From 51a7be2cb8e8b00a7758db0d6f730c8da72d291b Mon Sep 17 00:00:00 2001
From: Nixinova <nixinova@gmail.com>
Date: Tue, 19 Mar 2024 13:13:06 +1300
Subject: [PATCH] Apply gitignores as folders are traversed

Improves performance 3x.

Also, ensure gitignore reading is done to spec: https://git-scm.com/docs/gitignore.
---
 changelog.md                   |  4 ++++
 src/helpers/parse-gitignore.ts |  9 +++++++++
 src/helpers/walk-tree.ts       | 11 +++++++++++
 src/index.ts                   | 22 ----------------------
 4 files changed, 24 insertions(+), 22 deletions(-)
 create mode 100644 src/helpers/parse-gitignore.ts

diff --git a/changelog.md b/changelog.md
index 49922c0..4475b57 100644
--- a/changelog.md
+++ b/changelog.md
@@ -1,5 +1,9 @@
 # Changelog
 
+## Next
+- Fixed gitignore file reading not pertaining exactly to spec.
+- Improved performance by filtering out gitignored files during folder traversal.
+
 ## 2.7.0
 
 ### 2.7.0 (RC)
diff --git a/src/helpers/parse-gitignore.ts b/src/helpers/parse-gitignore.ts
new file mode 100644
index 0000000..fb0ccaf
--- /dev/null
+++ b/src/helpers/parse-gitignore.ts
@@ -0,0 +1,9 @@
+export default function parseGitignore(content: string): string[] {
+	const readableData = content
+		// Remove comments unless escaped
+		.replace(/(?<!\\)#.+/g, '')
+		// Remove whitespace unless escaped
+		.replace(/(?:(?<!\\)\s)+$/g, '')
+	const arrayData = readableData.split(/\r?\n/).filter(data => data);
+	return arrayData;
+}
diff --git a/src/helpers/walk-tree.ts b/src/helpers/walk-tree.ts
index 8ccbd82..5c0dda6 100644
--- a/src/helpers/walk-tree.ts
+++ b/src/helpers/walk-tree.ts
@@ -1,6 +1,7 @@
 import fs from 'fs';
 import paths from 'path';
 import ignore, { Ignore } from 'ignore';
+import parseGitignore from './parse-gitignore';
 
 let allFiles: Set<string>;
 let allFolders: Set<string>;
@@ -37,6 +38,7 @@ export default function walk(data: WalkInput): WalkOutput {
 	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
@@ -45,6 +47,15 @@ export default function walk(data: WalkInput): WalkOutput {
 			const isDir = fs.lstatSync(paths.resolve(commonRoot, base)).isDirectory();
 			return isDir ? `${base}/` : base;
 		});
+
+		// Read and apply gitignores
+		const gitignoreFilename = paths.join(folder, '.gitignore');
+		if (fs.existsSync(gitignoreFilename)) {
+			const gitignoreContents = fs.readFileSync(gitignoreFilename, 'utf-8');
+			const ignoredPaths = parseGitignore(gitignoreContents);
+			ignored.add(ignoredPaths);
+		}
+
 		// Loop through files and folders
 		for (const file of files) {
 			// Create absolute path for disc operations
diff --git a/src/index.ts b/src/index.ts
index 666096b..8dac672 100644
--- a/src/index.ts
+++ b/src/index.ts
@@ -72,36 +72,14 @@ async function analyse(rawPaths?: string | string[], opts: T.Options = {}): Prom
 
 	// Load file paths and folders
 	let files: T.AbsFile[];
-	let folders: T.AbsFolder[];
 	if (useRawContent) {
 		// Uses raw file content
 		files = input;
-		folders = [''];
 	}
 	else {
 		// Uses directory on disc
 		const data = walk({ init: true, commonRoot, folderRoots: resolvedInput, folders: resolvedInput, ignored });
 		files = data.files;
-		folders = data.folders;
-	}
-
-	// Load gitignore data and apply ignores rules
-	if (!useRawContent && opts.checkIgnored) {
-		const nestedIgnoreFiles = files.filter(file => file.endsWith('.gitignore'));
-		for (const ignoresFile of nestedIgnoreFiles) {
-			const relIgnoresFile = relPath(ignoresFile);
-			const relIgnoresFolder = paths.dirname(relIgnoresFile);
-			// Parse gitignores
-			const ignoresDataRaw = await readFile(ignoresFile);
-			const ignoresData = ignoresDataRaw.replace(/#.+|\s+$/gm, '');
-			const absoluteIgnoresData = ignoresData
-				// '.file' -> 'root/*/.file'
-				.replace(/^(?=[^\s\/\\])/gm, localRoot(relIgnoresFolder) + '/*/')
-				// '/folder' -> 'root/folder'
-				.replace(/^[\/\\]/gm, localRoot(relIgnoresFolder) + '/')
-			ignored.add(absoluteIgnoresData);
-			files = filterOutIgnored(files, ignored);
-		}
 	}
 
 	// Fetch and normalise gitattributes data of all subfolders and save to metadata