@whiteslove/parsing-lexicon 0.5.11 → 0.5.12

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@whiteslove/parsing-lexicon",
3
- "version": "0.5.11",
3
+ "version": "0.5.12",
4
4
  "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
5
5
  "repository": {
6
6
  "type": "git",
@@ -130,6 +130,7 @@ export function parseExperience(value) {
130
130
 
131
131
  const range = text.match(RANGE_RE);
132
132
  if (range) {
133
+ if (/(?:возраст|вік|age|yosh|ёш|vârsta|varsta)/iu.test(range[0]) && !/(?:опыт|experience|досвід|tajriba|тәжірибе)/iu.test(range[0])) return null;
133
134
  const minYears = Number(range[1].replace(',', '.'));
134
135
  const maxYears = Number(range[2].replace(',', '.'));
135
136
  return Object.freeze({ requirement: modifier || 'required', minYears: Math.min(minYears, maxYears), maxYears: Math.max(minYears, maxYears) });
@@ -137,6 +138,7 @@ export function parseExperience(value) {
137
138
 
138
139
  const single = text.match(MIN_RE);
139
140
  if (single) {
141
+ if (/(?:возраст|вік|age|yosh|ёш|vârsta|varsta)/iu.test(single[0]) && !/(?:опыт|experience|досвід|tajriba|тәжірибе)/iu.test(single[0])) return null;
140
142
  const years = Number(single[1].replace(',', '.'));
141
143
  const isExact = !/(?:от|не\s+менее|minimum|min\.?|at\s+least|від|kamida|кемінде|\+)/iu.test(single[0]);
142
144
  return Object.freeze({ requirement: modifier || 'required', minYears: years, maxYears: isExact ? years : null });
@@ -194,7 +194,7 @@ export const HIRING_SECTION_MARKERS = Object.freeze({
194
194
  export const HIRING_NON_CONTENT_TERMS = Object.freeze([
195
195
  group('vacancyDigest', { ru: ['дайджест вакансий', 'подборка вакансий', 'вакансии недели'], en: ['job digest', 'vacancy digest', 'jobs roundup'], uk: ['добірка вакансій'], ro: ['selecție de joburi'], uzLatn: ['vakansiyalar toplami'], uzCyrl: ['вакансиялар тўплами'], kk: ['вакансиялар топтамасы'] }),
196
196
  group('recruitmentAd', { ru: ['поможем подобрать сотрудника', 'услуги рекрутера', 'закроем вакансию'], en: ['recruitment services', 'we find candidates', 'staffing services'], uk: ['послуги рекрутера'], ro: ['servicii de recrutare'], uzLatn: ['rekruting xizmati'], uzCyrl: ['рекрутинг хизмати'], kk: ['рекрутинг қызметтері'] }),
197
- group('jobService', { ru: ['помогу найти работу', 'подбор вакансий', 'сервис поиска работы'], en: ['job search service', 'we help you find a job'], uk: ['допоможемо знайти роботу'], ro: ['serviciu de căutare job'], uzLatn: ['ish topishga yordam'], uzCyrl: ['иш топишга ёрдам'], kk: ['жұмыс табуға көмектесеміз'] }),
197
+ group('jobService', { ru: ['помогу найти работу', 'подбор вакансий', 'сервис поиска работы', 'каналы с вакансиями', 'выбирай свою сферу и подписывайся'], en: ['job search service', 'we help you find a job'], uk: ['допоможемо знайти роботу'], ro: ['serviciu de căutare job'], uzLatn: ['ish topishga yordam'], uzCyrl: ['иш топишга ёрдам'], kk: ['жұмыс табуға көмектесеміз'] }),
198
198
  group('spam', { ru: ['подпишись на канал', 'заработок без вложений', 'доход без вложений', 'легкий заработок'], en: ['subscribe to our channel', 'earn money without investment', 'easy money'], uk: ['підпишись на канал', 'заробіток без вкладень'], ro: ['abonează-te la canal', 'câștig fără investiții'], uzLatn: ['kanalga obuna boling', 'sarmoyasiz daromad'], uzCyrl: ['каналга обуна бўлинг'], kk: ['арнаға жазылыңыз', 'инвестициясыз табыс'] }),
199
199
  ]);
200
200
 
@@ -6,6 +6,13 @@ export type VisaSponsorshipWording = 'offered' | 'notOffered' | null;
6
6
  export function cleanHiringSourceText(value: unknown): string;
7
7
  export function extractCandidateStructuredField(value: unknown, key: string, maxLength?: number): string | null;
8
8
  export function extractJobStructuredField(value: unknown, key: string, maxLength?: number): string | null;
9
+ export type HiringVacancyEntry = Readonly<{ title: string; text: string }>;
10
+ export type SplitHiringVacancyPost = Readonly<{
11
+ prefix: string;
12
+ suffix: string;
13
+ entries: readonly HiringVacancyEntry[];
14
+ }>;
15
+ export function splitHiringVacancyEntries(value: unknown): SplitHiringVacancyPost;
9
16
  export function extractCandidateDisplayName(value: unknown): string;
10
17
  export function extractCandidateExperienceMentions(value: unknown): readonly CandidateExperienceMention[];
11
18
  export function parseHiringSourceSalary(value: unknown): SalaryParseResult | null;
@@ -62,6 +62,39 @@ export function extractJobStructuredField(value, key, maxLength = 220) {
62
62
  return extractField(value, JOB_FIELD_TERMS[key], FIELD_EXTRA_ALIASES.job[key] || [], maxLength);
63
63
  }
64
64
 
65
+ const INLINE_VACANCY_HEADING_RE = /✔️?\s*([A-ZА-ЯЁЎҚҒҲІЇЄ][A-ZА-ЯЁЎҚҒҲІЇЄ\s()\/-]{2,100}?)(?=\s*(?:[•⏳✔📍✈📞]|$))/gu;
66
+ const SHARED_HIRING_CONTACT_RE = /(?:✈️?\s*Telegram\s*:|📞\s*Контакт\s*:)/iu;
67
+
68
+ /** Split a multi-vacancy source post into role-local blocks and shared context. */
69
+ export function splitHiringVacancyEntries(value) {
70
+ const text = cleanHiringSourceText(value);
71
+ if (!text) return Object.freeze({ prefix: '', suffix: '', entries: Object.freeze([]) });
72
+
73
+ const headings = [...text.matchAll(INLINE_VACANCY_HEADING_RE)]
74
+ .map((match) => ({ start: match.index ?? 0, title: match[1].replace(/\s+/g, ' ').trim() }));
75
+ if (headings.length < 2) return Object.freeze({ prefix: text, suffix: '', entries: Object.freeze([]) });
76
+
77
+ let suffix = '';
78
+ let contentEnd = text.length;
79
+ const lastStart = headings.at(-1).start;
80
+ const contact = SHARED_HIRING_CONTACT_RE.exec(text.slice(lastStart));
81
+ if (contact?.index != null) {
82
+ contentEnd = lastStart + contact.index;
83
+ suffix = text.slice(contentEnd).trim();
84
+ }
85
+
86
+ const entries = headings.map((heading, index) => {
87
+ const end = Math.min(headings[index + 1]?.start ?? contentEnd, contentEnd);
88
+ return Object.freeze({ title: heading.title, text: text.slice(heading.start, end).trim() });
89
+ }).filter(({ text: entryText }) => Boolean(entryText));
90
+
91
+ return Object.freeze({
92
+ prefix: text.slice(0, headings[0].start).trim(),
93
+ suffix,
94
+ entries: Object.freeze(entries),
95
+ });
96
+ }
97
+
65
98
  export function extractCandidateDisplayName(value) {
66
99
  const text = String(value || '');
67
100
  const structured = extractCandidateName(text);
package/src/money.js CHANGED
@@ -101,6 +101,10 @@ function rangeSearchText(text) {
101
101
  for (const symbol of Object.keys(CURRENCY_SYMBOL_CANDIDATES)) {
102
102
  normalized = normalized.split(symbol).join(' '.repeat(symbol.length));
103
103
  }
104
+ // Some job boards render salary parts as separate UI tokens and join their
105
+ // accessible text with middle dots: "2 500 · - 3 000 · €". Treat those
106
+ // decorative separators as whitespace while preserving source offsets.
107
+ normalized = normalized.replace(/[·•]/gu, ' ');
104
108
  return normalized;
105
109
  }
106
110