@cyanheads/pubmed-mcp-server/dist/services/ncbi/parsing/article-parser.d.ts

Search PubMed/Europe PMC, fetch articles and full text (PMC/EPMC/Unpaywall), citations, MeSH terms via MCP. STDIO or Streamable HTTP.

/**
 * @fileoverview Helper functions for parsing detailed PubMed Article XML structures,
 * typically from EFetch results.
 * @module src/services/ncbi/parsing/article-parser
 */
import type { ParsedArticle, ParsedArticleAuthor, ParsedArticleDate, ParsedGrant, ParsedJournalInfo, ParsedMeshTerm, ParseFullArticleOptions, XmlArticle, XmlArticleIdList, XmlAuthorList, XmlGrantList, XmlJournal, XmlKeywordList, XmlMedlineCitation, XmlMeshHeadingList, XmlPublicationTypeList, XmlPubmedArticle } from '../types.js';
/**
 * Result of extracting authors with deduplicated affiliations.
 */
export interface ExtractedAuthors {
    affiliations: string[];
    authors: ParsedArticleAuthor[];
}
/**
 * Extracts and formats author information from XML, deduplicating affiliations.
 * Affiliations are collected into a single array; each author references them by index.
 * This avoids repeating identical institutional strings per-author (common in multi-center papers).
 * @param authorListXml - The XML AuthorList element.
 * @returns Authors and a deduplicated affiliations list.
 */
export declare function extractAuthors(authorListXml?: XmlAuthorList): ExtractedAuthors;
/**
 * Extracts and formats journal information from XML.
 * @param journalXml - The XML Journal element from an Article.
 * @param articleXml - The XML Article element (for Pagination).
 * @returns Formatted journal information.
 */
export declare function extractJournalInfo(journalXml?: XmlJournal, articleXml?: XmlArticle): ParsedJournalInfo | undefined;
/**
 * Extracts and formats MeSH terms from XML.
 * @param meshHeadingListXml - The XML MeshHeadingList element.
 * @returns An array of formatted MeSH term objects.
 */
export declare function extractMeshTerms(meshHeadingListXml?: XmlMeshHeadingList): ParsedMeshTerm[];
/**
 * Extracts and formats grant information from XML.
 * @param grantListXml - The XML GrantList element.
 * @returns An array of formatted grant objects.
 */
export declare function extractGrants(grantListXml?: XmlGrantList): ParsedGrant[];
/**
 * Extracts DOI from various possible locations in the XML.
 * Prioritizes ELocationID with ValidYN='Y', then any ELocationID, then ArticleIdList,
 * then PubmedData.ArticleIdList.
 * @param articleXml - The XML Article element.
 * @param pubmedDataArticleIdList - Optional ArticleIdList from PubmedData (sibling of MedlineCitation).
 * @returns The DOI string or undefined.
 */
export declare function extractDoi(articleXml?: XmlArticle, pubmedDataArticleIdList?: XmlArticleIdList): string | undefined;
/**
 * Extracts PMC ID from ArticleIdList locations in the XML.
 * Searches Article.ArticleIdList and PubmedData.ArticleIdList for IdType='pmc'.
 * @param articleXml - The XML Article element.
 * @param pubmedDataArticleIdList - Optional ArticleIdList from PubmedData.
 * @returns The PMC ID string (e.g. 'PMC1234567') or undefined.
 */
export declare function extractPmcId(articleXml?: XmlArticle, pubmedDataArticleIdList?: XmlArticleIdList): string | undefined;
/**
 * Extracts publication types from XML.
 * @param publicationTypeListXml - The XML PublicationTypeList element.
 * @returns An array of publication type strings.
 */
export declare function extractPublicationTypes(publicationTypeListXml?: XmlPublicationTypeList): string[];
/**
 * Extracts keywords from XML. Handles single or multiple KeywordList elements.
 * @param keywordListsXml - The XML KeywordList element or an array of them.
 * @returns An array of keyword strings.
 */
export declare function extractKeywords(keywordListsXml?: XmlKeywordList[] | XmlKeywordList): string[];
/**
 * Extracts abstract text from XML. Handles structured abstracts by concatenating sections.
 * If AbstractText is an array, joins them. If it's a single object/string, uses it directly.
 * Prefixes with Label if present.
 * @param abstractXml - The XML Abstract element from an Article.
 * @returns The abstract text string, or undefined if not found or empty.
 */
export declare function extractAbstractText(abstractXml?: XmlArticle['Abstract']): string | undefined;
/**
 * Extracts PMID from MedlineCitation.
 * @param medlineCitationXml - The XML MedlineCitation element.
 * @returns The PMID string or undefined.
 */
export declare function extractPmid(medlineCitationXml?: XmlMedlineCitation): string | undefined;
/**
 * Extracts article dates from XML.
 * @param articleXml - The XML Article element.
 * @returns An array of parsed article dates.
 */
export declare function extractArticleDates(articleXml?: XmlArticle): ParsedArticleDate[];
/**
 * Parses a full PubMed article XML structure into a ParsedArticle object,
 * combining all individual extractors.
 * @param xmlArticle - The raw XML PubmedArticle element.
 * @param options - Options controlling which optional sections to include.
 * @returns A fully parsed article object.
 */
export declare function parseFullArticle(xmlArticle: XmlPubmedArticle, options?: ParseFullArticleOptions): ParsedArticle;
//# sourceMappingURL=article-parser.d.ts.map