@ncukondo/search-hub 0.12.1 → 0.12.2

package/dist/node_modules/nth-check/lib/parse.js

@@ -0,0 +1,69 @@
+import { __exports as parse } from "../../../_virtual/parse2.js";
+var hasRequiredParse;
+function requireParse() {
+  if (hasRequiredParse) return parse;
+  hasRequiredParse = 1;
+  Object.defineProperty(parse, "__esModule", { value: true });
+  parse.parse = void 0;
+  var whitespace = /* @__PURE__ */ new Set([9, 10, 12, 13, 32]);
+  var ZERO = "0".charCodeAt(0);
+  var NINE = "9".charCodeAt(0);
+  function parse$1(formula) {
+    formula = formula.trim().toLowerCase();
+    if (formula === "even") {
+      return [2, 0];
+    } else if (formula === "odd") {
+      return [2, 1];
+    }
+    var idx = 0;
+    var a = 0;
+    var sign = readSign();
+    var number = readNumber();
+    if (idx < formula.length && formula.charAt(idx) === "n") {
+      idx++;
+      a = sign * (number !== null && number !== void 0 ? number : 1);
+      skipWhitespace();
+      if (idx < formula.length) {
+        sign = readSign();
+        skipWhitespace();
+        number = readNumber();
+      } else {
+        sign = number = 0;
+      }
+    }
+    if (number === null || idx < formula.length) {
+      throw new Error("n-th rule couldn't be parsed ('".concat(formula, "')"));
+    }
+    return [a, sign * number];
+    function readSign() {
+      if (formula.charAt(idx) === "-") {
+        idx++;
+        return -1;
+      }
+      if (formula.charAt(idx) === "+") {
+        idx++;
+      }
+      return 1;
+    }
+    function readNumber() {
+      var start = idx;
+      var value = 0;
+      while (idx < formula.length && formula.charCodeAt(idx) >= ZERO && formula.charCodeAt(idx) <= NINE) {
+        value = value * 10 + (formula.charCodeAt(idx) - ZERO);
+        idx++;
+      }
+      return idx === start ? null : value;
+    }
+    function skipWhitespace() {
+      while (idx < formula.length && whitespace.has(formula.charCodeAt(idx))) {
+        idx++;
+      }
+    }
+  }
+  parse.parse = parse$1;
+  return parse;
+}
+export {
+  requireParse as __require
+};
+//# sourceMappingURL=parse.js.map

package/dist/node_modules/nth-check/lib/parse.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"parse.js","sources":["../../../../node_modules/nth-check/lib/parse.js"],"sourcesContent":["\"use strict\";\n// Following http://www.w3.org/TR/css3-selectors/#nth-child-pseudo\nObject.defineProperty(exports, \"__esModule\", { value: true });\nexports.parse = void 0;\n// Whitespace as per https://www.w3.org/TR/selectors-3/#lex is \" \\t\\r\\n\\f\"\nvar whitespace = new Set([9, 10, 12, 13, 32]);\nvar ZERO = \"0\".charCodeAt(0);\nvar NINE = \"9\".charCodeAt(0);\n/**\n * Parses an expression.\n *\n * @throws An `Error` if parsing fails.\n * @returns An array containing the integer step size and the integer offset of the nth rule.\n * @example nthCheck.parse(\"2n+3\"); // returns [2, 3]\n */\nfunction parse(formula) {\n    formula = formula.trim().toLowerCase();\n    if (formula === \"even\") {\n        return [2, 0];\n    }\n    else if (formula === \"odd\") {\n        return [2, 1];\n    }\n    // Parse [ ['-'|'+']? INTEGER? {N} [ S* ['-'|'+'] S* INTEGER ]?\n    var idx = 0;\n    var a = 0;\n    var sign = readSign();\n    var number = readNumber();\n    if (idx < formula.length && formula.charAt(idx) === \"n\") {\n        idx++;\n        a = sign * (number !== null && number !== void 0 ? number : 1);\n        skipWhitespace();\n        if (idx < formula.length) {\n            sign = readSign();\n            skipWhitespace();\n            number = readNumber();\n        }\n        else {\n            sign = number = 0;\n        }\n    }\n    // Throw if there is anything else\n    if (number === null || idx < formula.length) {\n        throw new Error(\"n-th rule couldn't be parsed ('\".concat(formula, \"')\"));\n    }\n    return [a, sign * number];\n    function readSign() {\n        if (formula.charAt(idx) === \"-\") {\n            idx++;\n            return -1;\n        }\n        if (formula.charAt(idx) === \"+\") {\n            idx++;\n        }\n        return 1;\n    }\n    function readNumber() {\n        var start = idx;\n        var value = 0;\n        while (idx < formula.length &&\n            formula.charCodeAt(idx) >= ZERO &&\n            formula.charCodeAt(idx) <= NINE) {\n            value = value * 10 + (formula.charCodeAt(idx) - ZERO);\n            idx++;\n        }\n        // Return `null` if we didn't read anything.\n        return idx === start ? null : value;\n    }\n    function skipWhitespace() {\n        while (idx < formula.length &&\n            whitespace.has(formula.charCodeAt(idx))) {\n            idx++;\n        }\n    }\n}\nexports.parse = parse;\n//# sourceMappingURL=parse.js.map"],"names":["parse_1","parse"],"mappings":";;;;;AAEA,SAAO,eAAeA,OAAS,cAAc,EAAE,OAAO,MAAM;AAC5DA,QAAA,QAAgB;AAEhB,MAAI,aAAa,oBAAI,IAAI,CAAC,GAAG,IAAI,IAAI,IAAI,EAAE,CAAC;AAC5C,MAAI,OAAO,IAAI,WAAW,CAAC;AAC3B,MAAI,OAAO,IAAI,WAAW,CAAC;AAQ3B,WAASC,QAAM,SAAS;AACpB,cAAU,QAAQ,KAAI,EAAG,YAAW;AACpC,QAAI,YAAY,QAAQ;AACpB,aAAO,CAAC,GAAG,CAAC;AAAA,IACpB,WACa,YAAY,OAAO;AACxB,aAAO,CAAC,GAAG,CAAC;AAAA,IACpB;AAEI,QAAI,MAAM;AACV,QAAI,IAAI;AACR,QAAI,OAAO,SAAQ;AACnB,QAAI,SAAS,WAAU;AACvB,QAAI,MAAM,QAAQ,UAAU,QAAQ,OAAO,GAAG,MAAM,KAAK;AACrD;AACA,UAAI,QAAQ,WAAW,QAAQ,WAAW,SAAS,SAAS;AAC5D,qBAAc;AACd,UAAI,MAAM,QAAQ,QAAQ;AACtB,eAAO,SAAQ;AACf,uBAAc;AACd,iBAAS,WAAU;AAAA,MAC/B,OACa;AACD,eAAO,SAAS;AAAA,MAC5B;AAAA,IACA;AAEI,QAAI,WAAW,QAAQ,MAAM,QAAQ,QAAQ;AACzC,YAAM,IAAI,MAAM,kCAAkC,OAAO,SAAS,IAAI,CAAC;AAAA,IAC/E;AACI,WAAO,CAAC,GAAG,OAAO,MAAM;AACxB,aAAS,WAAW;AAChB,UAAI,QAAQ,OAAO,GAAG,MAAM,KAAK;AAC7B;AACA,eAAO;AAAA,MACnB;AACQ,UAAI,QAAQ,OAAO,GAAG,MAAM,KAAK;AAC7B;AAAA,MACZ;AACQ,aAAO;AAAA,IACf;AACI,aAAS,aAAa;AAClB,UAAI,QAAQ;AACZ,UAAI,QAAQ;AACZ,aAAO,MAAM,QAAQ,UACjB,QAAQ,WAAW,GAAG,KAAK,QAC3B,QAAQ,WAAW,GAAG,KAAK,MAAM;AACjC,gBAAQ,QAAQ,MAAM,QAAQ,WAAW,GAAG,IAAI;AAChD;AAAA,MACZ;AAEQ,aAAO,QAAQ,QAAQ,OAAO;AAAA,IACtC;AACI,aAAS,iBAAiB;AACtB,aAAO,MAAM,QAAQ,UACjB,WAAW,IAAI,QAAQ,WAAW,GAAG,CAAC,GAAG;AACzC;AAAA,MACZ;AAAA,IACA;AAAA,EACA;AACAD,QAAA,QAAgBC;;;","x_google_ignoreList":[0]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ncukondo/search-hub",
-  "version": "0.12.1",
+  "version": "0.12.2",
   "description": "A CLI tool for systematic literature searching across multiple academic databases",
   "type": "module",
   "engines": {
@@ -63,7 +63,7 @@
   "license": "MIT",
   "dependencies": {
     "@iarna/toml": "^3.0.0",
-    "any-ascii": "^0.3.3",
+    "@ncukondo/academic-fulltext": "^0.2.0",
     "cli-progress": "^3.12.0",
     "commander": "^13.0.0",
     "dotenv": "^17.2.3",

package/dist/fulltext/attach-shared.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"attach-shared.d.ts","sourceRoot":"","sources":["../../src/fulltext/attach-shared.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAKH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,YAAY,CAAC;AAC/C,OAAO,KAAK,EAAE,oBAAoB,EAAE,MAAM,yBAAyB,CAAC;AAEpE,iDAAiD;AACjD,eAAO,MAAM,gBAAgB,0CAA2C,CAAC;AAEzE;;;GAGG;AACH,wBAAgB,SAAS,CAAC,IAAI,EAAE,YAAY,EAAE,SAAS,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,MAAM,GAAG,SAAS,CAUhG;AAED,iEAAiE;AACjE,MAAM,WAAW,YAAY;IAC3B,OAAO,EAAE,MAAM,CAAC;IAChB,UAAU,EAAE,MAAM,CAAC;IACnB,IAAI,EAAE,YAAY,GAAG,IAAI,CAAC;CAC3B;AAED;;;GAGG;AACH,wBAAsB,mBAAmB,CAAC,WAAW,EAAE,MAAM,GAAG,OAAO,CAAC,YAAY,EAAE,CAAC,CAwBtF;AAED;;GAEG;AACH,wBAAsB,sBAAsB,CAC1C,UAAU,EAAE,MAAM,EAClB,IAAI,EAAE,YAAY,GACjB,OAAO,CAAC,MAAM,EAAE,CAAC,CAcnB;AAED,0CAA0C;AAC1C,MAAM,WAAW,6BAA6B;IAC5C,2DAA2D;IAC3D,WAAW,EAAE,MAAM,CAAC;IACpB,6CAA6C;IAC7C,SAAS,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAC/B,sDAAsD;IACtD,UAAU,EAAE,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC;IAC/D,+EAA+E;IAC/E,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,wBAAwB;IACxB,UAAU,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,KAAK,IAAI,CAAC;CACvD;AAED;;;GAGG;AACH,wBAAsB,sBAAsB,CAC1C,OAAO,EAAE,6BAA6B,GACrC,OAAO,CAAC,oBAAoB,CAAC,CAsE/B"}

package/dist/fulltext/attach-shared.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"attach-shared.js","sources":["../../src/fulltext/attach-shared.ts"],"sourcesContent":["/**\n * Shared fulltext attach utilities.\n * Used by both the integration attach (register flow) and standalone attach command.\n */\n\nimport { readdir, readFile, access } from 'node:fs/promises';\nimport { constants } from 'node:fs';\nimport { join } from 'node:path';\nimport type { FulltextMeta } from './types.js';\nimport type { FulltextAttachResult } from '../integration/types.js';\n\n/** Files we try to attach, in priority order. */\nexport const ATTACHABLE_FILES = ['fulltext.pdf', 'fulltext.md'] as const;\n\n/**\n * Find the ref ID matching a fulltext meta entry.\n * Tries DOI first, then PMID.\n */\nexport function findRefId(meta: FulltextMeta, refLookup: Map<string, string>): string | undefined {\n  if (meta.doi) {\n    const byDoi = refLookup.get(`doi:${meta.doi}`);\n    if (byDoi) return byDoi;\n  }\n  if (meta.pmid) {\n    const byPmid = refLookup.get(`pmid:${meta.pmid}`);\n    if (byPmid) return byPmid;\n  }\n  return undefined;\n}\n\n/** A single article directory entry with its loaded metadata. */\nexport interface ArticleEntry {\n  dirName: string;\n  articleDir: string;\n  meta: FulltextMeta | null;\n}\n\n/**\n * Scan a fulltext directory and load metadata for all article subdirectories.\n * Returns an empty array if the directory doesn't exist.\n */\nexport async function loadFulltextEntries(fulltextDir: string): Promise<ArticleEntry[]> {\n  let dirNames: string[];\n  try {\n    const dirEntries = await readdir(fulltextDir, { withFileTypes: true });\n    dirNames = dirEntries\n      .filter((e) => e.isDirectory())\n      .map((e) => e.name);\n  } catch {\n    return [];\n  }\n\n  const entries: ArticleEntry[] = [];\n  for (const dirName of dirNames) {\n    const articleDir = join(fulltextDir, dirName);\n    try {\n      const raw = await readFile(join(articleDir, 'meta.json'), 'utf-8');\n      const meta = JSON.parse(raw) as FulltextMeta;\n      entries.push({ dirName, articleDir, meta });\n    } catch {\n      // Will be handled as a failed entry by the caller\n      entries.push({ dirName, articleDir, meta: null });\n    }\n  }\n  return entries;\n}\n\n/**\n * Determine which attachable files exist for an article, verifying on disk.\n */\nexport async function resolveAttachableFiles(\n  articleDir: string,\n  meta: FulltextMeta,\n): Promise<string[]> {\n  const filesToAttach: string[] = [];\n  for (const filename of ATTACHABLE_FILES) {\n    const fileKey = filename === 'fulltext.pdf' ? 'pdf' : 'markdown';\n    if (meta.files[fileKey]) {\n      try {\n        await access(join(articleDir, filename), constants.R_OK);\n        filesToAttach.push(filename);\n      } catch {\n        // File recorded in meta but not on disk — skip\n      }\n    }\n  }\n  return filesToAttach;\n}\n\n/** Options for processFulltextEntries. */\nexport interface ProcessFulltextEntriesOptions {\n  /** Path to the fulltext directory (sessionDir/fulltext) */\n  fulltextDir: string;\n  /** Lookup map from identifiers to ref IDs */\n  refLookup: Map<string, string>;\n  /** Function to attach a single file to a ref entry */\n  attachFile: (refId: string, filePath: string) => Promise<void>;\n  /** If true, skip actual attach calls and just record what would be attached */\n  dryRun?: boolean;\n  /** Progress callback */\n  onProgress?: (current: number, total: number) => void;\n}\n\n/**\n * Shared attach loop: scan fulltext entries, match to refs, and attach files.\n * Both the standalone and integrated attach commands delegate to this function.\n */\nexport async function processFulltextEntries(\n  options: ProcessFulltextEntriesOptions,\n): Promise<FulltextAttachResult> {\n  const { fulltextDir, refLookup, attachFile, dryRun = false, onProgress } = options;\n\n  const result: FulltextAttachResult = {\n    summary: { total: 0, attached: 0, skipped: 0, failed: 0 },\n    attached: [],\n    skipped: [],\n    failed: [],\n  };\n\n  const entries = await loadFulltextEntries(fulltextDir);\n  result.summary.total = entries.length;\n\n  for (let i = 0; i < entries.length; i++) {\n    const entry = entries[i]!;\n    const { dirName, articleDir, meta } = entry;\n\n    if (onProgress) {\n      onProgress(i + 1, entries.length);\n    }\n\n    if (!meta) {\n      result.summary.failed++;\n      result.failed.push({\n        dirName,\n        reason: 'meta_read_error',\n        error: 'Failed to read meta.json',\n      });\n      continue;\n    }\n\n    const refId = findRefId(meta, refLookup);\n    if (!refId) {\n      result.summary.skipped++;\n      result.skipped.push({ dirName, reason: 'not_in_ref' });\n      continue;\n    }\n\n    const filesToAttach = await resolveAttachableFiles(articleDir, meta);\n\n    if (filesToAttach.length === 0) {\n      result.summary.skipped++;\n      result.skipped.push({ dirName, reason: 'no_files' });\n      continue;\n    }\n\n    if (dryRun) {\n      result.summary.attached++;\n      result.attached.push({ refId, files: filesToAttach });\n      continue;\n    }\n\n    try {\n      for (const filename of filesToAttach) {\n        const filePath = join(articleDir, filename);\n        await attachFile(refId, filePath);\n      }\n      result.summary.attached++;\n      result.attached.push({ refId, files: filesToAttach });\n    } catch (error) {\n      result.summary.failed++;\n      result.failed.push({\n        dirName,\n        reason: 'attach_error',\n        error: error instanceof Error ? error.message : 'Unknown error',\n      });\n    }\n  }\n\n  return result;\n}\n"],"names":[],"mappings":";;;AAYO,MAAM,mBAAmB,CAAC,gBAAgB,aAAa;AAMvD,SAAS,UAAU,MAAoB,WAAoD;AAChG,MAAI,KAAK,KAAK;AACZ,UAAM,QAAQ,UAAU,IAAI,OAAO,KAAK,GAAG,EAAE;AAC7C,QAAI,MAAO,QAAO;AAAA,EACpB;AACA,MAAI,KAAK,MAAM;AACb,UAAM,SAAS,UAAU,IAAI,QAAQ,KAAK,IAAI,EAAE;AAChD,QAAI,OAAQ,QAAO;AAAA,EACrB;AACA,SAAO;AACT;AAaA,eAAsB,oBAAoB,aAA8C;AACtF,MAAI;AACJ,MAAI;AACF,UAAM,aAAa,MAAM,QAAQ,aAAa,EAAE,eAAe,MAAM;AACrE,eAAW,WACR,OAAO,CAAC,MAAM,EAAE,YAAA,CAAa,EAC7B,IAAI,CAAC,MAAM,EAAE,IAAI;AAAA,EACtB,QAAQ;AACN,WAAO,CAAA;AAAA,EACT;AAEA,QAAM,UAA0B,CAAA;AAChC,aAAW,WAAW,UAAU;AAC9B,UAAM,aAAa,KAAK,aAAa,OAAO;AAC5C,QAAI;AACF,YAAM,MAAM,MAAM,SAAS,KAAK,YAAY,WAAW,GAAG,OAAO;AACjE,YAAM,OAAO,KAAK,MAAM,GAAG;AAC3B,cAAQ,KAAK,EAAE,SAAS,YAAY,MAAM;AAAA,IAC5C,QAAQ;AAEN,cAAQ,KAAK,EAAE,SAAS,YAAY,MAAM,MAAM;AAAA,IAClD;AAAA,EACF;AACA,SAAO;AACT;AAKA,eAAsB,uBACpB,YACA,MACmB;AACnB,QAAM,gBAA0B,CAAA;AAChC,aAAW,YAAY,kBAAkB;AACvC,UAAM,UAAU,aAAa,iBAAiB,QAAQ;AACtD,QAAI,KAAK,MAAM,OAAO,GAAG;AACvB,UAAI;AACF,cAAM,OAAO,KAAK,YAAY,QAAQ,GAAG,UAAU,IAAI;AACvD,sBAAc,KAAK,QAAQ;AAAA,MAC7B,QAAQ;AAAA,MAER;AAAA,IACF;AAAA,EACF;AACA,SAAO;AACT;AAoBA,eAAsB,uBACpB,SAC+B;AAC/B,QAAM,EAAE,aAAa,WAAW,YAAY,SAAS,OAAO,eAAe;AAE3E,QAAM,SAA+B;AAAA,IACnC,SAAS,EAAE,OAAO,GAAG,UAAU,GAAG,SAAS,GAAG,QAAQ,EAAA;AAAA,IACtD,UAAU,CAAA;AAAA,IACV,SAAS,CAAA;AAAA,IACT,QAAQ,CAAA;AAAA,EAAC;AAGX,QAAM,UAAU,MAAM,oBAAoB,WAAW;AACrD,SAAO,QAAQ,QAAQ,QAAQ;AAE/B,WAAS,IAAI,GAAG,IAAI,QAAQ,QAAQ,KAAK;AACvC,UAAM,QAAQ,QAAQ,CAAC;AACvB,UAAM,EAAE,SAAS,YAAY,KAAA,IAAS;AAEtC,QAAI,YAAY;AACd,iBAAW,IAAI,GAAG,QAAQ,MAAM;AAAA,IAClC;AAEA,QAAI,CAAC,MAAM;AACT,aAAO,QAAQ;AACf,aAAO,OAAO,KAAK;AAAA,QACjB;AAAA,QACA,QAAQ;AAAA,QACR,OAAO;AAAA,MAAA,CACR;AACD;AAAA,IACF;AAEA,UAAM,QAAQ,UAAU,MAAM,SAAS;AACvC,QAAI,CAAC,OAAO;AACV,aAAO,QAAQ;AACf,aAAO,QAAQ,KAAK,EAAE,SAAS,QAAQ,cAAc;AACrD;AAAA,IACF;AAEA,UAAM,gBAAgB,MAAM,uBAAuB,YAAY,IAAI;AAEnE,QAAI,cAAc,WAAW,GAAG;AAC9B,aAAO,QAAQ;AACf,aAAO,QAAQ,KAAK,EAAE,SAAS,QAAQ,YAAY;AACnD;AAAA,IACF;AAEA,QAAI,QAAQ;AACV,aAAO,QAAQ;AACf,aAAO,SAAS,KAAK,EAAE,OAAO,OAAO,eAAe;AACpD;AAAA,IACF;AAEA,QAAI;AACF,iBAAW,YAAY,eAAe;AACpC,cAAM,WAAW,KAAK,YAAY,QAAQ;AAC1C,cAAM,WAAW,OAAO,QAAQ;AAAA,MAClC;AACA,aAAO,QAAQ;AACf,aAAO,SAAS,KAAK,EAAE,OAAO,OAAO,eAAe;AAAA,IACtD,SAAS,OAAO;AACd,aAAO,QAAQ;AACf,aAAO,OAAO,KAAK;AAAA,QACjB;AAAA,QACA,QAAQ;AAAA,QACR,OAAO,iBAAiB,QAAQ,MAAM,UAAU;AAAA,MAAA,CACjD;AAAA,IACH;AAAA,EACF;AAEA,SAAO;AACT;"}

package/dist/fulltext/citation-key.d.ts

@@ -1,15 +0,0 @@
-/**
- * Citation key generation for fulltext directories.
- */
-/**
- * Generate a citation key from author and year.
- * Format: {family-name-lowercase}{year}
- * With collision handling via letter suffixes (a, b, c, ...).
- */
-export declare function generateCitationKey(author: string | undefined, year: string | undefined, existingKeys?: string[]): string;
-/**
- * Generate a directory name from a citation key.
- * Format: {citationKey}-{uuid8}
- */
-export declare function generateDirName(citationKey: string, uuid?: string): string;
-//# sourceMappingURL=citation-key.d.ts.map

package/dist/fulltext/citation-key.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"citation-key.d.ts","sourceRoot":"","sources":["../../src/fulltext/citation-key.ts"],"names":[],"mappings":"AAAA;;GAEG;AAiCH;;;;GAIG;AACH,wBAAgB,mBAAmB,CACjC,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,IAAI,EAAE,MAAM,GAAG,SAAS,EACxB,YAAY,CAAC,EAAE,MAAM,EAAE,GACtB,MAAM,CA4BR;AAED;;;GAGG;AACH,wBAAgB,eAAe,CAAC,WAAW,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,MAAM,GAAG,MAAM,CAI1E"}

package/dist/fulltext/citation-key.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"citation-key.js","sources":["../../src/fulltext/citation-key.ts"],"sourcesContent":["/**\n * Citation key generation for fulltext directories.\n */\n\nimport anyAscii from 'any-ascii';\nimport { randomUUID } from 'node:crypto';\n\n/**\n * Generate a collision suffix: a, b, ..., z, aa, ab, ...\n */\nfunction collisionSuffix(index: number): string {\n  let result = '';\n  let n = index;\n  do {\n    result = String.fromCodePoint(97 + (n % 26)) + result;\n    n = Math.floor(n / 26) - 1;\n  } while (n >= 0);\n  return result;\n}\n\n/** CJK Unified Ideographs range: U+4E00–U+9FFF */\nconst CJK_REGEX = /[\\u4e00-\\u9fff]/;\n\n/**\n * Extract the family name portion from an author string.\n * Handles formats like \"Smith, J.\" → \"Smith\", \"Smith\" → \"Smith\".\n */\nfunction extractFamilyName(author: string): string {\n  const commaIndex = author.indexOf(',');\n  if (commaIndex >= 0) {\n    return author.slice(0, commaIndex).trim();\n  }\n  return author.trim();\n}\n\n/**\n * Generate a citation key from author and year.\n * Format: {family-name-lowercase}{year}\n * With collision handling via letter suffixes (a, b, c, ...).\n */\nexport function generateCitationKey(\n  author: string | undefined,\n  year: string | undefined,\n  existingKeys?: string[],\n): string {\n  // Extract and normalize author\n  const rawFamily = author?.trim() ? extractFamilyName(author) : 'unknown';\n\n  // CJK characters cannot be accurately transliterated to the correct reading\n  // (any-ascii maps them to Chinese pinyin, not Japanese romaji etc.)\n  // Fall back to 'unknown' for names containing CJK ideographs.\n  const normalizedFamily = CJK_REGEX.test(rawFamily)\n    ? 'unknown'\n    : anyAscii(rawFamily).toLowerCase().replace(/[^a-z]/g, '') || 'unknown';\n\n  // Normalize year\n  const normalizedYear = year?.trim() || '0000';\n\n  const baseKey = `${normalizedFamily}${normalizedYear}`;\n\n  // Handle collisions\n  if (!existingKeys || !existingKeys.includes(baseKey)) {\n    return baseKey;\n  }\n\n  // Find the first available suffix\n  for (let i = 0; ; i++) {\n    const candidateKey = `${baseKey}${collisionSuffix(i)}`;\n    if (!existingKeys.includes(candidateKey)) {\n      return candidateKey;\n    }\n  }\n}\n\n/**\n * Generate a directory name from a citation key.\n * Format: {citationKey}-{uuid8}\n */\nexport function generateDirName(citationKey: string, uuid?: string): string {\n  const id = uuid ?? randomUUID();\n  const uuid8 = id.slice(0, 8);\n  return `${citationKey}-${uuid8}`;\n}\n"],"names":[],"mappings":";;AAUA,SAAS,gBAAgB,OAAuB;AAC9C,MAAI,SAAS;AACb,MAAI,IAAI;AACR,KAAG;AACD,aAAS,OAAO,cAAc,KAAM,IAAI,EAAG,IAAI;AAC/C,QAAI,KAAK,MAAM,IAAI,EAAE,IAAI;AAAA,EAC3B,SAAS,KAAK;AACd,SAAO;AACT;AAGA,MAAM,YAAY;AAMlB,SAAS,kBAAkB,QAAwB;AACjD,QAAM,aAAa,OAAO,QAAQ,GAAG;AACrC,MAAI,cAAc,GAAG;AACnB,WAAO,OAAO,MAAM,GAAG,UAAU,EAAE,KAAA;AAAA,EACrC;AACA,SAAO,OAAO,KAAA;AAChB;AAOO,SAAS,oBACd,QACA,MACA,cACQ;AAER,QAAM,YAAY,QAAQ,KAAA,IAAS,kBAAkB,MAAM,IAAI;AAK/D,QAAM,mBAAmB,UAAU,KAAK,SAAS,IAC7C,YACA,SAAS,SAAS,EAAE,YAAA,EAAc,QAAQ,WAAW,EAAE,KAAK;AAGhE,QAAM,iBAAiB,MAAM,KAAA,KAAU;AAEvC,QAAM,UAAU,GAAG,gBAAgB,GAAG,cAAc;AAGpD,MAAI,CAAC,gBAAgB,CAAC,aAAa,SAAS,OAAO,GAAG;AACpD,WAAO;AAAA,EACT;AAGA,WAAS,IAAI,KAAK,KAAK;AACrB,UAAM,eAAe,GAAG,OAAO,GAAG,gBAAgB,CAAC,CAAC;AACpD,QAAI,CAAC,aAAa,SAAS,YAAY,GAAG;AACxC,aAAO;AAAA,IACT;AAAA,EACF;AACF;AAMO,SAAS,gBAAgB,aAAqB,MAAuB;AAC1E,QAAM,KAAK,QAAQ,WAAA;AACnB,QAAM,QAAQ,GAAG,MAAM,GAAG,CAAC;AAC3B,SAAO,GAAG,WAAW,IAAI,KAAK;AAChC;"}

package/dist/fulltext/convert/index.d.ts

@@ -1,20 +0,0 @@
-/**
- * Conversion orchestrator for PMC XML to Markdown.
- *
- * Ties together the JATS parser and Markdown writer with file I/O.
- */
-export interface ConvertResult {
-    success: boolean;
-    error?: string;
-    title?: string;
-    sections?: number;
-    references?: number;
-}
-/**
- * Convert a PMC JATS XML file to Markdown.
- *
- * Reads the XML, parses it into a JatsDocument, writes Markdown,
- * and optionally updates meta.json.
- */
-export declare function convertPmcXmlToMarkdown(xmlPath: string, mdPath: string, metaPath?: string): Promise<ConvertResult>;
-//# sourceMappingURL=index.d.ts.map

package/dist/fulltext/convert/index.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/fulltext/convert/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAQH,MAAM,WAAW,aAAa;IAC5B,OAAO,EAAE,OAAO,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB;AAED;;;;;GAKG;AACH,wBAAsB,uBAAuB,CAC3C,OAAO,EAAE,MAAM,EACf,MAAM,EAAE,MAAM,EACd,QAAQ,CAAC,EAAE,MAAM,GAChB,OAAO,CAAC,aAAa,CAAC,CAqDxB"}

package/dist/fulltext/convert/index.js

@@ -1,50 +0,0 @@
-import { readFile, writeFile, stat } from "node:fs/promises";
-import { parseJatsMetadata, parseJatsBody, parseJatsReferences, parseJatsBackMatter } from "./jats-parser.js";
-import { writeMarkdown } from "./markdown-writer.js";
-async function convertPmcXmlToMarkdown(xmlPath, mdPath, metaPath) {
-  try {
-    const xml = await readFile(xmlPath, "utf-8");
-    const metadata = parseJatsMetadata(xml);
-    const sections = parseJatsBody(xml);
-    const references = parseJatsReferences(xml);
-    const backMatter = parseJatsBackMatter(xml);
-    const doc = { metadata, sections, references };
-    if (backMatter.acknowledgments) doc.acknowledgments = backMatter.acknowledgments;
-    if (backMatter.appendices) doc.appendices = backMatter.appendices;
-    if (backMatter.footnotes) doc.footnotes = backMatter.footnotes;
-    if (backMatter.floats) doc.floats = backMatter.floats;
-    if (backMatter.notes) doc.notes = backMatter.notes;
-    const md = writeMarkdown(doc);
-    await writeFile(mdPath, md, "utf-8");
-    if (metaPath) {
-      try {
-        await stat(metaPath);
-        const metaRaw = await readFile(metaPath, "utf-8");
-        const meta = JSON.parse(metaRaw);
-        const mdStat = await stat(mdPath);
-        meta.files.markdown = {
-          filename: "fulltext.md",
-          source: "conversion",
-          retrievedAt: (/* @__PURE__ */ new Date()).toISOString(),
-          size: mdStat.size,
-          convertedFrom: "fulltext.xml"
-        };
-        await writeFile(metaPath, JSON.stringify(meta, null, 2) + "\n", "utf-8");
-      } catch {
-      }
-    }
-    const result = { success: true };
-    result.title = metadata.title;
-    result.sections = sections.length;
-    result.references = references.length;
-    return result;
-  } catch (err) {
-    const result = { success: false };
-    result.error = err instanceof Error ? err.message : String(err);
-    return result;
-  }
-}
-export {
-  convertPmcXmlToMarkdown
-};
-//# sourceMappingURL=index.js.map

package/dist/fulltext/convert/index.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"index.js","sources":["../../../src/fulltext/convert/index.ts"],"sourcesContent":["/**\n * Conversion orchestrator for PMC XML to Markdown.\n *\n * Ties together the JATS parser and Markdown writer with file I/O.\n */\n\nimport { readFile, writeFile, stat } from 'node:fs/promises';\nimport { parseJatsMetadata, parseJatsBody, parseJatsReferences, parseJatsBackMatter } from './jats-parser.js';\nimport { writeMarkdown } from './markdown-writer.js';\nimport type { JatsDocument } from './types.js';\nimport type { FulltextMeta } from '../types.js';\n\nexport interface ConvertResult {\n  success: boolean;\n  error?: string;\n  title?: string;\n  sections?: number;\n  references?: number;\n}\n\n/**\n * Convert a PMC JATS XML file to Markdown.\n *\n * Reads the XML, parses it into a JatsDocument, writes Markdown,\n * and optionally updates meta.json.\n */\nexport async function convertPmcXmlToMarkdown(\n  xmlPath: string,\n  mdPath: string,\n  metaPath?: string,\n): Promise<ConvertResult> {\n  try {\n    const xml = await readFile(xmlPath, 'utf-8');\n\n    // Parse\n    const metadata = parseJatsMetadata(xml);\n    const sections = parseJatsBody(xml);\n    const references = parseJatsReferences(xml);\n    const backMatter = parseJatsBackMatter(xml);\n\n    const doc: JatsDocument = { metadata, sections, references };\n    if (backMatter.acknowledgments) doc.acknowledgments = backMatter.acknowledgments;\n    if (backMatter.appendices) doc.appendices = backMatter.appendices;\n    if (backMatter.footnotes) doc.footnotes = backMatter.footnotes;\n    if (backMatter.floats) doc.floats = backMatter.floats;\n    if (backMatter.notes) doc.notes = backMatter.notes;\n\n    // Write Markdown\n    const md = writeMarkdown(doc);\n    await writeFile(mdPath, md, 'utf-8');\n\n    // Update meta.json if path provided and file exists\n    if (metaPath) {\n      try {\n        await stat(metaPath);\n        const metaRaw = await readFile(metaPath, 'utf-8');\n        const meta = JSON.parse(metaRaw) as FulltextMeta;\n        const mdStat = await stat(mdPath);\n\n        meta.files.markdown = {\n          filename: 'fulltext.md',\n          source: 'conversion',\n          retrievedAt: new Date().toISOString(),\n          size: mdStat.size,\n          convertedFrom: 'fulltext.xml',\n        };\n\n        await writeFile(metaPath, JSON.stringify(meta, null, 2) + '\\n', 'utf-8');\n      } catch {\n        // meta.json doesn't exist or can't be read, skip update\n      }\n    }\n\n    const result: ConvertResult = { success: true };\n    result.title = metadata.title;\n    result.sections = sections.length;\n    result.references = references.length;\n    return result;\n  } catch (err) {\n    const result: ConvertResult = { success: false };\n    result.error = err instanceof Error ? err.message : String(err);\n    return result;\n  }\n}\n"],"names":[],"mappings":";;;AA0BA,eAAsB,wBACpB,SACA,QACA,UACwB;AACxB,MAAI;AACF,UAAM,MAAM,MAAM,SAAS,SAAS,OAAO;AAG3C,UAAM,WAAW,kBAAkB,GAAG;AACtC,UAAM,WAAW,cAAc,GAAG;AAClC,UAAM,aAAa,oBAAoB,GAAG;AAC1C,UAAM,aAAa,oBAAoB,GAAG;AAE1C,UAAM,MAAoB,EAAE,UAAU,UAAU,WAAA;AAChD,QAAI,WAAW,gBAAiB,KAAI,kBAAkB,WAAW;AACjE,QAAI,WAAW,WAAY,KAAI,aAAa,WAAW;AACvD,QAAI,WAAW,UAAW,KAAI,YAAY,WAAW;AACrD,QAAI,WAAW,OAAQ,KAAI,SAAS,WAAW;AAC/C,QAAI,WAAW,MAAO,KAAI,QAAQ,WAAW;AAG7C,UAAM,KAAK,cAAc,GAAG;AAC5B,UAAM,UAAU,QAAQ,IAAI,OAAO;AAGnC,QAAI,UAAU;AACZ,UAAI;AACF,cAAM,KAAK,QAAQ;AACnB,cAAM,UAAU,MAAM,SAAS,UAAU,OAAO;AAChD,cAAM,OAAO,KAAK,MAAM,OAAO;AAC/B,cAAM,SAAS,MAAM,KAAK,MAAM;AAEhC,aAAK,MAAM,WAAW;AAAA,UACpB,UAAU;AAAA,UACV,QAAQ;AAAA,UACR,cAAa,oBAAI,KAAA,GAAO,YAAA;AAAA,UACxB,MAAM,OAAO;AAAA,UACb,eAAe;AAAA,QAAA;AAGjB,cAAM,UAAU,UAAU,KAAK,UAAU,MAAM,MAAM,CAAC,IAAI,MAAM,OAAO;AAAA,MACzE,QAAQ;AAAA,MAER;AAAA,IACF;AAEA,UAAM,SAAwB,EAAE,SAAS,KAAA;AACzC,WAAO,QAAQ,SAAS;AACxB,WAAO,WAAW,SAAS;AAC3B,WAAO,aAAa,WAAW;AAC/B,WAAO;AAAA,EACT,SAAS,KAAK;AACZ,UAAM,SAAwB,EAAE,SAAS,MAAA;AACzC,WAAO,QAAQ,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAC9D,WAAO;AAAA,EACT;AACF;"}

package/dist/fulltext/convert/jats-parser.d.ts

@@ -1,36 +0,0 @@
-import { JatsMetadata, JatsSection, JatsReference, JatsFootnote, BackMatterNote, BlockElement } from './types.js';
-/**
- * Parse JATS XML front matter to extract article metadata.
- */
-export declare function parseJatsMetadata(xml: string): JatsMetadata;
-/**
- * Parse a <table-wrap> element into a table block.
- * Exported for standalone use and used internally by parseBlockContent.
- */
-export declare function parseJatsTable(xml: string): {
-    caption?: string;
-    headers: string[];
-    rows: string[][];
-};
-/**
- * Parse JATS XML body to extract sections and content.
- */
-export declare function parseJatsBody(xml: string): JatsSection[];
-/**
- * Parse JATS XML back matter to extract references.
- */
-export declare function parseJatsReferences(xml: string): JatsReference[];
-/** Result of parsing back matter and floats-group. */
-export interface BackMatterResult {
-    acknowledgments?: string;
-    appendices?: JatsSection[];
-    footnotes?: JatsFootnote[];
-    floats?: BlockElement[];
-    notes?: BackMatterNote[];
-}
-/**
- * Parse JATS XML back matter sections (ack, app-group, fn-group)
- * and top-level floats-group.
- */
-export declare function parseJatsBackMatter(xml: string): BackMatterResult;
-//# sourceMappingURL=jats-parser.d.ts.map

package/dist/fulltext/convert/jats-parser.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"jats-parser.d.ts","sourceRoot":"","sources":["../../../src/fulltext/convert/jats-parser.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAGH,OAAO,KAAK,EAEV,YAAY,EACZ,WAAW,EACX,aAAa,EACb,YAAY,EACZ,cAAc,EACd,YAAY,EAEb,MAAM,YAAY,CAAC;AAuMpB;;GAEG;AACH,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,MAAM,GAAG,YAAY,CA2M3D;AAuLD;;;GAGG;AACH,wBAAgB,cAAc,CAAC,GAAG,EAAE,MAAM,GAAG;IAC3C,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,IAAI,EAAE,MAAM,EAAE,EAAE,CAAC;CAClB,CAQA;AA4PD;;GAEG;AACH,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,GAAG,WAAW,EAAE,CAwBxD;AA8JD;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,aAAa,EAAE,CAkDhE;AAID,sDAAsD;AACtD,MAAM,WAAW,gBAAgB;IAC/B,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,UAAU,CAAC,EAAE,WAAW,EAAE,CAAC;IAC3B,SAAS,CAAC,EAAE,YAAY,EAAE,CAAC;IAC3B,MAAM,CAAC,EAAE,YAAY,EAAE,CAAC;IACxB,KAAK,CAAC,EAAE,cAAc,EAAE,CAAC;CAC1B;AAED;;;GAGG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,gBAAgB,CAuIjE"}