@whiteslove/parsing-lexicon 0.9.4 → 0.9.6

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.9.4",
+  "version": "0.9.6",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/housing-commercial.js

@@ -63,3 +63,22 @@ export function isDirectOwner(value) {
 export function hasZeroCommissionSignal(value) {
   return isDirectOwner(value) || (Boolean(value) && EXPLICIT_ZERO_COMMISSION_RE.test(String(value)));
 }
+
+// A Telegram group's own "welcome"/intro post routinely names the group
+// itself, and a real-estate group is typically named after what it rents
+// out ("Оренда квартир Одеса"). That alone can satisfy HOUSING_RE-style
+// content gates even though the message describes no property at all -- it
+// is a greeting, not a listing. No genuine single-property ad opens with
+// "welcome to the group", so this is safe to treat as a strong, low-noise
+// signal on its own.
+const GROUP_WELCOME_RE = /(?:добро\s+пожаловать\s+в\s+групп\p{L}*|ласкаво\s+просимо\s+(?:до|в)\s+груп\p{L}*|welcome\s+to\s+(?:the\s+|our\s+)?group|bine\s+a[țt]i\s+venit\s+[îi]n\s+grup|guruhga\s+xush\s+kelibsiz|топқа\s+қош\s+келдіңіз)/iu;
+
+/**
+ * A group's channel welcome/intro post, or an off-topic channel
+ * cross-promotion riding on the group's housing-flavored name -- neither
+ * describes a specific property and both should be excluded from listing
+ * feeds even when they incidentally contain housing keywords.
+ */
+export function looksLikeGroupWelcomeMessage(value) {
+  return Boolean(value) && GROUP_WELCOME_RE.test(String(value));
+}

package/src/location-merge.js

@@ -49,6 +49,18 @@ export function locationEntries(rows = []) {
   return Object.freeze(rows.map(([name, ...aliases]) => locationEntry(name, ...aliases)));
 }
 
+// Copy an entry's data fields without touching its lazy `re` getter. Object
+// spread invokes getters, which would compile every alias regex eagerly at
+// merge time (tens of seconds at import) instead of on first match.
+function entryData(entry) {
+  const data = {};
+  if (!entry) return data;
+  for (const key of Object.keys(entry)) {
+    if (key !== 're') data[key] = entry[key];
+  }
+  return data;
+}
+
 function mergeEntry(existing, incoming) {
   const aliases = [...new Set([
     ...(existing?.aliases || []),
@@ -56,7 +68,7 @@ function mergeEntry(existing, incoming) {
     existing?.name,
     incoming?.name,
   ].filter(Boolean))];
-  const base = { ...(existing || {}), ...(incoming || {}) };
+  const base = { ...entryData(existing), ...entryData(incoming) };
   // A map-data entry may enrich an existing reviewed owner with additional
   // aliases. It remains fallback-only only when every merged source is map
   // data; otherwise the reviewed owner must retain its matching precedence.