@cyanheads/pubmed-mcp-server 2.10.19 → 2.10.21

package/dist/services/ncbi/parsing/pmc-xml-helpers.js

@@ -45,6 +45,16 @@ export function textOf(node) {
 const TEX_MATH_TAG = 'tex-math';
 /** JATS container holding equivalent renderings of one object. */
 const ALTERNATIVES_TAG = 'alternatives';
+/**
+ * Elements whose text identifies the element around them and is never prose:
+ * the `<institution-id>` (a ROR URL, a GRID or ISNI id, a Crossref Funder DOI)
+ * an `<institution-wrap>` carries beside the `<institution>` name. Read as text
+ * it fuses onto that name, in an affiliation
+ * (`1https://ror.org/034t3zs45grid.454711.2School of …`) and in a funding
+ * statement (`sponsored by the 10.13039/501100001809National Natural Science
+ * Foundation of China`) alike, so no reader takes its text. (#196, #208)
+ */
+export const IDENTIFIER_TAGS = new Set(['institution-id']);
 /**
  * JATS pointers to an external rendering: a file reference, not a rendering of
  * the object itself.
@@ -77,6 +87,166 @@ const TEX_DOCUMENT_BODY = /\\begin\{document\}([\s\S]*?)(?:\\end\{document\}|$)/
 function texMathExpression(raw) {
     return TEX_DOCUMENT_BODY.exec(raw)?.[1] ?? raw;
 }
+/** An element's tag without its namespace prefix: `math` for `mml:math`. */
+const localName = (tag) => tag.slice(tag.indexOf(':') + 1);
+/**
+ * True for a MathML `<math>` element, matched by local name whatever its
+ * namespace prefix — `<mml:math>` in PMC deposits, `<math>` elsewhere. (#207)
+ */
+export function isMathTag(tag) {
+    return localName(tag) === 'math';
+}
+/** MathML token elements, which contribute their text as deposited. */
+const MATH_TOKEN_TAGS = new Set(['mi', 'mn', 'mo', 'mtext', 'ms']);
+/**
+ * A MathML `<math>` element in a TeX-style linear form — the notation a
+ * formula deposited as `<tex-math>` already reads in — so a formula keeps the
+ * structure that carries its meaning instead of reading as the run of its
+ * leaves (`∑j=1Nlij`, a matrix's numeric cells fused into `−4.401.20`). (#207)
+ *
+ * | MathML | Linear form |
+ * |:--|:--|
+ * | `mi`, `mn`, `mo`, `mtext`, `ms` | their text as deposited (`−` stays U+2212) |
+ * | `msub`, `munder` / `msup`, `mover` | base`_{…}` / base`^{…}` |
+ * | `msubsup`, `munderover`, `mmultiscripts` | base`_{…}^{…}`, per script pair; `mmultiscripts` prescripts as `{}_{…}^{…}` before the base |
+ * | `mfrac` / `msqrt` / `mroot` | `\frac{…}{…}` / `\sqrt{…}` / `\sqrt[index]{…}` |
+ * | `mfenced` | `open`, the children split by `separators`, `close` (defaults `(`, `,`, `)`) |
+ * | `mtable` | cells split by ` & `, rows by ` \\ ` |
+ * | `mspace` / `mphantom` | one space / nothing |
+ * | anything else | its children in order |
+ *
+ * An empty script slot is left out (`x^{2}`, not `x_{}^{2}`). Whitespace-only
+ * text between elements contributes nothing; any other text reads verbatim.
+ * Elements match by local name, whatever their prefix. Pieces are written to
+ * one buffer rather than returned and concatenated level by level, so the cost
+ * stays linear in the size of the formula however deeply it nests.
+ */
+export function linearMath(math) {
+    const out = [];
+    const emit = (text) => {
+        if (text)
+            out.push(text);
+    };
+    /** Element children only: text between MathML arguments is never one of them. */
+    const argsOf = (node) => childrenOf(node).filter((child) => !isTextNode(child));
+    /** Write `pieces` after `open`, or nothing at all when they write nothing. */
+    const unlessEmpty = (open, pieces, close = '') => {
+        const start = out.length;
+        out.push(open);
+        pieces();
+        if (out.length === start + 1)
+            out.length = start;
+        else
+            emit(close);
+    };
+    const script = (mark, node) => unlessEmpty(`${mark}{`, () => write(node), '}');
+    /** Script pairs — subscript, then superscript — in document order. */
+    const scriptPairs = (scripts) => {
+        for (let i = 0; i < scripts.length; i += 2) {
+            script('_', scripts[i]);
+            script('^', scripts[i + 1]);
+        }
+    };
+    const write = (node) => {
+        if (!node)
+            return;
+        if (isTextNode(node)) {
+            const text = textOf(node);
+            if (VISIBLE_CHAR.test(text))
+                emit(text);
+            return;
+        }
+        const name = localName(tagNameOf(node) ?? '');
+        if (MATH_TOKEN_TAGS.has(name)) {
+            for (const child of childrenOf(node)) {
+                if (isTextNode(child))
+                    emit(textOf(child));
+                else
+                    write(child);
+            }
+            return;
+        }
+        const args = argsOf(node);
+        switch (name) {
+            case 'msub':
+            case 'munder':
+                write(args[0]);
+                script('_', args[1]);
+                return;
+            case 'msup':
+            case 'mover':
+                write(args[0]);
+                script('^', args[1]);
+                return;
+            case 'msubsup':
+            case 'munderover':
+                write(args[0]);
+                script('_', args[1]);
+                script('^', args[2]);
+                return;
+            case 'mmultiscripts': {
+                const split = args.findIndex((arg) => localName(tagNameOf(arg) ?? '') === 'mprescripts');
+                if (split !== -1)
+                    unlessEmpty('{}', () => scriptPairs(args.slice(split + 1)));
+                write(args[0]);
+                scriptPairs(args.slice(1, split === -1 ? undefined : split));
+                return;
+            }
+            case 'mfrac':
+                emit('\\frac{');
+                write(args[0]);
+                emit('}{');
+                write(args[1]);
+                emit('}');
+                return;
+            case 'msqrt':
+                emit('\\sqrt{');
+                for (const arg of args)
+                    write(arg);
+                emit('}');
+                return;
+            case 'mroot':
+                emit('\\sqrt[');
+                write(args[1]);
+                emit(']{');
+                write(args[0]);
+                emit('}');
+                return;
+            case 'mfenced': {
+                const separators = [...(attrOf(node, 'separators') ?? ',').replace(/\s/g, '')];
+                emit(attrOf(node, 'open') ?? '(');
+                for (const [i, arg] of args.entries()) {
+                    if (i > 0)
+                        emit(separators[Math.min(i - 1, separators.length - 1)] ?? '');
+                    write(arg);
+                }
+                emit(attrOf(node, 'close') ?? ')');
+                return;
+            }
+            case 'mtable':
+                for (const [r, row] of args.entries()) {
+                    if (r > 0)
+                        emit(' \\\\ ');
+                    for (const [c, cell] of argsOf(row).entries()) {
+                        if (c > 0)
+                            emit(' & ');
+                        write(cell);
+                    }
+                }
+                return;
+            case 'mspace':
+                emit(' ');
+                return;
+            case 'mphantom':
+                return;
+            default:
+                for (const child of childrenOf(node))
+                    write(child);
+        }
+    };
+    write(math);
+    return out.join('');
+}
 /**
  * The single child of an `<alternatives>` whose text stands for the whole
  * element.
@@ -97,67 +267,246 @@ function texMathExpression(raw) {
  */
 export function selectAlternative(node, excluded) {
     const children = childrenOf(node);
-    const carriesText = (child) => concatText(child, excluded).trim() !== '';
+    const carriesText = (child) => readText(child, excluded).trim() !== '';
     return (children.find((child) => tagNameOf(child) === TEX_MATH_TAG && carriesText(child)) ??
         children.find((child) => !POINTER_TAGS.has(tagNameOf(child) ?? '') && carriesText(child)) ??
         children.find(carriesText));
 }
 /**
- * Concatenate text content in document order without normalizing whitespace.
- * Internal helper so recursion preserves the original spacing between siblings.
- * `excluded` skips a whole subtree by tag name; omitting it reads everything.
+ * Elements whose start and end are line boundaries: each holds a statement, a
+ * footnote, a list or list item, or a heading of its own. A deposit marks these
+ * boundaries with the element alone — `RATIOS<break/>IRR`, `</fn><fn>` — so
+ * reading straight through fuses the words and numbers on either side
+ * (`57 ± 2095`, `1.082.44`). Inline markup (`<sub>`, `<sup>`, `<italic>`,
+ * `<xref>`) is never one: `H<sub>2</sub>O` must read `H2O`. (#185)
+ */
+const LINE_BOUNDARY_TAGS = new Set([
+    'attrib',
+    'def-item',
+    'def-list',
+    'disp-quote',
+    'fn',
+    'list',
+    'list-item',
+    'p',
+    'title',
+]);
+/** Empty elements that are a line boundary in themselves. (#185) */
+const LINE_BREAK_TAGS = new Set(['break', 'hr']);
+/**
+ * A marker printed beside the text it labels — a footnote's `*`, an
+ * affiliation's `1`, a list item's `2.` — with nothing between them in the
+ * source. (#185)
+ */
+const LABEL_TAG = 'label';
+/** A cross-reference: in prose, usually a citation marker. (#197) */
+const XREF_TAG = 'xref';
+/** No separator owed. */
+const NO_SEPARATOR = 0;
+/** One space owed: what follows a `<label>`. */
+const SPACE_SEPARATOR = 1;
+/** A line boundary owed. */
+const LINE_SEPARATOR = 2;
+/** One character of whitespace — the same class `\s+` collapses. */
+const WHITESPACE_CHAR = /\s/;
+/** Any character other than whitespace. */
+const VISIBLE_CHAR = /\S/;
+/**
+ * Accumulates JATS text in document order and decides every separator the
+ * markup implies in the same pass, from state it carries rather than by
+ * re-reading text already emitted — so the cost stays linear in the size of the
+ * subtree. A reader drives it with {@link text} for text nodes and with
+ * {@link open} / {@link close} around each element's children.
+ *
+ * - A line-boundary element ({@link LINE_BOUNDARY_TAGS}, `<break/>`, `<hr/>`)
+ *   owes a line boundary before the next content.
+ * - A `<label>` that carried text owes one space, and a line boundary arriving
+ *   before any further content becomes that space: `<fn><label>*</label><p>…`
+ *   reads `* …`, the marker leading its line rather than standing alone on it.
+ * - An `<xref>` that carries text and follows another with no source text
+ *   between them is preceded by a comma: `[1,2,3,4]`, not `[1234]`, which reads
+ *   as one reference. Any source text between them, whitespace included, means
+ *   the source already separated them. (#197)
+ *
+ * An owed separator is only written between two pieces of content, so a reader
+ * adds nothing at the edges of what it reads. By default the source text is
+ * kept verbatim and a separator is written only where neither side carries
+ * whitespace — `a<break/>b` gains a `\n`, `a <break/>b` stays as it was — so a
+ * caller collapsing whitespace afterwards reads every boundary as one space. In
+ * `lines` mode each text node's whitespace collapses to a space and every line
+ * boundary is written as `\n`, replacing whatever spacing the source had there,
+ * so the result keeps the source's line structure and nothing else. (#185)
+ */
+export class TextAssembler {
+    #lines;
+    #out = [];
+    /** How many non-whitespace text pieces have been written. */
+    #written = 0;
+    /** The last character written is whitespace. */
+    #endsWithSpace = false;
+    #owed = NO_SEPARATOR;
+    /** A `<label>` closed and nothing but whitespace has arrived since. */
+    #afterLabel = false;
+    /** An `<xref>` carrying text closed and no source text has arrived since. */
+    #afterXref = false;
+    /** A comma is owed before the open `<xref>`'s first text. */
+    #commaOwed = false;
+    /** {@link #written} at each open `<label>` and `<xref>`, innermost last. */
+    #marks = [];
+    constructor(options = {}) {
+        this.#lines = options.lines ?? false;
+    }
+    /** Append one text node's value. */
+    text(value) {
+        if (value === '')
+            return;
+        this.#afterXref = false;
+        if (!VISIBLE_CHAR.test(value)) {
+            this.#out.push(this.#lines ? ' ' : value);
+            this.#endsWithSpace = true;
+            return;
+        }
+        if (this.#written > 0 && this.#owed !== NO_SEPARATOR) {
+            const separator = this.#owed === LINE_SEPARATOR ? '\n' : ' ';
+            const sourceSpaced = this.#endsWithSpace || WHITESPACE_CHAR.test(value.charAt(0));
+            if (this.#lines || !sourceSpaced)
+                this.#out.push(separator);
+        }
+        else if (this.#commaOwed) {
+            this.#out.push(',');
+        }
+        this.#owed = NO_SEPARATOR;
+        this.#afterLabel = false;
+        this.#commaOwed = false;
+        this.#out.push(this.#lines ? value.replace(/\s+/g, ' ') : value);
+        this.#endsWithSpace = WHITESPACE_CHAR.test(value.charAt(value.length - 1));
+        this.#written += 1;
+    }
+    /** Enter an element, before its children are read. */
+    open(tag) {
+        if (LINE_BREAK_TAGS.has(tag) || LINE_BOUNDARY_TAGS.has(tag)) {
+            this.#oweLine();
+        }
+        else if (tag === LABEL_TAG) {
+            this.#marks.push(this.#written);
+        }
+        else if (tag === XREF_TAG) {
+            this.#marks.push(this.#written);
+            this.#commaOwed = this.#afterXref;
+        }
+    }
+    /**
+     * Leave an element, after its children are read. A close with no matching
+     * mark — the run was restarted inside the element, as a prose walk does at a
+     * block nested in inline markup — owes nothing.
+     */
+    close(tag) {
+        if (LINE_BOUNDARY_TAGS.has(tag)) {
+            this.#oweLine();
+            return;
+        }
+        if (tag !== LABEL_TAG && tag !== XREF_TAG)
+            return;
+        const mark = this.#marks.pop();
+        if (mark === undefined)
+            return;
+        const carriedText = mark !== this.#written;
+        if (tag === LABEL_TAG) {
+            if (!carriedText)
+                return;
+            this.#owed = SPACE_SEPARATOR;
+            this.#afterLabel = true;
+            return;
+        }
+        this.#commaOwed = false;
+        if (carriedText)
+            this.#afterXref = true;
+    }
+    /** The text assembled so far. */
+    toString() {
+        const text = this.#out.join('');
+        if (!this.#lines)
+            return text;
+        return text
+            .replace(/ {2,}/g, ' ')
+            .replace(/ ?\n ?/g, '\n')
+            .trim();
+    }
+    #oweLine() {
+        if (!this.#afterLabel)
+            this.#owed = LINE_SEPARATOR;
+    }
+}
+/**
+ * Drive `sink` through a node or sibling list in document order. `excluded`
+ * skips a whole subtree by tag name; omitting it reads everything else.
  *
- * Two JATS elements do not contribute the plain concatenation of their subtree:
- * a `<tex-math>` contributes only its LaTeX document body, and an
- * `<alternatives>` contributes exactly one child. Both rules live here so every
- * prose consumer inherits them — paragraphs, table cells, captions, abstracts.
- * (#135)
- */
-function concatText(input, excluded) {
-    const nodes = Array.isArray(input) ? input : [input];
-    const parts = [];
-    for (const node of nodes) {
+ * Four JATS elements do not contribute the plain concatenation of their
+ * subtree: a `<tex-math>` contributes only its LaTeX document body, a MathML
+ * `<math>` its {@link linearMath} form, an `<alternatives>` exactly one child,
+ * and an {@link IDENTIFIER_TAGS} element nothing. The rules live here, beside
+ * the boundary rules {@link TextAssembler} applies, so every consumer inherits
+ * them — table cells, footnotes, titles, affiliations, references. (#135, #185,
+ * #207, #208)
+ */
+function readInto(input, sink, excluded) {
+    for (const node of Array.isArray(input) ? input : [input]) {
         if (isTextNode(node)) {
-            parts.push(textOf(node));
+            sink.text(textOf(node));
             continue;
         }
         const tag = tagNameOf(node) ?? '';
-        if (excluded?.has(tag))
+        if (IDENTIFIER_TAGS.has(tag) || excluded?.has(tag))
             continue;
         if (tag === TEX_MATH_TAG) {
-            parts.push(texMathExpression(concatText(childrenOf(node))));
+            sink.text(texMathExpression(readText(childrenOf(node))));
+            continue;
+        }
+        if (isMathTag(tag)) {
+            sink.text(linearMath(node));
             continue;
         }
         if (tag === ALTERNATIVES_TAG) {
             const chosen = selectAlternative(node, excluded);
             if (chosen)
-                parts.push(concatText(chosen, excluded));
+                readInto(chosen, sink, excluded);
             continue;
         }
-        parts.push(concatText(childrenOf(node), excluded));
+        sink.open(tag);
+        readInto(childrenOf(node), sink, excluded);
+        sink.close(tag);
     }
-    return parts.join('');
+}
+/** Read a subtree through a fresh {@link TextAssembler}. */
+function readText(input, excluded, lines = false) {
+    const sink = new TextAssembler({ lines });
+    readInto(input, sink, excluded);
+    return sink.toString();
 }
 /**
  * Extract all text from a node or sibling list in document order with the
- * source spacing intact — no whitespace collapsing, no trimming. Use it when
- * the caller assembles several fragments itself and needs to collapse once over
- * the joined result; {@link textContent} is the normalizing form for everything
- * else.
+ * source spacing intact — no whitespace collapsing, no trimming. A line boundary
+ * the source marks with an element alone gains a `\n`, and a `<label>` beside
+ * its text a space; a boundary the source already spaced is left as it is, and
+ * nothing is added at the edges of the node read. Use it when the caller assembles several
+ * fragments itself and needs to collapse once over the joined result;
+ * {@link textContent} is the normalizing form for everything else.
  */
 export function rawTextContent(input) {
-    return input ? concatText(input) : '';
+    return input ? readText(input) : '';
 }
 /**
  * Extract all text from a node or sibling list in document order, collapsing
  * runs of whitespace to a single space and trimming the result. Use this for
  * mixed-content elements (`<p>`, `<title>`, `<abstract>`, …) where inline
- * children must read back in the order they appear in the source.
+ * children must read back in the order they appear in the source. A structural
+ * boundary — `<break/>`, a sibling `<p>` or `<fn>`, a `<label>` beside its text —
+ * reads as one space; inline markup never gains one. (#185)
  */
 export function textContent(input) {
     if (!input)
         return '';
-    return concatText(input).replace(/\s+/g, ' ').trim();
+    return readText(input).replace(/\s+/g, ' ').trim();
 }
 /**
  * {@link textContent}, with the subtree of any element named in `excludedTags`
@@ -169,7 +518,19 @@ export function textContent(input) {
 export function textContentExcluding(input, excludedTags) {
     if (!input)
         return '';
-    return concatText(input, excludedTags).replace(/\s+/g, ' ').trim();
+    return readText(input, excludedTags).replace(/\s+/g, ' ').trim();
+}
+/**
+ * {@link textContent}, keeping every structural line boundary as `\n`: one line
+ * per `<break/>`-separated value, paragraph, footnote, or list item, a
+ * `<label>` leading its line by one space. Source whitespace inside a line —
+ * pretty-printing newlines included — collapses to one space, and blank lines
+ * drop out. For table cells and `<table-wrap-foot>`, whose line structure is
+ * part of the value: `1.08<break/>2.44` is two estimates, not `1.082.44`.
+ * (#185)
+ */
+export function lineTextContent(input) {
+    return input ? readText(input, undefined, true) : '';
 }
 /** First direct child with the given tag name. */
 export function findOne(input, tagName) {

package/dist/services/ncbi/parsing/pmc-xml-helpers.js.map

@@ -1 +1 @@
-{"version":3,"file":"pmc-xml-helpers.js","sourceRoot":"","sources":["../../../../src/services/ncbi/parsing/pmc-xml-helpers.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAgBH,MAAM,QAAQ,GAAG,IAAI,CAAC;AACtB,MAAM,QAAQ,GAAG,OAAO,CAAC;AAEzB,uFAAuF;AACvF,MAAM,UAAU,SAAS,CAAC,IAAc;IACtC,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,QAAQ;YAAE,OAAO,GAAG,CAAC;IACvD,CAAC;IACD,OAAO;AACT,CAAC;AAED,iFAAiF;AACjF,MAAM,UAAU,UAAU,CAAC,IAAc;IACvC,MAAM,GAAG,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAC5B,IAAI,CAAC,GAAG;QAAE,OAAO,EAAE,CAAC;IACpB,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC;IACxB,OAAO,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAE,KAAsB,CAAC,CAAC,CAAC,EAAE,CAAC;AAC7D,CAAC;AAED,sDAAsD;AACtD,MAAM,UAAU,MAAM,CAAC,IAAc,EAAE,IAAY;IACjD,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAwC,CAAC;IACpE,MAAM,GAAG,GAAG,KAAK,EAAE,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IACjC,OAAO,GAAG,IAAI,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;AAC/C,CAAC;AAED,yCAAyC;AACzC,MAAM,UAAU,UAAU,CAAC,IAAc;IACvC,OAAO,QAAQ,IAAI,IAAI,CAAC;AAC1B,CAAC;AAED,6CAA6C;AAC7C,MAAM,UAAU,MAAM,CAAC,IAAc;IACnC,MAAM,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC;IACzB,OAAO,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;AACpC,CAAC;AAED,4DAA4D;AAC5D,MAAM,YAAY,GAAG,UAAU,CAAC;AAEhC,kEAAkE;AAClE,MAAM,gBAAgB,GAAG,cAAc,CAAC;AAExC;;;GAGG;AACH,MAAM,YAAY,GAAwB,IAAI,GAAG,CAAC;IAChD,SAAS;IACT,gBAAgB;IAChB,OAAO;IACP,cAAc;CACf,CAAC,CAAC;AAEH;;;;;GAKG;AACH,MAAM,iBAAiB,GAAG,sDAAsD,CAAC;AAEjF;;;;;;;;;;;GAWG;AACH,SAAS,iBAAiB,CAAC,GAAW;IACpC,OAAO,iBAAiB,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC;AACjD,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,iBAAiB,CAC/B,IAAc,EACd,QAA8B;IAE9B,MAAM,QAAQ,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;IAClC,MAAM,WAAW,GAAG,CAAC,KAAe,EAAW,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC;IAC5F,OAAO,CACL,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,SAAS,CAAC,KAAK,CAAC,KAAK,YAAY,IAAI,WAAW,CAAC,KAAK,CAAC,CAAC;QACjF,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,WAAW,CAAC,KAAK,CAAC,CAAC;QACzF,QAAQ,CAAC,IAAI,CAAC,WAAW,CAAC,CAC3B,CAAC;AACJ,CAAC;AAED;;;;;;;;;;GAUG;AACH,SAAS,UAAU,CAAC,KAA8B,EAAE,QAA8B;IAChF,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;IACrD,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;YACrB,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;YACzB,SAAS;QACX,CAAC;QACD,MAAM,GAAG,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QAClC,IAAI,QAAQ,EAAE,GAAG,CAAC,GAAG,CAAC;YAAE,SAAS;QACjC,IAAI,GAAG,KAAK,YAAY,EAAE,CAAC;YACzB,KAAK,CAAC,IAAI,CAAC,iBAAiB,CAAC,UAAU,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YAC5D,SAAS;QACX,CAAC;QACD,IAAI,GAAG,KAAK,gBAAgB,EAAE,CAAC;YAC7B,MAAM,MAAM,GAAG,iBAAiB,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;YACjD,IAAI,MAAM;gBAAE,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC,CAAC;YACrD,SAAS;QACX,CAAC;QACD,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,QAAQ,CAAC,CAAC,CAAC;IACrD,CAAC;IACD,OAAO,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACxB,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,cAAc,CAAC,KAA0C;IACvE,OAAO,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;AACxC,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,WAAW,CAAC,KAA0C;IACpE,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,OAAO,UAAU,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACvD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,oBAAoB,CAClC,KAA0C,EAC1C,YAAiC;IAEjC,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,OAAO,UAAU,CAAC,KAAK,EAAE,YAAY,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACrE,CAAC;AAED,kDAAkD;AAClD,MAAM,UAAU,OAAO,CACrB,KAA0C,EAC1C,OAAe;IAEf,IAAI,CAAC,KAAK;QAAE,OAAO;IACnB,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;IAClE,OAAO,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,OAAO,CAAC,CAAC;AACxD,CAAC;AAED,mDAAmD;AACnD,MAAM,UAAU,OAAO,CAAC,KAA0C,EAAE,OAAe;IACjF,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;IAClE,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,OAAO,CAAC,CAAC;AAC1D,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,kBAAkB,CAChC,KAA0C,EAC1C,OAAe;IAEf,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,MAAM,KAAK,GAAe,EAAE,CAAC;IAC7B,MAAM,KAAK,GAAG,CAAC,KAAmB,EAAQ,EAAE;QAC1C,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,OAAO;gBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAClD,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC;QAC1B,CAAC;IACH,CAAC,CAAC;IACF,KAAK,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC;IACxD,OAAO,KAAK,CAAC;AACf,CAAC"}
+{"version":3,"file":"pmc-xml-helpers.js","sourceRoot":"","sources":["../../../../src/services/ncbi/parsing/pmc-xml-helpers.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAgBH,MAAM,QAAQ,GAAG,IAAI,CAAC;AACtB,MAAM,QAAQ,GAAG,OAAO,CAAC;AAEzB,uFAAuF;AACvF,MAAM,UAAU,SAAS,CAAC,IAAc;IACtC,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,QAAQ;YAAE,OAAO,GAAG,CAAC;IACvD,CAAC;IACD,OAAO;AACT,CAAC;AAED,iFAAiF;AACjF,MAAM,UAAU,UAAU,CAAC,IAAc;IACvC,MAAM,GAAG,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAC5B,IAAI,CAAC,GAAG;QAAE,OAAO,EAAE,CAAC;IACpB,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC;IACxB,OAAO,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAE,KAAsB,CAAC,CAAC,CAAC,EAAE,CAAC;AAC7D,CAAC;AAED,sDAAsD;AACtD,MAAM,UAAU,MAAM,CAAC,IAAc,EAAE,IAAY;IACjD,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAwC,CAAC;IACpE,MAAM,GAAG,GAAG,KAAK,EAAE,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IACjC,OAAO,GAAG,IAAI,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;AAC/C,CAAC;AAED,yCAAyC;AACzC,MAAM,UAAU,UAAU,CAAC,IAAc;IACvC,OAAO,QAAQ,IAAI,IAAI,CAAC;AAC1B,CAAC;AAED,6CAA6C;AAC7C,MAAM,UAAU,MAAM,CAAC,IAAc;IACnC,MAAM,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC;IACzB,OAAO,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;AACpC,CAAC;AAED,4DAA4D;AAC5D,MAAM,YAAY,GAAG,UAAU,CAAC;AAEhC,kEAAkE;AAClE,MAAM,gBAAgB,GAAG,cAAc,CAAC;AAExC;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,eAAe,GAAwB,IAAI,GAAG,CAAC,CAAC,gBAAgB,CAAC,CAAC,CAAC;AAEhF;;;GAGG;AACH,MAAM,YAAY,GAAwB,IAAI,GAAG,CAAC;IAChD,SAAS;IACT,gBAAgB;IAChB,OAAO;IACP,cAAc;CACf,CAAC,CAAC;AAEH;;;;;GAKG;AACH,MAAM,iBAAiB,GAAG,sDAAsD,CAAC;AAEjF;;;;;;;;;;;GAWG;AACH,SAAS,iBAAiB,CAAC,GAAW;IACpC,OAAO,iBAAiB,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC;AACjD,CAAC;AAED,4EAA4E;AAC5E,MAAM,SAAS,GAAG,CAAC,GAAW,EAAU,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;AAE3E;;;GAGG;AACH,MAAM,UAAU,SAAS,CAAC,GAAW;IACnC,OAAO,SAAS,CAAC,GAAG,CAAC,KAAK,MAAM,CAAC;AACnC,CAAC;AAED,uEAAuE;AACvE,MAAM,eAAe,GAAwB,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC,CAAC;AAExF;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,UAAU,UAAU,CAAC,IAAc;IACvC,MAAM,GAAG,GAAa,EAAE,CAAC;IACzB,MAAM,IAAI,GAAG,CAAC,IAAY,EAAQ,EAAE;QAClC,IAAI,IAAI;YAAE,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC3B,CAAC,CAAC;IACF,iFAAiF;IACjF,MAAM,MAAM,GAAG,CAAC,IAAc,EAAgB,EAAE,CAC9C,UAAU,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC;IACzD,8EAA8E;IAC9E,MAAM,WAAW,GAAG,CAAC,IAAY,EAAE,MAAkB,EAAE,KAAK,GAAG,EAAE,EAAQ,EAAE;QACzE,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC;QACzB,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACf,MAAM,EAAE,CAAC;QACT,IAAI,GAAG,CAAC,MAAM,KAAK,KAAK,GAAG,CAAC;YAAE,GAAG,CAAC,MAAM,GAAG,KAAK,CAAC;;YAC5C,IAAI,CAAC,KAAK,CAAC,CAAC;IACnB,CAAC,CAAC;IACF,MAAM,MAAM,GAAG,CAAC,IAAe,EAAE,IAA0B,EAAQ,EAAE,CACnE,WAAW,CAAC,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC;IAClD,sEAAsE;IACtE,MAAM,WAAW,GAAG,CAAC,OAAqB,EAAQ,EAAE;QAClD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC;YAC3C,MAAM,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;YACxB,MAAM,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QAC9B,CAAC;IACH,CAAC,CAAC;IACF,MAAM,KAAK,GAAG,CAAC,IAA0B,EAAQ,EAAE;QACjD,IAAI,CAAC,IAAI;YAAE,OAAO;QAClB,IAAI,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;YACrB,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC;YAC1B,IAAI,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC;gBAAE,IAAI,CAAC,IAAI,CAAC,CAAC;YACxC,OAAO;QACT,CAAC;QACD,MAAM,IAAI,GAAG,SAAS,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;QAC9C,IAAI,eAAe,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YAC9B,KAAK,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;gBACrC,IAAI,UAAU,CAAC,KAAK,CAAC;oBAAE,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC;;oBACtC,KAAK,CAAC,KAAK,CAAC,CAAC;YACpB,CAAC;YACD,OAAO;QACT,CAAC;QACD,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC;QAC1B,QAAQ,IAAI,EAAE,CAAC;YACb,KAAK,MAAM,CAAC;YACZ,KAAK,QAAQ;gBACX,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACrB,OAAO;YACT,KAAK,MAAM,CAAC;YACZ,KAAK,OAAO;gBACV,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACrB,OAAO;YACT,KAAK,SAAS,CAAC;YACf,KAAK,YAAY;gBACf,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACrB,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACrB,OAAO;YACT,KAAK,eAAe,EAAE,CAAC;gBACrB,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,SAAS,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,KAAK,aAAa,CAAC,CAAC;gBACzF,IAAI,KAAK,KAAK,CAAC,CAAC;oBAAE,WAAW,CAAC,IAAI,EAAE,GAAG,EAAE,CAAC,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;gBAC9E,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC;gBAC7D,OAAO;YACT,CAAC;YACD,KAAK,OAAO;gBACV,IAAI,CAAC,SAAS,CAAC,CAAC;gBAChB,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,IAAI,CAAC,IAAI,CAAC,CAAC;gBACX,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,IAAI,CAAC,GAAG,CAAC,CAAC;gBACV,OAAO;YACT,KAAK,OAAO;gBACV,IAAI,CAAC,SAAS,CAAC,CAAC;gBAChB,KAAK,MAAM,GAAG,IAAI,IAAI;oBAAE,KAAK,CAAC,GAAG,CAAC,CAAC;gBACnC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACV,OAAO;YACT,KAAK,OAAO;gBACV,IAAI,CAAC,SAAS,CAAC,CAAC;gBAChB,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,IAAI,CAAC,IAAI,CAAC,CAAC;gBACX,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;gBACf,IAAI,CAAC,GAAG,CAAC,CAAC;gBACV,OAAO;YACT,KAAK,SAAS,EAAE,CAAC;gBACf,MAAM,UAAU,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,YAAY,CAAC,IAAI,GAAG,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,CAAC;gBAC/E,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,CAAC;gBAClC,KAAK,MAAM,CAAC,CAAC,EAAE,GAAG,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,EAAE,CAAC;oBACtC,IAAI,CAAC,GAAG,CAAC;wBAAE,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,EAAE,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;oBAC1E,KAAK,CAAC,GAAG,CAAC,CAAC;gBACb,CAAC;gBACD,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,OAAO,CAAC,IAAI,GAAG,CAAC,CAAC;gBACnC,OAAO;YACT,CAAC;YACD,KAAK,QAAQ;gBACX,KAAK,MAAM,CAAC,CAAC,EAAE,GAAG,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,EAAE,CAAC;oBACtC,IAAI,CAAC,GAAG,CAAC;wBAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;oBAC1B,KAAK,MAAM,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC;wBAC9C,IAAI,CAAC,GAAG,CAAC;4BAAE,IAAI,CAAC,KAAK,CAAC,CAAC;wBACvB,KAAK,CAAC,IAAI,CAAC,CAAC;oBACd,CAAC;gBACH,CAAC;gBACD,OAAO;YACT,KAAK,QAAQ;gBACX,IAAI,CAAC,GAAG,CAAC,CAAC;gBACV,OAAO;YACT,KAAK,UAAU;gBACb,OAAO;YACT;gBACE,KAAK,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC;oBAAE,KAAK,CAAC,KAAK,CAAC,CAAC;QACvD,CAAC;IACH,CAAC,CAAC;IACF,KAAK,CAAC,IAAI,CAAC,CAAC;IACZ,OAAO,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACtB,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,iBAAiB,CAC/B,IAAc,EACd,QAA8B;IAE9B,MAAM,QAAQ,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;IAClC,MAAM,WAAW,GAAG,CAAC,KAAe,EAAW,EAAE,CAAC,QAAQ,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC;IAC1F,OAAO,CACL,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,SAAS,CAAC,KAAK,CAAC,KAAK,YAAY,IAAI,WAAW,CAAC,KAAK,CAAC,CAAC;QACjF,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,WAAW,CAAC,KAAK,CAAC,CAAC;QACzF,QAAQ,CAAC,IAAI,CAAC,WAAW,CAAC,CAC3B,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,kBAAkB,GAAwB,IAAI,GAAG,CAAC;IACtD,QAAQ;IACR,UAAU;IACV,UAAU;IACV,YAAY;IACZ,IAAI;IACJ,MAAM;IACN,WAAW;IACX,GAAG;IACH,OAAO;CACR,CAAC,CAAC;AAEH,oEAAoE;AACpE,MAAM,eAAe,GAAwB,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,CAAC;AAEtE;;;;GAIG;AACH,MAAM,SAAS,GAAG,OAAO,CAAC;AAE1B,qEAAqE;AACrE,MAAM,QAAQ,GAAG,MAAM,CAAC;AAExB,yBAAyB;AACzB,MAAM,YAAY,GAAG,CAAC,CAAC;AACvB,gDAAgD;AAChD,MAAM,eAAe,GAAG,CAAC,CAAC;AAC1B,4BAA4B;AAC5B,MAAM,cAAc,GAAG,CAAC,CAAC;AAGzB,oEAAoE;AACpE,MAAM,eAAe,GAAG,IAAI,CAAC;AAC7B,2CAA2C;AAC3C,MAAM,YAAY,GAAG,IAAI,CAAC;AAE1B;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,MAAM,OAAO,aAAa;IACf,MAAM,CAAU;IAChB,IAAI,GAAa,EAAE,CAAC;IAC7B,6DAA6D;IAC7D,QAAQ,GAAG,CAAC,CAAC;IACb,gDAAgD;IAChD,cAAc,GAAG,KAAK,CAAC;IACvB,KAAK,GAAc,YAAY,CAAC;IAChC,uEAAuE;IACvE,WAAW,GAAG,KAAK,CAAC;IACpB,6EAA6E;IAC7E,UAAU,GAAG,KAAK,CAAC;IACnB,6DAA6D;IAC7D,UAAU,GAAG,KAAK,CAAC;IACnB,4EAA4E;IACnE,MAAM,GAAa,EAAE,CAAC;IAE/B,YAAY,OAAO,GAAwB,EAAE;QAC3C,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK,IAAI,KAAK,CAAC;IACvC,CAAC;IAED,oCAAoC;IACpC,IAAI,CAAC,KAAa;QAChB,IAAI,KAAK,KAAK,EAAE;YAAE,OAAO;QACzB,IAAI,CAAC,UAAU,GAAG,KAAK,CAAC;QACxB,IAAI,CAAC,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAC9B,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;YAC1C,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC;YAC3B,OAAO;QACT,CAAC;QACD,IAAI,IAAI,CAAC,QAAQ,GAAG,CAAC,IAAI,IAAI,CAAC,KAAK,KAAK,YAAY,EAAE,CAAC;YACrD,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,KAAK,cAAc,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC;YAC7D,MAAM,YAAY,GAAG,IAAI,CAAC,cAAc,IAAI,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;YAClF,IAAI,IAAI,CAAC,MAAM,IAAI,CAAC,YAAY;gBAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;QAC9D,CAAC;aAAM,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;YAC3B,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACtB,CAAC;QACD,IAAI,CAAC,KAAK,GAAG,YAAY,CAAC;QAC1B,IAAI,CAAC,WAAW,GAAG,KAAK,CAAC;QACzB,IAAI,CAAC,UAAU,GAAG,KAAK,CAAC;QACxB,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QACjE,IAAI,CAAC,cAAc,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;QAC3E,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;IACrB,CAAC;IAED,sDAAsD;IACtD,IAAI,CAAC,GAAW;QACd,IAAI,eAAe,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,kBAAkB,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;YAC5D,IAAI,CAAC,QAAQ,EAAE,CAAC;QAClB,CAAC;aAAM,IAAI,GAAG,KAAK,SAAS,EAAE,CAAC;YAC7B,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAClC,CAAC;aAAM,IAAI,GAAG,KAAK,QAAQ,EAAE,CAAC;YAC5B,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;YAChC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU,CAAC;QACpC,CAAC;IACH,CAAC;IAED;;;;OAIG;IACH,KAAK,CAAC,GAAW;QACf,IAAI,kBAAkB,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;YAChC,IAAI,CAAC,QAAQ,EAAE,CAAC;YAChB,OAAO;QACT,CAAC;QACD,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,QAAQ;YAAE,OAAO;QAClD,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC;QAC/B,IAAI,IAAI,KAAK,SAAS;YAAE,OAAO;QAC/B,MAAM,WAAW,GAAG,IAAI,KAAK,IAAI,CAAC,QAAQ,CAAC;QAC3C,IAAI,GAAG,KAAK,SAAS,EAAE,CAAC;YACtB,IAAI,CAAC,WAAW;gBAAE,OAAO;YACzB,IAAI,CAAC,KAAK,GAAG,eAAe,CAAC;YAC7B,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;YACxB,OAAO;QACT,CAAC;QACD,IAAI,CAAC,UAAU,GAAG,KAAK,CAAC;QACxB,IAAI,WAAW;YAAE,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC;IAC1C,CAAC;IAED,iCAAiC;IACjC,QAAQ;QACN,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAChC,IAAI,CAAC,IAAI,CAAC,MAAM;YAAE,OAAO,IAAI,CAAC;QAC9B,OAAO,IAAI;aACR,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC;aACtB,OAAO,CAAC,SAAS,EAAE,IAAI,CAAC;aACxB,IAAI,EAAE,CAAC;IACZ,CAAC;IAED,QAAQ;QACN,IAAI,CAAC,IAAI,CAAC,WAAW;YAAE,IAAI,CAAC,KAAK,GAAG,cAAc,CAAC;IACrD,CAAC;CACF;AAED;;;;;;;;;;;GAWG;AACH,SAAS,QAAQ,CACf,KAA8B,EAC9B,IAAmB,EACnB,QAA8B;IAE9B,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,EAAE,CAAC;QAC1D,IAAI,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;YACrB,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;YACxB,SAAS;QACX,CAAC;QACD,MAAM,GAAG,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QAClC,IAAI,eAAe,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,QAAQ,EAAE,GAAG,CAAC,GAAG,CAAC;YAAE,SAAS;QAC7D,IAAI,GAAG,KAAK,YAAY,EAAE,CAAC;YACzB,IAAI,CAAC,IAAI,CAAC,iBAAiB,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YACzD,SAAS;QACX,CAAC;QACD,IAAI,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC;YACnB,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC;YAC5B,SAAS;QACX,CAAC;QACD,IAAI,GAAG,KAAK,gBAAgB,EAAE,CAAC;YAC7B,MAAM,MAAM,GAAG,iBAAiB,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;YACjD,IAAI,MAAM;gBAAE,QAAQ,CAAC,MAAM,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAC;YAC7C,SAAS;QACX,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACf,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAC;QAC3C,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAClB,CAAC;AACH,CAAC;AAED,4DAA4D;AAC5D,SAAS,QAAQ,CACf,KAA8B,EAC9B,QAA8B,EAC9B,KAAK,GAAG,KAAK;IAEb,MAAM,IAAI,GAAG,IAAI,aAAa,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC;IAC1C,QAAQ,CAAC,KAAK,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAC;IAChC,OAAO,IAAI,CAAC,QAAQ,EAAE,CAAC;AACzB,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,cAAc,CAAC,KAA0C;IACvE,OAAO,KAAK,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;AACtC,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CAAC,KAA0C;IACpE,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,OAAO,QAAQ,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACrD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,oBAAoB,CAClC,KAA0C,EAC1C,YAAiC;IAEjC,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,OAAO,QAAQ,CAAC,KAAK,EAAE,YAAY,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACnE,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,eAAe,CAAC,KAA0C;IACxE,OAAO,KAAK,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,EAAE,SAAS,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;AACvD,CAAC;AAED,kDAAkD;AAClD,MAAM,UAAU,OAAO,CACrB,KAA0C,EAC1C,OAAe;IAEf,IAAI,CAAC,KAAK;QAAE,OAAO;IACnB,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;IAClE,OAAO,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,OAAO,CAAC,CAAC;AACxD,CAAC;AAED,mDAAmD;AACnD,MAAM,UAAU,OAAO,CAAC,KAA0C,EAAE,OAAe;IACjF,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;IAClE,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,OAAO,CAAC,CAAC;AAC1D,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,kBAAkB,CAChC,KAA0C,EAC1C,OAAe;IAEf,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IACtB,MAAM,KAAK,GAAe,EAAE,CAAC;IAC7B,MAAM,KAAK,GAAG,CAAC,KAAmB,EAAQ,EAAE;QAC1C,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,OAAO;gBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAClD,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC;QAC1B,CAAC;IACH,CAAC,CAAC;IACF,KAAK,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC;IACxD,OAAO,KAAK,CAAC;AACf,CAAC"}

package/dist/services/ncbi/parsing/text-helpers.d.ts

@@ -18,12 +18,14 @@
 export declare function decodeHtmlEntities(text: string): string;
 /**
  * Turn an upstream string that may carry JATS/HTML markup into display-ready
- * plain text: strip structural tags, decode character references, drop soft
- * hyphens (U+00AD — invisible, and it corrupts token matching mid-word), and
- * collapse the resulting whitespace. Tags are stripped before decoding so a
- * structural `<bold>` is removed while an entity-escaped `&lt;gene&gt;` survives
- * as literal text. Used for Europe PMC abstract snippets before truncation, so
- * the character budget is spent on text rather than markup.
+ * plain text: strip tags, decode character references, drop soft hyphens
+ * (U+00AD — invisible, and it corrupts token matching mid-word), and collapse
+ * the resulting whitespace. Tags are stripped in one pass before decoding, so a
+ * raw `<bold>` and an encoded `&lt;i&gt;` are removed while an escaped
+ * `&lt;gene&gt;` survives as literal text — and nothing a removal joins or a
+ * decode produces is scanned again (`&lt;&lt;i&gt;i&gt;` reads `<i>`). Used for
+ * Europe PMC titles, authors, journals, and abstracts before truncation, so the
+ * character budget is spent on text rather than markup.
  */
 export declare function toDisplayText(text: string): string;
 //# sourceMappingURL=text-helpers.d.ts.map

package/dist/services/ncbi/parsing/text-helpers.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"text-helpers.d.ts","sourceRoot":"","sources":["../../../../src/services/ncbi/parsing/text-helpers.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAiCH;;;;;;;GAOG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAcvD;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAKlD"}
+{"version":3,"file":"text-helpers.d.ts","sourceRoot":"","sources":["../../../../src/services/ncbi/parsing/text-helpers.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAsHH;;;;;;;GAOG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAcvD;AAED;;;;;;;;;;GAUG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAKlD"}