@mountainpass/addressr 2.2.0 → 2.4.0

package/lib/client/elasticsearch.js

@@ -111,6 +111,14 @@ async function initIndex(esClient, clear, synonyms) {
             }
           }
         },
+        // ADR 026: range-number address expansion. Multi-valued text field
+        // populated with one expanded address per in-range number for range-
+        // numbered G-NAF docs (span cap 20). Absent on non-range docs
+        // (asymmetric population). No .raw sub-field — never sorted on.
+        sla_range_expanded: {
+          type: 'text',
+          analyzer: 'my_analyzer'
+        },
         confidence: {
           type: 'integer'
         },

package/lib/service/address-service.js

@@ -30,6 +30,7 @@ var _unzipStream = _interopRequireDefault(require("unzip-stream"));
 var _elasticsearch = require("../client/elasticsearch");
 var _streamDown = _interopRequireDefault(require("../utils/stream-down"));
 var _setLinkOptions = require("./set-link-options");
+var _rangeExpansion = require("./range-expansion");
 var _keyv = require("keyv");
 var _keyvFile = require("keyv-file");
 var _nodeCrypto = _interopRequireDefault(require("node:crypto"));
@@ -683,14 +684,26 @@ function mapAddressDetails(d, context, index, count) {
     rval.smla = mapToShortMla(rval.structured);
     rval.ssla = mapToSla(rval.smla);
   }
+  // ADR 026: asymmetric population of sla_range_expanded — one expanded
+  // alias per in-range number for range-numbered records up to SPAN_CAP.
+  // Non-range docs leave the field absent. expandRangeAliases returns []
+  // for span>cap or invalid inputs; in that case we keep the field absent
+  // rather than storing an empty array so OpenSearch _source stays clean.
+  if (rval.structured.number?.last?.number !== undefined) {
+    const s = rval.structured;
+    const streetType = s.street.type ? ` ${s.street.type.name}` : '';
+    const streetSuffix = s.street.suffix ? ` ${s.street.suffix.name}` : '';
+    rval.sla_range_expanded = (0, _rangeExpansion.expandRangeAliases)(Number(s.number.number), Number(s.number.last.number), `${s.street.name}${streetType}${streetSuffix}`, `${s.locality.name} ${s.state.abbreviation} ${s.postcode}`);
+    if (rval.sla_range_expanded.length === 0) {
+      delete rval.sla_range_expanded;
+    }
+  }
   if (count) {
     if (index % Math.ceil(count / 100) === 0) {
-      logger('addr', JSON.stringify(rval, undefined, 2));
       logger(`${index / count * 100}%`);
     }
   } else {
     if (index % 10_000 === 0) {
-      logger('addr', JSON.stringify(rval, undefined, 2));
       logger(`${index} rows`);
     }
   }
@@ -849,7 +862,13 @@ async function searchForAddress(searchString, p, pageSize = PAGE_SIZE) {
               multi_match: {
                 fields: ['sla', 'ssla'],
                 query: searchString,
-                fuzziness: 'AUTO',
+                // ADR 027: AUTO:5,8 (not default AUTO / AUTO:3,6) so that
+                // 3-4 digit street numbers and postcodes require exact
+                // match. Default AUTO lets `138` fuzzy-match `137`, `135`
+                // etc., which tf-inflates adjacent-number docs above the
+                // actual target (P026). 5+ char tokens still get 1 edit
+                // (Muray → Murray), 8+ char tokens get 2 edits.
+                fuzziness: 'AUTO:5,8',
                 type: 'bool_prefix',
                 lenient: true,
                 auto_generate_synonyms_phrase_query: false,
@@ -857,7 +876,14 @@ async function searchForAddress(searchString, p, pageSize = PAGE_SIZE) {
               }
             }, {
               multi_match: {
-                fields: ['sla', 'ssla'],
+                // ADR 026: sla_range_expanded added HERE ONLY (not in the
+                // bool_prefix clause above). phrase_prefix uses best_fields
+                // max with tie_breaker default 0.0, so an absent field on
+                // non-range docs contributes 0 to the max — no P007-shape
+                // asymmetry. Adding sla_range_expanded to the bool_prefix
+                // fields would reintroduce the summation asymmetry ADR 025
+                // resolved. DO NOT move this field into the clause above.
+                fields: ['sla', 'ssla', 'sla_range_expanded'],
                 query: searchString,
                 // fuzziness: 'AUTO',
                 type: 'phrase_prefix',
@@ -1594,20 +1620,27 @@ async function getAddress(addressId) {
     };
   } catch (error_) {
     error('error getting record from elastic search', error_);
-    if (error_.body.found === false) {
+    if (error_.body && error_.body.found === false) {
       return {
         statusCode: 404,
         json: {
           error: 'not found'
         }
       };
-    } else if (error_.body.error.type === 'index_not_found_exception') {
+    } else if (error_.body && error_.body.error && error_.body.error.type === 'index_not_found_exception') {
       return {
         statusCode: 503,
         json: {
           error: 'service unavailable'
         }
       };
+    } else if (error_.displayName === 'RequestTimeout') {
+      return {
+        statusCode: 504,
+        json: {
+          error: 'gateway timeout'
+        }
+      };
     } else {
       return {
         statusCode: 500,

package/lib/service/range-expansion.js

@@ -0,0 +1,31 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.expandRangeAliases = expandRangeAliases;
+// ADR 028 (supersedes ADR 026): endpoint-only range-address expansion.
+// Pure helper — no I/O, no OpenSearch dependency. Invoked by `mapToMla` in
+// `service/address-service.js` when a G-NAF address has NUMBER_LAST set.
+//
+// For a G-NAF range like `103-107 GAZE RD` (NUMBER_FIRST=103, NUMBER_LAST=107),
+// emit exactly two aliases — the first and last endpoints — and nothing else.
+// Mid-range numbers (104, 105, 106) do NOT match the range document:
+//
+//   - Even-indexed mid-range numbers (104, 106) typically belong to properties
+//     on the opposite side of the street under Australian addressing convention.
+//   - Odd-indexed mid-range numbers (105) could be separate properties the
+//     range record absorbed, or a single contiguous frontage — G-NAF does not
+//     tell us which. Returning a range doc for a mid-range query would be a
+//     false positive either way.
+//
+// See ADR 028 and P015 / #367 reporter comments 2022-06-24 (`138-144`) and
+// 2022-07-10 (`225-245`). The earlier full-interpolation approach from ADR 026
+// shipped in v2.3.0 produced these false positives and is superseded.
+
+function expandRangeAliases(first, last, streetPart, localityPart) {
+  if (!Number.isInteger(first) || !Number.isInteger(last)) return [];
+  if (first <= 0 || last <= 0) return [];
+  if (first >= last) return [];
+  return [`${first} ${streetPart}, ${localityPart}`, `${last} ${streetPart}, ${localityPart}`];
+}

package/lib/src/waycharter-server.js

@@ -999,6 +999,12 @@ function startRest2Server() {
         }],
         headers: {
           etag: `"${_version.version}"`,
+          // Long-lived by design (P018 parked). New rels are added
+          // infrequently and every client page load fetches this for
+          // HATEOAS discovery; a short TTL would cost an origin
+          // round-trip per request. When the rel set does change,
+          // request a RapidAPI CF purge (natural expiry up to 7 days
+          // per P017 close notes).
           'cache-control': `public, max-age=${ONE_WEEK}`
         }
       };

package/lib/version.js

@@ -5,4 +5,4 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.version = void 0;
 // Generated by genversion.
-const version = exports.version = '2.2.0';
+const version = exports.version = '2.4.0';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mountainpass/addressr",
-  "version": "2.2.0",
+  "version": "2.4.0",
   "description": "Australian Address Validation, Search and Autocomplete",
   "author": {
     "name": "Mountain Pass",
@@ -108,6 +108,7 @@
     "cover:cli:geo": "nyc --report-dir coverage/cli --temp-dir coverage/cli/.nyc_output npm run test:cli:nogeo",
     "test:mcp:smoke": "node --test test/mcp/smoke.test.mjs",
     "test:precommit": "node --test test/precommit/*.test.mjs",
+    "test:js": "node --test test/js/__tests__/*.test.mjs",
     "test:nodejs:QLD:nogeo": "PORT=$npm_package_config_localport ES_INDEX_NAME=test COVERED_STATES=QLD DEBUG=error,api,express:*,swagger-tools*,test,es TEST_PROFILE=default cucumber-js -p default -- --harmony_async_iteration",
     "test:nodejs:QLD:geo": "PORT=$npm_package_config_localport ADDRESSR_ENABLE_GEO=1 ES_INDEX_NAME=test-geo COVERED_STATES=QLD DEBUG=error,api,express:*,swagger-tools*,test,es TEST_PROFILE=default NODE_OPTIONS=--max_old_space_size=8196 cucumber-js -p default -- --harmony_async_iteration",
     "prebuildX": "npm run genversion && cat ./templates/LICENSE.md | envsubst '${PRODUCT},${VERSION},${COMPANY},${YEAR}' > ./LICENSE.md",
@@ -123,7 +124,7 @@
     "docker:push": "docker push \"mountainpass/addressr:${npm_package_version}\"",
     "postdocker:push": "docker push \"mountainpass/addressr:latest\"",
     "check-licenses": "license-checker --production --onlyAllow 'MIT;Apache-2.0;ISC;Custom: http://github.com/substack/node-bufferlist;Unlicense;BSD-2-Clause;BSD-3-Clause;WTFPL;0BSD;MIT*;Python-2.0;MPL-2.0;BlueOak-1.0.0' --summary",
-    "pre-commit": "lint-staged && npm run check-licenses && npm run check:not-cli2-tags",
+    "pre-commit": "lint-staged && npm run check-licenses && npm run check:not-cli2-tags && npm run test:js",
     "check:not-cli2-tags": "node scripts/check-not-cli2-tags.mjs",
     "check-deps": "dry-aged-deps --check",
     "test:performance": "k6 run --out csv=target/stress.csv test/k6/script.js",