apache-autoindex-parse/dist/traverse.cjs

parse apache's autoindex html files

const require_src = require('./src-DqwlaVzi.cjs');

//#region src/traverse.ts
/**
* Recursively traverses an Apache autoindex directory structure.
*
* This function fetches the HTML content from the provided URL, parses it to extract directory entries,
* and then recursively traverses any subdirectories found.
*
* @param {string} rootUrl - The URL of the Apache autoindex directory to traverse
* @param {TraverseOptions?} options - Optional configuration for the traversal process
* @returns {Promise<TraverseEntry[]>} A promise that resolves to a RootEntry object representing the directory structure, or null if parsing failed
*
* @example
* ```typescript
* import { traverse } from 'apache-autoindex-parse/traverse';
*
* const directoryStructure = await traverse('https://example.com/files');
* ```
*/
async function traverse(rootUrl, options) {
	return traverseInternal(rootUrl, "", options);
}
async function traverseInternal(rootUrl, pathPrefix, options) {
	try {
		const res = await fetch(rootUrl, {
			headers: {
				"User-Agent": "github.com/luxass/apache-autoindex-parse",
				...options?.extraHeaders
			},
			signal: options?.abortSignal
		});
		if (!res.ok) throw new Error(`failed to fetch directory listing from ${rootUrl}: ${res.status} ${res.statusText}`);
		const rootEntries = require_src.parse(await res.text(), options?.format);
		if (!rootEntries) return [];
		return await Promise.all(rootEntries.map(async (entry) => {
			let fullPath = pathPrefix ? `${pathPrefix}/${entry.path}` : entry.path;
			fullPath = require_src.trimTrailingSlash(require_src.trimLeadingSlash(fullPath));
			if (entry.type === "file") {
				const newFileEntry = {
					...entry,
					path: fullPath
				};
				await options?.onFile?.(newFileEntry);
				return newFileEntry;
			}
			const child = await traverseInternal(rootUrl.endsWith("/") ? rootUrl + (entry.path.startsWith("/") ? entry.path.slice(1) : entry.path) : rootUrl + (entry.path.startsWith("/") ? entry.path : `/${entry.path}`), fullPath, options);
			entry.name = require_src.trimTrailingSlash(entry.name);
			const dirEntry = {
				...entry,
				path: fullPath,
				children: child
			};
			await options?.onDirectory?.(dirEntry);
			return dirEntry;
		}));
	} catch {
		return [];
	}
}

//#endregion
exports.traverse = traverse;