glossarist 0.4.0 → 0.4.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "glossarist",
-  "version": "0.4.0",
+  "version": "0.4.2",
   "description": "JavaScript SDK for Glossarist GCR packages — read, write, validate, and manage terminology concepts",
   "type": "module",
   "main": "src/index.js",
@@ -68,7 +68,7 @@
     "access": "public"
   },
   "dependencies": {
-    "js-yaml": "^4.1.0",
+    "js-yaml": "^5.0.0",
     "jszip": "^3.10.1"
   },
   "devDependencies": {

package/src/concept-collection.js

@@ -86,22 +86,14 @@ export class ConceptCollection {
         for (const t of lc.terms) {
           if ((t.designation ?? '').toLowerCase().includes(q)) return true;
         }
-        if (this._searchDefs(lc.definitions, q)) return true;
-        if (this._searchDefs(lc.notes, q)) return true;
-        if (this._searchDefs(lc.examples, q)) return true;
-        if (this._searchDefs(lc.annotations, q)) return true;
+      }
+      for (const { text } of c.walkTexts()) {
+        if (text.toLowerCase().includes(q)) return true;
       }
       return false;
     }));
   }
 
-  _searchDefs(arr, q) {
-    for (const d of arr) {
-      if ((d.content ?? '').toLowerCase().includes(q)) return true;
-    }
-    return false;
-  }
-
   allLanguages() {
     const set = new Set();
     for (const c of this[_items]) {

package/src/concept-parser.js

@@ -1,4 +1,4 @@
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { Concept } from './models/concept.js';
 import { RelatedConcept } from './models/related-concept.js';
 import { InvalidInputError, YamlParseError } from './errors.js';

package/src/concept-reader.js

@@ -1,6 +1,6 @@
 import fs from 'fs';
 import path from 'path';
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { conceptParser } from './concept-parser.js';
 import { naturalSort } from './sort.js';
 import { InvalidInputError } from './errors.js';

package/src/concept-serializer.js

@@ -1,4 +1,4 @@
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 
 const DUMP_OPTS = { lineWidth: -1, noRefs: true, sortKeys: false, skipInvalid: true };
 

package/src/gcr-reader.js

@@ -1,5 +1,5 @@
 import JSZip from 'jszip';
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { conceptParser } from './concept-parser.js';
 import { InvalidInputError } from './errors.js';
 import { COMPILED_FORMATS, parseCompiledPath, compiledPath } from './compiled-format.js';
@@ -37,7 +37,7 @@ const BASE64_RE = /^[A-Za-z0-9+/]{100,}={0,2}$/;
  * @property {Term[]} terms - term designations
  * @property {Definition[]} [definition] - definition content
  * @property {{ content: string }[]} [notes] - editorial notes
- * @property {{ content: string }[]} [examples] - usage examples
+ * @property {Array<{ content: string, examples?: object[], sources?: object[] }>} [examples] - usage examples (may themselves nest examples)
  * @property {Source[]} [sources] - bibliographic sources
  * @property {string} [entry_status] - e.g. 'valid', 'draft'
  * @property {string} [normative_status] - e.g. 'preferred', 'admitted'

package/src/models/bibliography-data.js

@@ -1,4 +1,4 @@
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { GlossaristModel } from './base.js';
 import { BibliographyEntry } from './bibliography-entry.js';
 

package/src/models/concept.js

@@ -131,6 +131,23 @@ export class Concept extends GlossaristModel {
       r => FormulaReference.fromJSON(r));
   }
 
+  /**
+   * Yield every content-text fragment in this concept across all
+   * localizations, recursing through nested examples. Each yielded
+   * `{ text, source }` carries a dotted path rooted at
+   * `localizations.<lang>.<slot>[i]...content`.
+   *
+   * Designations are not included; they live on `LocalizedConcept.terms`
+   * and have a different shape.
+   */
+  *walkTexts() {
+    for (const lang of this.languages) {
+      const lc = this.localization(lang);
+      if (!lc) continue;
+      yield* lc.walkTexts(`localizations.${lang}`);
+    }
+  }
+
   toJSON() {
     const obj = { id: this.id };
     if (this.term != null) obj.term = this.term;

package/src/models/detailed-definition.js

@@ -5,19 +5,43 @@ export class DetailedDefinition extends GlossaristModel {
   constructor(data = {}) {
     super();
     this.content = data.content ?? '';
-    this.sources = (data.sources ?? []).map(
-      s => s instanceof Citation ? s : new Citation(s)
-    );
+    this._rawSources = data.sources ?? [];
+    this._rawExamples = data.examples ?? [];
+    this._sources = null;
+    this._examples = null;
+  }
+
+  get sources() {
+    return this._lazy('_sources', '_rawSources',
+      s => s instanceof Citation ? s : new Citation(s));
+  }
+
+  get examples() {
+    return this._lazy('_examples', '_rawExamples',
+      e => e instanceof DetailedDefinition ? e : new DetailedDefinition(e));
   }
 
   toJSON() {
     const obj = { content: this.content };
-    if (this.sources.length > 0) {
-      obj.sources = this.sources.map(s => s.toJSON());
-    }
+    this._serialize(obj, 'sources', '_sources', '_rawSources');
+    this._serialize(obj, 'examples', '_examples', '_rawExamples');
     return obj;
   }
 
+  /**
+   * Yield this definition's content and the content of every nested
+   * example (recursively). Each item carries `{ text, source }` where
+   * `source` is `<path>.content` rooted at the `path` argument.
+   */
+  *walkTexts(path) {
+    if (typeof this.content === 'string' && this.content.length > 0) {
+      yield { text: this.content, source: `${path}.content` };
+    }
+    for (let i = 0; i < this.examples.length; i++) {
+      yield* this.examples[i].walkTexts(`${path}.examples[${i}]`);
+    }
+  }
+
   static fromJSON(data) {
     return new DetailedDefinition(data);
   }

package/src/models/gcr-metadata.js

@@ -1,4 +1,4 @@
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { GlossaristModel } from './base.js';
 import { GcrStatistics } from './gcr-statistics.js';
 

package/src/models/index.d.ts

@@ -65,6 +65,7 @@ export class Concept extends GlossaristModel {
   definition(lang: string): string | null;
   setLocalization(lang: string, lc: LocalizedConcept | Record<string, unknown>): this;
   hasLocalization(lang: string): boolean;
+  walkTexts(): Iterable<{ text: string; source: string }>;
   static fromJSON(data: Record<string, unknown>): Concept;
 }
 
@@ -96,6 +97,7 @@ export class LocalizedConcept extends GlossaristModel {
   readonly related: RelatedConcept[];
   readonly primaryDesignation: string | null;
   readonly primaryDefinition: string | null;
+  walkTexts(basePath: string): Iterable<{ text: string; source: string }>;
   static fromJSON(data: Record<string, unknown>): LocalizedConcept;
 }
 
@@ -243,6 +245,9 @@ export class ConceptDate extends GlossaristModel {
 export class DetailedDefinition extends GlossaristModel {
   readonly content: string;
   readonly sources: ConceptSource[];
+  readonly examples: DetailedDefinition[];
+  walkTexts(path: string): Iterable<{ text: string; source: string }>;
+  static fromJSON(data: Record<string, unknown>): DetailedDefinition;
 }
 
 export class NonVerbRep extends GlossaristModel {

package/src/models/localized-concept.js

@@ -99,6 +99,27 @@ export class LocalizedConcept extends GlossaristModel {
     return this.definitions[0]?.content ?? null;
   }
 
+  *_textSlots() {
+    yield ['definitions', this.definitions];
+    yield ['notes', this.notes];
+    yield ['examples', this.examples];
+    yield ['annotations', this.annotations];
+  }
+
+  /**
+   * Yield every content-text fragment in this localization, recursing
+   * through nested examples. `basePath` prefixes every emitted
+   * `source` path; pass `localizations.<lang>` to get paths consistent
+   * with the rest of the codebase. Designations are not included.
+   */
+  *walkTexts(basePath) {
+    for (const [name, arr] of this._textSlots()) {
+      for (let i = 0; i < arr.length; i++) {
+        yield* arr[i].walkTexts(`${basePath}.${name}[${i}]`);
+      }
+    }
+  }
+
   toJSON() {
     const obj = {};
     if (this.languageCode) obj.language_code = this.languageCode;

package/src/reference-resolver.js

@@ -101,8 +101,7 @@ export class ReferenceResolver {
         }
       }
 
-      const texts = this._collectTexts(lc, lang);
-      for (const { text, source } of texts) {
+      for (const { text, source } of lc.walkTexts(`localizations.${lang}`)) {
         for (const ref of this._extractFromText(text, source, concept)) {
           refs.push(ref);
         }
@@ -122,35 +121,6 @@ export class ReferenceResolver {
     });
   }
 
-  _collectTexts(lc, lang) {
-    const out = [];
-    for (let i = 0; i < lc.definitions.length; i++) {
-      const content = lc.definitions[i]?.content;
-      if (typeof content === 'string') {
-        out.push({ text: content, source: `localizations.${lang}.definitions[${i}].content` });
-      }
-    }
-    for (let i = 0; i < lc.notes.length; i++) {
-      const content = lc.notes[i]?.content;
-      if (content) {
-        out.push({ text: content, source: `localizations.${lang}.notes[${i}].content` });
-      }
-    }
-    for (let i = 0; i < lc.examples.length; i++) {
-      const content = lc.examples[i]?.content;
-      if (typeof content === 'string') {
-        out.push({ text: content, source: `localizations.${lang}.examples[${i}].content` });
-      }
-    }
-    for (let i = 0; i < lc.annotations.length; i++) {
-      const content = lc.annotations[i]?.content;
-      if (typeof content === 'string') {
-        out.push({ text: content, source: `localizations.${lang}.annotations[${i}].content` });
-      }
-    }
-    return out;
-  }
-
   _extractFromText(text, source, concept) {
     const refs = [];
     const re = /\{\{([^{}]*?)\}\}/g;

package/src/v1-reader.js

@@ -1,6 +1,6 @@
 import fs from 'fs';
 import path from 'path';
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { Concept } from './models/concept.js';
 import { InvalidInputError } from './errors.js';
 

package/src/validators/gcr-validator.js

@@ -1,4 +1,4 @@
-import yaml from 'js-yaml';
+import * as yaml from 'js-yaml';
 import { DATASET_ASSETS } from '../dataset-asset.js';
 import { ValidationResult } from './validation-result.js';
 

package/src/validators/v3-rules.js

@@ -162,36 +162,14 @@ const CITE_MENTION_RE = /\{\{\s*cite:([^,}\s]+)[^}]*?\}\}/g;
 
 function _findCiteMentions(concept) {
   const mentions = [];
-  const walkText = (text, source) => {
-    if (typeof text !== 'string' || text.length === 0) return;
+  for (const { text, source } of concept.walkTexts()) {
+    if (typeof text !== 'string' || text.length === 0) continue;
     CITE_MENTION_RE.lastIndex = 0;
     let m;
     while ((m = CITE_MENTION_RE.exec(text)) !== null) {
       mentions.push({ key: m[1].trim(), source });
     }
-  };
-
-  for (const lang of concept.languages) {
-    const lc = concept.localization(lang);
-    if (!lc) continue;
-
-    for (let i = 0; lc.definitions[i]; i++) {
-      walkText(lc.definitions[i]?.content, `localizations.${lang}.definitions[${i}].content`);
-    }
-    for (let i = 0; lc.notes[i]; i++) {
-      const content = typeof lc.notes[i] === 'object'
-        ? (lc.notes[i]?.content ?? '')
-        : String(lc.notes[i] ?? '');
-      walkText(content, `localizations.${lang}.notes[${i}].content`);
-    }
-    for (let i = 0; lc.examples[i]; i++) {
-      walkText(lc.examples[i]?.content, `localizations.${lang}.examples[${i}].content`);
-    }
-    for (let i = 0; lc.annotations[i]; i++) {
-      walkText(lc.annotations[i]?.content, `localizations.${lang}.annotations[${i}].content`);
-    }
   }
-
   return mentions;
 }
 
@@ -297,25 +275,8 @@ function _findNvrMentions(concept) {
     }
   };
 
-  for (const lang of concept.languages) {
-    const lc = concept.localization(lang);
-    if (!lc) continue;
-    for (let i = 0; i < lc.definitions.length; i++) {
-      walkText(lc.definitions[i]?.content,
-        `localizations.${lang}.definitions[${i}].content`);
-    }
-    for (let i = 0; i < lc.notes.length; i++) {
-      walkText(lc.notes[i]?.content,
-        `localizations.${lang}.notes[${i}].content`);
-    }
-    for (let i = 0; i < lc.examples.length; i++) {
-      walkText(lc.examples[i]?.content,
-        `localizations.${lang}.examples[${i}].content`);
-    }
-    for (let i = 0; i < lc.annotations.length; i++) {
-      walkText(lc.annotations[i]?.content,
-        `localizations.${lang}.annotations[${i}].content`);
-    }
+  for (const { text, source } of concept.walkTexts()) {
+    walkText(text, source);
   }
   return mentions;
 }