Properly fix file path resolutions

Author: Nixinova

changelog.md

@@ -1,7 +1,7 @@
 # Changelog
 
 ## Next
-- Fixed the `.git` folder not being ignored when analysing multiple folders ([#25](https://github.com/Nixinova/LinguistJS/issues/25)).
+- Fixed file paths not resolving properly when analysing multiple folders ([#25](https://github.com/Nixinova/LinguistJS/issues/25)).
 
 ## 2.6.0
 *2023-06-29*
@@ -21,7 +21,7 @@
 *2023-01-11*
 - Fixed gitattributes wildcards not being applied into subfolders ([#17](https://github.com/Nixinova/LinguistJS/issues/17)).
 
-# 2.5.3
+## 2.5.3
 *2022-09-03*
 - Fixed a crash occurring when parsing heuristics for `.txt` files ([#16](https://github.com/Nixinova/LinguistJS/issues/16)).
 

src/helpers/walk-tree.ts

@@ -5,8 +5,26 @@ import { Ignore } from 'ignore';
 let allFiles: Set<string>;
 let allFolders: Set<string>;
 
+interface WalkInput {
+	/** Whether this is walking the tree from the root */
+	init: boolean,
+	/** The common root absolute path of all folders being checked */
+	commonRoot: string,
+	/** The absolute path that each folder is relative to */
+	folderRoots: string[],
+	/** The absolute path of folders being checked */
+	folders: string[],
+	gitignores: Ignore,
+	regexIgnores: RegExp[],
+};
+interface WalkOutput {
+	files: string[],
+	folders: string[],
+};
+
 /** Generate list of files in a directory. */
-export default function walk(init: boolean, root: string, folders: string[], gitignores: Ignore, regexIgnores: RegExp[]): { files: string[], folders: string[] } {
+export default function walk(data: WalkInput): WalkOutput {
+	const { init, commonRoot, folderRoots, folders, gitignores, regexIgnores } = data;
 
 	// Initialise files and folders lists
 	if (init) {
@@ -17,46 +35,48 @@ export default function walk(init: boolean, root: string, folders: string[], git
 	// Walk tree of a folder
 	if (folders.length === 1) {
 		const folder = folders[0];
+		const localRoot = folderRoots[0].replace(commonRoot, '').replace(/^\//, '');
 		// Get list of files and folders inside this folder
 		const files = fs.readdirSync(folder).map(file => {
 			// Create path relative to root
-			const base = paths.resolve(folder, file).replace(/\\/g, '/').replace(root, '.');
+			const base = paths.resolve(folder, file).replace(/\\/g, '/').replace(commonRoot, '.');
 			// Add trailing slash to mark directories
-			const isDir = fs.lstatSync(paths.resolve(root, base)).isDirectory();
+			const isDir = fs.lstatSync(paths.resolve(commonRoot, base)).isDirectory();
 			return isDir ? `${base}/` : base;
 		});
 		// Loop through files and folders
 		for (const file of files) {
 			// Create absolute path for disc operations
-			const path = paths.resolve(root, file).replace(/\\/g, '/');
+			const path = paths.resolve(commonRoot, file).replace(/\\/g, '/');
+			const localPath = localRoot ? file.replace(`./${localRoot}/`, '') : file.replace('./', '');
 			// Skip if nonexistant or ignored
 			const nonExistant = !fs.existsSync(path);
-			const isGitIgnored = gitignores.test(file.replace('./', '')).ignored;
-			const isRegexIgnored = regexIgnores.find(match => file.replace('./', '').match(match));
+			const isGitIgnored = gitignores.test(localPath).ignored;
+			const isRegexIgnored = regexIgnores.find(match => localPath.match(match));
 			if (nonExistant || isGitIgnored || isRegexIgnored) continue;
 			// Add absolute folder path to list
 			allFolders.add(paths.resolve(folder).replace(/\\/g, '/'));
 			// Check if this is a folder or file
 			if (file.endsWith('/')) {
 				// Recurse into subfolders
 				allFolders.add(path);
-				walk(false, root, [path], gitignores, regexIgnores);
+				walk({ init: false, commonRoot: commonRoot, folderRoots, folders: [path], gitignores, regexIgnores });
 			}
 			else {
-				// Add relative file path to list
+				// Add file path to list
 				allFiles.add(path);
 			}
 		}
 	}
 	// Recurse into all folders
 	else {
-		for (const path of folders) {
-			walk(false, root, [path], gitignores, regexIgnores);
+		for (const i in folders) {
+			walk({ init: false, commonRoot: commonRoot, folderRoots: [folderRoots[i]], folders: [folders[i]], gitignores, regexIgnores });
 		}
 	}
 	// Return absolute files and folders lists
 	return {
-		files: [...allFiles].map(file => file.replace(/^\./, root)),
+		files: [...allFiles].map(file => file.replace(/^\./, commonRoot)),
 		folders: [...allFolders],
 	};
 }

src/index.ts

@@ -46,10 +46,13 @@ async function analyse(input?: string | string[], opts: T.Options = {}): Promise
 	if (opts.ignoredFiles) gitignores.add(opts.ignoredFiles);
 
 	// Set a common root path so that vendor paths do not incorrectly match parent folders
-	const resolvedInput = input.map(path => paths.resolve(path).replace(/\\/g, '/'));
+	const normPath = (file: string) => file.replace(/\\/g, '/');
+	const resolvedInput = input.map(path => normPath(paths.resolve(path)));
 	const commonRoot = (input.length > 1 ? commonPrefix(resolvedInput) : resolvedInput[0]).replace(/\/?$/, '');
-	const relPath = (file: string) => paths.relative(commonRoot, file).replace(/\\/g, '/');
-	const unRelPath = (file: string) => paths.resolve(commonRoot, file).replace(/\\/g, '/');
+	const localRoot = (folder: string) => folder.replace(commonRoot, '').replace(/^\//, '');
+	const relPath = (file: string) => normPath(paths.relative(commonRoot, file));
+	const unRelPath = (file: string) => normPath(paths.resolve(commonRoot, file));
+	const localPath = (file: string) => localRoot(unRelPath(file));
 
 	// Load file paths and folders
 	let files, folders;
@@ -60,7 +63,7 @@ async function analyse(input?: string | string[], opts: T.Options = {}): Promise
 	}
 	else {
 		// Uses directory on disc
-		const data = walk(true, commonRoot, input, gitignores, regexIgnores);
+		const data = walk({ init: true, commonRoot, folderRoots: resolvedInput, folders: resolvedInput, gitignores, regexIgnores });
 		files = data.files;
 		folders = data.folders;
 	}
@@ -90,6 +93,8 @@ async function analyse(input?: string | string[], opts: T.Options = {}): Promise
 	const customText = ignore();
 	if (!useRawContent && opts.checkAttributes) {
 		for (const folder of folders) {
+			// TODO FIX: this is absolute when only 1 path given
+			const localFilePath = (path: string) => localRoot(folder) ? localRoot(folder) + '/' + localPath(path) : path;
 
 			// Skip if folder is marked in gitattributes
 			if (relPath(folder) && gitignores.ignores(relPath(folder))) {
@@ -100,7 +105,8 @@ async function analyse(input?: string | string[], opts: T.Options = {}): Promise
 			const ignoresFile = paths.join(folder, '.gitignore');
 			if (opts.checkIgnored && fs.existsSync(ignoresFile)) {
 				const ignoresData = await readFile(ignoresFile);
-				gitignores.add(ignoresData);
+				const localIgnoresData = ignoresData.replace(/^[\/\\]/g, localRoot(folder) + '/');
+				gitignores.add(localIgnoresData);
 			}
 
 			// Parse gitattributes
@@ -111,16 +117,16 @@ async function analyse(input?: string | string[], opts: T.Options = {}): Promise
 				const contentTypeMatches = attributesData.matchAll(/^(\S+).*?(-?binary|-?text)(?!=auto)/gm);
 				for (const [_line, path, type] of contentTypeMatches) {
 					if (['text', '-binary'].includes(type)) {
-						customText.add(path);
+						customText.add(localFilePath(path));
 					}
 					if (['-text', 'binary'].includes(type)) {
-						customBinary.add(path);
+						customBinary.add(localFilePath(path));
 					}
 				}
 				// Custom vendor options
 				const vendorMatches = attributesData.matchAll(/^(\S+).*[^-]linguist-(vendored|generated|documentation)(?!=false)/gm);
 				for (const [_line, path] of vendorMatches) {
-					gitignores.add(path);
+					gitignores.add(localFilePath(path));
 				}
 				// Custom file associations
 				const customLangMatches = attributesData.matchAll(/^(\S+).*[^-]linguist-language=(\S+)/gm);
@@ -147,7 +153,7 @@ async function analyse(input?: string | string[], opts: T.Options = {}): Promise
 			files = files.filter(file => !regexIgnores.find(match => match.test(file)));
 		}
 		else {
-			files = gitignores.filter(files.map(relPath)).map(unRelPath);
+			files = gitignores.filter(files.map(localPath)).map(unRelPath);
 		}
 	}