citeclaw 2.0.1 → 2.0.3

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "citeclaw",
-	"version": "2.0.1",
+	"version": "2.0.3",
 	"description": "Citation and bibliography toolkit for DOI, URL, arXiv, PDF, Zotero, and MCP workflows.",
 	"homepage": "https://github.com/trotsky1997/citeclaw",
 	"license": "Apache-2.0",

package/scripts/botcite.js

@@ -28,13 +28,45 @@ const vendoredOfficialStylesDir = path.join( rootDir, 'vendor', 'styles-official
 const localDir = path.join( rootDir, '.local' );
 const logDir = path.join( localDir, 'logs' );
 const stateDir = path.join( localDir, 'state' );
+const localTranslatorSourcesDir = path.join( localDir, 'translator-sources' );
 const mergedTranslatorsDir = process.env.LOCAL_TRANSLATORS_DIR ||
 	path.join( localDir, 'translators' );
 const stylesRootDir = process.env.LOCAL_STYLES_DIR ||
 	path.join( localDir, 'styles' );
+const localStyleSourcesDir = path.join( localDir, 'style-sources' );
 const cslDir = path.join( stylesRootDir, 'csl' );
 const localeDir = path.join( stylesRootDir, 'locales' );
 const defaultStyleSources = [ vendoredOfficialStylesDir, vendoredStylesDir ];
+const defaultStyleGitSources = [
+	{
+		url: 'https://github.com/citation-style-language/styles.git',
+		dir: path.join( localStyleSourcesDir, 'styles-official' )
+	},
+	{
+		url: 'https://github.com/zotero-chinese/styles.git',
+		dir: path.join( localStyleSourcesDir, 'styles' )
+	}
+];
+const defaultTranslatorGitSources = [
+	{
+		url: 'https://github.com/zotero/translators.git',
+		dir: path.join( localTranslatorSourcesDir, 'translators-official' )
+	},
+	{
+		url: 'https://github.com/l0o0/translators_CN.git',
+		dir: path.join( localTranslatorSourcesDir, 'translators_CN' )
+	}
+];
+
+function resolvedOfficialTranslatorsDir() {
+	const fallback = defaultTranslatorGitSources[ 0 ].dir;
+	return fileExists( officialTranslatorsDir ) ? officialTranslatorsDir : fallback;
+}
+
+function resolvedCnTranslatorsDir() {
+	const fallback = defaultTranslatorGitSources[ 1 ].dir;
+	return fileExists( cnTranslatorsDir ) ? cnTranslatorsDir : fallback;
+}
 const defaultPdfFetchIntervalMs = parseInt( process.env.CITOID_LOCAL_FETCH_INTERVAL_MS || '800', 10 );
 const defaultRequestTimeoutMs = parseInt( process.env.CITOID_LOCAL_FETCH_TIMEOUT_MS || '15000', 10 );
 const defaultProbeBodyBytes = parseInt( process.env.CITOID_LOCAL_PROBE_BODY_BYTES || '1572864', 10 );
@@ -469,13 +501,15 @@ function repoHeadOrMissing( repoPath ) {
 
 function translatorsNeedSync() {
 	const stampPath = path.join( stateDir, 'translators-sync.stamp' );
+	const activeOfficialDir = resolvedOfficialTranslatorsDir();
+	const activeCnDir = resolvedCnTranslatorsDir();
 	const hasMerged = fileExists( mergedTranslatorsDir ) &&
 		fs.readdirSync( mergedTranslatorsDir ).some( ( name ) => name.endsWith( '.js' ) );
 	if ( !hasMerged ) {
 		return true;
 	}
 
-	const currentStamp = `zotero=${ repoHeadOrMissing( path.join( zoteroDir, 'modules', 'translators' ) ) };official=${ repoHeadOrMissing( officialTranslatorsDir ) };cn=${ repoHeadOrMissing( cnTranslatorsDir ) }`;
+	const currentStamp = `zotero=${ repoHeadOrMissing( path.join( zoteroDir, 'modules', 'translators' ) ) };official=${ repoHeadOrMissing( activeOfficialDir ) };cn=${ repoHeadOrMissing( activeCnDir ) }`;
 	let previousStamp = '';
 	if ( fileExists( stampPath ) ) {
 		previousStamp = fs.readFileSync( stampPath, 'utf8' ).trim();
@@ -485,7 +519,7 @@ function translatorsNeedSync() {
 
 function writeTranslatorStamp() {
 	const stampPath = path.join( stateDir, 'translators-sync.stamp' );
-	const stamp = `zotero=${ repoHeadOrMissing( path.join( zoteroDir, 'modules', 'translators' ) ) };official=${ repoHeadOrMissing( officialTranslatorsDir ) };cn=${ repoHeadOrMissing( cnTranslatorsDir ) }`;
+	const stamp = `zotero=${ repoHeadOrMissing( path.join( zoteroDir, 'modules', 'translators' ) ) };official=${ repoHeadOrMissing( resolvedOfficialTranslatorsDir() ) };cn=${ repoHeadOrMissing( resolvedCnTranslatorsDir() ) }`;
 	fs.writeFileSync( stampPath, `${ stamp }\n` );
 }
 
@@ -493,6 +527,8 @@ function syncMergedTranslators() {
 	fs.rmSync( mergedTranslatorsDir, { recursive: true, force: true } );
 	fs.mkdirSync( mergedTranslatorsDir, { recursive: true } );
 	const zoteroTranslators = path.join( zoteroDir, 'modules', 'translators' );
+	const activeOfficialDir = resolvedOfficialTranslatorsDir();
+	const activeCnDir = resolvedCnTranslatorsDir();
 	if ( fileExists( zoteroTranslators ) ) {
 		fs.readdirSync( zoteroTranslators )
 			.filter( ( name ) => name.endsWith( '.js' ) )
@@ -503,22 +539,22 @@ function syncMergedTranslators() {
 				);
 			} );
 	}
-	if ( fileExists( officialTranslatorsDir ) ) {
-		fs.readdirSync( officialTranslatorsDir )
+	if ( fileExists( activeOfficialDir ) ) {
+		fs.readdirSync( activeOfficialDir )
 			.filter( ( name ) => name.endsWith( '.js' ) )
 			.forEach( ( name ) => {
 				fs.copyFileSync(
-					path.join( officialTranslatorsDir, name ),
+					path.join( activeOfficialDir, name ),
 					path.join( mergedTranslatorsDir, name )
 				);
 			} );
 	}
-	if ( fileExists( cnTranslatorsDir ) ) {
-		fs.readdirSync( cnTranslatorsDir )
+	if ( fileExists( activeCnDir ) ) {
+		fs.readdirSync( activeCnDir )
 			.filter( ( name ) => name.endsWith( '.js' ) )
 			.forEach( ( name ) => {
 				fs.copyFileSync(
-					path.join( cnTranslatorsDir, name ),
+					path.join( activeCnDir, name ),
 					path.join( mergedTranslatorsDir, name )
 				);
 			} );
@@ -526,10 +562,39 @@ function syncMergedTranslators() {
 	writeTranslatorStamp();
 }
 
+function ensureDefaultTranslatorSources() {
+	const existingOfficial = fileExists( officialTranslatorsDir ) ? officialTranslatorsDir : null;
+	const existingCn = fileExists( cnTranslatorsDir ) ? cnTranslatorsDir : null;
+	if ( existingOfficial && existingCn ) {
+		return {
+			officialDir: existingOfficial,
+			cnDir: existingCn
+		};
+	}
+	if ( !commandExists( 'git' ) ) {
+		throw new Error(
+			'missing translator sources and git is not installed; install git or provide vendored translator directories'
+		);
+	}
+	fs.mkdirSync( localTranslatorSourcesDir, { recursive: true } );
+	defaultTranslatorGitSources.forEach( ( source ) => {
+		if ( fileExists( source.dir ) ) {
+			runCommandOrThrow( 'git', [ '-C', source.dir, 'pull', '--ff-only' ] );
+			return;
+		}
+		runCommandOrThrow( 'git', [ 'clone', '--depth', '1', source.url, source.dir ] );
+	} );
+	return {
+		officialDir: defaultTranslatorGitSources[ 0 ].dir,
+		cnDir: defaultTranslatorGitSources[ 1 ].dir
+	};
+}
+
 function bootstrapLocalEnvironment() {
 	ensureDirs();
 	const installer = resolveInstallCommand();
-	if ( !repoReady( zoteroDir ) || !repoReady( cnTranslatorsDir ) || !repoReady( officialTranslatorsDir ) ) {
+	const translatorSources = ensureDefaultTranslatorSources();
+	if ( !repoReady( zoteroDir ) || !repoReady( translatorSources.cnDir ) || !repoReady( translatorSources.officialDir ) ) {
 		throw new Error(
 			'missing vendored repos under vendor/. pull the complete repository content.'
 		);
@@ -572,6 +637,27 @@ function ensureStyleRuntime() {
 	throw new Error( 'Styles are not synced. Run: citeclaw styles sync' );
 }
 
+function ensureDefaultStyleSources() {
+	const existing = defaultStyleSources.filter( ( sourceDir ) => fileExists( sourceDir ) );
+	if ( existing.length ) {
+		return existing;
+	}
+	if ( !commandExists( 'git' ) ) {
+		throw new Error(
+			'no bundled style sources found in this package, and git is not installed; install git or pass --repo <styles-dir>'
+		);
+	}
+	fs.mkdirSync( localStyleSourcesDir, { recursive: true } );
+	defaultStyleGitSources.forEach( ( source ) => {
+		if ( fileExists( source.dir ) ) {
+			runCommandOrThrow( 'git', [ '-C', source.dir, 'pull', '--ff-only' ] );
+			return;
+		}
+		runCommandOrThrow( 'git', [ 'clone', '--depth', '1', source.url, source.dir ] );
+	} );
+	return defaultStyleGitSources.map( ( source ) => source.dir );
+}
+
 function walkFiles( dirPath ) {
 	const entries = fs.readdirSync( dirPath, { withFileTypes: true } );
 	let result = [];
@@ -1393,16 +1479,11 @@ async function syncStyles( options ) {
 				options.repo :
 				path.join( rootDir, options.repo || 'vendor/styles' )
 		] :
-		defaultStyleSources;
+		ensureDefaultStyleSources();
 	const sourceDirs = configuredSourceDirs.filter( ( sourceDir ) => fileExists( sourceDir ) );
 	if ( options.repo && !sourceDirs.length ) {
 		throw new Error( `styles source not found: ${ configuredSourceDirs.join( ', ' ) }` );
 	}
-	if ( !options.repo && !sourceDirs.length ) {
-		throw new Error(
-			'no bundled style sources found in this package; provide --repo <styles-dir> or run from a full source checkout'
-		);
-	}
 
 	let copiedCount = 0;
 	sourceDirs.forEach( ( sourceDir ) => {
@@ -1578,7 +1659,7 @@ function isHttpUrl( raw ) {
 function normalizeArxivId( raw ) {
 	const value = String( raw || '' ).trim();
 	const id = value
-		.replace( /^https?:\/\/arxiv\.org\/(?:abs|pdf)\//i, '' )
+		.replace( /^https?:\/\/arxiv\.org\/(?:abs|pdf|html)\//i, '' )
 		.replace( /\.pdf$/i, '' )
 		.replace( /^arxiv:/i, '' )
 		.trim();
@@ -5012,6 +5093,7 @@ module.exports = {
 	extractBestDoiCandidate,
 	extractPdfCandidates,
 	main,
+	normalizeArxivId,
 	normalizeDoi,
 	shouldAttemptPdfOcr
 };