jats-xml 1.1.4 → 1.1.5

package/dist/index.d.ts

@@ -1,5 +1,8 @@
 export { default as version } from './version.js';
 export { Jats } from './jats.js';
+export { recordJatsMessage } from './messages.js';
+export { knownXmlDefectRepairMessage, repairKnownXmlDefects, type AppliedXmlDefectRepair, type KnownXmlDefectRepair, type RepairKnownXmlDefectsResult, } from './repairKnownXmlDefects.js';
+export { sanitizeXmlEntities, type SanitizeXmlEntitiesResult } from './sanitizeXmlEntities.js';
 export * from './session.js';
 export * from './types.js';
 export * from './utils.js';

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,IAAI,OAAO,EAAE,MAAM,cAAc,CAAC;AAClD,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,cAAc,cAAc,CAAC;AAC7B,cAAc,YAAY,CAAC;AAC3B,cAAc,YAAY,CAAC;AAC3B,cAAc,qBAAqB,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,IAAI,OAAO,EAAE,MAAM,cAAc,CAAC;AAClD,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC;AAClD,OAAO,EACL,2BAA2B,EAC3B,qBAAqB,EACrB,KAAK,sBAAsB,EAC3B,KAAK,oBAAoB,EACzB,KAAK,2BAA2B,GACjC,MAAM,4BAA4B,CAAC;AACpC,OAAO,EAAE,mBAAmB,EAAE,KAAK,yBAAyB,EAAE,MAAM,0BAA0B,CAAC;AAC/F,cAAc,cAAc,CAAC;AAC7B,cAAc,YAAY,CAAC;AAC3B,cAAc,YAAY,CAAC;AAC3B,cAAc,qBAAqB,CAAC"}

package/dist/index.js

@@ -1,5 +1,8 @@
 export { default as version } from './version.js';
 export { Jats } from './jats.js';
+export { recordJatsMessage } from './messages.js';
+export { knownXmlDefectRepairMessage, repairKnownXmlDefects, } from './repairKnownXmlDefects.js';
+export { sanitizeXmlEntities } from './sanitizeXmlEntities.js';
 export * from './session.js';
 export * from './types.js';
 export * from './utils.js';

package/dist/jats.d.ts

@@ -3,10 +3,12 @@ import type { Element, DeclarationAttributes } from 'xml-js';
 import { type PageFrontmatter } from 'myst-frontmatter';
 import type { Front, Body, Back, SubArticle, RefList, Reference, TitleGroup, ArticleTitle, Subtitle, Permissions, PubDate, License, Abstract, ContribGroup, Contrib, Affiliation, KeywordGroup, Keyword, ArticleCategories, ArticleMeta } from 'jats-tags';
 import type { Logger } from 'myst-cli-utils';
+import type { VFile } from 'vfile';
 import { type SerializationOptions } from 'jats-utils';
 type Options = {
     log?: Logger;
     source?: string;
+    vfile?: VFile;
 };
 type WriteOptions = SerializationOptions & {
     bodyOnly?: boolean;
@@ -18,6 +20,8 @@ export declare class Jats {
     log?: Logger;
     tree: GenericParent;
     source?: string;
+    vfile?: VFile;
+    private frontmatterMessagesRecorded;
     constructor(data: string, opts?: Options);
     get frontmatter(): PageFrontmatter;
     get front(): Front | undefined;
@@ -33,7 +37,24 @@ export declare class Jats {
      * often including an "accepted" date that is a better default for project/page `date`.
      */
     get historyDates(): GenericParent[];
+    /**
+     * Prefer a fully-specified publication date (day/month/year).
+     * If publication date is incomplete (e.g. year-only), fall back through history dates.
+     */
+    private resolvePublicationDate;
+    private recordFrontmatterMessages;
+    private pickPublicationDate;
+    private static readonly PUB_HISTORY_PRIMARY_DATE_TYPES;
+    private static readonly HISTORY_PRIMARY_DATE_TYPES;
+    private jatsDateType;
+    private dateHasDay;
+    /** Last `<date>` with a `<day>` whose `date-type` is not in `excludeTypes`. */
+    private lastDateWithDay;
+    private fallbackDatePick;
+    private findHistoryDateByType;
     get historyAcceptedDate(): GenericParent | undefined;
+    get pubHistoryFallbackDate(): GenericParent | undefined;
+    get historyFallbackDate(): GenericParent | undefined;
     /**
      * Some sources store important dates under:
      * `<article-meta><pub-history><event><date date-type="...">...</date></event></pub-history>`.

package/dist/jats.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"jats.d.ts","sourceRoot":"","sources":["../src/jats.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAe,aAAa,EAAE,MAAM,aAAa,CAAC;AAI9D,OAAO,KAAK,EAAE,OAAO,EAAE,qBAAqB,EAAE,MAAM,QAAQ,CAAC;AAC7D,OAAO,EAA2B,KAAK,eAAe,EAAE,MAAM,kBAAkB,CAAC;AAIjF,OAAO,KAAK,EACV,KAAK,EACL,IAAI,EACJ,IAAI,EACJ,UAAU,EACV,OAAO,EACP,SAAS,EACT,UAAU,EACV,YAAY,EACZ,QAAQ,EACR,WAAW,EACX,OAAO,EACP,OAAO,EACP,QAAQ,EACR,YAAY,EACZ,OAAO,EACP,WAAW,EACX,YAAY,EACZ,OAAO,EACP,iBAAiB,EACjB,WAAW,EAEZ,MAAM,WAAW,CAAC;AACnB,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,gBAAgB,CAAC;AAG7C,OAAO,EAEL,KAAK,oBAAoB,EAI1B,MAAM,YAAY,CAAC;AAEpB,KAAK,OAAO,GAAG;IAAE,GAAG,CAAC,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAiBjD,KAAK,YAAY,GAAG,oBAAoB,GAAG;IACzC,QAAQ,CAAC,EAAE,OAAO,CAAC;CACpB,CAAC;AAiCF,qBAAa,IAAI;IACf,WAAW,CAAC,EAAE,qBAAqB,CAAC;IACpC,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,GAAG,EAAE,OAAO,CAAC;IACb,GAAG,CAAC,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,aAAa,CAAC;IACpB,MAAM,CAAC,EAAE,MAAM,CAAC;gBAEJ,IAAI,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,OAAO;IAoCxC,IAAI,WAAW,IAAI,eAAe,CA+EjC;IAED,IAAI,KAAK,IAAI,KAAK,GAAG,SAAS,CAE7B;IAED,IAAI,WAAW,IAAI,WAAW,GAAG,SAAS,CAEzC;IAED,IAAI,WAAW,IAAI,WAAW,GAAG,SAAS,CAEzC;IAED,IAAI,GAAG,IAAI,MAAM,GAAG,SAAS,CAE5B;IAED,IAAI,GAAG,IAAI,MAAM,GAAG,SAAS,CAE5B;IAED,IAAI,IAAI,IAAI,MAAM,GAAG,SAAS,CAE7B;IAED,IAAI,gBAAgB,IAAI,OAAO,EAAE,CAEhC;IAED,IAAI,eAAe,IAAI,OAAO,GAAG,SAAS,CAEzC;IAED;;;OAGG;IACH,IAAI,YAAY,IAAI,aAAa,EAAE,CAElC;IAED,IAAI,mBAAmB,IAAI,aAAa,GAAG,SAAS,CAOnD;IAED;;;OAGG;IACH,IAAI,eAAe,IAAI,aAAa,EAAE,CAErC;IAED,OAAO,CAAC,wBAAwB;IAUhC,IAAI,iBAAiB,IAAI,aAAa,GAAG,SAAS,CAEjD;IAED,IAAI,sBAAsB,IAAI,aAAa,GAAG,SAAS,CAEtD;IAED,IAAI,OAAO,IAAI,OAAO,GAAG,SAAS,CAEjC;IAED,IAAI,YAAY,IAAI,YAAY,GAAG,SAAS,CAE3C;IAED,yBAAyB;IACzB,IAAI,QAAQ,IAAI,OAAO,EAAE,CAExB;IAED,IAAI,aAAa,IAAI,YAAY,EAAE,CAElC;IAED,IAAI,iBAAiB,IAAI,iBAAiB,GAAG,SAAS,CAErD;IAED,IAAI,UAAU,IAAI,UAAU,GAAG,SAAS,CAEvC;IAED,IAAI,YAAY,IAAI,YAAY,GAAG,SAAS,CAE3C;IAED,IAAI,eAAe,IAAI,QAAQ,GAAG,SAAS,CAE1C;IAED,IAAI,eAAe,IAAI,QAAQ,GAAG,SAAS,CAE1C;IAED,IAAI,QAAQ,IAAI,QAAQ,GAAG,SAAS,CAEnC;IAED,IAAI,SAAS,IAAI,QAAQ,EAAE,CAE1B;IAED,IAAI,YAAY,IAAI,YAAY,GAAG,SAAS,CAE3C;IAED,IAAI,aAAa,IAAI,YAAY,EAAE,CAElC;IAED,IAAI,cAAc,IAAI,OAAO,EAAE,CAU9B;IAED,IAAI,mBAAmB,IAAI,WAAW,EAAE,CAEvC;IAED,IAAI,IAAI,IAAI,IAAI,GAAG,SAAS,CAE3B;IAED,IAAI,IAAI,IAAI,IAAI,GAAG,SAAS,CAE3B;IAED,IAAI,WAAW,IAAI,UAAU,EAAE,CAE9B;IAED,uCAAuC;IACvC,IAAI,OAAO,IAAI,OAAO,GAAG,SAAS,CAEjC;IAED,+DAA+D;IAC/D,IAAI,QAAQ,IAAI,OAAO,EAAE,CAGxB;IAED,yEAAyE;IACzE,IAAI,UAAU,IAAI,SAAS,EAAE,CAE5B;IAED,IAAI;IAeJ,SAAS,CAAC,IAAI,CAAC,EAAE,YAAY,GAAG,MAAM;CAoBvC"}
+{"version":3,"file":"jats.d.ts","sourceRoot":"","sources":["../src/jats.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAe,aAAa,EAAE,MAAM,aAAa,CAAC;AAI9D,OAAO,KAAK,EAAE,OAAO,EAAE,qBAAqB,EAAE,MAAM,QAAQ,CAAC;AAC7D,OAAO,EAA2B,KAAK,eAAe,EAAE,MAAM,kBAAkB,CAAC;AAIjF,OAAO,KAAK,EACV,KAAK,EACL,IAAI,EACJ,IAAI,EACJ,UAAU,EACV,OAAO,EACP,SAAS,EACT,UAAU,EACV,YAAY,EACZ,QAAQ,EACR,WAAW,EACX,OAAO,EACP,OAAO,EACP,QAAQ,EACR,YAAY,EACZ,OAAO,EACP,WAAW,EACX,YAAY,EACZ,OAAO,EACP,iBAAiB,EACjB,WAAW,EAEZ,MAAM,WAAW,CAAC;AACnB,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,gBAAgB,CAAC;AAE7C,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,OAAO,CAAC;AAKnC,OAAO,EAEL,KAAK,oBAAoB,EAI1B,MAAM,YAAY,CAAC;AAEpB,KAAK,OAAO,GAAG;IAAE,GAAG,CAAC,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IAAC,KAAK,CAAC,EAAE,KAAK,CAAA;CAAE,CAAC;AAiBhE,KAAK,YAAY,GAAG,oBAAoB,GAAG;IACzC,QAAQ,CAAC,EAAE,OAAO,CAAC;CACpB,CAAC;AAkCF,qBAAa,IAAI;IACf,WAAW,CAAC,EAAE,qBAAqB,CAAC;IACpC,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,GAAG,EAAE,OAAO,CAAC;IACb,GAAG,CAAC,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,aAAa,CAAC;IACpB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,KAAK,CAAC,EAAE,KAAK,CAAC;IACd,OAAO,CAAC,2BAA2B,CAAS;gBAEhC,IAAI,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,OAAO;IAiExC,IAAI,WAAW,IAAI,eAAe,CAoEjC;IAED,IAAI,KAAK,IAAI,KAAK,GAAG,SAAS,CAE7B;IAED,IAAI,WAAW,IAAI,WAAW,GAAG,SAAS,CAEzC;IAED,IAAI,WAAW,IAAI,WAAW,GAAG,SAAS,CAEzC;IAED,IAAI,GAAG,IAAI,MAAM,GAAG,SAAS,CAE5B;IAED,IAAI,GAAG,IAAI,MAAM,GAAG,SAAS,CAE5B;IAED,IAAI,IAAI,IAAI,MAAM,GAAG,SAAS,CAE7B;IAED,IAAI,gBAAgB,IAAI,OAAO,EAAE,CAEhC;IAED,IAAI,eAAe,IAAI,OAAO,GAAG,SAAS,CAEzC;IAED;;;OAGG;IACH,IAAI,YAAY,IAAI,aAAa,EAAE,CAElC;IAED;;;OAGG;IACH,OAAO,CAAC,sBAAsB;IAkB9B,OAAO,CAAC,yBAAyB;IA6BjC,OAAO,CAAC,mBAAmB;IAsB3B,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,8BAA8B,CAMnD;IAEH,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,0BAA0B,CAAmC;IAErF,OAAO,CAAC,YAAY;IAMpB,OAAO,CAAC,UAAU;IAIlB,+EAA+E;IAC/E,OAAO,CAAC,eAAe;IAWvB,OAAO,CAAC,gBAAgB;IAgBxB,OAAO,CAAC,qBAAqB;IAS7B,IAAI,mBAAmB,IAAI,aAAa,GAAG,SAAS,CAEnD;IAED,IAAI,sBAAsB,IAAI,aAAa,GAAG,SAAS,CAEtD;IAED,IAAI,mBAAmB,IAAI,aAAa,GAAG,SAAS,CAEnD;IAED;;;OAGG;IACH,IAAI,eAAe,IAAI,aAAa,EAAE,CAErC;IAED,OAAO,CAAC,wBAAwB;IAOhC,IAAI,iBAAiB,IAAI,aAAa,GAAG,SAAS,CAEjD;IAED,IAAI,sBAAsB,IAAI,aAAa,GAAG,SAAS,CAEtD;IAED,IAAI,OAAO,IAAI,OAAO,GAAG,SAAS,CAEjC;IAED,IAAI,YAAY,IAAI,YAAY,GAAG,SAAS,CAE3C;IAED,yBAAyB;IACzB,IAAI,QAAQ,IAAI,OAAO,EAAE,CAExB;IAED,IAAI,aAAa,IAAI,YAAY,EAAE,CAElC;IAED,IAAI,iBAAiB,IAAI,iBAAiB,GAAG,SAAS,CAErD;IAED,IAAI,UAAU,IAAI,UAAU,GAAG,SAAS,CAEvC;IAED,IAAI,YAAY,IAAI,YAAY,GAAG,SAAS,CAE3C;IAED,IAAI,eAAe,IAAI,QAAQ,GAAG,SAAS,CAE1C;IAED,IAAI,eAAe,IAAI,QAAQ,GAAG,SAAS,CAE1C;IAED,IAAI,QAAQ,IAAI,QAAQ,GAAG,SAAS,CAEnC;IAED,IAAI,SAAS,IAAI,QAAQ,EAAE,CAE1B;IAED,IAAI,YAAY,IAAI,YAAY,GAAG,SAAS,CAE3C;IAED,IAAI,aAAa,IAAI,YAAY,EAAE,CAElC;IAED,IAAI,cAAc,IAAI,OAAO,EAAE,CAU9B;IAED,IAAI,mBAAmB,IAAI,WAAW,EAAE,CAEvC;IAED,IAAI,IAAI,IAAI,IAAI,GAAG,SAAS,CAE3B;IAED,IAAI,IAAI,IAAI,IAAI,GAAG,SAAS,CAE3B;IAED,IAAI,WAAW,IAAI,UAAU,EAAE,CAE9B;IAED,uCAAuC;IACvC,IAAI,OAAO,IAAI,OAAO,GAAG,SAAS,CAEjC;IAED,+DAA+D;IAC/D,IAAI,QAAQ,IAAI,OAAO,EAAE,CAGxB;IAED,yEAAyE;IACzE,IAAI,UAAU,IAAI,SAAS,EAAE,CAE5B;IAED,IAAI;IAeJ,SAAS,CAAC,IAAI,CAAC,EAAE,YAAY,GAAG,MAAM;CAoBvC"}

package/dist/jats.js

@@ -6,6 +6,9 @@ import { select as unistSelect, selectAll } from 'unist-util-select';
 import { Tags } from 'jats-tags';
 import { findArticleId, processAffiliation, processContributor } from './utils.js';
 import { tic } from 'myst-cli-utils';
+import { recordJatsMessage } from './messages.js';
+import { knownXmlDefectRepairMessage, repairKnownXmlDefects } from './repairKnownXmlDefects.js';
+import { sanitizeXmlEntities } from './sanitizeXmlEntities.js';
 import { articleMetaOrder, tableWrapOrder } from './order.js';
 import { serializeJatsXml, convertToUnist, convertToXml, toDate, } from 'jats-utils';
 function select(selector, node) {
@@ -38,27 +41,45 @@ function significantChildElements(elements) {
     });
 }
 /**
- * Drop a leading processing instruction when it precedes doctype and the root element.
+ * Drop top-level processing instructions (e.g. xml-stylesheet) from the prolog.
  * Malformed prologs such as <?version xml="1.0"?> are parsed this way by xml-js.
  */
-function dropLeadingPrologInstruction(elements) {
-    if ((elements === null || elements === void 0 ? void 0 : elements.length) === 3 &&
-        elements[0].type === 'instruction' &&
-        elements[1].type === 'doctype' &&
-        elements[2].type === 'element') {
-        return elements.slice(1);
-    }
-    return elements;
+function dropTopLevelInstructions(elements, onInstruction) {
+    return elements === null || elements === void 0 ? void 0 : elements.filter((elem) => {
+        if (elem.type === 'instruction') {
+            onInstruction === null || onInstruction === void 0 ? void 0 : onInstruction(elem);
+            return false;
+        }
+        return true;
+    });
 }
 export class Jats {
     constructor(data, opts) {
-        var _a, _b;
+        var _a;
+        this.frontmatterMessagesRecorded = false;
         const toc = tic();
         this.log = opts === null || opts === void 0 ? void 0 : opts.log;
         if (opts === null || opts === void 0 ? void 0 : opts.source)
             this.source = opts.source;
+        this.vfile = opts === null || opts === void 0 ? void 0 : opts.vfile;
+        const warnProlog = (reason, note) => {
+            recordJatsMessage(this.vfile, reason, { note });
+        };
+        const { xml: afterDefectRepairs, applied: defectRepairs } = repairKnownXmlDefects(data);
+        defectRepairs.forEach(({ repair, count }) => {
+            const reason = knownXmlDefectRepairMessage(repair.from, repair.to);
+            const note = count === 1 ? undefined : `${count} occurrences`;
+            warnProlog(reason, note);
+        });
+        const { xml: parseInput, escapedBareAmpersandCount } = sanitizeXmlEntities(afterDefectRepairs);
+        if (escapedBareAmpersandCount > 0) {
+            const note = escapedBareAmpersandCount === 1
+                ? '1 bare & rewritten to &#38;'
+                : `${escapedBareAmpersandCount} bare & rewritten to &#38;`;
+            warnProlog('Escaped bare ampersand(s) before XML parse', note);
+        }
         try {
-            this.raw = xml2js(data, {
+            this.raw = xml2js(parseInput, {
                 compact: false,
                 // Preserve whitespace-only text nodes between elements. This is usually unnecessary except inside <preformat>.
                 // convertToUnist drops these outside preformat so other content is processed independent of arbitrary xml whitespace.
@@ -70,9 +91,15 @@ export class Jats {
         }
         const { declaration, elements } = this.raw;
         this.declaration = declaration === null || declaration === void 0 ? void 0 : declaration.attributes;
-        const filteredElements = dropLeadingPrologInstruction(significantChildElements(elements));
+        const filteredElements = dropTopLevelInstructions(significantChildElements(elements), (instruction) => {
+            var _a, _b;
+            const name = (_a = instruction.name) !== null && _a !== void 0 ? _a : '(unnamed)';
+            const body = String((_b = instruction.instruction) !== null && _b !== void 0 ? _b : '').trim();
+            const note = body ? `name=${name} ${body}` : `name=${name}`;
+            warnProlog('Removed top-level XML processing instruction from prolog', note);
+        });
         if ((filteredElements === null || filteredElements === void 0 ? void 0 : filteredElements.length) && filteredElements[0].type !== 'doctype') {
-            (_a = this.log) === null || _a === void 0 ? void 0 : _a.warn('JATS is missing DOCTYPE declaration');
+            warnProlog('JATS is missing DOCTYPE declaration; inserted empty doctype');
             filteredElements.unshift({ type: 'doctype' });
         }
         if (!((filteredElements === null || filteredElements === void 0 ? void 0 : filteredElements.length) === 2 &&
@@ -81,36 +108,32 @@ export class Jats {
             throw new Error('JATS must be structured as <!DOCTYPE><article>...</article>');
         }
         this.doctype = filteredElements[0].doctype;
+        if (filteredElements[1].name === 'pmc-articleset') {
+            warnProlog('JATS root is pmc-articleset wrapper', 'Using nested article element');
+        }
         const converted = convertToUnist(filteredElements[1]);
         this.tree = select('article', converted);
-        (_b = this.log) === null || _b === void 0 ? void 0 : _b.debug(toc('Parsed and converted JATS to unist tree in %s'));
+        (_a = this.log) === null || _a === void 0 ? void 0 : _a.debug(toc('Parsed and converted JATS to unist tree in %s'));
     }
     get frontmatter() {
-        var _a, _b, _c, _d, _e, _f, _g, _h, _j, _k, _l;
+        var _a, _b, _c, _d, _e, _f, _g, _h;
         const title = this.articleTitle;
         const subtitle = this.articleSubtitle;
         const short_title = this.articleAltTitle;
-        let date;
-        // Prefer a fully-specified publication date (day/month/year).
-        // If publication date is incomplete (e.g. year-only), fall back to history accepted date.
-        const preferredDateNode = (_c = (_b = (_a = this.publicationDate) !== null && _a !== void 0 ? _a : this.pubHistoryPubDate) !== null && _b !== void 0 ? _b : this.pubHistoryAcceptedDate) !== null && _c !== void 0 ? _c : this.historyAcceptedDate;
-        if (preferredDateNode) {
-            const d = toDate(preferredDateNode);
-            if (d) {
-                const year = d.getUTCFullYear();
-                const month = (d.getUTCMonth() + 1).toString().padStart(2, '0');
-                const day = d.getUTCDate().toString().padStart(2, '0');
-                date = `${year}-${month}-${day}`;
-            }
-        }
-        const authors = (_d = this.articleAuthors) === null || _d === void 0 ? void 0 : _d.map((auth) => {
-            return processContributor(auth);
+        const titleText = title ? toText(title).trim() : '';
+        const authors = ((_a = this.articleAuthors) !== null && _a !== void 0 ? _a : []).map((auth) => processContributor(auth));
+        const { date, datePick } = this.resolvePublicationDate();
+        this.recordFrontmatterMessages({
+            date,
+            datePick,
+            titleText,
+            authorCount: authors.length,
         });
-        const affiliations = (_e = this.articleAffiliations) === null || _e === void 0 ? void 0 : _e.map((aff) => {
+        const affiliations = (_b = this.articleAffiliations) === null || _b === void 0 ? void 0 : _b.map((aff) => {
             return processAffiliation(aff);
         });
-        const keywords = (_g = (_f = this.keywords) === null || _f === void 0 ? void 0 : _f.map((k) => toText(k))) !== null && _g !== void 0 ? _g : [];
-        const subjectScope = (_h = this.articleCategories) !== null && _h !== void 0 ? _h : this.front;
+        const keywords = (_d = (_c = this.keywords) === null || _c === void 0 ? void 0 : _c.map((k) => toText(k))) !== null && _d !== void 0 ? _d : [];
+        const subjectScope = (_e = this.articleCategories) !== null && _e !== void 0 ? _e : this.front;
         const journalCollSubject = select(`${Tags.subjGroup}[subj-group-type="hwp-journal-coll"] ${Tags.subject}`, subjectScope);
         const articleSubject = journalCollSubject !== null && journalCollSubject !== void 0 ? journalCollSubject : select(Tags.subject, subjectScope);
         const journalTitle = select(Tags.journalTitle, this.front);
@@ -124,13 +147,13 @@ export class Jats {
         }
         else if (selectAll('ext-link', license).length === 1) {
             // this should only happen if there is only one ext-link
-            licenseString = (_j = select('ext-link', license)['xlink:href']) !== null && _j !== void 0 ? _j : null;
+            licenseString = (_f = select('ext-link', license)['xlink:href']) !== null && _f !== void 0 ? _f : null;
         }
         else if (license) {
             licenseString = toText(license);
         }
         let openAccess;
-        const licenseType = (_k = license === null || license === void 0 ? void 0 : license['license-type']) === null || _k === void 0 ? void 0 : _k.toLowerCase();
+        const licenseType = (_g = license === null || license === void 0 ? void 0 : license['license-type']) === null || _g === void 0 ? void 0 : _g.toLowerCase();
         if (licenseType && ['openaccess', 'open-access'].includes(licenseType)) {
             openAccess = true;
         }
@@ -144,10 +167,10 @@ export class Jats {
         const pmc = this.pmc;
         const identifiers = pmc ? { pmcid: `PMC${pmc}` } : undefined;
         const frontmatter = validatePageFrontmatter({
-            title: title ? toText(title) : undefined,
+            title: titleText || undefined,
             subtitle: subtitle ? toText(subtitle) : undefined,
             short_title: short_title ? toText(short_title) : undefined,
-            doi: (_l = this.doi) !== null && _l !== void 0 ? _l : undefined,
+            doi: (_h = this.doi) !== null && _h !== void 0 ? _h : undefined,
             identifiers,
             date,
             authors: authors.length ? authors : undefined,
@@ -195,16 +218,116 @@ export class Jats {
         var _a;
         return selectAll('history date', (_a = this.articleMeta) !== null && _a !== void 0 ? _a : this.front);
     }
-    get historyAcceptedDate() {
-        const accepted = this.historyDates.find((d) => {
-            var _a;
-            const dt = String((_a = d['date-type']) !== null && _a !== void 0 ? _a : '').toLowerCase();
-            return dt === 'accepted' || dt === 'accept';
+    /**
+     * Prefer a fully-specified publication date (day/month/year).
+     * If publication date is incomplete (e.g. year-only), fall back through history dates.
+     */
+    resolvePublicationDate() {
+        const datePick = this.pickPublicationDate();
+        let date;
+        if (datePick) {
+            const d = toDate(datePick.node);
+            if (d) {
+                const year = d.getUTCFullYear();
+                const month = (d.getUTCMonth() + 1).toString().padStart(2, '0');
+                const day = d.getUTCDate().toString().padStart(2, '0');
+                date = `${year}-${month}-${day}`;
+            }
+        }
+        return { date, datePick };
+    }
+    recordFrontmatterMessages(input) {
+        var _a;
+        if (this.frontmatterMessagesRecorded)
+            return;
+        this.frontmatterMessagesRecorded = true;
+        if ((_a = input.datePick) === null || _a === void 0 ? void 0 : _a.warn) {
+            recordJatsMessage(this.vfile, input.datePick.warn, { note: input.datePick.note });
+        }
+        if (!input.date) {
+            recordJatsMessage(this.vfile, 'No publication date found in JATS', {
+                note: 'article-meta/pub-date or history/pub-history dates',
+            });
+        }
+        if (!input.titleText) {
+            recordJatsMessage(this.vfile, 'No article title found in JATS', {
+                note: 'article-meta/title-group/article-title',
+            });
+        }
+        if (input.authorCount === 0) {
+            recordJatsMessage(this.vfile, 'No authors found in JATS', {
+                note: 'article-meta/contrib-group/contrib[@contrib-type="author"]',
+            });
+        }
+    }
+    pickPublicationDate() {
+        if (this.publicationDate) {
+            return { node: this.publicationDate };
+        }
+        if (this.pubHistoryPubDate) {
+            return { node: this.pubHistoryPubDate };
+        }
+        if (this.pubHistoryAcceptedDate) {
+            return { node: this.pubHistoryAcceptedDate };
+        }
+        if (this.historyAcceptedDate) {
+            return { node: this.historyAcceptedDate };
+        }
+        if (this.pubHistoryFallbackDate) {
+            return this.fallbackDatePick(this.pubHistoryFallbackDate, 'pub-history');
+        }
+        if (this.historyFallbackDate) {
+            return this.fallbackDatePick(this.historyFallbackDate, 'history');
+        }
+        return undefined;
+    }
+    jatsDateType(node) {
+        var _a;
+        return String((_a = node['date-type']) !== null && _a !== void 0 ? _a : '')
+            .trim()
+            .toLowerCase();
+    }
+    dateHasDay(node) {
+        return !!select(Tags.day, node);
+    }
+    /** Last `<date>` with a `<day>` whose `date-type` is not in `excludeTypes`. */
+    lastDateWithDay(dates, excludeTypes) {
+        const eligible = dates.filter((d) => {
+            if (!this.dateHasDay(d))
+                return false;
+            return !excludeTypes.has(this.jatsDateType(d));
         });
-        if (accepted && select(Tags.day, accepted))
-            return accepted;
+        return eligible.at(-1);
+    }
+    fallbackDatePick(node, source) {
+        const dateType = this.jatsDateType(node) || 'unknown';
+        const xpath = source === 'pub-history'
+            ? `article-meta/pub-history/event/date[@date-type="${dateType}"]`
+            : `article-meta/history/date[@date-type="${dateType}"]`;
+        return {
+            node,
+            warn: `Using JATS ${source} date as publication date`,
+            note: xpath,
+        };
+    }
+    findHistoryDateByType(types) {
+        const wanted = new Set(types.map((t) => t.toLowerCase()));
+        const found = this.historyDates.find((d) => {
+            return wanted.has(this.jatsDateType(d));
+        });
+        if (found && this.dateHasDay(found))
+            return found;
         return undefined;
     }
+    get historyAcceptedDate() {
+        return this.findHistoryDateByType(['accepted', 'accept']);
+    }
+    get pubHistoryFallbackDate() {
+        return this.lastDateWithDay(this.pubHistoryDates, Jats.PUB_HISTORY_PRIMARY_DATE_TYPES);
+    }
+    get historyFallbackDate() {
+        return this.lastDateWithDay(this.historyDates, Jats.HISTORY_PRIMARY_DATE_TYPES);
+    }
     /**
      * Some sources store important dates under:
      * `<article-meta><pub-history><event><date date-type="...">...</date></event></pub-history>`.
@@ -215,12 +338,8 @@ export class Jats {
     }
     findPubHistoryDateByType(types) {
         const wanted = new Set(types.map((t) => t.toLowerCase()));
-        const found = this.pubHistoryDates.find((d) => {
-            var _a;
-            const dt = String((_a = d['date-type']) !== null && _a !== void 0 ? _a : '').toLowerCase();
-            return wanted.has(dt);
-        });
-        if (found && select(Tags.day, found))
+        const found = this.pubHistoryDates.find((d) => wanted.has(this.jatsDateType(d)));
+        if (found && this.dateHasDay(found))
             return found;
         return undefined;
     }
@@ -338,6 +457,14 @@ export class Jats {
         return xml;
     }
 }
+Jats.PUB_HISTORY_PRIMARY_DATE_TYPES = new Set([
+    'pub',
+    'published',
+    'publication',
+    'accepted',
+    'accept',
+]);
+Jats.HISTORY_PRIMARY_DATE_TYPES = new Set(['accepted', 'accept']);
 function hasSingleArticle(element) {
     if (element.name === 'article') {
         return true;

package/dist/messages.d.ts

@@ -0,0 +1,5 @@
+import type { VFile } from 'vfile';
+export declare function recordJatsMessage(vfile: VFile | undefined, reason: string, opts?: {
+    note?: string;
+}): void;
+//# sourceMappingURL=messages.d.ts.map

package/dist/messages.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"messages.d.ts","sourceRoot":"","sources":["../src/messages.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,OAAO,CAAC;AAEnC,wBAAgB,iBAAiB,CAC/B,KAAK,EAAE,KAAK,GAAG,SAAS,EACxB,MAAM,EAAE,MAAM,EACd,IAAI,CAAC,EAAE;IAAE,IAAI,CAAC,EAAE,MAAM,CAAA;CAAE,QAIzB"}

package/dist/messages.js

@@ -0,0 +1,6 @@
+import { fileWarn, RuleId } from 'myst-common';
+export function recordJatsMessage(vfile, reason, opts) {
+    if (!vfile)
+        return;
+    fileWarn(vfile, reason, { source: 'jats-xml', ruleId: RuleId.jatsParses, note: opts === null || opts === void 0 ? void 0 : opts.note });
+}

package/dist/repairKnownXmlDefects.d.ts

@@ -0,0 +1,23 @@
+/**
+ * Literal find/replace fixes for malformed JATS seen in the wild.
+ * Each entry emits at most one vfile message when it matches.
+ */
+export type KnownXmlDefectRepair = {
+    from: string;
+    to: string;
+};
+export declare function knownXmlDefectRepairMessage(from: string, to: string): string;
+export type AppliedXmlDefectRepair = {
+    repair: KnownXmlDefectRepair;
+    count: number;
+};
+export type RepairKnownXmlDefectsResult = {
+    xml: string;
+    applied: AppliedXmlDefectRepair[];
+};
+/**
+ * Apply KNOWN_XML_DEFECT_REPAIRS in order. Each rule uses a single literal
+ * `from` → `to` substitution (all occurrences of `from`).
+ */
+export declare function repairKnownXmlDefects(xml: string): RepairKnownXmlDefectsResult;
+//# sourceMappingURL=repairKnownXmlDefects.d.ts.map

package/dist/repairKnownXmlDefects.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"repairKnownXmlDefects.d.ts","sourceRoot":"","sources":["../src/repairKnownXmlDefects.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,MAAM,MAAM,oBAAoB,GAAG;IACjC,IAAI,EAAE,MAAM,CAAC;IACb,EAAE,EAAE,MAAM,CAAC;CACZ,CAAC;AASF,wBAAgB,2BAA2B,CAAC,IAAI,EAAE,MAAM,EAAE,EAAE,EAAE,MAAM,GAAG,MAAM,CAE5E;AAED,MAAM,MAAM,sBAAsB,GAAG;IACnC,MAAM,EAAE,oBAAoB,CAAC;IAC7B,KAAK,EAAE,MAAM,CAAC;CACf,CAAC;AAEF,MAAM,MAAM,2BAA2B,GAAG;IACxC,GAAG,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,sBAAsB,EAAE,CAAC;CACnC,CAAC;AAEF;;;GAGG;AACH,wBAAgB,qBAAqB,CAAC,GAAG,EAAE,MAAM,GAAG,2BAA2B,CAW9E"}

package/dist/repairKnownXmlDefects.js

@@ -0,0 +1,26 @@
+const KNOWN_XML_DEFECT_REPAIRS = [
+    {
+        from: '<fn id="n1"fn-type="equal">',
+        to: '<fn id="n1" fn-type="equal">',
+    },
+];
+export function knownXmlDefectRepairMessage(from, to) {
+    return `replaced ${from} with ${to}`;
+}
+/**
+ * Apply KNOWN_XML_DEFECT_REPAIRS in order. Each rule uses a single literal
+ * `from` → `to` substitution (all occurrences of `from`).
+ */
+export function repairKnownXmlDefects(xml) {
+    const applied = [];
+    let next = xml;
+    for (const repair of KNOWN_XML_DEFECT_REPAIRS) {
+        if (!next.includes(repair.from))
+            continue;
+        const parts = next.split(repair.from);
+        const count = parts.length - 1;
+        next = parts.join(repair.to);
+        applied.push({ repair, count });
+    }
+    return { xml: next, applied };
+}

package/dist/sanitizeXmlEntities.d.ts

@@ -0,0 +1,14 @@
+export type SanitizeXmlEntitiesResult = {
+    xml: string;
+    /** How many bare `&` were rewritten to `&`. */
+    escapedBareAmpersandCount: number;
+};
+/**
+ * Escape bare ampersands so xml-js (and other XML parsers) do not fail with
+ * "Invalid character in entity name" / unterminated reference errors.
+ *
+ * Existing `&`, `–`, `{`, etc. are left unchanged; xml-js expands
+ * those when parsing.
+ */
+export declare function sanitizeXmlEntities(xml: string): SanitizeXmlEntitiesResult;
+//# sourceMappingURL=sanitizeXmlEntities.d.ts.map

package/dist/sanitizeXmlEntities.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"sanitizeXmlEntities.d.ts","sourceRoot":"","sources":["../src/sanitizeXmlEntities.ts"],"names":[],"mappings":"AAmBA,MAAM,MAAM,yBAAyB,GAAG;IACtC,GAAG,EAAE,MAAM,CAAC;IACZ,mDAAmD;IACnD,yBAAyB,EAAE,MAAM,CAAC;CACnC,CAAC;AAEF;;;;;;GAMG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,yBAAyB,CAO1E"}

package/dist/sanitizeXmlEntities.js

@@ -0,0 +1,26 @@
+import { characterEntities } from 'character-entities';
+/**
+ * Names of HTML/XML character references recognized when deciding whether `&` starts
+ * a valid entity (e.g. `&`, `–`, `&`, `&`).
+ */
+const VALID_ENTITY_NAMES = Object.keys(characterEntities).map((name) => name.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'));
+/**
+ * `&` not followed by a known named entity or numeric reference — e.g. "Bill & Melinda".
+ * Matches the escape step from common JATS preprocessors before XML parse.
+ */
+const BARE_AMPERSAND_RE = new RegExp(`&(?!(?:${VALID_ENTITY_NAMES.join('|')}|#(?:x[0-9a-fA-F]+|\\d+));)`, 'g');
+/**
+ * Escape bare ampersands so xml-js (and other XML parsers) do not fail with
+ * "Invalid character in entity name" / unterminated reference errors.
+ *
+ * Existing `&`, `–`, `{`, etc. are left unchanged; xml-js expands
+ * those when parsing.
+ */
+export function sanitizeXmlEntities(xml) {
+    let escapedBareAmpersandCount = 0;
+    const sanitized = xml.replace(BARE_AMPERSAND_RE, () => {
+        escapedBareAmpersandCount += 1;
+        return '&';
+    });
+    return { xml: sanitized, escapedBareAmpersandCount };
+}

package/dist/utils.js

@@ -50,7 +50,7 @@ function markForDeletion(nodes) {
     });
 }
 export function processAffiliation(aff) {
-    var _a, _b;
+    var _a, _b, _c, _d;
     const id = aff.id;
     let ror;
     let isni;
@@ -84,7 +84,7 @@ export function processAffiliation(aff) {
         countryNode,
     ]);
     remove(aff, '__delete__');
-    const affChildren = aff.children.filter((child) => child.type !== 'label');
+    const affChildren = (_d = (_c = aff.children) === null || _c === void 0 ? void 0 : _c.filter((child) => child.type !== 'label')) !== null && _d !== void 0 ? _d : [];
     let institution;
     if (affChildren.filter((child) => ['text', 'institution-wrap', 'institution'].includes(child.type))
         .length === affChildren.length) {

package/dist/version.d.ts

@@ -1,3 +1,3 @@
-declare const version = "1.1.4";
+declare const version = "1.1.5";
 export default version;
 //# sourceMappingURL=version.d.ts.map

package/dist/version.js

@@ -1,2 +1,2 @@
-const version = '1.1.4';
+const version = '1.1.5';
 export default version;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jats-xml",
-  "version": "1.1.4",
+  "version": "1.1.5",
   "description": "Types and utilities for working with JATS in Typescript",
   "author": "Rowan Cockett <rowan@continuousfoundation.org>",
   "homepage": "https://github.com/continuous-foundation/jats",
@@ -39,15 +39,17 @@
   },
   "dependencies": {
     "adm-zip": "^0.5.10",
+    "character-entities": "^2.0.2",
     "doi-utils": "^2.0.6",
-    "jats-tags": "^1.1.4",
-    "jats-utils": "^1.1.4",
+    "jats-tags": "^1.1.5",
+    "jats-utils": "^1.1.5",
     "myst-cli-utils": "^2.0.11",
     "myst-common": "^1.7.5",
     "myst-frontmatter": "^1.7.5",
     "node-fetch": "^3.3.1",
-    "unist-util-select": "^4.0.0",
     "unist-util-remove": "4.0.0",
+    "unist-util-select": "^4.0.0",
+    "vfile": "^5.0.0",
     "which": "^3.0.1",
     "xml-js": "^1.6.11"
   },