renovate/lib/modules/manager/pre-commit/extract.ts

import {
  isEmptyObject,
  isNonEmptyObject,
  isPlainObject,
} from '@sindresorhus/is';
import type { PlatformFamilyId } from '../../../constants/index.ts';
import { PLATFORM_FAMILIES } from '../../../constants/index.ts';
import { logger } from '../../../logger/index.ts';
import type { SkipReason } from '../../../types/index.ts';
import { coerceArray } from '../../../util/array.ts';
import { detectPlatform } from '../../../util/common.ts';
import { find } from '../../../util/host-rules.ts';
import { newlineRegex, regEx } from '../../../util/regex.ts';
import { parseSingleYaml } from '../../../util/yaml.ts';
import { parseLine } from '../gomod/line-parser.ts';
import { extractDependency as npmExtractDependency } from '../npm/extract/common/dependency.ts';
import { pep508ToPackageDependency } from '../pep621/utils.ts';
import type { PackageDependency, PackageFileContent } from '../types.ts';
import {
  matchesPrecommitConfigHeuristic,
  matchesPrecommitDependencyHeuristic,
} from './parsing.ts';
import type { PreCommitConfig } from './types.ts';

/** The platform families pre-commit can resolve `rev:` tags on. */
const supportedFamilies = [
  'forgejo',
  'github',
  'gitlab',
] as const satisfies PlatformFamilyId[];

/**
 * Determines the datasource(id) to be used for this dependency
 * @param repository the full git url, ie git@github.com/user/project.
 *        Used in debug statements to clearly indicate the related dependency.
 * @param hostname the hostname (ie github.com)
 *        Used to determine which renovate datasource should be used.
 *        Is matched literally against `github.com` and `gitlab.com`.
 *        If that doesn't match, `hostRules.find()` is used to find related sources.
 *        In that case, the hostname is passed on as registryUrl to the corresponding datasource.
 */
function determineDatasource(
  repository: string,
  hostname: string,
): { datasource?: string; registryUrls?: string[]; skipReason?: SkipReason } {
  const platform = detectPlatform(repository);

  if (platform === 'forgejo') {
    logger.debug(
      { repository, hostname },
      'Found forgejo dependency with custom registryUrl',
    );
    return {
      datasource: PLATFORM_FAMILIES.forgejo.tagsDatasource,
      registryUrls: [`https://${hostname}`],
    };
  }
  if (hostname === 'github.com' || platform === 'github') {
    logger.debug({ repository, hostname }, 'Found github dependency');
    return { datasource: PLATFORM_FAMILIES.github.tagsDatasource };
  }
  if (hostname === 'gitlab.com') {
    logger.debug({ repository, hostname }, 'Found gitlab dependency');
    return { datasource: PLATFORM_FAMILIES.gitlab.tagsDatasource };
  }
  if (platform === 'gitlab') {
    logger.debug(
      { repository, hostname },
      'Found gitlab dependency with custom registryUrl',
    );
    return {
      datasource: PLATFORM_FAMILIES.gitlab.tagsDatasource,
      registryUrls: [`https://${hostname}`],
    };
  }
  const hostUrl = `https://${hostname}`;
  const res = find({ url: hostUrl });
  if (isEmptyObject(res)) {
    // 1 check, to possibly prevent 3 failures in combined query of hostType & url.
    logger.debug(
      { repository, hostUrl },
      'Provided hostname does not match any hostRules. Ignoring',
    );
    return { skipReason: 'unknown-registry', registryUrls: [hostname] };
  }
  for (const hostType of supportedFamilies) {
    if (isNonEmptyObject(find({ hostType, url: hostUrl }))) {
      logger.debug(
        { repository, hostUrl, hostType },
        `Provided hostname matches a ${hostType} hostrule.`,
      );
      return {
        datasource: PLATFORM_FAMILIES[hostType].tagsDatasource,
        registryUrls: [hostname],
      };
    }
  }
  logger.debug(
    { repository, registry: hostUrl },
    'Provided hostname did not match any of the hostRules of hostType forgejo, github nor gitlab',
  );
  return { skipReason: 'unknown-registry', registryUrls: [hostname] };
}

const gitUrlRegex = regEx(/\.git$/i);

// Matches: rev: <digest><whitespace># frozen: <version>
const revLineWithFrozenCommentRegex = regEx(
  /^\s*rev:\s*(?<replaceString>(?<currentDigest>[a-f0-9]{40})(?<commentWhiteSpaces>\s+)#\s*frozen:\s*(?<currentValue>\S+))/,
);

interface RegexDep {
  currentDigest: string;
  currentValue: string;
  replaceString: string;
  autoReplaceStringTemplate: string;
}

function extractWithRegex(content: string): Map<string, RegexDep> {
  logger.trace('pre-commit.extractWithRegex()');
  const regexDeps = new Map<string, RegexDep>();

  for (const line of content.split(newlineRegex)) {
    if (line.trim().startsWith('#')) {
      continue;
    }

    const match = revLineWithFrozenCommentRegex.exec(line);
    if (match?.groups) {
      const { currentDigest, currentValue, replaceString, commentWhiteSpaces } =
        match.groups;

      // Store by digest to correlate with YAML-extracted deps later
      regexDeps.set(currentDigest, {
        currentDigest,
        currentValue,
        replaceString,
        autoReplaceStringTemplate: `{{newDigest}}${commentWhiteSpaces}# frozen: {{newValue}}`,
      });
    }
  }

  return regexDeps;
}

function extractDependency(tag: string, repository: string): PackageDependency {
  logger.debug(`Found version ${tag}`);

  const urlMatchers = [
    // This splits "http://my.github.com/user/repo" -> "my.github.com" "user/repo
    regEx('^https?://(?<hostname>[^/]+)/(?<depName>\\S*)'),
    // This splits "git@private.registry.com:user/repo" -> "private.registry.com" "user/repo
    regEx('^git@(?<hostname>[^:]+):(?<depName>\\S*)'),
    // This split "git://github.com/pre-commit/pre-commit-hooks" -> "github.com" "pre-commit/pre-commit-hooks"
    regEx(/^git:\/\/(?<hostname>[^/]+)\/(?<depName>\S*)/),
    // This splits "ssh://git@github.com/pre-commit/pre-commit-hooks" -> "github.com" "pre-commit/pre-commit-hooks"
    regEx(/^ssh:\/\/git@(?<hostname>[^/]+)\/(?<depName>\S*)/),
  ];
  for (const urlMatcher of urlMatchers) {
    const match = urlMatcher.exec(repository);
    if (match?.groups) {
      const hostname = match.groups.hostname;
      const depName = match.groups.depName.replace(gitUrlRegex, '');
      const sourceDef = determineDatasource(repository, hostname);
      return {
        ...sourceDef,
        depName,
        depType: 'repository',
        packageName: depName,
        currentValue: tag,
      };
    }
  }
  logger.info(
    { repository },
    'Could not separate hostname from full dependency url.',
  );
  return {
    depName: undefined,
    depType: 'repository',
    datasource: undefined,
    packageName: undefined,
    skipReason: 'invalid-url',
    currentValue: tag,
  };
}

/**
 * Find all supported dependencies in the pre-commit yaml object.
 *
 * @param precommitFile the parsed yaml config file
 * @param regexDeps Map of regex-extracted deps keyed by digest for enrichment
 */
function findDependencies(
  precommitFile: PreCommitConfig,
  regexDeps: Map<string, RegexDep>,
): PackageDependency[] {
  if (!precommitFile.repos) {
    logger.debug(`No repos section found, skipping file`);
    return [];
  }
  const packageDependencies: PackageDependency[] = [];

  for (const item of precommitFile.repos) {
    // meta hooks is defined from pre-commit and doesn't support `additional_dependencies`
    if (item.repo !== 'meta') {
      for (const hook of coerceArray(item.hooks)) {
        // normally language are not defined in yaml
        // only support it when it's explicitly defined.
        // this avoid to parse hooks from pre-commit-hooks.yaml from git repo
        if (hook.language === 'node') {
          hook.additional_dependencies?.map((req) => {
            const match = regEx('^(?<name>.+)@(?<range>.+)$').exec(req);
            if (!match?.groups) {
              return;
            }

            const depType = 'pre-commit-node';
            const dep = npmExtractDependency(
              depType,
              match.groups.name,
              match.groups.range,
            );
            packageDependencies.push({
              depType,
              depName: match.groups.name,
              packageName: match.groups.name,
              ...dep,
            });
          });
        } else if (hook.language === 'python') {
          hook.additional_dependencies?.map((req) => {
            const dep = pep508ToPackageDependency('pre-commit-python', req);
            if (dep) {
              packageDependencies.push(dep);
            }
          });
        } else if (hook.language === 'golang') {
          hook.additional_dependencies?.map((req) => {
            // Convert dependency into a gomod require line to use the gomod line parser
            const requireLine = `require ${req.replace('@', ' ')}`;
            const dep = parseLine(requireLine);
            if (dep) {
              const depType = 'pre-commit-golang';
              packageDependencies.push({
                ...dep,
                depType,
              });
            }
          });
        }
      }
    }

    if (matchesPrecommitDependencyHeuristic(item)) {
      logger.trace(item, 'Matched pre-commit dependency spec');
      const repository = String(item.repo);
      const tag = String(item.rev);
      const dep = extractDependency(tag, repository);

      // Check if this rev has regex-extracted formatting info
      const regexDep = regexDeps.get(tag);
      if (regexDep) {
        // Enrich with formatting info from regex extraction
        dep.currentDigest = regexDep.currentDigest;
        dep.currentValue = regexDep.currentValue;
        dep.replaceString = regexDep.replaceString;
        dep.autoReplaceStringTemplate = regexDep.autoReplaceStringTemplate;
      }

      packageDependencies.push(dep);
    } else {
      logger.trace(item, 'Did not find pre-commit repo spec');
    }
  }
  return packageDependencies;
}

export function extractPackageFile(
  content: string,
  packageFile: string,
): PackageFileContent | null {
  type ParsedContent = Record<string, unknown> | PreCommitConfig;
  let parsedContent: ParsedContent;
  try {
    // TODO: use schema (#9610)
    parsedContent = parseSingleYaml(content);
  } catch (err) {
    logger.debug(
      { filename: packageFile, err },
      'Failed to parse pre-commit config YAML',
    );
    return null;
  }
  if (!isPlainObject<Record<string, unknown>>(parsedContent)) {
    logger.debug(
      { packageFile },
      `Parsing of pre-commit config YAML returned invalid result`,
    );
    return null;
  }
  if (!matchesPrecommitConfigHeuristic(parsedContent)) {
    logger.debug(
      { packageFile },
      `File does not look like a pre-commit config file`,
    );
    return null;
  }
  try {
    const regexDeps = extractWithRegex(content);
    const deps = findDependencies(parsedContent, regexDeps);
    if (deps.length) {
      logger.trace({ deps }, 'Found dependencies in pre-commit config');
      return { deps };
    }
  } catch (err) /* istanbul ignore next */ {
    logger.debug(
      { packageFile, err },
      'Error scanning parsed pre-commit config',
    );
  }
  return null;
}