makelore/electron/coding-projects/project-files.ts

import { open, lstat, readdir, realpath } from 'node:fs/promises';
import path from 'node:path';
import type {
  CodingProjectFileContent,
  CodingProjectFileEntry,
  CodingProjectFileStatus,
  CodingTextSearchResult,
} from '../../shared/coding-product-tools';
import {
  ProcessConversationGitAdapter,
  type ConversationGitAdapter,
  type GitCommandResult,
} from './conversation-change-tracker';

const MAX_STATUS_RESULTS = 200;
const DEFAULT_FIND_RESULTS = 20;
const MAX_FIND_RESULTS = 200;
const MAX_DISCOVERED_FILES = 20_000;
const MAX_CONTENT_BYTES = 256 * 1024;
const MAX_SEARCH_FILE_BYTES = 1024 * 1024;
const MAX_SEARCH_RESULTS = 200;
const MAX_SEARCH_LINE_CHARS = 2_048;
const SKIPPED_DIRECTORIES = new Set([
  '.git',
  '.next',
  '.nuxt',
  '.svelte-kit',
  'build',
  'coverage',
  'dist',
  'node_modules',
  'out',
]);

function normalizeRelativePath(value: string): string {
  const raw = value.trim();
  const slashPath = raw.replaceAll('\\', '/');
  if (!slashPath || slashPath.includes('\0') || path.isAbsolute(raw)
    || path.win32.isAbsolute(raw) || path.posix.isAbsolute(slashPath)) {
    throw new Error('Project file path must be relative');
  }
  const normalized = path.posix.normalize(slashPath).replace(/^\.\//, '');
  if (!normalized || normalized === '.' || normalized === '..' || normalized.startsWith('../')) {
    throw new Error('Project file path escapes the active project');
  }
  return normalized;
}

function projectTarget(projectPath: string, relativePath: string): string {
  const root = path.resolve(projectPath);
  const target = path.resolve(root, ...relativePath.split('/'));
  const relative = path.relative(root, target);
  if (!relative || relative.startsWith('..') || path.isAbsolute(relative)) {
    throw new Error('Project file path escapes the active project');
  }
  return target;
}

async function containedExistingTarget(projectPath: string, relativePath: string): Promise<string> {
  const root = await realpath(path.resolve(projectPath));
  const target = await realpath(projectTarget(root, relativePath));
  const relative = path.relative(root, target);
  if (!relative || relative.startsWith('..') || path.isAbsolute(relative)) {
    throw new Error('Project file path escapes the active project');
  }
  return target;
}

function statusFromSignature(
  signature: string,
  renamed: boolean,
  conflicted = false,
): CodingProjectFileStatus {
  if (signature === '??') return 'untracked';
  if (conflicted) return 'conflicted';
  if (renamed || signature.includes('R')) return 'renamed';
  if (signature.includes('D')) return 'deleted';
  if (signature.includes('A')) return 'added';
  return 'modified';
}

function parseStatus(value: string): CodingProjectFileEntry[] {
  const records = value.split('\0');
  const files: CodingProjectFileEntry[] = [];
  for (let index = 0; index < records.length; index += 1) {
    const record = records[index];
    if (!record || record.startsWith('! ')) continue;
    let filePath: string | undefined;
    let status: CodingProjectFileStatus;
    if (record.startsWith('? ')) {
      filePath = record.slice(2);
      status = 'untracked';
    } else {
      const renamed = record.startsWith('2 ');
      const conflicted = record.startsWith('u ');
      const match = conflicted
        ? record.match(/^u ([^ ]+) [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ (.*)$/s)
        : renamed
        ? record.match(/^2 ([^ ]+) [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ (.*)$/s)
        : record.match(/^1 ([^ ]+) [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ [^ ]+ (.*)$/s);
      if (!match) continue;
      filePath = match[2];
      status = statusFromSignature(match[1], renamed, conflicted);
      if (renamed) index += 1;
    }
    const normalized = normalizeRelativePath(filePath);
    files.push({
      path: normalized,
      name: path.posix.basename(normalized),
      type: 'file',
      status,
    });
    if (files.length >= MAX_STATUS_RESULTS) break;
  }
  return files;
}

async function readBoundedFile(
  projectPath: string,
  relativePath: string,
  maxBytes: number,
): Promise<{ data: Buffer; truncated: boolean }> {
  const target = await containedExistingTarget(projectPath, relativePath);
  const metadata = await lstat(target);
  if (!metadata.isFile()) throw new Error('Project file path is not a file');
  const handle = await open(target, 'r');
  const data = Buffer.alloc(Math.min(metadata.size, maxBytes + 1));
  let bytesRead: number;
  try {
    ({ bytesRead } = await handle.read(data, 0, data.byteLength, 0));
  } finally {
    await handle.close();
  }
  return {
    data: data.subarray(0, Math.min(bytesRead, maxBytes)),
    truncated: metadata.size > maxBytes || bytesRead > maxBytes,
  };
}

function decodeText(data: Buffer, allowIncompleteSuffix = false): string {
  if (data.includes(0)) throw new Error('Binary project files cannot be previewed');
  try {
    return new TextDecoder('utf-8', { fatal: true }).decode(
      data,
      allowIncompleteSuffix ? { stream: true } : undefined,
    );
  } catch {
    throw new Error('Project file is not valid UTF-8 text');
  }
}

function foldedTextWithOffsets(value: string): {
  text: string;
  starts: number[];
  ends: number[];
} {
  let text = '';
  const starts: number[] = [];
  const ends: number[] = [];
  for (let index = 0; index < value.length;) {
    const codePoint = value.codePointAt(index);
    if (codePoint === undefined) break;
    const character = String.fromCodePoint(codePoint);
    const end = index + character.length;
    const folded = character.toLowerCase();
    text += folded;
    for (let foldedIndex = 0; foldedIndex < folded.length; foldedIndex += 1) {
      starts.push(index);
      ends.push(end);
    }
    index = end;
  }
  return { text, starts, ends };
}

async function fallbackFileList(projectPath: string): Promise<string[]> {
  const root = path.resolve(projectPath);
  const files: string[] = [];
  const pending = [''];
  while (pending.length > 0 && files.length < MAX_DISCOVERED_FILES) {
    const relativeDirectory = pending.shift() as string;
    const target = relativeDirectory
      ? path.join(root, ...relativeDirectory.split('/'))
      : root;
    let entries;
    try {
      entries = await readdir(target, { withFileTypes: true });
    } catch {
      continue;
    }
    entries.sort((left, right) => left.name.localeCompare(right.name));
    for (const entry of entries) {
      const relativePath = relativeDirectory
        ? `${relativeDirectory}/${entry.name}`
        : entry.name;
      if (entry.isDirectory()) {
        if (!SKIPPED_DIRECTORIES.has(entry.name)) pending.push(relativePath);
      } else if (entry.isFile()) {
        files.push(relativePath);
        if (files.length >= MAX_DISCOVERED_FILES) break;
      }
    }
  }
  return files;
}

export class CodingProjectFileService {
  constructor(
    private readonly git: ConversationGitAdapter = new ProcessConversationGitAdapter(),
  ) {}

  async status(projectPath: string): Promise<CodingProjectFileEntry[]> {
    const result = await this.gitResult(projectPath, [
      'status',
      '--porcelain=v2',
      '-z',
      '--untracked-files=all',
      '--',
      '.',
    ]);
    return result?.code === 0 ? parseStatus(result.stdout) : [];
  }

  async find(
    projectPath: string,
    query: string,
    requestedLimit = DEFAULT_FIND_RESULTS,
  ): Promise<CodingProjectFileEntry[]> {
    const normalizedQuery = query.trim().toLocaleLowerCase();
    if (!normalizedQuery) throw new Error('File query is required');
    if (normalizedQuery.length > 200) throw new Error('File query is too long');
    const limit = Math.min(
      Number.isSafeInteger(requestedLimit) && requestedLimit > 0 ? requestedLimit : DEFAULT_FIND_RESULTS,
      MAX_FIND_RESULTS,
    );
    const paths = await this.listFiles(projectPath);
    const matches = paths
      .filter((filePath) => filePath.toLocaleLowerCase().includes(normalizedQuery))
      .sort((left, right) => {
        const leftName = path.posix.basename(left).toLocaleLowerCase();
        const rightName = path.posix.basename(right).toLocaleLowerCase();
        const leftPrefix = leftName.startsWith(normalizedQuery) ? 0 : 1;
        const rightPrefix = rightName.startsWith(normalizedQuery) ? 0 : 1;
        return leftPrefix - rightPrefix || left.length - right.length || left.localeCompare(right);
      })
      .slice(0, limit);
    return await Promise.all(matches.map(async (filePath) => {
      let size: number | undefined;
      try {
        const metadata = await lstat(projectTarget(projectPath, filePath));
        if (metadata.isFile()) size = metadata.size;
      } catch {
        // A concurrently removed result remains useful by path.
      }
      return {
        path: filePath,
        name: path.posix.basename(filePath),
        type: 'file' as const,
        ...(size === undefined ? {} : { size }),
      };
    }));
  }

  async content(projectPath: string, requestedPath: string): Promise<CodingProjectFileContent> {
    const relativePath = normalizeRelativePath(requestedPath);
    const result = await readBoundedFile(projectPath, relativePath, MAX_CONTENT_BYTES);
    return {
      path: relativePath,
      content: decodeText(result.data, result.truncated),
      truncated: result.truncated,
    };
  }

  async search(projectPath: string, pattern: string): Promise<CodingTextSearchResult[]> {
    const needle = pattern.trim();
    if (!needle) throw new Error('Search pattern is required');
    if (needle.length > 512) throw new Error('Search pattern is too long');
    const foldedNeedle = foldedTextWithOffsets(needle).text;
    const matches: CodingTextSearchResult[] = [];
    for (const filePath of await this.listFiles(projectPath)) {
      if (matches.length >= MAX_SEARCH_RESULTS) break;
      let content: string;
      try {
        const bounded = await readBoundedFile(projectPath, filePath, MAX_SEARCH_FILE_BYTES);
        if (bounded.truncated) continue;
        content = decodeText(bounded.data);
      } catch {
        continue;
      }
      const lines = content.split(/\r?\n/);
      for (let lineIndex = 0; lineIndex < lines.length && matches.length < MAX_SEARCH_RESULTS; lineIndex += 1) {
        const line = lines[lineIndex];
        const foldedLine = foldedTextWithOffsets(line);
        const first = foldedLine.text.indexOf(foldedNeedle);
        if (first < 0) continue;
        const firstOriginalIndex = foldedLine.starts[first] ?? 0;
        const windowStart = Math.max(0, firstOriginalIndex - 256);
        const lineText = line.slice(windowStart, windowStart + MAX_SEARCH_LINE_CHARS);
        const foldedText = foldedTextWithOffsets(lineText);
        const submatches = [];
        let offset = 0;
        while (submatches.length < 20) {
          const foldedStart = foldedText.text.indexOf(foldedNeedle, offset);
          if (foldedStart < 0) break;
          const foldedEnd = foldedStart + foldedNeedle.length;
          const start = foldedText.starts[foldedStart];
          const end = foldedText.ends[foldedEnd - 1];
          if (start === undefined || end === undefined) break;
          submatches.push({ text: lineText.slice(start, end), start, end });
          offset = Math.max(foldedEnd, foldedStart + 1);
        }
        matches.push({
          path: filePath,
          name: path.posix.basename(filePath),
          lineNumber: lineIndex + 1,
          lineText,
          submatches,
        });
      }
    }
    return matches;
  }

  private async listFiles(projectPath: string): Promise<string[]> {
    const repository = await this.gitResult(projectPath, ['rev-parse', '--is-inside-work-tree']);
    if (repository?.code === 0 && repository.stdout.trim() === 'true') {
      const listed = await this.gitResult(projectPath, [
        'ls-files',
        '-co',
        '--exclude-standard',
        '-z',
        '--',
        '.',
      ]);
      if (listed?.code === 0) {
        return [...new Set(listed.stdout
          .split('\0')
          .filter(Boolean)
          .map(normalizeRelativePath))]
          .sort()
          .slice(0, MAX_DISCOVERED_FILES);
      }
    }
    return await fallbackFileList(projectPath);
  }

  private async gitResult(projectPath: string, args: readonly string[]): Promise<GitCommandResult | null> {
    try {
      return await this.git.run(projectPath, args);
    } catch {
      return null;
    }
  }
}