@mantra-ai/core 0.1.0

package/dist/normalization/jats/utils/category.js

@@ -0,0 +1,431 @@
+import { cleanHeading } from "./strings";
+const SEC_TYPE_MAP = {
+    "supplementary-material": "supplementary",
+    supplementary: "supplementary",
+    "data-availability": "data_availability",
+    funding: "funding",
+    "financial-disclosure": "funding",
+    "coi-statement": "competing_interests",
+    conflict: "competing_interests",
+    "competing-interests": "competing_interests",
+    "author-notes": "author_contributions",
+    contrib: "author_contributions",
+    contributors: "author_contributions",
+    abbreviations: "abbreviations",
+    glossary: "glossary",
+    appendix: "appendices",
+    app: "appendices",
+};
+export function categoryFromSecType(secType) {
+    if (!secType)
+        return undefined;
+    const t = cleanHeading(secType);
+    if (!t)
+        return undefined;
+    // Exact match on standard JATS sec-type values
+    const exact = SEC_TYPE_MAP[t];
+    if (exact)
+        return exact;
+    // Body-section patterns
+    if (t.includes("results") && t.includes("discussion"))
+        return "results-discussion";
+    if (/(^|\b)(intro|background)(\b|$)/.test(t))
+        return "introduction";
+    if (/method|material|procedure|experimental|study[-\s]?design/.test(t))
+        return "methods";
+    if (/result|finding/.test(t))
+        return "results";
+    if (/discussion|interpretation|comment/.test(t))
+        return "discussion";
+    if (/conclusion|implication|outlook|summary/.test(t))
+        return "conclusions";
+    if (/limitation/.test(t))
+        return "limitations";
+    if (/ethic|ethical|approval|consent/.test(t))
+        return "ethics";
+    if (/acknowledge/.test(t))
+        return "acknowledgements";
+    // Back-matter patterns (non-standard sec-type forms)
+    if (/supplement/.test(t))
+        return "supplementary";
+    if (/data.?avail/.test(t))
+        return "data_availability";
+    if (/fund/.test(t))
+        return "funding";
+    if (/conflict|competing|coi|disclosure/.test(t))
+        return "competing_interests";
+    if (/contrib|author.?note/.test(t))
+        return "author_contributions";
+    if (/abbreviat/.test(t))
+        return "abbreviations";
+    if (/glossary/.test(t))
+        return "glossary";
+    if (/appendix|appendices/.test(t))
+        return "appendices";
+    if (/\bnotes?\b/.test(t))
+        return "notes";
+    return undefined;
+}
+export function categoryFromDirectHeading(t) {
+    const exactMap = [
+        [/^results?\s+and\s+discussion(s)?$/, "results-discussion"],
+        [/^discussion(s)?\s+and\s+results?$/, "results-discussion"],
+        [/^results?\s*&\s*discussion(s)?$/, "results-discussion"],
+        [/^discussion(s)?\s*&\s*results?$/, "results-discussion"],
+        // Introduction and equivalents
+        [/^introduction$/, "introduction"],
+        [/^background(s)?$/, "introduction"],
+        [/^overview$/, "introduction"],
+        [/^context$/, "introduction"],
+        [/^rationale$/, "introduction"],
+        [/^motivation$/, "introduction"],
+        [/^aims?$/, "introduction"],
+        [/^objectives?$/, "introduction"],
+        [/^purpose$/, "introduction"],
+        [/^scope$/, "introduction"],
+        [/^preface$/, "introduction"],
+        [/^problem\s+statement$/, "introduction"],
+        [/^related\s+work$/, "introduction"],
+        [/^literature\s+review$/, "introduction"],
+        // Methods and equivalents
+        [/^methods?$/, "methods"],
+        [/^methodology$/, "methods"],
+        [/^procedure(s)?$/, "methods"],
+        [/^steps?$/, "methods"],
+        [/^workflow(s)?$/, "methods"],
+        [/^approach(es)?$/, "methods"],
+        [/^implementation$/, "methods"],
+        [/^materials?\s*&\s*methods?$/, "methods"],
+        [/^materials?\s+and\s+methods?$/, "methods"],
+        [/^methods?\s+and\s+materials?$/, "methods"],
+        [/^methods?\s*&\s*materials?$/, "methods"],
+        [/^patients?\s+and\s+methods?$/, "methods"],
+        [/^patients?\s*&\s*methods?$/, "methods"],
+        [/^subjects?\s+and\s+methods?$/, "methods"],
+        [/^subjects?\s*&\s*methods?$/, "methods"],
+        [/^participants?\s+and\s+methods?$/, "methods"],
+        [/^participants?\s*&\s*methods?$/, "methods"],
+        [/^experimental\s+procedures?$/, "methods"],
+        [/^experimental\s+methods?$/, "methods"],
+        [/^study\s+design$/, "methods"],
+        [/^research\s+design$/, "methods"],
+        [/^experimental\s+design$/, "methods"],
+        [/^data\s+and\s+methods?$/, "methods"],
+        [/^data\s*&\s*methods?$/, "methods"],
+        [/^data\s+and\s+methodology$/, "methods"],
+        [/^data\s+collection$/, "methods"],
+        [/^data\s+acquisition$/, "methods"],
+        [/^data\s+processing$/, "methods"],
+        [/^data\s+preprocessing$/, "methods"],
+        [/^data\s+analysis$/, "methods"],
+        [/^analysis\s+methods?$/, "methods"],
+        [/^analytical\s+methods?$/, "methods"],
+        [/^experimental\s+setup$/, "methods"],
+        [/^protocols?$/, "methods"],
+        [/^technique(s)?$/, "methods"],
+        [/^statistical\s+methods?$/, "methods"],
+        [/^measurement\s+methods?$/, "methods"],
+        // Materials-only headings mapped to methods for normalization
+        [/^materials?$/, "methods"],
+        [/^experimental\s+materials?$/, "methods"],
+        [/^reagents?$/, "methods"],
+        [/^equipment$/, "methods"],
+        [/^instrumentation$/, "methods"],
+        [/^apparatus$/, "methods"],
+        [/^study\s+materials?$/, "methods"],
+        // Results and equivalents
+        [/^results?$/, "results"],
+        [/^findings?$/, "results"],
+        [/^outcomes?$/, "results"],
+        [/^observations?$/, "results"],
+        [/^experimental\s+results?$/, "results"],
+        [/^empirical\s+results?$/, "results"],
+        [/^experiments?$/, "results"],
+        [/^experimental\s+evaluation$/, "results"],
+        [/^empirical\s+evaluation$/, "results"],
+        [/^results?\s+and\s+evaluation$/, "results"],
+        [/^results?\s*&\s*evaluation$/, "results"],
+        [/^results?\s+and\s+analysis$/, "results"],
+        [/^results?\s*&\s*analysis$/, "results"],
+        [/^results?\s+analysis$/, "results"],
+        [/^evaluation$/, "results"],
+        [/^performance\s+evaluation$/, "results"],
+        [/^case\s+stud(y|ies)$/, "results"],
+        // Discussion and equivalents
+        [/^discussion(s)?$/, "discussion"],
+        [/^general\s+discussion$/, "discussion"],
+        [/^discussion\s+of\s+results?$/, "discussion"],
+        [/^discussion\s+and\s+outlook$/, "discussion"],
+        [/^discussion\s*&\s*outlook$/, "discussion"],
+        [/^discussion\s+and\s+perspectives$/, "discussion"],
+        [/^discussion\s*&\s*perspectives$/, "discussion"],
+        [/^interpretation(s)?$/, "discussion"],
+        [/^implications?$/, "discussion"],
+        [/^commentary$/, "discussion"],
+        [/^critical\s+discussion$/, "discussion"],
+        [/^discussion(s)?\s+and\s+conclusion(s)?$/, "discussion"],
+        [/^discussion(s)?\s*&\s*conclusion(s)?$/, "discussion"],
+        // Conclusions and equivalents
+        [/^conclusion(s)?$/, "conclusions"],
+        [/^summary\s+and\s+conclusion(s)?$/, "conclusions"],
+        [/^concluding\s+remarks?$/, "conclusions"],
+        [/^future\s+work$/, "conclusions"],
+        [/^summary\s+and\s+future\s+work$/, "conclusions"],
+        [/^outlook\s+and\s+future\s+work$/, "conclusions"],
+        [/^conclusion(s)?\s+and\s+future\s+work$/, "conclusions"],
+        [/^final\s+remarks?$/, "conclusions"],
+        [/^closing\s+remarks?$/, "conclusions"],
+        [/^outlook$/, "conclusions"],
+        [/^perspective(s)?$/, "conclusions"],
+        [/^summary$/, "conclusions"],
+        [/^limitations?$/, "limitations"],
+        [/^ethics(\s+statement)?$/, "ethics"],
+        [/^ethical\s+approval$/, "ethics"],
+        [/^ethical\s+considerations?$/, "ethics"],
+        [/^ethics\s+approval\s+and\s+consent\s+to\s+participate$/, "ethics"],
+        [/^informed\s+consent$/, "ethics"],
+        [/^consent\s+for\s+publication$/, "ethics"],
+        [/^acknowledg(e)?ments?$/, "acknowledgements"],
+        // Non‑English common headings (ES/PT/FR/DE/IT/PL + CJK)
+        [
+            /^introducci[óo]n$|^introduccion$|^introducao$|^einleitung$|^einfuhrung$|^einführung$|^introduzione$|^introduction$|^序論$|^序论$|^はじめに$|^導入$|^서론$|^배경$|^简介$|^背景$/,
+            "introduction",
+        ],
+        [
+            /^metodos$|^m[ée]todos$|^methoden$|^methodik$|^m[ée]thodes$|^metodolog(ia|y)$|^metodi$|^方法$|^研究方法$|^実験方法$|^實驗方法$|^재료(\s*및\s*)?방법$|^방법$/,
+            "methods",
+        ],
+        [
+            /^material(es)?\s+y\s+m[ée]todos$|^materiais\s+e\s+m[ée]todos$|^materiali\s+e\s+metodi$|^materiaux\s+et\s+m[ée]thodes$|^材料(与|及)方法$/,
+            "methods",
+        ],
+        [
+            /^ergebnisse$|^resultados$|^r[ée]sultats$|^risultati$|^wyniki$|^结果$|^結果$|^결과$/,
+            "results",
+        ],
+        [
+            /^diskussion$|^discusi[óo]n$|^discuss[aã]o$|^discussione$|^besprechung$|^讨论$|^討論$|^考察$|^논의$|^고찰$/,
+            "discussion",
+        ],
+        [
+            /^schlussfolgerung(en)?$|^conclusi[óo]n(es)?$|^conclus[aã]o(es)?$|^conclusioni$|^resumen$|^zusammenfassung$|^结论$|^結論$|^결론$|^요약$/,
+            "conclusions",
+        ],
+        [
+            /^(resultados|risultati|ergebnisse)\s+y\s+(discusi[óo]n|discussione|diskussion)$|^结果(与|及)讨论$|^結果と考察$|^결과\s*및\s*고찰$/,
+            "results-discussion",
+        ],
+        // Back-matter headings
+        // Data availability
+        [/^data\s+availability(\s+statement)?$/, "data_availability"],
+        [/^availability\s+of\s+data(\s+and\s+materials?)?$/, "data_availability"],
+        [/^data\s+sharing(\s+statement)?$/, "data_availability"],
+        [/^data\s+access(\s+statement)?$/, "data_availability"],
+        // Funding
+        [/^funding(\s+statement)?$/, "funding"],
+        [/^financial\s+(support|disclosure)$/, "funding"],
+        [/^grant\s+support$/, "funding"],
+        [/^sources?\s+of\s+funding$/, "funding"],
+        // Competing interests
+        [/^competing\s+interests?(\s+statement)?$/, "competing_interests"],
+        [/^conflicts?\s+of\s+interest(\s+statement)?$/, "competing_interests"],
+        [/^declaration\s+of\s+interests?$/, "competing_interests"],
+        [/^disclosure(\s+statement)?$/, "competing_interests"],
+        // Author contributions
+        [/^authors?('s?)?\s+contributions?$/, "author_contributions"],
+        [/^contributors?$/, "author_contributions"],
+        [
+            /^credit\s+authorship\s+contribution\s+statement$/,
+            "author_contributions",
+        ],
+        // Abbreviations
+        [/^abbreviations?$/, "abbreviations"],
+        [/^list\s+of\s+abbreviations?$/, "abbreviations"],
+        // Glossary
+        [/^glossary$/, "glossary"],
+        [/^definitions?$/, "glossary"],
+        // Supplementary
+        [/^supplementary?\s+(materials?|information|data|files?)$/, "supplementary"],
+        [/^supporting\s+information$/, "supplementary"],
+        [/^online\s+supplement(ary)?$/, "supplementary"],
+        [/^additional\s+files?$/, "supplementary"],
+        // Appendix
+        [/^appendix$/, "appendices"],
+        [/^appendices$/, "appendices"],
+        [/^annex(es)?$/, "appendices"],
+        // Publisher note
+        [/^publisher'?s?\s+note$/, "publisher_note"],
+        [/^editor'?s?\s+note$/, "publisher_note"],
+        // Provenance
+        [/^provenance(\s+and\s+peer\s+review)?$/, "provenance"],
+        // Trial registration
+        [/^trial\s+registration$/, "trial_registration"],
+        [/^clinical\s+trial\s+registration$/, "trial_registration"],
+        // Code availability
+        [/^code\s+availability(\s+statement)?$/, "code_availability"],
+        [/^software\s+availability$/, "code_availability"],
+        // Materials availability
+        [/^materials?\s+availability(\s+statement)?$/, "materials_availability"],
+        // Notes
+        [/^notes?$/, "notes"],
+        [/^footnotes?$/, "notes"],
+    ];
+    for (const [re, cat] of exactMap) {
+        if (re.test(t))
+            return cat;
+    }
+    return undefined;
+}
+export function categoryFromKeywords(t) {
+    if (t.includes("results") && t.includes("discussion"))
+        return "results-discussion";
+    if (/(^|\b)(intro|background)(\b|$)/.test(t))
+        return "introduction";
+    if (/method|material|procedure|experimental|participants?|subjects?|statistical\s+analysis|study[-\s]?design|data\s+(and\s+)?methods?/.test(t))
+        return "methods";
+    if (/result|finding|outcome/.test(t))
+        return "results";
+    if (/discussion|interpretation|commentary/.test(t))
+        return "discussion";
+    if (/conclusion|implication|outlook|summary|future\s+work|perspective/.test(t))
+        return "conclusions";
+    if (/limitation|strengths?\s+and\s+limitations?/.test(t))
+        return "limitations";
+    if (/ethic|approval|consent/.test(t))
+        return "ethics";
+    if (/acknowledge/.test(t))
+        return "acknowledgements";
+    // Back-matter keyword patterns
+    if (/(data\s*availability|availability\s*of\s*data|data\s+sharing)/.test(t))
+        return "data_availability";
+    if (/fund(ing|s)|financial\s+(support|disclosure)|grant\s+support/.test(t))
+        return "funding";
+    if (/competing\s+interests?|conflicts?\s+of\s+interest|declaration\s+of\s+interest/.test(t))
+        return "competing_interests";
+    if (/author(s)?\s+contributions?|credit\s+authorship/.test(t))
+        return "author_contributions";
+    if (/abbreviations?|nomenclature/.test(t))
+        return "abbreviations";
+    if (/glossary|definitions?/.test(t))
+        return "glossary";
+    if (/supplementary|supporting\s+information|additional\s+files?/.test(t))
+        return "supplementary";
+    if (/appendix|appendices|annex/.test(t))
+        return "appendices";
+    if (/publisher'?s?\s+note/.test(t))
+        return "publisher_note";
+    if (/provenance|peer\s+review/.test(t))
+        return "provenance";
+    if (/trial\s+registration|clinical\s+trial/.test(t))
+        return "trial_registration";
+    if (/code\s+availability|software\s+availability/.test(t))
+        return "code_availability";
+    if (/materials?\s+availability/.test(t))
+        return "materials_availability";
+    return undefined;
+}
+export function classifySection(rawTitle, secType, level, parentCategory) {
+    // Prefer strong, explicit headings when present; otherwise use sec-type as a hint.
+    const t = cleanHeading(rawTitle);
+    if (t) {
+        const direct = categoryFromDirectHeading(t);
+        if (direct)
+            return { category: direct, normalized_title: normalizedTitleFor(direct) };
+    }
+    // Use sec-type next
+    const bySecType = categoryFromSecType(secType);
+    if (bySecType) {
+        return {
+            category: bySecType,
+            normalized_title: normalizedTitleFor(bySecType),
+        };
+    }
+    if (t) {
+        const keyword = categoryFromKeywords(t);
+        if (keyword)
+            return {
+                category: keyword,
+                normalized_title: normalizedTitleFor(keyword),
+            };
+    }
+    // Inherit from parent for nested sections when uncertain
+    if (parentCategory) {
+        return {
+            category: parentCategory,
+            normalized_title: normalizedTitleFor(parentCategory),
+        };
+    }
+    // Fallback
+    return { category: "other", normalized_title: undefined };
+}
+export function normalizedTitleFor(cat) {
+    switch (cat) {
+        case "introduction":
+            return "Introduction";
+        case "methods":
+            return "Methods";
+        case "results":
+            return "Results";
+        case "discussion":
+            return "Discussion";
+        case "conclusions":
+            return "Conclusions";
+        case "results-discussion":
+            return "Results and Discussion";
+        case "limitations":
+            return "Limitations";
+        case "ethics":
+            return "Ethics";
+        case "acknowledgements":
+            return "Acknowledgements";
+        case "abstract":
+            return "Abstract";
+        case "data_availability":
+            return "Data Availability";
+        case "funding":
+            return "Funding";
+        case "competing_interests":
+            return "Competing Interests";
+        case "author_contributions":
+            return "Author Contributions";
+        case "abbreviations":
+            return "Abbreviations";
+        case "glossary":
+            return "Glossary";
+        case "supplementary":
+            return "Supplementary Material";
+        case "publisher_note":
+            return "Publisher's Note";
+        case "provenance":
+            return "Provenance and Peer Review";
+        case "trial_registration":
+            return "Trial Registration";
+        case "code_availability":
+            return "Code Availability";
+        case "materials_availability":
+            return "Materials Availability";
+        case "appendices":
+            return "Appendix";
+        case "notes":
+            return "Notes";
+        default:
+            return "Other";
+    }
+}
+export const BACK_MATTER_CATEGORIES = [
+    "acknowledgements",
+    "data_availability",
+    "funding",
+    "competing_interests",
+    "author_contributions",
+    "abbreviations",
+    "glossary",
+    "publisher_note",
+    "provenance",
+    "trial_registration",
+    "code_availability",
+    "materials_availability",
+    "appendices",
+    "notes",
+];
+//# sourceMappingURL=category.js.map

package/dist/normalization/jats/utils/category.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"category.js","sourceRoot":"","sources":["../../../../src/normalization/jats/utils/category.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AAEzC,MAAM,YAAY,GAAoC;IACpD,wBAAwB,EAAE,eAAe;IACzC,aAAa,EAAE,eAAe;IAC9B,mBAAmB,EAAE,mBAAmB;IACxC,OAAO,EAAE,SAAS;IAClB,sBAAsB,EAAE,SAAS;IACjC,eAAe,EAAE,qBAAqB;IACtC,QAAQ,EAAE,qBAAqB;IAC/B,qBAAqB,EAAE,qBAAqB;IAC5C,cAAc,EAAE,sBAAsB;IACtC,OAAO,EAAE,sBAAsB;IAC/B,YAAY,EAAE,sBAAsB;IACpC,aAAa,EAAE,eAAe;IAC9B,QAAQ,EAAE,UAAU;IACpB,QAAQ,EAAE,YAAY;IACtB,GAAG,EAAE,YAAY;CAClB,CAAC;AAEF,MAAM,UAAU,mBAAmB,CACjC,OAAgB;IAEhB,IAAI,CAAC,OAAO;QAAE,OAAO,SAAS,CAAC;IAC/B,MAAM,CAAC,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;IAChC,IAAI,CAAC,CAAC;QAAE,OAAO,SAAS,CAAC;IACzB,+CAA+C;IAC/C,MAAM,KAAK,GAAG,YAAY,CAAC,CAAC,CAAC,CAAC;IAC9B,IAAI,KAAK;QAAE,OAAO,KAAK,CAAC;IACxB,wBAAwB;IACxB,IAAI,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,YAAY,CAAC;QACnD,OAAO,oBAAoB,CAAC;IAC9B,IAAI,gCAAgC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,cAAc,CAAC;IACpE,IAAI,0DAA0D,CAAC,IAAI,CAAC,CAAC,CAAC;QACpE,OAAO,SAAS,CAAC;IACnB,IAAI,gBAAgB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,SAAS,CAAC;IAC/C,IAAI,mCAAmC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,YAAY,CAAC;IACrE,IAAI,wCAAwC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,aAAa,CAAC;IAC3E,IAAI,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,aAAa,CAAC;IAC/C,IAAI,gCAAgC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,QAAQ,CAAC;IAC9D,IAAI,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,kBAAkB,CAAC;IACrD,qDAAqD;IACrD,IAAI,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,eAAe,CAAC;IACjD,IAAI,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,mBAAmB,CAAC;IACtD,IAAI,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,SAAS,CAAC;IACrC,IAAI,mCAAmC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,qBAAqB,CAAC;IAC9E,IAAI,sBAAsB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,sBAAsB,CAAC;IAClE,IAAI,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,eAAe,CAAC;IAChD,IAAI,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,UAAU,CAAC;IAC1C,IAAI,qBAAqB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,YAAY,CAAC;IACvD,IAAI,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,OAAO,CAAC;IACzC,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,yBAAyB,CACvC,CAAS;IAET,MAAM,QAAQ,GAAqC;QACjD,CAAC,mCAAmC,EAAE,oBAAoB,CAAC;QAC3D,CAAC,mCAAmC,EAAE,oBAAoB,CAAC;QAC3D,CAAC,iCAAiC,EAAE,oBAAoB,CAAC;QACzD,CAAC,iCAAiC,EAAE,oBAAoB,CAAC;QACzD,+BAA+B;QAC/B,CAAC,gBAAgB,EAAE,cAAc,CAAC;QAClC,CAAC,kBAAkB,EAAE,cAAc,CAAC;QACpC,CAAC,YAAY,EAAE,cAAc,CAAC;QAC9B,CAAC,WAAW,EAAE,cAAc,CAAC;QAC7B,CAAC,aAAa,EAAE,cAAc,CAAC;QAC/B,CAAC,cAAc,EAAE,cAAc,CAAC;QAChC,CAAC,SAAS,EAAE,cAAc,CAAC;QAC3B,CAAC,eAAe,EAAE,cAAc,CAAC;QACjC,CAAC,WAAW,EAAE,cAAc,CAAC;QAC7B,CAAC,SAAS,EAAE,cAAc,CAAC;QAC3B,CAAC,WAAW,EAAE,cAAc,CAAC;QAC7B,CAAC,uBAAuB,EAAE,cAAc,CAAC;QACzC,CAAC,kBAAkB,EAAE,cAAc,CAAC;QACpC,CAAC,uBAAuB,EAAE,cAAc,CAAC;QACzC,0BAA0B;QAC1B,CAAC,YAAY,EAAE,SAAS,CAAC;QACzB,CAAC,eAAe,EAAE,SAAS,CAAC;QAC5B,CAAC,iBAAiB,EAAE,SAAS,CAAC;QAC9B,CAAC,UAAU,EAAE,SAAS,CAAC;QACvB,CAAC,gBAAgB,EAAE,SAAS,CAAC;QAC7B,CAAC,iBAAiB,EAAE,SAAS,CAAC;QAC9B,CAAC,kBAAkB,EAAE,SAAS,CAAC;QAC/B,CAAC,6BAA6B,EAAE,SAAS,CAAC;QAC1C,CAAC,+BAA+B,EAAE,SAAS,CAAC;QAC5C,CAAC,+BAA+B,EAAE,SAAS,CAAC;QAC5C,CAAC,6BAA6B,EAAE,SAAS,CAAC;QAC1C,CAAC,8BAA8B,EAAE,SAAS,CAAC;QAC3C,CAAC,4BAA4B,EAAE,SAAS,CAAC;QACzC,CAAC,8BAA8B,EAAE,SAAS,CAAC;QAC3C,CAAC,4BAA4B,EAAE,SAAS,CAAC;QACzC,CAAC,kCAAkC,EAAE,SAAS,CAAC;QAC/C,CAAC,gCAAgC,EAAE,SAAS,CAAC;QAC7C,CAAC,8BAA8B,EAAE,SAAS,CAAC;QAC3C,CAAC,2BAA2B,EAAE,SAAS,CAAC;QACxC,CAAC,kBAAkB,EAAE,SAAS,CAAC;QAC/B,CAAC,qBAAqB,EAAE,SAAS,CAAC;QAClC,CAAC,yBAAyB,EAAE,SAAS,CAAC;QACtC,CAAC,yBAAyB,EAAE,SAAS,CAAC;QACtC,CAAC,uBAAuB,EAAE,SAAS,CAAC;QACpC,CAAC,4BAA4B,EAAE,SAAS,CAAC;QACzC,CAAC,qBAAqB,EAAE,SAAS,CAAC;QAClC,CAAC,sBAAsB,EAAE,SAAS,CAAC;QACnC,CAAC,qBAAqB,EAAE,SAAS,CAAC;QAClC,CAAC,wBAAwB,EAAE,SAAS,CAAC;QACrC,CAAC,mBAAmB,EAAE,SAAS,CAAC;QAChC,CAAC,uBAAuB,EAAE,SAAS,CAAC;QACpC,CAAC,yBAAyB,EAAE,SAAS,CAAC;QACtC,CAAC,wBAAwB,EAAE,SAAS,CAAC;QACrC,CAAC,cAAc,EAAE,SAAS,CAAC;QAC3B,CAAC,iBAAiB,EAAE,SAAS,CAAC;QAC9B,CAAC,0BAA0B,EAAE,SAAS,CAAC;QACvC,CAAC,0BAA0B,EAAE,SAAS,CAAC;QACvC,8DAA8D;QAC9D,CAAC,cAAc,EAAE,SAAS,CAAC;QAC3B,CAAC,6BAA6B,EAAE,SAAS,CAAC;QAC1C,CAAC,aAAa,EAAE,SAAS,CAAC;QAC1B,CAAC,aAAa,EAAE,SAAS,CAAC;QAC1B,CAAC,mBAAmB,EAAE,SAAS,CAAC;QAChC,CAAC,aAAa,EAAE,SAAS,CAAC;QAC1B,CAAC,sBAAsB,EAAE,SAAS,CAAC;QACnC,0BAA0B;QAC1B,CAAC,YAAY,EAAE,SAAS,CAAC;QACzB,CAAC,aAAa,EAAE,SAAS,CAAC;QAC1B,CAAC,aAAa,EAAE,SAAS,CAAC;QAC1B,CAAC,iBAAiB,EAAE,SAAS,CAAC;QAC9B,CAAC,2BAA2B,EAAE,SAAS,CAAC;QACxC,CAAC,wBAAwB,EAAE,SAAS,CAAC;QACrC,CAAC,gBAAgB,EAAE,SAAS,CAAC;QAC7B,CAAC,6BAA6B,EAAE,SAAS,CAAC;QAC1C,CAAC,0BAA0B,EAAE,SAAS,CAAC;QACvC,CAAC,+BAA+B,EAAE,SAAS,CAAC;QAC5C,CAAC,6BAA6B,EAAE,SAAS,CAAC;QAC1C,CAAC,6BAA6B,EAAE,SAAS,CAAC;QAC1C,CAAC,2BAA2B,EAAE,SAAS,CAAC;QACxC,CAAC,uBAAuB,EAAE,SAAS,CAAC;QACpC,CAAC,cAAc,EAAE,SAAS,CAAC;QAC3B,CAAC,4BAA4B,EAAE,SAAS,CAAC;QACzC,CAAC,sBAAsB,EAAE,SAAS,CAAC;QACnC,6BAA6B;QAC7B,CAAC,kBAAkB,EAAE,YAAY,CAAC;QAClC,CAAC,wBAAwB,EAAE,YAAY,CAAC;QACxC,CAAC,8BAA8B,EAAE,YAAY,CAAC;QAC9C,CAAC,8BAA8B,EAAE,YAAY,CAAC;QAC9C,CAAC,4BAA4B,EAAE,YAAY,CAAC;QAC5C,CAAC,mCAAmC,EAAE,YAAY,CAAC;QACnD,CAAC,iCAAiC,EAAE,YAAY,CAAC;QACjD,CAAC,sBAAsB,EAAE,YAAY,CAAC;QACtC,CAAC,iBAAiB,EAAE,YAAY,CAAC;QACjC,CAAC,cAAc,EAAE,YAAY,CAAC;QAC9B,CAAC,yBAAyB,EAAE,YAAY,CAAC;QACzC,CAAC,yCAAyC,EAAE,YAAY,CAAC;QACzD,CAAC,uCAAuC,EAAE,YAAY,CAAC;QACvD,8BAA8B;QAC9B,CAAC,kBAAkB,EAAE,aAAa,CAAC;QACnC,CAAC,kCAAkC,EAAE,aAAa,CAAC;QACnD,CAAC,yBAAyB,EAAE,aAAa,CAAC;QAC1C,CAAC,iBAAiB,EAAE,aAAa,CAAC;QAClC,CAAC,iCAAiC,EAAE,aAAa,CAAC;QAClD,CAAC,iCAAiC,EAAE,aAAa,CAAC;QAClD,CAAC,wCAAwC,EAAE,aAAa,CAAC;QACzD,CAAC,oBAAoB,EAAE,aAAa,CAAC;QACrC,CAAC,sBAAsB,EAAE,aAAa,CAAC;QACvC,CAAC,WAAW,EAAE,aAAa,CAAC;QAC5B,CAAC,mBAAmB,EAAE,aAAa,CAAC;QACpC,CAAC,WAAW,EAAE,aAAa,CAAC;QAC5B,CAAC,gBAAgB,EAAE,aAAa,CAAC;QACjC,CAAC,yBAAyB,EAAE,QAAQ,CAAC;QACrC,CAAC,sBAAsB,EAAE,QAAQ,CAAC;QAClC,CAAC,6BAA6B,EAAE,QAAQ,CAAC;QACzC,CAAC,wDAAwD,EAAE,QAAQ,CAAC;QACpE,CAAC,sBAAsB,EAAE,QAAQ,CAAC;QAClC,CAAC,+BAA+B,EAAE,QAAQ,CAAC;QAC3C,CAAC,wBAAwB,EAAE,kBAAkB,CAAC;QAC9C,wDAAwD;QACxD;YACE,8JAA8J;YAC9J,cAAc;SACf;QACD;YACE,uIAAuI;YACvI,SAAS;SACV;QACD;YACE,mIAAmI;YACnI,SAAS;SACV;QACD;YACE,8EAA8E;YAC9E,SAAS;SACV;QACD;YACE,iGAAiG;YACjG,YAAY;SACb;QACD;YACE,gIAAgI;YAChI,aAAa;SACd;QACD;YACE,mHAAmH;YACnH,oBAAoB;SACrB;QACD,uBAAuB;QACvB,oBAAoB;QACpB,CAAC,sCAAsC,EAAE,mBAAmB,CAAC;QAC7D,CAAC,kDAAkD,EAAE,mBAAmB,CAAC;QACzE,CAAC,iCAAiC,EAAE,mBAAmB,CAAC;QACxD,CAAC,gCAAgC,EAAE,mBAAmB,CAAC;QACvD,UAAU;QACV,CAAC,0BAA0B,EAAE,SAAS,CAAC;QACvC,CAAC,oCAAoC,EAAE,SAAS,CAAC;QACjD,CAAC,mBAAmB,EAAE,SAAS,CAAC;QAChC,CAAC,2BAA2B,EAAE,SAAS,CAAC;QACxC,sBAAsB;QACtB,CAAC,yCAAyC,EAAE,qBAAqB,CAAC;QAClE,CAAC,6CAA6C,EAAE,qBAAqB,CAAC;QACtE,CAAC,iCAAiC,EAAE,qBAAqB,CAAC;QAC1D,CAAC,6BAA6B,EAAE,qBAAqB,CAAC;QACtD,uBAAuB;QACvB,CAAC,mCAAmC,EAAE,sBAAsB,CAAC;QAC7D,CAAC,iBAAiB,EAAE,sBAAsB,CAAC;QAC3C;YACE,kDAAkD;YAClD,sBAAsB;SACvB;QACD,gBAAgB;QAChB,CAAC,kBAAkB,EAAE,eAAe,CAAC;QACrC,CAAC,8BAA8B,EAAE,eAAe,CAAC;QACjD,WAAW;QACX,CAAC,YAAY,EAAE,UAAU,CAAC;QAC1B,CAAC,gBAAgB,EAAE,UAAU,CAAC;QAC9B,gBAAgB;QAChB,CAAC,yDAAyD,EAAE,eAAe,CAAC;QAC5E,CAAC,4BAA4B,EAAE,eAAe,CAAC;QAC/C,CAAC,6BAA6B,EAAE,eAAe,CAAC;QAChD,CAAC,uBAAuB,EAAE,eAAe,CAAC;QAC1C,WAAW;QACX,CAAC,YAAY,EAAE,YAAY,CAAC;QAC5B,CAAC,cAAc,EAAE,YAAY,CAAC;QAC9B,CAAC,cAAc,EAAE,YAAY,CAAC;QAC9B,iBAAiB;QACjB,CAAC,wBAAwB,EAAE,gBAAgB,CAAC;QAC5C,CAAC,qBAAqB,EAAE,gBAAgB,CAAC;QACzC,aAAa;QACb,CAAC,uCAAuC,EAAE,YAAY,CAAC;QACvD,qBAAqB;QACrB,CAAC,wBAAwB,EAAE,oBAAoB,CAAC;QAChD,CAAC,mCAAmC,EAAE,oBAAoB,CAAC;QAC3D,oBAAoB;QACpB,CAAC,sCAAsC,EAAE,mBAAmB,CAAC;QAC7D,CAAC,2BAA2B,EAAE,mBAAmB,CAAC;QAClD,yBAAyB;QACzB,CAAC,4CAA4C,EAAE,wBAAwB,CAAC;QACxE,QAAQ;QACR,CAAC,UAAU,EAAE,OAAO,CAAC;QACrB,CAAC,cAAc,EAAE,OAAO,CAAC;KAC1B,CAAC;IACF,KAAK,MAAM,CAAC,EAAE,EAAE,GAAG,CAAC,IAAI,QAAQ,EAAE,CAAC;QACjC,IAAI,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;YAAE,OAAO,GAAG,CAAC;IAC7B,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,oBAAoB,CAAC,CAAS;IAC5C,IAAI,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,YAAY,CAAC;QACnD,OAAO,oBAAoB,CAAC;IAC9B,IAAI,gCAAgC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,cAAc,CAAC;IACpE,IACE,kIAAkI,CAAC,IAAI,CACrI,CAAC,CACF;QAED,OAAO,SAAS,CAAC;IACnB,IAAI,wBAAwB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,SAAS,CAAC;IACvD,IAAI,sCAAsC,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,YAAY,CAAC;IACxE,IACE,kEAAkE,CAAC,IAAI,CAAC,CAAC,CAAC;QAE1E,OAAO,aAAa,CAAC;IACvB,IAAI,4CAA4C,CAAC,IAAI,CAAC,CAAC,CAAC;QACtD,OAAO,aAAa,CAAC;IACvB,IAAI,wBAAwB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,QAAQ,CAAC;IACtD,IAAI,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,kBAAkB,CAAC;IACrD,+BAA+B;IAC/B,IAAI,+DAA+D,CAAC,IAAI,CAAC,CAAC,CAAC;QACzE,OAAO,mBAAmB,CAAC;IAC7B,IAAI,8DAA8D,CAAC,IAAI,CAAC,CAAC,CAAC;QACxE,OAAO,SAAS,CAAC;IACnB,IACE,+EAA+E,CAAC,IAAI,CAClF,CAAC,CACF;QAED,OAAO,qBAAqB,CAAC;IAC/B,IAAI,iDAAiD,CAAC,IAAI,CAAC,CAAC,CAAC;QAC3D,OAAO,sBAAsB,CAAC;IAChC,IAAI,6BAA6B,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,eAAe,CAAC;IAClE,IAAI,uBAAuB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,UAAU,CAAC;IACvD,IAAI,4DAA4D,CAAC,IAAI,CAAC,CAAC,CAAC;QACtE,OAAO,eAAe,CAAC;IACzB,IAAI,2BAA2B,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,YAAY,CAAC;IAC7D,IAAI,sBAAsB,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,gBAAgB,CAAC;IAC5D,IAAI,0BAA0B,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,YAAY,CAAC;IAC5D,IAAI,uCAAuC,CAAC,IAAI,CAAC,CAAC,CAAC;QACjD,OAAO,oBAAoB,CAAC;IAC9B,IAAI,6CAA6C,CAAC,IAAI,CAAC,CAAC,CAAC;QACvD,OAAO,mBAAmB,CAAC;IAC7B,IAAI,2BAA2B,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,wBAAwB,CAAC;IACzE,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,eAAe,CAC7B,QAA4B,EAC5B,OAA2B,EAC3B,KAAa,EACb,cAA2C;IAE3C,mFAAmF;IACnF,MAAM,CAAC,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;IACjC,IAAI,CAAC,EAAE,CAAC;QACN,MAAM,MAAM,GAAG,yBAAyB,CAAC,CAAC,CAAC,CAAC;QAC5C,IAAI,MAAM;YACR,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,gBAAgB,EAAE,kBAAkB,CAAC,MAAM,CAAC,EAAE,CAAC;IAC9E,CAAC;IAED,oBAAoB;IACpB,MAAM,SAAS,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAC;IAC/C,IAAI,SAAS,EAAE,CAAC;QACd,OAAO;YACL,QAAQ,EAAE,SAAS;YACnB,gBAAgB,EAAE,kBAAkB,CAAC,SAAS,CAAC;SAChD,CAAC;IACJ,CAAC;IAED,IAAI,CAAC,EAAE,CAAC;QACN,MAAM,OAAO,GAAG,oBAAoB,CAAC,CAAC,CAAC,CAAC;QACxC,IAAI,OAAO;YACT,OAAO;gBACL,QAAQ,EAAE,OAAO;gBACjB,gBAAgB,EAAE,kBAAkB,CAAC,OAAO,CAAC;aAC9C,CAAC;IACN,CAAC;IAED,yDAAyD;IACzD,IAAI,cAAc,EAAE,CAAC;QACnB,OAAO;YACL,QAAQ,EAAE,cAAc;YACxB,gBAAgB,EAAE,kBAAkB,CAAC,cAAc,CAAC;SACrD,CAAC;IACJ,CAAC;IAED,WAAW;IACX,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,gBAAgB,EAAE,SAAS,EAAE,CAAC;AAC5D,CAAC;AAED,MAAM,UAAU,kBAAkB,CAAC,GAAoB;IACrD,QAAQ,GAAG,EAAE,CAAC;QACZ,KAAK,cAAc;YACjB,OAAO,cAAc,CAAC;QACxB,KAAK,SAAS;YACZ,OAAO,SAAS,CAAC;QACnB,KAAK,SAAS;YACZ,OAAO,SAAS,CAAC;QACnB,KAAK,YAAY;YACf,OAAO,YAAY,CAAC;QACtB,KAAK,aAAa;YAChB,OAAO,aAAa,CAAC;QACvB,KAAK,oBAAoB;YACvB,OAAO,wBAAwB,CAAC;QAClC,KAAK,aAAa;YAChB,OAAO,aAAa,CAAC;QACvB,KAAK,QAAQ;YACX,OAAO,QAAQ,CAAC;QAClB,KAAK,kBAAkB;YACrB,OAAO,kBAAkB,CAAC;QAC5B,KAAK,UAAU;YACb,OAAO,UAAU,CAAC;QACpB,KAAK,mBAAmB;YACtB,OAAO,mBAAmB,CAAC;QAC7B,KAAK,SAAS;YACZ,OAAO,SAAS,CAAC;QACnB,KAAK,qBAAqB;YACxB,OAAO,qBAAqB,CAAC;QAC/B,KAAK,sBAAsB;YACzB,OAAO,sBAAsB,CAAC;QAChC,KAAK,eAAe;YAClB,OAAO,eAAe,CAAC;QACzB,KAAK,UAAU;YACb,OAAO,UAAU,CAAC;QACpB,KAAK,eAAe;YAClB,OAAO,wBAAwB,CAAC;QAClC,KAAK,gBAAgB;YACnB,OAAO,kBAAkB,CAAC;QAC5B,KAAK,YAAY;YACf,OAAO,4BAA4B,CAAC;QACtC,KAAK,oBAAoB;YACvB,OAAO,oBAAoB,CAAC;QAC9B,KAAK,mBAAmB;YACtB,OAAO,mBAAmB,CAAC;QAC7B,KAAK,wBAAwB;YAC3B,OAAO,wBAAwB,CAAC;QAClC,KAAK,YAAY;YACf,OAAO,UAAU,CAAC;QACpB,KAAK,OAAO;YACV,OAAO,OAAO,CAAC;QACjB;YACE,OAAO,OAAO,CAAC;IACnB,CAAC;AACH,CAAC;AAED,MAAM,CAAC,MAAM,sBAAsB,GAAsB;IACvD,kBAAkB;IAClB,mBAAmB;IACnB,SAAS;IACT,qBAAqB;IACrB,sBAAsB;IACtB,eAAe;IACf,UAAU;IACV,gBAAgB;IAChB,YAAY;IACZ,oBAAoB;IACpB,mBAAmB;IACnB,wBAAwB;IACxB,YAAY;IACZ,OAAO;CACR,CAAC"}

package/dist/normalization/jats/utils/collectors/abstracts.d.ts

@@ -0,0 +1,3 @@
+import type { NormalizedJats } from "../types";
+export declare function collectAbstracts(articleMeta: any, domPO: any): NonNullable<NormalizedJats["abstracts"]>;
+//# sourceMappingURL=abstracts.d.ts.map

package/dist/normalization/jats/utils/collectors/abstracts.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"abstracts.d.ts","sourceRoot":"","sources":["../../../../../src/normalization/jats/utils/collectors/abstracts.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,cAAc,EAKf,MAAM,UAAU,CAAC;AAelB,wBAAgB,gBAAgB,CAC9B,WAAW,EAAE,GAAG,EAChB,KAAK,EAAE,GAAG,GACT,WAAW,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CA2J1C"}

package/dist/normalization/jats/utils/collectors/abstracts.js

@@ -0,0 +1,168 @@
+import { normalizeArray, asString } from "../strings";
+import { getText } from "../xml";
+import { poFindFirstByLocal, poFindDescendantsByLocal } from "../po/query";
+import { poChildren, poIsTextNode, poTagName, poMatchesLocal, } from "../po/nodes";
+import { poText } from "../po/text";
+import { poSerialize, poSerializeMinimal } from "../po/serialize";
+import { nsArray } from "../xml";
+import { collectSectionNode } from "./sections";
+export function collectAbstracts(articleMeta, domPO) {
+    const abs = normalizeArray(articleMeta?.abstract);
+    const out = [];
+    // Try to align with preserveOrder <abstract> nodes by index order
+    const frontPO = poFindFirstByLocal(domPO, "front");
+    const amPO = frontPO
+        ? poFindFirstByLocal(frontPO, "article-meta")
+        : poFindFirstByLocal(domPO, "article-meta");
+    const absPOList = amPO
+        ? poFindDescendantsByLocal(amPO, "abstract")
+        : poFindDescendantsByLocal(domPO, "abstract");
+    for (const a of abs) {
+        const type = asString(a?.["@_abstract-type"]) || undefined;
+        const title = getText(a?.title) || undefined;
+        const pNodesObj = normalizeArray(a?.p);
+        const absSecsObj = nsArray(a, "sec");
+        // Build blocks preserving interleaving using preserveOrder when available
+        let blocks = [];
+        if (absPOList.length) {
+            const idx = out.length; // current abstract index
+            const absPO = absPOList[idx];
+            if (absPO) {
+                let poSecIdx = 0;
+                for (const ch of poChildren(absPO)) {
+                    if (poIsTextNode(ch)) {
+                        const t = (poText(ch) || "").trim();
+                        if (t)
+                            blocks.push({ kind: "paragraph", text: t });
+                        continue;
+                    }
+                    const nm = poTagName(ch) || "";
+                    if (poMatchesLocal(nm, "title") || poMatchesLocal(nm, "label"))
+                        continue;
+                    if (poMatchesLocal(nm, "sec")) {
+                        const os = absSecsObj[poSecIdx];
+                        const sec = os
+                            ? collectSectionNode(os, 1, undefined, domPO, absPO, poSecIdx)
+                            : { paragraphs: [] };
+                        blocks.push({ kind: "section", section: sec });
+                        poSecIdx += 1;
+                        continue;
+                    }
+                    if (poMatchesLocal(nm, "p")) {
+                        const t = (poText(poChildren(ch)) || poText(ch) || "").trim();
+                        if (!t)
+                            continue;
+                        const x = poSerialize(ch);
+                        const xm = poSerializeMinimal(ch);
+                        blocks.push({
+                            kind: "paragraph",
+                            text: t,
+                            ...(x && x.trim() ? { xml: x } : {}),
+                            ...(xm && xm.trim() ? { xml_minimal: xm } : {}),
+                        });
+                        continue;
+                    }
+                    // Optionally handle lists/boxed-text/def-list similarly to sections
+                    if (poMatchesLocal(nm, "list")) {
+                        for (const it of poFindDescendantsByLocal(ch, "list-item")) {
+                            const p = poFindFirstByLocal(it, "p");
+                            const t = ((p && (poText(poChildren(p)) || poText(p))) ||
+                                poText(it) ||
+                                "").trim();
+                            if (!t)
+                                continue;
+                            const x = p ? poSerialize(p) : undefined;
+                            const xm = p ? poSerializeMinimal(p) : undefined;
+                            blocks.push({
+                                kind: "paragraph",
+                                text: t,
+                                ...(x ? { xml: x } : {}),
+                                ...(xm ? { xml_minimal: xm } : {}),
+                            });
+                        }
+                        continue;
+                    }
+                    if (poMatchesLocal(nm, "boxed-text")) {
+                        for (const p of poFindDescendantsByLocal(ch, "p")) {
+                            const t = (poText(poChildren(p)) || poText(p) || "").trim();
+                            if (!t)
+                                continue;
+                            const x = poSerialize(p);
+                            const xm = poSerializeMinimal(p);
+                            blocks.push({
+                                kind: "paragraph",
+                                text: t,
+                                ...(x ? { xml: x } : {}),
+                                ...(xm ? { xml_minimal: xm } : {}),
+                            });
+                        }
+                        continue;
+                    }
+                    if (poMatchesLocal(nm, "def-list")) {
+                        for (const di of poFindDescendantsByLocal(ch, "def-item")) {
+                            const term = poFindFirstByLocal(di, "term");
+                            const defn = poFindFirstByLocal(di, "def");
+                            const line = [
+                                poText(poChildren(term)) || poText(term),
+                                poText(poChildren(defn)) || poText(defn),
+                            ]
+                                .filter(Boolean)
+                                .map((s) => String(s).trim())
+                                .filter(Boolean)
+                                .join(" – ");
+                            if (line)
+                                blocks.push({ kind: "paragraph", text: line });
+                        }
+                        continue;
+                    }
+                }
+            }
+        }
+        // Fallback: object-mode blocks when PO is not available
+        if (!blocks.length) {
+            for (const p of pNodesObj) {
+                const t = getText(p);
+                if (t && t.trim())
+                    blocks.push({ kind: "paragraph", text: t.trim() });
+            }
+            for (const s of absSecsObj) {
+                const sec = collectSectionNode(s, 1, undefined, domPO);
+                blocks.push({ kind: "section", section: sec });
+            }
+        }
+        // Derive legacy arrays from blocks
+        const paragraphs = blocks
+            .filter((b) => b.kind === "paragraph")
+            .map((b) => b.text);
+        const paragraphs_xml = blocks
+            .filter((b) => b.kind === "paragraph" && b.xml)
+            .map((b) => b.xml);
+        const paragraphs_xml_minimal = blocks
+            .filter((b) => b.kind === "paragraph" && b.xml_minimal)
+            .map((b) => b.xml_minimal);
+        const absSecs = blocks
+            .filter((b) => b.kind === "section")
+            .map((b) => b.section);
+        // Some abstracts may be plain text without explicit p; keep a final fallback
+        if (!paragraphs.length && !absSecs.length) {
+            const text = getText(a);
+            if (text && text.trim()) {
+                blocks.push({ kind: "paragraph", text: text.trim() });
+                paragraphs.push(text.trim());
+            }
+        }
+        const item = { type, title, paragraphs };
+        if (paragraphs_xml.length)
+            item.paragraphs_xml = paragraphs_xml;
+        if (paragraphs_xml_minimal.length)
+            item.paragraphs_xml_minimal = paragraphs_xml_minimal;
+        if (absSecs.length)
+            item.sections = absSecs;
+        if (blocks.length)
+            item.blocks = blocks;
+        if (paragraphs.length || absSecs.length)
+            out.push(item);
+    }
+    return out;
+}
+//# sourceMappingURL=abstracts.js.map

package/dist/normalization/jats/utils/collectors/abstracts.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"abstracts.js","sourceRoot":"","sources":["../../../../../src/normalization/jats/utils/collectors/abstracts.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,cAAc,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAC;AACtD,OAAO,EAAE,OAAO,EAAE,MAAM,QAAQ,CAAC;AACjC,OAAO,EAAE,kBAAkB,EAAE,wBAAwB,EAAE,MAAM,aAAa,CAAC;AAC3E,OAAO,EACL,UAAU,EACV,YAAY,EACZ,SAAS,EACT,cAAc,GACf,MAAM,aAAa,CAAC;AACrB,OAAO,EAAE,MAAM,EAAE,MAAM,YAAY,CAAC;AACpC,OAAO,EAAE,WAAW,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AAClE,OAAO,EAAE,OAAO,EAAE,MAAM,QAAQ,CAAC;AACjC,OAAO,EAAE,kBAAkB,EAAE,MAAM,YAAY,CAAC;AAEhD,MAAM,UAAU,gBAAgB,CAC9B,WAAgB,EAChB,KAAU;IAEV,MAAM,GAAG,GAAG,cAAc,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAC;IAClD,MAAM,GAAG,GAA6C,EAAE,CAAC;IACzD,kEAAkE;IAClE,MAAM,OAAO,GAAG,kBAAkB,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACnD,MAAM,IAAI,GAAG,OAAO;QAClB,CAAC,CAAC,kBAAkB,CAAC,OAAO,EAAE,cAAc,CAAC;QAC7C,CAAC,CAAC,kBAAkB,CAAC,KAAK,EAAE,cAAc,CAAC,CAAC;IAC9C,MAAM,SAAS,GAAG,IAAI;QACpB,CAAC,CAAC,wBAAwB,CAAC,IAAI,EAAE,UAAU,CAAC;QAC5C,CAAC,CAAC,wBAAwB,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;IAChD,KAAK,MAAM,CAAC,IAAI,GAAG,EAAE,CAAC;QACpB,MAAM,IAAI,GAAG,QAAQ,CAAC,CAAC,EAAE,CAAC,iBAAiB,CAAC,CAAC,IAAI,SAAS,CAAC;QAC3D,MAAM,KAAK,GAAG,OAAO,CAAC,CAAC,EAAE,KAAK,CAAC,IAAI,SAAS,CAAC;QAC7C,MAAM,SAAS,GAAG,cAAc,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;QACvC,MAAM,UAAU,GAAG,OAAO,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;QACrC,0EAA0E;QAC1E,IAAI,MAAM,GAAuD,EAAE,CAAC;QACpE,IAAI,SAAS,CAAC,MAAM,EAAE,CAAC;YACrB,MAAM,GAAG,GAAG,GAAG,CAAC,MAAM,CAAC,CAAC,yBAAyB;YACjD,MAAM,KAAK,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC;YAC7B,IAAI,KAAK,EAAE,CAAC;gBACV,IAAI,QAAQ,GAAG,CAAC,CAAC;gBACjB,KAAK,MAAM,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;oBACnC,IAAI,YAAY,CAAC,EAAE,CAAC,EAAE,CAAC;wBACrB,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;wBACpC,IAAI,CAAC;4BAAE,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC;wBACnD,SAAS;oBACX,CAAC;oBACD,MAAM,EAAE,GAAG,SAAS,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC;oBAC/B,IAAI,cAAc,CAAC,EAAE,EAAE,OAAO,CAAC,IAAI,cAAc,CAAC,EAAE,EAAE,OAAO,CAAC;wBAC5D,SAAS;oBACX,IAAI,cAAc,CAAC,EAAE,EAAE,KAAK,CAAC,EAAE,CAAC;wBAC9B,MAAM,EAAE,GAAG,UAAU,CAAC,QAAQ,CAAC,CAAC;wBAChC,MAAM,GAAG,GAAG,EAAE;4BACZ,CAAC,CAAC,kBAAkB,CAAC,EAAE,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,QAAQ,CAAC;4BAC9D,CAAC,CAAE,EAAE,UAAU,EAAE,EAAE,EAAc,CAAC;wBACpC,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,GAAG,EAAE,CAAC,CAAC;wBAC/C,QAAQ,IAAI,CAAC,CAAC;wBACd,SAAS;oBACX,CAAC;oBACD,IAAI,cAAc,CAAC,EAAE,EAAE,GAAG,CAAC,EAAE,CAAC;wBAC5B,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC,IAAI,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;wBAC9D,IAAI,CAAC,CAAC;4BAAE,SAAS;wBACjB,MAAM,CAAC,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC;wBAC1B,MAAM,EAAE,GAAG,kBAAkB,CAAC,EAAE,CAAC,CAAC;wBAClC,MAAM,CAAC,IAAI,CAAC;4BACV,IAAI,EAAE,WAAW;4BACjB,IAAI,EAAE,CAAC;4BACP,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;4BACpC,GAAG,CAAC,EAAE,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;yBAChD,CAAC,CAAC;wBACH,SAAS;oBACX,CAAC;oBACD,oEAAoE;oBACpE,IAAI,cAAc,CAAC,EAAE,EAAE,MAAM,CAAC,EAAE,CAAC;wBAC/B,KAAK,MAAM,EAAE,IAAI,wBAAwB,CAAC,EAAE,EAAE,WAAW,CAAC,EAAE,CAAC;4BAC3D,MAAM,CAAC,GAAG,kBAAkB,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC;4BACtC,MAAM,CAAC,GAAG,CACR,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;gCAC3C,MAAM,CAAC,EAAE,CAAC;gCACV,EAAE,CACH,CAAC,IAAI,EAAE,CAAC;4BACT,IAAI,CAAC,CAAC;gCAAE,SAAS;4BACjB,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;4BACzC,MAAM,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;4BACjD,MAAM,CAAC,IAAI,CAAC;gCACV,IAAI,EAAE,WAAW;gCACjB,IAAI,EAAE,CAAC;gCACP,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gCACxB,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;6BACnC,CAAC,CAAC;wBACL,CAAC;wBACD,SAAS;oBACX,CAAC;oBACD,IAAI,cAAc,CAAC,EAAE,EAAE,YAAY,CAAC,EAAE,CAAC;wBACrC,KAAK,MAAM,CAAC,IAAI,wBAAwB,CAAC,EAAE,EAAE,GAAG,CAAC,EAAE,CAAC;4BAClD,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;4BAC5D,IAAI,CAAC,CAAC;gCAAE,SAAS;4BACjB,MAAM,CAAC,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC;4BACzB,MAAM,EAAE,GAAG,kBAAkB,CAAC,CAAC,CAAC,CAAC;4BACjC,MAAM,CAAC,IAAI,CAAC;gCACV,IAAI,EAAE,WAAW;gCACjB,IAAI,EAAE,CAAC;gCACP,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gCACxB,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;6BACnC,CAAC,CAAC;wBACL,CAAC;wBACD,SAAS;oBACX,CAAC;oBACD,IAAI,cAAc,CAAC,EAAE,EAAE,UAAU,CAAC,EAAE,CAAC;wBACnC,KAAK,MAAM,EAAE,IAAI,wBAAwB,CAAC,EAAE,EAAE,UAAU,CAAC,EAAE,CAAC;4BAC1D,MAAM,IAAI,GAAG,kBAAkB,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;4BAC5C,MAAM,IAAI,GAAG,kBAAkB,CAAC,EAAE,EAAE,KAAK,CAAC,CAAC;4BAC3C,MAAM,IAAI,GAAG;gCACX,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,IAAI,MAAM,CAAC,IAAI,CAAC;gCACxC,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,IAAI,MAAM,CAAC,IAAI,CAAC;6BACzC;iCACE,MAAM,CAAC,OAAO,CAAC;iCACf,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;iCAC5B,MAAM,CAAC,OAAO,CAAC;iCACf,IAAI,CAAC,KAAK,CAAC,CAAC;4BACf,IAAI,IAAI;gCAAE,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC;wBAC3D,CAAC;wBACD,SAAS;oBACX,CAAC;gBACH,CAAC;YACH,CAAC;QACH,CAAC;QACD,wDAAwD;QACxD,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;YACnB,KAAK,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;gBAC1B,MAAM,CAAC,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;gBACrB,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE;oBAAE,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;YACxE,CAAC;YACD,KAAK,MAAM,CAAC,IAAI,UAAU,EAAE,CAAC;gBAC3B,MAAM,GAAG,GAAG,kBAAkB,CAAC,CAAC,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,CAAC,CAAC;gBACvD,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,GAAG,EAAE,CAAC,CAAC;YACjD,CAAC;QACH,CAAC;QAED,mCAAmC;QACnC,MAAM,UAAU,GAAG,MAAM;aACtB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,CAAC;aACrC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAE,CAAoB,CAAC,IAAI,CAAC,CAAC;QAC1C,MAAM,cAAc,GAAG,MAAM;aAC1B,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,IAAK,CAAoB,CAAC,GAAG,CAAC;aAClE,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAE,CAAoB,CAAC,GAAI,CAAa,CAAC;QACtD,MAAM,sBAAsB,GAAG,MAAM;aAClC,MAAM,CACL,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,IAAK,CAAoB,CAAC,WAAW,CACnE;aACA,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAE,CAAoB,CAAC,WAAY,CAAa,CAAC;QAC9D,MAAM,OAAO,GAAG,MAAM;aACnB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,SAAS,CAAC;aACnC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAE,CAAkB,CAAC,OAAO,CAAC,CAAC;QAE3C,6EAA6E;QAC7E,IAAI,CAAC,UAAU,CAAC,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;YAC1C,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;YACxB,IAAI,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;gBACxB,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;gBACtD,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;YAC/B,CAAC;QACH,CAAC;QAED,MAAM,IAAI,GAAQ,EAAE,IAAI,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC;QAC9C,IAAI,cAAc,CAAC,MAAM;YAAE,IAAI,CAAC,cAAc,GAAG,cAAc,CAAC;QAChE,IAAI,sBAAsB,CAAC,MAAM;YAC/B,IAAI,CAAC,sBAAsB,GAAG,sBAAsB,CAAC;QACvD,IAAI,OAAO,CAAC,MAAM;YAAE,IAAI,CAAC,QAAQ,GAAG,OAAO,CAAC;QAC5C,IAAI,MAAM,CAAC,MAAM;YAAE,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACxC,IAAI,UAAU,CAAC,MAAM,IAAI,OAAO,CAAC,MAAM;YAAE,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC1D,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC"}