelignore"
   - ".bzrignore"
+  - ".ckignore"
   - ".coffeelintignore"
   - ".cvsignore"
   - ".dockerignore"
   - ".easignore"
   - ".eleventyignore"
+  - ".eslint-ignore"
   - ".eslintignore"
   - ".gitignore"
   - ".ignore"
@@ -3315,6 +3364,7 @@ Ignore List:
   - ".nodemonignore"
   - ".npmignore"
   - ".prettierignore"
+  - ".rgignore"
   - ".stylelintignore"
   - ".vercelignore"
   - ".vscodeignore"
@@ -3471,6 +3521,7 @@ JSON:
   - ".jsonl"
   - ".mcmeta"
   - ".sarif"
+  - ".slnlaunch"
   - ".tact"
   - ".tfstate"
   - ".tfstate.backup"
@@ -3696,8 +3747,10 @@ JavaScript:
   filenames:
   - Jakefile
   interpreters:
+  - bun
   - chakra
   - d8
+  - deno
   - gjs
   - js
   - node
@@ -4526,6 +4579,7 @@ Makefile:
   - Makefile.frag
   - Makefile.in
   - Makefile.inc
+  - Makefile.pc
   - Makefile.wat
   - makefile
   - makefile.sco
@@ -5487,6 +5541,19 @@ Org:
   tm_scope: none
   ace_mode: text
   language_id: 267
+OverPy:
+  type: programming
+  color: "#78b355"
+  wrap: true
+  extensions:
+  - ".opy"
+  tm_scope: source.opy
+  aliases:
+  - opy
+  ace_mode: python
+  codemirror_mode: python
+  codemirror_mime_type: text/x-python
+  language_id: 492781155
 OverpassQL:
   type: programming
   color: "#cce2aa"
@@ -5912,6 +5979,16 @@ PostScript:
   - postscr
   ace_mode: text
   language_id: 291
+Power Query:
+  type: programming
+  aliases:
+    - powerquery
+  extensions:
+    - ".pq"
+  tm_scope: source.powerquery
+  ace_mode: text
+  color: "#d38e0d"
+  language_id: 37496382
 PowerBuilder:
   type: programming
   color: "#8f0f8d"
@@ -5956,6 +6033,17 @@ Prisma:
   tm_scope: source.prisma
   ace_mode: prisma
   language_id: 499933428
+Pro*C:
+  fs_name: ProC
+  color: "#bb8368"
+  type: programming
+  extensions:
+  - ".pc"
+  tm_scope: source.c
+  ace_mode: c_cpp
+  codemirror_mode: clike
+  codemirror_mime_type: text/x-csrc
+  language_id: 991047534
 Processing:
   type: programming
   color: "#0096D8"
@@ -6577,6 +6665,14 @@ Redirect Rules:
   tm_scope: source.redirects
   ace_mode: text
   language_id: 1020148948
+Redscript:
+  type: programming
+  color: "#f44336"
+  extensions:
+  - ".reds"
+  tm_scope: source.redscript
+  ace_mode: text
+  language_id: 686691365
 Regular Expression:
   type: data
   color: "#009a00"
@@ -6647,6 +6743,17 @@ RobotFramework:
   tm_scope: text.robot
   ace_mode: robot
   language_id: 324
+Robots Exclusion Rules:
+  type: data
+  aliases:
+  - robots
+  - robots txt
+  - robotx.txt
+  filenames:
+  - robots.txt
+  ace_mode: text
+  tm_scope: text.robots-txt
+  language_id: 674736065
 Roc:
   type: programming
   color: "#7c38f5"
@@ -7901,6 +8008,7 @@ Text:
   - README.me
   - README.mysql
   - README.nss
+  - README.pc
   - click.me
   - delete.me
   - keep.me
@@ -8217,6 +8325,7 @@ Valve Data Format:
   - vdf
   extensions:
   - ".vdf"
+  - ".vmf"
   ace_mode: text
   tm_scope: source.keyvalues
   language_id: 544060961
@@ -9050,11 +9159,11 @@ crontab:
   ace_mode: tcl
   language_id: 705203557
 desktop:
+  group: INI
   type: data
   extensions:
   - ".desktop"
   - ".desktop.in"
-  - ".service"
   tm_scope: source.desktop
   ace_mode: text
   language_id: 412
@@ -9208,6 +9317,19 @@ ooc:
   tm_scope: source.ooc
   ace_mode: text
   language_id: 418
+pkg-config:
+  type: data
+  color: "#2b5e82"
+  aliases:
+  - pkgconf
+  extensions:
+  - ".pc"
+  - ".pc.in"
+  tm_scope: source.pkgconf
+  ace_mode: properties
+  codemirror_mode: properties
+  codemirror_mime_type: text/x-properties
+  language_id: 925023573
 q:
   type: programming
   extensions:
@@ -9234,16 +9356,6 @@ reStructuredText:
   codemirror_mode: rst
   codemirror_mime_type: text/x-rst
   language_id: 419
-robots.txt:
-  type: data
-  aliases:
-  - robots
-  - robots txt
-  filenames:
-  - robots.txt
-  ace_mode: text
-  tm_scope: text.robots-txt
-  language_id: 674736065
 sed:
   type: programming
   color: "#64b970"

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "linguist-js",
-	"version": "3.0.0",
+	"version": "3.0.2",
 	"description": "Analyse the programming languages used in a folder or from raw content, using the same rules that GitHub Linguist does.",
 	"main": "dist/index.js",
 	"type": "module",
@@ -45,14 +45,14 @@
 		"common-path-prefix": "^3.0.0",
 		"ignore": "^7.0.5",
 		"isbinaryfile": "^6.0.0",
-		"js-yaml": "^4.1.1",
+		"js-yaml": "^4.3.0",
 		"node-cache": "^5.1.2"
 	},
 	"devDependencies": {
 		"@types/js-yaml": "^4.0.9",
-		"@types/node": "^25.9.1",
+		"@types/node": "^25.9.4",
 		"deep-object-diff": "^1.1.9",
-		"prettier": "^3.8.3",
+		"prettier": "^3.9.4",
 		"typescript": "~6.0.3"
 	}
 }

package/readme.md

@@ -50,8 +50,8 @@ Running LinguistJS on this will return the following JSON:
 {
   "files": {
     "count": 4,
-    "bytes": 31,
-    "lines": { "total": 8, "content": 4, "code": 4 },
+    "bytes": 6010,
+    "lines": { "total": 150, "content": 75 },
     "results": {
       "/info.md": "Markdown",
       "/no-lang": null,
@@ -61,35 +61,25 @@ Running LinguistJS on this will return the following JSON:
   },
   "languages": {
     "count": 3,
-    "bytes": 18,
-    "lines": { "total": 6, "content": 3, "code": 3 },
+    "bytes": 6000,
+    "lines": { "total": 147, "content": 74 },
     "results": {
-      "Markdown": {
-        "type": "prose",
-        "bytes": 6,
-        "lines": { "total": 2, "content": 1, "code": 1 },
-        "color": "#083fa1"
-      },
-      "JavaScript": {
-        "type": "programming",
-        "bytes": 6,
-        "lines": { "total": 2, "content": 1, "code": 1 },
-        "color": "#f1e05a"
-      },
-      "TypeScript": {
-        "type": "programming",
-        "bytes": 6,
-        "lines": { "total": 2, "content": 1, "code": 1 },
-        "color": "#3178c6"
-      }
+      "Markdown": { "count": 1, "bytes": 3000, "lines": { "total": 10, "content": 5 } },
+      "JavaScript": { "count": 1, "bytes": 1000, "lines": { "total": 46, "content": 23 } },
+      "TypeScript": { "count": 1, "bytes": 2000, "lines": { "total": 91, "content": 46 } }
     }
   },
   "unknown": {
     "count": 1,
-    "bytes": 13,
-    "lines": { "total": 2, "content": 1, "code": 1 },
+    "bytes": 10,
+    "lines": { "total": 3, "content": 1 },
     "extensions": {},
-    "filenames": { "no-lang": 13 }
+    "filenames": { "no-lang": 10 }
+  },
+  "repository": {
+    "Markdown": { "type": "prose", "color": "#083fa1" },
+    "JavaScript": { "type": "programming", "color": "#f1e05a" },
+    "TypeScript": { "type": "programming", "color": "#3178c6" }
   }
 }
 ```
@@ -126,15 +116,15 @@ const { files, languages, unknown, repository } = await linguist.analyseRawConte
 **Exports:**
 
 - `analyseFolders(folders?, opts?)`:
-  Analyse the language of all files found in a folder or folders.
+  Analyse the language of all files found in a list of folders.
   - `folders` (optional; string array):
     A list of folders to analyse (defaults to `['./']`).
   - `opts` (optional; object):
     An object containing analyser options.
-- `analyseRawContent(folders?, opts?)`:
-  Analyse the language of all files found in a folder or folders.
-  - `entry` (optional; string or string array):
-    A list of folders to analyse (defaults to `['./']`).
+- `analyseRawContent(fileContent, opts?)`:
+  Analyse the language of of preloaded file content.
+  - `fileContent` (string-string map):
+    A mapping of file name to file content to analyse.
   - `opts` (optional; object):
     An object containing analyser options.