git-codereview/app/lib/ocr.js

/** Runs the OpenCodeReview CLI and parses its JSON output. */
import { spawn } from "node:child_process";
import { existsSync } from "node:fs";
import { mkdtemp, rm, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import { join } from "node:path";

export class OcrError extends Error {
  constructor(message, { exitCode, stderr, stdout } = {}) {
    super(message);
    this.name = "OcrError";
    this.exitCode = exitCode;
    this.stderr = stderr;
    this.stdout = stdout;
  }
}

export const CATEGORY_VALUES = [
  "bug", "security", "performance", "maintainability",
  "test", "style", "documentation", "other",
];
export const SEVERITY_RANK = { critical: 4, high: 3, medium: 2, low: 1 };

const MAX_CAPTURE = 2 * 1024 * 1024;

function run(command, args, { cwd, env, timeoutMs, onOutput } = {}) {
  return new Promise((resolve) => {
    const child = spawn(command, args, { cwd, env, windowsHide: true });
    let stdout = "";
    let stderr = "";
    let settled = false;

    const finish = (result) => {
      if (settled) return;
      settled = true;
      clearTimeout(timer);
      resolve(result);
    };

    const timer = timeoutMs
      ? setTimeout(() => {
          try { child.kill("SIGKILL"); } catch { /* already gone */ }
          finish({ code: 124, stdout, stderr: `${stderr}\n[timeout after ${timeoutMs} ms]` });
        }, timeoutMs)
      : null;

    child.stdout.on("data", (chunk) => {
      const text = chunk.toString();
      if (stdout.length < MAX_CAPTURE) stdout += text;
      onOutput?.("stdout", text);
    });
    child.stderr.on("data", (chunk) => {
      const text = chunk.toString();
      if (stderr.length < MAX_CAPTURE) stderr += text;
      onOutput?.("stderr", text);
    });
    child.on("error", (err) => {
      finish({ code: 127, stdout, stderr: `${stderr}\n${err.message}` });
    });
    child.on("close", (code) => finish({ code, stdout, stderr }));
  });
}

function git(args, { cwd, timeoutMs = 300000 } = {}) {
  return run("git", args, { cwd, timeoutMs });
}

export class OcrRunner {
  constructor({
    command = "ocr",
    workspaceDir,
    llm = {},
    timeoutMs = 45 * 60 * 1000,
    gitTimeoutMs = 10 * 60 * 1000,
    logger = () => {},
  } = {}) {
    this.command = command;
    this.workspaceDir = workspaceDir;
    this.llm = llm;
    this.timeoutMs = timeoutMs;
    this.gitTimeoutMs = gitTimeoutMs;
    this.logger = logger;
  }

  ocrEnv(extra = {}) {
    const env = { ...process.env, ...extra };
    if (this.llm.url) env.OCR_LLM_URL = this.llm.url;
    if (this.llm.token) env.OCR_LLM_TOKEN = this.llm.token;
    if (this.llm.model) env.OCR_LLM_MODEL = this.llm.model;
    if (this.llm.protocol) env.OCR_LLM_PROTOCOL = this.llm.protocol;
    if (this.llm.authHeader) env.OCR_LLM_AUTH_HEADER = this.llm.authHeader;
    if (this.llm.extraHeaders) env.OCR_LLM_EXTRA_HEADERS = this.llm.extraHeaders;
    if (this.llm.timeoutSeconds) env.OCR_LLM_TIMEOUT = String(this.llm.timeoutSeconds);
    env.OCR_ENABLE_TELEMETRY = "0";
    return env;
  }

  /** Verify the OCR binary and the configured LLM endpoint. */
  async selfTest() {
    const version = await run(this.command, ["version"], {
      env: this.ocrEnv(), timeoutMs: 60000,
    });
    if (version.code !== 0) {
      throw new OcrError(`cannot run '${this.command} version'`, {
        exitCode: version.code, stderr: version.stderr, stdout: version.stdout,
      });
    }
    const test = await run(this.command, ["llm", "test"], {
      env: this.ocrEnv(), timeoutMs: 120000,
    });
    return {
      version: version.stdout.trim(),
      llmOk: test.code === 0,
      llmOutput: `${test.stdout}\n${test.stderr}`.trim(),
    };
  }

  async gitClone({ cloneUrl, token, dir }) {
    // No partial clone: a blob:none clone defers blob downloads to later
    // `git show` / `git grep` calls, and those lazily-fetched requests do not
    // inherit the per-command extraHeader, so private repositories fail with
    // "could not read Username". A full clone keeps every read local.
    const args = ["clone", "--no-tags"];
    const env = { ...process.env, GIT_TERMINAL_PROMPT: "0" };
    if (token) {
      // Written into .git/config by `git clone`, so every later fetch/rebase in
      // this workspace authenticates without extra plumbing.
      env.GIT_CONFIG_COUNT = "1";
      env.GIT_CONFIG_KEY_0 = "http.extraHeader";
      env.GIT_CONFIG_VALUE_0 = `Authorization: token ${token}`;
    }
    args.push(cloneUrl, dir);
    const res = await run("git", args, { env, timeoutMs: this.gitTimeoutMs });
    if (res.code !== 0) {
      throw new OcrError(`git clone failed for ${cloneUrl}`, {
        exitCode: res.code, stderr: res.stderr, stdout: res.stdout,
      });
    }
    // Persist the auth header for this workspace so later fetches keep working
    // even if the caller forgets to pass a token.
    if (token) {
      const cfg = await run("git", ["config", "--local", "http.extraHeader", `Authorization: token ${token}`], {
        cwd: dir, timeoutMs: 60000,
      });
      if (cfg.code !== 0) {
        throw new OcrError("failed to persist repository credentials", {
          exitCode: cfg.code, stderr: cfg.stderr, stdout: cfg.stdout,
        });
      }
    }
    return dir;
  }

  async fetch(dir, { refs = [], token } = {}) {
    // GIT_TERMINAL_PROMPT=0 keeps a missing credential a hard error instead of
    // a hanging prompt inside the container.
    const env = { ...process.env, GIT_TERMINAL_PROMPT: "0" };
    if (token) {
      env.GIT_CONFIG_COUNT = "1";
      env.GIT_CONFIG_KEY_0 = "http.extraHeader";
      env.GIT_CONFIG_VALUE_0 = `Authorization: token ${token}`;
    }
    const args = ["fetch", "--prune", "--no-tags", "origin"];
    for (const ref of refs) args.push(ref);
    const res = await run("git", args, { cwd: dir, env, timeoutMs: this.gitTimeoutMs });
    if (res.code !== 0) {
      throw new OcrError(`git fetch failed in ${dir}`, {
        exitCode: res.code, stderr: res.stderr, stdout: res.stdout,
      });
    }
    return res;
  }

  async revParse(dir, ref) {
    const res = await git(["rev-parse", "--verify", `${ref}^{commit}`], { cwd: dir, timeoutMs: 60000 });
    if (res.code !== 0) return null;
    return res.stdout.trim().split("\n")[0];
  }

  async mergeBase(dir, a, b) {
    const res = await git(["merge-base", a, b], { cwd: dir, timeoutMs: 120000 });
    if (res.code !== 0) return null;
    return res.stdout.trim().split("\n")[0];
  }

  /** Parent commit count for a commit; >1 means it is a merge commit. */
  async parentCount(dir, sha) {
    const res = await git(["rev-list", "--parents", "-n", "1", sha], { cwd: dir, timeoutMs: 60000 });
    if (res.code !== 0) return 0;
    const parts = res.stdout.trim().split(/\s+/).filter(Boolean);
    return Math.max(0, parts.length - 1);
  }

  async changedFiles(dir, fromSha, toSha) {
    const res = await git(
      ["diff", "--name-only", "--diff-filter=ACMRTUXB", fromSha, toSha],
      { cwd: dir, timeoutMs: this.gitTimeoutMs },
    );
    if (res.code !== 0) return [];
    return res.stdout.split("\n").map((s) => s.trim()).filter(Boolean);
  }

  /**
   * Run a diff review.
   * @returns {{ comments: object[], summary: object|null, raw: object, stderr: string }}
   */
  async review({
    dir, fromSha, toSha, excludes = [], background, concurrency = 4,
    maxComments = 30, maxTokensBudget = 0, rulePath, onOutput,
  }) {
    const outFile = join(await mkdtemp(join(tmpdir(), "ocr-out-")), "result.json");
    const args = [
      "review",
      "--repo", dir,
      "--from", fromSha,
      "--to", toSha,
      "--format", "json",
      "--audience", "agent",
      "--concurrency", String(concurrency),
      "--output", outFile,
    ];
    if (excludes.length) args.push("--exclude", excludes.join(","));
    if (background) args.push("--background", background);
    if (rulePath && existsSync(rulePath)) args.push("--rule", rulePath);
    if (maxTokensBudget > 0) args.push("--max-tokens-budget", String(maxTokensBudget));

    this.logger(`ocr review --from ${fromSha} --to ${toSha} (cwd=${dir})`);
    const res = await run(this.command, args, {
      cwd: dir, env: this.ocrEnv(), timeoutMs: this.timeoutMs, onOutput,
    });

    let raw = null;
    try {
      const { readFile } = await import("node:fs/promises");
      raw = JSON.parse(await readFile(outFile, "utf8"));
    } catch (err) {
      if (res.code !== 0) {
        throw new OcrError(`ocr review failed (exit ${res.code})`, {
          exitCode: res.code, stderr: res.stderr, stdout: res.stdout,
        });
      }
      throw new OcrError(`cannot parse ocr JSON output: ${err.message}`, {
        exitCode: res.code, stderr: res.stderr, stdout: res.stdout,
      });
    } finally {
      await rm(outFile, { force: true }).catch(() => {});
    }

    const comments = Array.isArray(raw?.comments) ? raw.comments : [];
    const selected = comments
      .filter((c) => c && typeof c.path === "string" && c.path.length > 0)
      .slice(0, Math.max(1, maxComments));
    return {
      comments: selected,
      totalComments: comments.length,
      summary: raw?.summary ?? null,
      status: raw?.status ?? null,
      projectSummary: raw?.project_summary ?? "",
      warnings: raw?.warnings ?? [],
      raw,
      stderr: res.stderr,
      exitCode: res.code,
    };
  }

  async preview({ dir, fromSha, toSha, excludes = [], onOutput }) {
    const args = [
      "review", "--repo", dir, "--from", fromSha, "--to", toSha,
      "--format", "json", "--audience", "agent", "--preview",
    ];
    if (excludes.length) args.push("--exclude", excludes.join(","));
    const res = await run(this.command, args, {
      cwd: dir, env: this.ocrEnv(), timeoutMs: 300000, onOutput,
    });
    if (res.code !== 0) {
      throw new OcrError(`ocr review --preview failed (exit ${res.code})`, {
        exitCode: res.code, stderr: res.stderr, stdout: res.stdout,
      });
    }
    return JSON.parse(res.stdout);
  }
}

export function severityAtLeast(severity, threshold) {
  const a = SEVERITY_RANK[String(severity || "").toLowerCase()] ?? 0;
  const b = SEVERITY_RANK[String(threshold || "").toLowerCase()] ?? 0;
  return a > 0 && b > 0 && a >= b;
}

export function parseList(value) {
  if (!value) return [];
  return String(value).split(",").map((s) => s.trim()).filter(Boolean);
}