jats-xml 0.0.15 → 0.0.16

package/dist/cjs/cli/validate.js

@@ -7,11 +7,18 @@ const session_1 = require("../session");
 const validate_1 = require("../validate");
 function makeValidateCLI(program) {
     const command = new commander_1.Command('validate')
-        .description('Fetch JATS DTD schema file from nih.gov ftp server')
+        .description(`
+      Validate JATS file against DTD schema.
+      
+      The JATS DTD schema file is fetched from nih.gov ftp server if not available locally.
+      This will attempt to infer the specific JATS DTD version, library, etc from the file header,
+      but options are available to override the inferred values.
+      `)
         .argument('<file>', 'JATS file to validate')
-        .addOption(new commander_1.Option('--jats <version>', 'JATS version, must be 1.2 or later').default('1.3'))
-        .addOption(new commander_1.Option('--mathml <version>', 'MathML version, 2 or 3').default('3'))
-        .addOption(new commander_1.Option('--oasis', 'Use OASIS table model').default(false))
+        .addOption(new commander_1.Option('--library <value>', 'JATS library - archiving, publishing, or authoring (default: archiving, if value cannot be inferred from file)'))
+        .addOption(new commander_1.Option('--jats <version>', 'JATS version, must be 1.1 or later (default: 1.3, if value cannot be inferred from file)'))
+        .addOption(new commander_1.Option('--mathml <version>', 'MathML version, 2 or 3 (default: 3, if value cannot be inferred from file)'))
+        .addOption(new commander_1.Option('--oasis', 'Use OASIS table model (default: false, if value cannot be inferred from file)'))
         .addOption(new commander_1.Option('--directory <value>', 'Directory to save DTD file'))
         .action((0, myst_cli_utils_1.clirun)(validate_1.validateJatsAgainstDtdWrapper, { program, getSession: session_1.getSession }));
     return command;

package/dist/cjs/validate/dtd.js

@@ -35,7 +35,7 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.validateJatsAgainstDtdWrapper = exports.validateJatsAgainstDtd = void 0;
+exports.validateJatsAgainstDtdWrapper = exports.validateJatsAgainstDtd = exports.inferOptions = void 0;
 const fs_1 = __importStar(require("fs"));
 const path_1 = __importDefault(require("path"));
 const node_fetch_1 = __importDefault(require("node-fetch"));
@@ -43,15 +43,54 @@ const unzipper_1 = __importDefault(require("unzipper"));
 const which_1 = require("which");
 const myst_cli_utils_1 = require("myst-cli-utils");
 const chalk_1 = __importDefault(require("chalk"));
-const JATS_VERSIONS = ['1.2', '1.2d1', '1.2d2', '1.3', '1.3d1', '1.3d2'];
+const JATS_VERSIONS = [
+    '1.1',
+    '1.1d1',
+    '1.1d2',
+    '1.1d3',
+    '1.2',
+    '1.2d1',
+    '1.2d2',
+    '1.3',
+    '1.3d1',
+    '1.3d2',
+];
 const DEFAULT_JATS_VERSION = '1.3';
 const MATHML_VERSIONS = ['2', '3'];
 const DEFAULT_MATHML_VERSION = '3';
-function validateOptions(opts) {
-    var _a;
+const JATS_LIBRARIES = ['authoring', 'publishing', 'archiving'];
+const DEFAULT_JATS_LIBRARY = 'archiving';
+/**
+ * Return static/ directory adjacent to the code
+ *
+ * This provides a standard location to cache DTD files, minimizing re-downloading.
+ */
+function defaultDirectory() {
+    return path_1.default.join(__dirname, 'static');
+}
+function warnOnOptionsMismatch(session, opts, inferredOpts) {
+    if (opts.jats && inferredOpts.jats && opts.jats !== inferredOpts.jats) {
+        session.log.warn(`Using JATS version ${opts.jats}; does not match version inferred from file ${inferredOpts.jats}`);
+    }
+    if (opts.library && inferredOpts.library && opts.library !== inferredOpts.library) {
+        session.log.warn(`Using JATS library ${opts.library}; does not match library inferred from file ${inferredOpts.library}`);
+    }
+    if (opts.mathml && inferredOpts.mathml && opts.mathml !== inferredOpts.mathml) {
+        session.log.warn(`Using MathML version ${opts.mathml}; does not match version inferred from file ${inferredOpts.mathml}`);
+    }
+    if (opts.oasis && !inferredOpts.oasis) {
+        session.log.warn('Using OASIS table model; does not match non-OASIS inferred from file');
+    }
+}
+/**
+ * Validate input value as JATS options and fill in defaults
+ */
+function validateOptions(session, opts, inferredOpts) {
+    var _a, _b, _c, _d, _e;
+    warnOnOptionsMismatch(session, opts, inferredOpts);
     let jats;
     if (!opts.jats) {
-        jats = DEFAULT_JATS_VERSION;
+        jats = (_a = inferredOpts.jats) !== null && _a !== void 0 ? _a : DEFAULT_JATS_VERSION;
     }
     else if (!JATS_VERSIONS.includes(opts.jats)) {
         throw new Error(`Invalid JATS version "${opts.jats}" - must be one of [${JATS_VERSIONS.join(', ')}]`);
@@ -61,7 +100,7 @@ function validateOptions(opts) {
     }
     let mathml;
     if (!opts.mathml) {
-        mathml = DEFAULT_MATHML_VERSION;
+        mathml = (_b = inferredOpts.mathml) !== null && _b !== void 0 ? _b : DEFAULT_MATHML_VERSION;
     }
     else if (!MATHML_VERSIONS.includes(opts.mathml)) {
         throw new Error(`Invalid MathML version "${opts.mathml}" - must be one of [${MATHML_VERSIONS.join(', ')}]`);
@@ -69,41 +108,128 @@ function validateOptions(opts) {
     else {
         mathml = opts.mathml;
     }
+    let library;
+    if (!opts.library) {
+        library = (_c = inferredOpts.library) !== null && _c !== void 0 ? _c : DEFAULT_JATS_LIBRARY;
+    }
+    else if (typeof opts.library !== 'string' ||
+        !JATS_LIBRARIES.includes(opts.library.toLowerCase())) {
+        throw new Error(`Invalid JATS library "${opts.library}" - must be one of [${JATS_LIBRARIES.join(', ')}]`);
+    }
+    else {
+        library = opts.library.toLowerCase();
+    }
+    const oasis = (_d = inferredOpts.oasis) !== null && _d !== void 0 ? _d : !!opts.oasis;
+    if (library === 'authoring' && oasis) {
+        throw new Error('JATS article authoring library cannot use OASIS table model');
+    }
     const out = {
+        library,
         jats,
         mathml,
-        oasis: !!opts.oasis,
-        directory: (_a = opts.directory) !== null && _a !== void 0 ? _a : defaultDirectory(),
+        oasis,
+        directory: (_e = opts.directory) !== null && _e !== void 0 ? _e : defaultDirectory(),
     };
     return out;
 }
+/**
+ * DTD folder name
+ */
 function dtdFolder(opts) {
     const version = opts.jats.replace('.', '-');
     const oasis = opts.oasis ? '-OASIS' : '';
     const mathml = `MathML${opts.mathml}`;
-    return `JATS-Archiving-${version}${oasis}-${mathml}-DTD`;
+    const library = opts.library.charAt(0).toUpperCase() + opts.library.slice(1);
+    return `JATS-${library}-${version}${oasis}-${mathml}-DTD`;
 }
+/**
+ * DTD zip file name on FTP server
+ */
 function dtdZipFile(opts) {
     return `${dtdFolder(opts)}.zip`;
 }
+/**
+ * Local location of DTD zip file
+ */
 function localDtdZipFile(opts) {
     return path_1.default.join(opts.directory, dtdZipFile(opts));
 }
+/**
+ * Extracted DTD file name
+ */
 function dtdFile(opts) {
-    const version = opts.jats.startsWith('1.2') ? '1' : opts.jats.replace('.', '-');
-    const article = opts.oasis ? 'archive-oasis-article' : 'archivearticle';
+    const version = opts.jats.startsWith('1.3') ? opts.jats.replace('.', '-') : '1';
+    let article;
+    if (opts.library === 'archiving') {
+        article = opts.oasis ? 'archive-oasis-article' : 'archivearticle';
+    }
+    else if (opts.library === 'publishing') {
+        article = opts.oasis ? 'journalpublishing-oasis-article' : 'journalpublishing';
+    }
+    else {
+        article = 'articleauthoring';
+    }
     const mathml = opts.mathml === '3' ? '-mathml3' : '';
     return `JATS-${article}${version}${mathml}.dtd`;
 }
+/**
+ * Local location of extracted DTD file
+ */
 function localDtdFile(opts) {
     return path_1.default.join(opts.directory, dtdFolder(opts), dtdFile(opts));
 }
+/**
+ * NIH FTP server and path for downloading JATS DTD files
+ *
+ * This is accessed by node-fetch over https.
+ */
 function ftpUrl(opts) {
-    return `https://ftp.ncbi.nih.gov/pub/jats/archiving/${opts.jats}/${dtdZipFile(opts)}`;
+    const library = opts.library === 'authoring' ? 'articleauthoring' : opts.library;
+    return `https://ftp.ncbi.nih.gov/pub/jats/${library}/${opts.jats}/${dtdZipFile(opts)}`;
 }
-function defaultDirectory() {
-    return path_1.default.join(__dirname, 'static');
+/**
+ * Create a DTS-filename-options lookup for implicitly setting options based on JATS header content
+ */
+function buildDtdFileLookup() {
+    const lookup = {};
+    JATS_VERSIONS.filter((jats) => jats === '1.2' || jats.startsWith('1.3')).forEach((jats) => {
+        MATHML_VERSIONS.forEach((mathml) => {
+            JATS_LIBRARIES.forEach((library) => {
+                (library === 'authoring' ? [false] : [true, false]).forEach((oasis) => {
+                    const opts = { jats, mathml, library, oasis };
+                    lookup[dtdFile(opts)] = opts;
+                });
+            });
+        });
+    });
+    return lookup;
+}
+/**
+ * Infer DTD options from file content
+ *
+ * This looks at DTD file name in DOCTYPE as well as dtd-version in article element
+ */
+function inferOptions(file) {
+    var _a, _b;
+    const data = fs_1.default.readFileSync(file).toString();
+    const doctype = (_a = data.match(/<!DOCTYPE [\s\S]+?">/g)) === null || _a === void 0 ? void 0 : _a[0];
+    const lookup = buildDtdFileLookup();
+    let opts = {};
+    Object.entries(lookup).forEach(([key, value]) => {
+        if (doctype === null || doctype === void 0 ? void 0 : doctype.includes(key))
+            opts = Object.assign({}, value);
+    });
+    const article = (_b = data.match(/<article [\s\S]+?>/g)) === null || _b === void 0 ? void 0 : _b[0];
+    JATS_VERSIONS.forEach((jats) => {
+        if (article === null || article === void 0 ? void 0 : article.includes(`dtd-version="${jats}"`))
+            opts.jats = jats;
+    });
+    return opts;
 }
+exports.inferOptions = inferOptions;
+/**
+ * Download DTD zip file from NIH FTP server
+ */
 function dtdDownload(session, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!fs_1.default.existsSync(opts.directory)) {
@@ -115,6 +241,9 @@ function dtdDownload(session, opts) {
         (0, myst_cli_utils_1.writeFileToFolder)(localDtdZipFile(opts), yield resp.buffer());
     });
 }
+/**
+ * Download DTD zip file from NIH FTP server if it does not yet exist
+ */
 function ensureDtdZipExists(session, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!fs_1.default.existsSync(path_1.default.join(opts.directory, dtdZipFile(opts)))) {
@@ -122,32 +251,48 @@ function ensureDtdZipExists(session, opts) {
         }
     });
 }
+/**
+ * Download and extract DTD file if it does not yet exist
+ */
 function ensureDtdExists(session, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!fs_1.default.existsSync(localDtdFile(opts))) {
             yield ensureDtdZipExists(session, opts);
             const zipFile = localDtdZipFile(opts);
-            session.log.info(`🤐 Unzipping template on disk ${zipFile}`);
+            session.log.info(`🤐 Unzipping template: ${zipFile}`);
             yield (0, fs_1.createReadStream)(zipFile)
                 .pipe(unzipper_1.default.Extract({ path: opts.directory }))
                 .promise();
         }
-        session.log.debug(`Validating against ${localDtdFile(opts)}`);
     });
 }
+/**
+ * Test if xmllint is available as a cli command
+ */
 function isXmllintAvailable() {
     return (0, which_1.sync)('xmllint', { nothrow: true });
 }
+/**
+ * Check if JATS file is valid based on JATS version/library/etc.
+ *
+ * Returns true if valid and false if invalid.
+ */
 function validateJatsAgainstDtd(session, file, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!isXmllintAvailable()) {
             session.log.error(`JATS validation against DTD requires xmllint\n\n${chalk_1.default.dim('To install:\n  mac:    brew install xmlstarlet\n  debian: apt install libxml2-utils')}`);
             return;
         }
-        const validatedOpts = validateOptions(opts !== null && opts !== void 0 ? opts : {});
+        const inferredOpts = inferOptions(file);
+        const validatedOpts = validateOptions(session, opts !== null && opts !== void 0 ? opts : {}, inferredOpts);
         yield ensureDtdExists(session, validatedOpts);
+        session.log.debug(`Validating against: ${localDtdFile(validatedOpts)}`);
+        session.log.info(`🧐 Validating against: ${dtdFolder(validatedOpts)}`);
         try {
-            yield (0, myst_cli_utils_1.makeExecutable)(`xmllint --dtdvalid ${localDtdFile(validatedOpts)} --noout ${file}`, session.log)();
+            // First drop DOCTYPE with DTD in it - we have already fetched the DTD
+            const dropDtdCommand = `xmllint --dropdtd`;
+            const validateCommand = `xmllint --noout --dtdvalid ${localDtdFile(validatedOpts)}`;
+            yield (0, myst_cli_utils_1.makeExecutable)(`${dropDtdCommand} ${file} | ${validateCommand} -`, session.log)();
         }
         catch (_a) {
             return false;
@@ -156,6 +301,11 @@ function validateJatsAgainstDtd(session, file, opts) {
     });
 }
 exports.validateJatsAgainstDtd = validateJatsAgainstDtd;
+/**
+ * Check if JATS file is valid based on JATS version/library/etc.
+ *
+ * Logs confirmation message if valid and throws an error if invalid.
+ */
 function validateJatsAgainstDtdWrapper(session, file, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         const success = yield validateJatsAgainstDtd(session, file, opts);

package/dist/cjs/version.js

@@ -1,4 +1,4 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-const version = '0.0.15';
+const version = '0.0.16';
 exports.default = version;

package/dist/esm/cli/validate.js

@@ -4,11 +4,18 @@ import { getSession } from '../session';
 import { validateJatsAgainstDtdWrapper } from '../validate';
 function makeValidateCLI(program) {
     const command = new Command('validate')
-        .description('Fetch JATS DTD schema file from nih.gov ftp server')
+        .description(`
+      Validate JATS file against DTD schema.
+      
+      The JATS DTD schema file is fetched from nih.gov ftp server if not available locally.
+      This will attempt to infer the specific JATS DTD version, library, etc from the file header,
+      but options are available to override the inferred values.
+      `)
         .argument('<file>', 'JATS file to validate')
-        .addOption(new Option('--jats <version>', 'JATS version, must be 1.2 or later').default('1.3'))
-        .addOption(new Option('--mathml <version>', 'MathML version, 2 or 3').default('3'))
-        .addOption(new Option('--oasis', 'Use OASIS table model').default(false))
+        .addOption(new Option('--library <value>', 'JATS library - archiving, publishing, or authoring (default: archiving, if value cannot be inferred from file)'))
+        .addOption(new Option('--jats <version>', 'JATS version, must be 1.1 or later (default: 1.3, if value cannot be inferred from file)'))
+        .addOption(new Option('--mathml <version>', 'MathML version, 2 or 3 (default: 3, if value cannot be inferred from file)'))
+        .addOption(new Option('--oasis', 'Use OASIS table model (default: false, if value cannot be inferred from file)'))
         .addOption(new Option('--directory <value>', 'Directory to save DTD file'))
         .action(clirun(validateJatsAgainstDtdWrapper, { program, getSession }));
     return command;

package/dist/esm/validate/dtd.js

@@ -14,15 +14,54 @@ import unzipper from 'unzipper';
 import { sync as which } from 'which';
 import { makeExecutable, writeFileToFolder } from 'myst-cli-utils';
 import chalk from 'chalk';
-const JATS_VERSIONS = ['1.2', '1.2d1', '1.2d2', '1.3', '1.3d1', '1.3d2'];
+const JATS_VERSIONS = [
+    '1.1',
+    '1.1d1',
+    '1.1d2',
+    '1.1d3',
+    '1.2',
+    '1.2d1',
+    '1.2d2',
+    '1.3',
+    '1.3d1',
+    '1.3d2',
+];
 const DEFAULT_JATS_VERSION = '1.3';
 const MATHML_VERSIONS = ['2', '3'];
 const DEFAULT_MATHML_VERSION = '3';
-function validateOptions(opts) {
-    var _a;
+const JATS_LIBRARIES = ['authoring', 'publishing', 'archiving'];
+const DEFAULT_JATS_LIBRARY = 'archiving';
+/**
+ * Return static/ directory adjacent to the code
+ *
+ * This provides a standard location to cache DTD files, minimizing re-downloading.
+ */
+function defaultDirectory() {
+    return path.join(__dirname, 'static');
+}
+function warnOnOptionsMismatch(session, opts, inferredOpts) {
+    if (opts.jats && inferredOpts.jats && opts.jats !== inferredOpts.jats) {
+        session.log.warn(`Using JATS version ${opts.jats}; does not match version inferred from file ${inferredOpts.jats}`);
+    }
+    if (opts.library && inferredOpts.library && opts.library !== inferredOpts.library) {
+        session.log.warn(`Using JATS library ${opts.library}; does not match library inferred from file ${inferredOpts.library}`);
+    }
+    if (opts.mathml && inferredOpts.mathml && opts.mathml !== inferredOpts.mathml) {
+        session.log.warn(`Using MathML version ${opts.mathml}; does not match version inferred from file ${inferredOpts.mathml}`);
+    }
+    if (opts.oasis && !inferredOpts.oasis) {
+        session.log.warn('Using OASIS table model; does not match non-OASIS inferred from file');
+    }
+}
+/**
+ * Validate input value as JATS options and fill in defaults
+ */
+function validateOptions(session, opts, inferredOpts) {
+    var _a, _b, _c, _d, _e;
+    warnOnOptionsMismatch(session, opts, inferredOpts);
     let jats;
     if (!opts.jats) {
-        jats = DEFAULT_JATS_VERSION;
+        jats = (_a = inferredOpts.jats) !== null && _a !== void 0 ? _a : DEFAULT_JATS_VERSION;
     }
     else if (!JATS_VERSIONS.includes(opts.jats)) {
         throw new Error(`Invalid JATS version "${opts.jats}" - must be one of [${JATS_VERSIONS.join(', ')}]`);
@@ -32,7 +71,7 @@ function validateOptions(opts) {
     }
     let mathml;
     if (!opts.mathml) {
-        mathml = DEFAULT_MATHML_VERSION;
+        mathml = (_b = inferredOpts.mathml) !== null && _b !== void 0 ? _b : DEFAULT_MATHML_VERSION;
     }
     else if (!MATHML_VERSIONS.includes(opts.mathml)) {
         throw new Error(`Invalid MathML version "${opts.mathml}" - must be one of [${MATHML_VERSIONS.join(', ')}]`);
@@ -40,41 +79,127 @@ function validateOptions(opts) {
     else {
         mathml = opts.mathml;
     }
+    let library;
+    if (!opts.library) {
+        library = (_c = inferredOpts.library) !== null && _c !== void 0 ? _c : DEFAULT_JATS_LIBRARY;
+    }
+    else if (typeof opts.library !== 'string' ||
+        !JATS_LIBRARIES.includes(opts.library.toLowerCase())) {
+        throw new Error(`Invalid JATS library "${opts.library}" - must be one of [${JATS_LIBRARIES.join(', ')}]`);
+    }
+    else {
+        library = opts.library.toLowerCase();
+    }
+    const oasis = (_d = inferredOpts.oasis) !== null && _d !== void 0 ? _d : !!opts.oasis;
+    if (library === 'authoring' && oasis) {
+        throw new Error('JATS article authoring library cannot use OASIS table model');
+    }
     const out = {
+        library,
         jats,
         mathml,
-        oasis: !!opts.oasis,
-        directory: (_a = opts.directory) !== null && _a !== void 0 ? _a : defaultDirectory(),
+        oasis,
+        directory: (_e = opts.directory) !== null && _e !== void 0 ? _e : defaultDirectory(),
     };
     return out;
 }
+/**
+ * DTD folder name
+ */
 function dtdFolder(opts) {
     const version = opts.jats.replace('.', '-');
     const oasis = opts.oasis ? '-OASIS' : '';
     const mathml = `MathML${opts.mathml}`;
-    return `JATS-Archiving-${version}${oasis}-${mathml}-DTD`;
+    const library = opts.library.charAt(0).toUpperCase() + opts.library.slice(1);
+    return `JATS-${library}-${version}${oasis}-${mathml}-DTD`;
 }
+/**
+ * DTD zip file name on FTP server
+ */
 function dtdZipFile(opts) {
     return `${dtdFolder(opts)}.zip`;
 }
+/**
+ * Local location of DTD zip file
+ */
 function localDtdZipFile(opts) {
     return path.join(opts.directory, dtdZipFile(opts));
 }
+/**
+ * Extracted DTD file name
+ */
 function dtdFile(opts) {
-    const version = opts.jats.startsWith('1.2') ? '1' : opts.jats.replace('.', '-');
-    const article = opts.oasis ? 'archive-oasis-article' : 'archivearticle';
+    const version = opts.jats.startsWith('1.3') ? opts.jats.replace('.', '-') : '1';
+    let article;
+    if (opts.library === 'archiving') {
+        article = opts.oasis ? 'archive-oasis-article' : 'archivearticle';
+    }
+    else if (opts.library === 'publishing') {
+        article = opts.oasis ? 'journalpublishing-oasis-article' : 'journalpublishing';
+    }
+    else {
+        article = 'articleauthoring';
+    }
     const mathml = opts.mathml === '3' ? '-mathml3' : '';
     return `JATS-${article}${version}${mathml}.dtd`;
 }
+/**
+ * Local location of extracted DTD file
+ */
 function localDtdFile(opts) {
     return path.join(opts.directory, dtdFolder(opts), dtdFile(opts));
 }
+/**
+ * NIH FTP server and path for downloading JATS DTD files
+ *
+ * This is accessed by node-fetch over https.
+ */
 function ftpUrl(opts) {
-    return `https://ftp.ncbi.nih.gov/pub/jats/archiving/${opts.jats}/${dtdZipFile(opts)}`;
+    const library = opts.library === 'authoring' ? 'articleauthoring' : opts.library;
+    return `https://ftp.ncbi.nih.gov/pub/jats/${library}/${opts.jats}/${dtdZipFile(opts)}`;
 }
-function defaultDirectory() {
-    return path.join(__dirname, 'static');
+/**
+ * Create a DTS-filename-options lookup for implicitly setting options based on JATS header content
+ */
+function buildDtdFileLookup() {
+    const lookup = {};
+    JATS_VERSIONS.filter((jats) => jats === '1.2' || jats.startsWith('1.3')).forEach((jats) => {
+        MATHML_VERSIONS.forEach((mathml) => {
+            JATS_LIBRARIES.forEach((library) => {
+                (library === 'authoring' ? [false] : [true, false]).forEach((oasis) => {
+                    const opts = { jats, mathml, library, oasis };
+                    lookup[dtdFile(opts)] = opts;
+                });
+            });
+        });
+    });
+    return lookup;
+}
+/**
+ * Infer DTD options from file content
+ *
+ * This looks at DTD file name in DOCTYPE as well as dtd-version in article element
+ */
+export function inferOptions(file) {
+    var _a, _b;
+    const data = fs.readFileSync(file).toString();
+    const doctype = (_a = data.match(/<!DOCTYPE [\s\S]+?">/g)) === null || _a === void 0 ? void 0 : _a[0];
+    const lookup = buildDtdFileLookup();
+    let opts = {};
+    Object.entries(lookup).forEach(([key, value]) => {
+        if (doctype === null || doctype === void 0 ? void 0 : doctype.includes(key))
+            opts = Object.assign({}, value);
+    });
+    const article = (_b = data.match(/<article [\s\S]+?>/g)) === null || _b === void 0 ? void 0 : _b[0];
+    JATS_VERSIONS.forEach((jats) => {
+        if (article === null || article === void 0 ? void 0 : article.includes(`dtd-version="${jats}"`))
+            opts.jats = jats;
+    });
+    return opts;
 }
+/**
+ * Download DTD zip file from NIH FTP server
+ */
 function dtdDownload(session, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!fs.existsSync(opts.directory)) {
@@ -86,6 +211,9 @@ function dtdDownload(session, opts) {
         writeFileToFolder(localDtdZipFile(opts), yield resp.buffer());
     });
 }
+/**
+ * Download DTD zip file from NIH FTP server if it does not yet exist
+ */
 function ensureDtdZipExists(session, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!fs.existsSync(path.join(opts.directory, dtdZipFile(opts)))) {
@@ -93,32 +221,48 @@ function ensureDtdZipExists(session, opts) {
         }
     });
 }
+/**
+ * Download and extract DTD file if it does not yet exist
+ */
 function ensureDtdExists(session, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!fs.existsSync(localDtdFile(opts))) {
             yield ensureDtdZipExists(session, opts);
             const zipFile = localDtdZipFile(opts);
-            session.log.info(`🤐 Unzipping template on disk ${zipFile}`);
+            session.log.info(`🤐 Unzipping template: ${zipFile}`);
             yield createReadStream(zipFile)
                 .pipe(unzipper.Extract({ path: opts.directory }))
                 .promise();
         }
-        session.log.debug(`Validating against ${localDtdFile(opts)}`);
     });
 }
+/**
+ * Test if xmllint is available as a cli command
+ */
 function isXmllintAvailable() {
     return which('xmllint', { nothrow: true });
 }
+/**
+ * Check if JATS file is valid based on JATS version/library/etc.
+ *
+ * Returns true if valid and false if invalid.
+ */
 export function validateJatsAgainstDtd(session, file, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         if (!isXmllintAvailable()) {
             session.log.error(`JATS validation against DTD requires xmllint\n\n${chalk.dim('To install:\n  mac:    brew install xmlstarlet\n  debian: apt install libxml2-utils')}`);
             return;
         }
-        const validatedOpts = validateOptions(opts !== null && opts !== void 0 ? opts : {});
+        const inferredOpts = inferOptions(file);
+        const validatedOpts = validateOptions(session, opts !== null && opts !== void 0 ? opts : {}, inferredOpts);
         yield ensureDtdExists(session, validatedOpts);
+        session.log.debug(`Validating against: ${localDtdFile(validatedOpts)}`);
+        session.log.info(`🧐 Validating against: ${dtdFolder(validatedOpts)}`);
         try {
-            yield makeExecutable(`xmllint --dtdvalid ${localDtdFile(validatedOpts)} --noout ${file}`, session.log)();
+            // First drop DOCTYPE with DTD in it - we have already fetched the DTD
+            const dropDtdCommand = `xmllint --dropdtd`;
+            const validateCommand = `xmllint --noout --dtdvalid ${localDtdFile(validatedOpts)}`;
+            yield makeExecutable(`${dropDtdCommand} ${file} | ${validateCommand} -`, session.log)();
         }
         catch (_a) {
             return false;
@@ -126,6 +270,11 @@ export function validateJatsAgainstDtd(session, file, opts) {
         return true;
     });
 }
+/**
+ * Check if JATS file is valid based on JATS version/library/etc.
+ *
+ * Logs confirmation message if valid and throws an error if invalid.
+ */
 export function validateJatsAgainstDtdWrapper(session, file, opts) {
     return __awaiter(this, void 0, void 0, function* () {
         const success = yield validateJatsAgainstDtd(session, file, opts);

package/dist/esm/version.js

@@ -1,2 +1,2 @@
-const version = '0.0.15';
+const version = '0.0.16';
 export default version;

package/dist/jats.js

@@ -26375,7 +26375,7 @@ var require_which = __commonJS({
 var import_commander3 = __toESM(require_commander());
 
 // src/version.ts
-var version = "0.0.15";
+var version = "0.0.16";
 var version_default = version;
 
 // src/cli/parse.ts
@@ -27238,15 +27238,52 @@ var import_node_fetch3 = __toESM(require_lib4());
 var import_unzipper = __toESM(require_unzip2());
 var import_which = __toESM(require_which());
 var import_chalk3 = __toESM(require_source());
-var JATS_VERSIONS = ["1.2", "1.2d1", "1.2d2", "1.3", "1.3d1", "1.3d2"];
+var JATS_VERSIONS = [
+  "1.1",
+  "1.1d1",
+  "1.1d2",
+  "1.1d3",
+  "1.2",
+  "1.2d1",
+  "1.2d2",
+  "1.3",
+  "1.3d1",
+  "1.3d2"
+];
 var DEFAULT_JATS_VERSION = "1.3";
 var MATHML_VERSIONS = ["2", "3"];
 var DEFAULT_MATHML_VERSION = "3";
-function validateOptions(opts) {
-  var _a;
+var JATS_LIBRARIES = ["authoring", "publishing", "archiving"];
+var DEFAULT_JATS_LIBRARY = "archiving";
+function defaultDirectory() {
+  return import_path4.default.join(__dirname, "static");
+}
+function warnOnOptionsMismatch(session, opts, inferredOpts) {
+  if (opts.jats && inferredOpts.jats && opts.jats !== inferredOpts.jats) {
+    session.log.warn(
+      `Using JATS version ${opts.jats}; does not match version inferred from file ${inferredOpts.jats}`
+    );
+  }
+  if (opts.library && inferredOpts.library && opts.library !== inferredOpts.library) {
+    session.log.warn(
+      `Using JATS library ${opts.library}; does not match library inferred from file ${inferredOpts.library}`
+    );
+  }
+  if (opts.mathml && inferredOpts.mathml && opts.mathml !== inferredOpts.mathml) {
+    session.log.warn(
+      `Using MathML version ${opts.mathml}; does not match version inferred from file ${inferredOpts.mathml}`
+    );
+  }
+  if (opts.oasis && !inferredOpts.oasis) {
+    session.log.warn("Using OASIS table model; does not match non-OASIS inferred from file");
+  }
+}
+function validateOptions(session, opts, inferredOpts) {
+  var _a, _b, _c, _d, _e;
+  warnOnOptionsMismatch(session, opts, inferredOpts);
   let jats;
   if (!opts.jats) {
-    jats = DEFAULT_JATS_VERSION;
+    jats = (_a = inferredOpts.jats) != null ? _a : DEFAULT_JATS_VERSION;
   } else if (!JATS_VERSIONS.includes(opts.jats)) {
     throw new Error(
       `Invalid JATS version "${opts.jats}" - must be one of [${JATS_VERSIONS.join(", ")}]`
@@ -27256,7 +27293,7 @@ function validateOptions(opts) {
   }
   let mathml;
   if (!opts.mathml) {
-    mathml = DEFAULT_MATHML_VERSION;
+    mathml = (_b = inferredOpts.mathml) != null ? _b : DEFAULT_MATHML_VERSION;
   } else if (!MATHML_VERSIONS.includes(opts.mathml)) {
     throw new Error(
       `Invalid MathML version "${opts.mathml}" - must be one of [${MATHML_VERSIONS.join(", ")}]`
@@ -27264,11 +27301,26 @@ function validateOptions(opts) {
   } else {
     mathml = opts.mathml;
   }
+  let library;
+  if (!opts.library) {
+    library = (_c = inferredOpts.library) != null ? _c : DEFAULT_JATS_LIBRARY;
+  } else if (typeof opts.library !== "string" || !JATS_LIBRARIES.includes(opts.library.toLowerCase())) {
+    throw new Error(
+      `Invalid JATS library "${opts.library}" - must be one of [${JATS_LIBRARIES.join(", ")}]`
+    );
+  } else {
+    library = opts.library.toLowerCase();
+  }
+  const oasis = (_d = inferredOpts.oasis) != null ? _d : !!opts.oasis;
+  if (library === "authoring" && oasis) {
+    throw new Error("JATS article authoring library cannot use OASIS table model");
+  }
   const out = {
+    library,
     jats,
     mathml,
-    oasis: !!opts.oasis,
-    directory: (_a = opts.directory) != null ? _a : defaultDirectory()
+    oasis,
+    directory: (_e = opts.directory) != null ? _e : defaultDirectory()
   };
   return out;
 }
@@ -27276,7 +27328,8 @@ function dtdFolder(opts) {
   const version2 = opts.jats.replace(".", "-");
   const oasis = opts.oasis ? "-OASIS" : "";
   const mathml = `MathML${opts.mathml}`;
-  return `JATS-Archiving-${version2}${oasis}-${mathml}-DTD`;
+  const library = opts.library.charAt(0).toUpperCase() + opts.library.slice(1);
+  return `JATS-${library}-${version2}${oasis}-${mathml}-DTD`;
 }
 function dtdZipFile(opts) {
   return `${dtdFolder(opts)}.zip`;
@@ -27285,8 +27338,15 @@ function localDtdZipFile(opts) {
   return import_path4.default.join(opts.directory, dtdZipFile(opts));
 }
 function dtdFile(opts) {
-  const version2 = opts.jats.startsWith("1.2") ? "1" : opts.jats.replace(".", "-");
-  const article = opts.oasis ? "archive-oasis-article" : "archivearticle";
+  const version2 = opts.jats.startsWith("1.3") ? opts.jats.replace(".", "-") : "1";
+  let article;
+  if (opts.library === "archiving") {
+    article = opts.oasis ? "archive-oasis-article" : "archivearticle";
+  } else if (opts.library === "publishing") {
+    article = opts.oasis ? "journalpublishing-oasis-article" : "journalpublishing";
+  } else {
+    article = "articleauthoring";
+  }
   const mathml = opts.mathml === "3" ? "-mathml3" : "";
   return `JATS-${article}${version2}${mathml}.dtd`;
 }
@@ -27294,10 +27354,39 @@ function localDtdFile(opts) {
   return import_path4.default.join(opts.directory, dtdFolder(opts), dtdFile(opts));
 }
 function ftpUrl(opts) {
-  return `https://ftp.ncbi.nih.gov/pub/jats/archiving/${opts.jats}/${dtdZipFile(opts)}`;
+  const library = opts.library === "authoring" ? "articleauthoring" : opts.library;
+  return `https://ftp.ncbi.nih.gov/pub/jats/${library}/${opts.jats}/${dtdZipFile(opts)}`;
 }
-function defaultDirectory() {
-  return import_path4.default.join(__dirname, "static");
+function buildDtdFileLookup() {
+  const lookup = {};
+  JATS_VERSIONS.filter((jats) => jats === "1.2" || jats.startsWith("1.3")).forEach((jats) => {
+    MATHML_VERSIONS.forEach((mathml) => {
+      JATS_LIBRARIES.forEach((library) => {
+        (library === "authoring" ? [false] : [true, false]).forEach((oasis) => {
+          const opts = { jats, mathml, library, oasis };
+          lookup[dtdFile(opts)] = opts;
+        });
+      });
+    });
+  });
+  return lookup;
+}
+function inferOptions(file) {
+  var _a, _b;
+  const data = import_fs3.default.readFileSync(file).toString();
+  const doctype = (_a = data.match(/<!DOCTYPE [\s\S]+?">/g)) == null ? void 0 : _a[0];
+  const lookup = buildDtdFileLookup();
+  let opts = {};
+  Object.entries(lookup).forEach(([key, value]) => {
+    if (doctype == null ? void 0 : doctype.includes(key))
+      opts = __spreadValues({}, value);
+  });
+  const article = (_b = data.match(/<article [\s\S]+?>/g)) == null ? void 0 : _b[0];
+  JATS_VERSIONS.forEach((jats) => {
+    if (article == null ? void 0 : article.includes(`dtd-version="${jats}"`))
+      opts.jats = jats;
+  });
+  return opts;
 }
 function dtdDownload(session, opts) {
   return __async(this, null, function* () {
@@ -27322,10 +27411,9 @@ function ensureDtdExists(session, opts) {
     if (!import_fs3.default.existsSync(localDtdFile(opts))) {
       yield ensureDtdZipExists(session, opts);
       const zipFile = localDtdZipFile(opts);
-      session.log.info(`\u{1F910} Unzipping template on disk ${zipFile}`);
+      session.log.info(`\u{1F910} Unzipping template: ${zipFile}`);
       yield (0, import_fs3.createReadStream)(zipFile).pipe(import_unzipper.default.Extract({ path: opts.directory })).promise();
     }
-    session.log.debug(`Validating against ${localDtdFile(opts)}`);
   });
 }
 function isXmllintAvailable() {
@@ -27343,13 +27431,15 @@ ${import_chalk3.default.dim(
       );
       return;
     }
-    const validatedOpts = validateOptions(opts != null ? opts : {});
+    const inferredOpts = inferOptions(file);
+    const validatedOpts = validateOptions(session, opts != null ? opts : {}, inferredOpts);
     yield ensureDtdExists(session, validatedOpts);
+    session.log.debug(`Validating against: ${localDtdFile(validatedOpts)}`);
+    session.log.info(`\u{1F9D0} Validating against: ${dtdFolder(validatedOpts)}`);
     try {
-      yield makeExecutable(
-        `xmllint --dtdvalid ${localDtdFile(validatedOpts)} --noout ${file}`,
-        session.log
-      )();
+      const dropDtdCommand = `xmllint --dropdtd`;
+      const validateCommand = `xmllint --noout --dtdvalid ${localDtdFile(validatedOpts)}`;
+      yield makeExecutable(`${dropDtdCommand} ${file} | ${validateCommand} -`, session.log)();
     } catch (e) {
       return false;
     }
@@ -27369,7 +27459,35 @@ function validateJatsAgainstDtdWrapper(session, file, opts) {
 
 // src/cli/validate.ts
 function makeValidateCLI(program2) {
-  const command = new import_commander2.Command("validate").description("Fetch JATS DTD schema file from nih.gov ftp server").argument("<file>", "JATS file to validate").addOption(new import_commander2.Option("--jats <version>", "JATS version, must be 1.2 or later").default("1.3")).addOption(new import_commander2.Option("--mathml <version>", "MathML version, 2 or 3").default("3")).addOption(new import_commander2.Option("--oasis", "Use OASIS table model").default(false)).addOption(new import_commander2.Option("--directory <value>", "Directory to save DTD file")).action(clirun(validateJatsAgainstDtdWrapper, { program: program2, getSession }));
+  const command = new import_commander2.Command("validate").description(
+    `
+      Validate JATS file against DTD schema.
+      
+      The JATS DTD schema file is fetched from nih.gov ftp server if not available locally.
+      This will attempt to infer the specific JATS DTD version, library, etc from the file header,
+      but options are available to override the inferred values.
+      `
+  ).argument("<file>", "JATS file to validate").addOption(
+    new import_commander2.Option(
+      "--library <value>",
+      "JATS library - archiving, publishing, or authoring (default: archiving, if value cannot be inferred from file)"
+    )
+  ).addOption(
+    new import_commander2.Option(
+      "--jats <version>",
+      "JATS version, must be 1.1 or later (default: 1.3, if value cannot be inferred from file)"
+    )
+  ).addOption(
+    new import_commander2.Option(
+      "--mathml <version>",
+      "MathML version, 2 or 3 (default: 3, if value cannot be inferred from file)"
+    )
+  ).addOption(
+    new import_commander2.Option(
+      "--oasis",
+      "Use OASIS table model (default: false, if value cannot be inferred from file)"
+    )
+  ).addOption(new import_commander2.Option("--directory <value>", "Directory to save DTD file")).action(clirun(validateJatsAgainstDtdWrapper, { program: program2, getSession }));
   return command;
 }
 function addValidateCLI(program2) {

package/dist/types/cli/validate.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"validate.d.ts","sourceRoot":"","sources":["../../../src/cli/validate.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAU,MAAM,WAAW,CAAC;AAiB5C,wBAAgB,cAAc,CAAC,OAAO,EAAE,OAAO,QAE9C"}
+{"version":3,"file":"validate.d.ts","sourceRoot":"","sources":["../../../src/cli/validate.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAU,MAAM,WAAW,CAAC;AA8C5C,wBAAgB,cAAc,CAAC,OAAO,EAAE,OAAO,QAE9C"}

package/dist/types/validate/dtd.d.ts

@@ -3,9 +3,26 @@ type Options = {
     jats: string;
     mathml: '2' | '3';
     oasis: boolean;
+    library: string;
     directory: string;
 };
-export declare function validateJatsAgainstDtd(session: ISession, file: string, opts?: Options): Promise<boolean | undefined>;
-export declare function validateJatsAgainstDtdWrapper(session: ISession, file: string, opts?: Options): Promise<void>;
+/**
+ * Infer DTD options from file content
+ *
+ * This looks at DTD file name in DOCTYPE as well as dtd-version in article element
+ */
+export declare function inferOptions(file: string): Partial<Options>;
+/**
+ * Check if JATS file is valid based on JATS version/library/etc.
+ *
+ * Returns true if valid and false if invalid.
+ */
+export declare function validateJatsAgainstDtd(session: ISession, file: string, opts?: Partial<Options>): Promise<boolean | undefined>;
+/**
+ * Check if JATS file is valid based on JATS version/library/etc.
+ *
+ * Logs confirmation message if valid and throws an error if invalid.
+ */
+export declare function validateJatsAgainstDtdWrapper(session: ISession, file: string, opts?: Partial<Options>): Promise<void>;
 export {};
 //# sourceMappingURL=dtd.d.ts.map

package/dist/types/validate/dtd.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"dtd.d.ts","sourceRoot":"","sources":["../../../src/validate/dtd.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,UAAU,CAAC;AASzC,KAAK,OAAO,GAAG;IACb,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,GAAG,GAAG,GAAG,CAAC;IAClB,KAAK,EAAE,OAAO,CAAC;IAEf,SAAS,EAAE,MAAM,CAAC;CACnB,CAAC;AAkGF,wBAAsB,sBAAsB,CAAC,OAAO,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,OAAO,gCAoB3F;AAED,wBAAsB,6BAA6B,CACjD,OAAO,EAAE,QAAQ,EACjB,IAAI,EAAE,MAAM,EACZ,IAAI,CAAC,EAAE,OAAO,iBAQf"}
+{"version":3,"file":"dtd.d.ts","sourceRoot":"","sources":["../../../src/validate/dtd.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,UAAU,CAAC;AAwBzC,KAAK,OAAO,GAAG;IACb,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,GAAG,GAAG,GAAG,CAAC;IAClB,KAAK,EAAE,OAAO,CAAC;IACf,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,MAAM,CAAC;CACnB,CAAC;AAgKF;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,oBAaxC;AA6CD;;;;GAIG;AACH,wBAAsB,sBAAsB,CAC1C,OAAO,EAAE,QAAQ,EACjB,IAAI,EAAE,MAAM,EACZ,IAAI,CAAC,EAAE,OAAO,CAAC,OAAO,CAAC,gCAwBxB;AAED;;;;GAIG;AACH,wBAAsB,6BAA6B,CACjD,OAAO,EAAE,QAAQ,EACjB,IAAI,EAAE,MAAM,EACZ,IAAI,CAAC,EAAE,OAAO,CAAC,OAAO,CAAC,iBAQxB"}

package/dist/types/version.d.ts

@@ -1,3 +1,3 @@
-declare const version = "0.0.15";
+declare const version = "0.0.16";
 export default version;
 //# sourceMappingURL=version.d.ts.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jats-xml",
-  "version": "0.0.15",
+  "version": "0.0.16",
   "description": "Types and utilities for working with JATS in Typescript",
   "author": "Rowan Cockett <rowan@curvenote.com>",
   "homepage": "https://github.com/curvenote/jats-xml",