universal-parcel-scraper 0.7.0-main.326 → 0.7.2-main.328

package/dist/carriers/india-post/adapter.js

@@ -116,16 +116,23 @@ const AIRPORT_PLACES = {
     FRA: { name: 'Frankfurt Airport', country: 'Germany' },
     CDG: { name: 'Paris Charles de Gaulle Airport', country: 'France' },
 };
-function departureAirport(office) {
+function officeAirport(office) {
     return /^Office - ([A-Z]{3})\b/.exec(office)?.[1];
 }
 /** Flight remarks have a narrow format; unrelated remarks can contain recipient details. */
-function flightDescription(remarks, office) {
+function flightDetails(providerCode, remarks, office) {
+    if (providerCode !== TAKE_OFF_CODE && providerCode !== 'MailArrived')
+        return null;
     const match = /^Flight No:\s*([A-Z0-9]{2}\d{1,4}[A-Z]?)\s*\(From ([A-Z]{3}) To ([A-Z]{3})\)$/.exec(clean(remarks, 500));
-    if (!match || match[2] !== departureAirport(office))
-        return 'Aircraft Departure';
+    if (!match || match[2] !== officeAirport(office))
+        return null;
     const airport = (code) => AIRPORT_PLACES[code] ? `${AIRPORT_PLACES[code].name} (${code})` : code;
-    return `Flight ${match[1]} departed: ${airport(match[2])} → ${airport(match[3])}`;
+    const departed = providerCode === TAKE_OFF_CODE;
+    // MailArrived retains the sending office. Its flight route names the arrival airport.
+    return {
+        description: `Flight ${match[1]} ${departed ? 'departed' : 'arrived'}: ${airport(match[2])} → ${airport(match[3])}`,
+        airport: match[departed ? 2 : 3],
+    };
 }
 /**
  * The zone a take-off row's clock is kept in. Its `tracked_at` is the
@@ -138,7 +145,7 @@ function flightDescription(remarks, office) {
 function takeOffZone(providerCode, office, trackedAt) {
     if (providerCode !== TAKE_OFF_CODE || !UTC_LABEL.test(trackedAt))
         return null;
-    const airport = departureAirport(office);
+    const airport = officeAirport(office);
     return airport ? AIRPORT_ZONES[airport] ?? null : null;
 }
 /**
@@ -191,14 +198,15 @@ export function parseIndiaPostTrackingHtml(html, trackingNumber) {
             ? mislabeledLocalTime(rawEvent.tracked_at, zone, 100)
             : isoTime(rawEvent.tracked_at, 'Asia/Kolkata', 100);
         const takeOff = providerCode === TAKE_OFF_CODE;
-        const description = takeOff ? flightDescription(rawEvent.remarks, office) : eventText(clean(rawEvent.event));
+        const flight = flightDetails(providerCode, rawEvent.remarks, office);
+        const description = flight?.description ?? (takeOff ? 'Aircraft Departure' : eventText(clean(rawEvent.event)));
         if (!time || !description)
             return;
         const pincode = /^\d{6}$/.test(clean(rawEvent.pincode, 6))
             ? clean(rawEvent.pincode, 6)
             : '';
-        const point = officePoint(rawEvent, office);
-        const airportCode = takeOff ? departureAirport(office) : undefined;
+        const airportCode = flight?.airport ?? (takeOff ? officeAirport(office) : undefined);
+        const point = airportCode ? null : officePoint(rawEvent, office);
         const airport = airportCode ? AIRPORT_PLACES[airportCode] : undefined;
         const identity = JSON.stringify([time.iso, description, office, pincode, providerCode]);
         if (seen.has(identity))
@@ -211,7 +219,8 @@ export function parseIndiaPostTrackingHtml(html, trackingNumber) {
             // deliberately never retained.
             event: {
                 time: time.iso,
-                location: airport ? `${airport.name} (${airportCode}), ${airport.country}` : [office, pincode].filter(Boolean).join(' '),
+                location: airport ? `${airport.name} (${airportCode}), ${airport.country}`
+                    : flight && !takeOff ? `${flight.airport} Airport` : [office, pincode].filter(Boolean).join(' '),
                 description,
                 stage: classified.stage,
                 ...(providerCode ? { provider_code: providerCode } : {}),

package/dist/carriers/swiss-post/adapter.js

@@ -3,6 +3,7 @@ import makeFetchCookie from 'fetch-cookie';
 import { CookieJar } from 'tough-cookie';
 import { lookupBudget } from '../../core/adapter/index.js';
 import { NotFoundError, SchemaError } from '../../core/errors/index.js';
+import { countryCode } from '../../core/time/index.js';
 import { fetchBounded, parseJsonBytes } from '../../core/transport/index.js';
 import { isRecord } from '../../core/types.js';
 import { FALLBACK_EVENT_LABELS, STAGE_STATUS, STATUS_MAP, swissPostEventStage } from './status.js';
@@ -117,9 +118,14 @@ function eventDescription(event, translations, internationalType) {
     const $ = load(`<span>${value}</span>`);
     return $('span').text().trim().slice(0, 500) || 'Tracking update';
 }
-/** Operational scan location only: the city and its postcode, never a street. */
+/** Operational scan location: city and postcode, or an explicit country when city is absent. */
 function eventLocation(event) {
-    return [text(event.city, 100), text(event.zip, 30)].filter(Boolean).join(' ').slice(0, 160);
+    const city = text(event.city, 100);
+    const postcode = text(event.zip, 30);
+    const country = countryCode(event.country);
+    if (!city)
+        return country && country !== 'ZZ' ? country : postcode;
+    return [city, postcode].filter(Boolean).join(' ').slice(0, 160);
 }
 /**
  * Local ordering policy: event timestamps are read only to sort. Values without

package/dist/carriers/swiss-post/app.d.ts

@@ -0,0 +1,2 @@
+import type { SameInstantIdentityPolicy } from '../../core/catalog/eventIdentity.js';
+export declare const sameInstantIdentityPolicy: SameInstantIdentityPolicy;

package/dist/carriers/swiss-post/app.js

@@ -0,0 +1,6 @@
+import { matchesLocationEnrichment } from '../../core/catalog/locationIdentity.js';
+export const sameInstantIdentityPolicy = {
+    sourceCarrierId: 'swiss-post', storedSources: ['swiss-post'], requireProviderCode: true,
+    matchEachScan: true,
+    matches: matchesLocationEnrichment,
+};

package/dist/core/catalog/eventIdentity.d.ts

@@ -10,6 +10,8 @@ export interface SameInstantIdentityPolicy {
     readonly sourceCarrierId: string;
     readonly storedSources: readonly string[];
     readonly requireProviderCode: boolean;
+    /** Match distinct scans sharing an instant by their evidence, with unique matches in both directions. */
+    readonly matchEachScan?: boolean;
     readonly matches?: (incoming: SameInstantScan, stored: SameInstantScan) => boolean;
 }
 /** Unlisted sources cannot identify a reworded scan by its instant alone. */

package/dist/core/catalog/eventIdentity.js

@@ -1,7 +1,9 @@
 import { sameInstantIdentityPolicy as dpd } from '../../carriers/dpd/app.js';
 import { sameInstantIdentityPolicy as indiaPost } from '../../carriers/india-post/app.js';
 import { sameInstantIdentityPolicy as ups } from '../../carriers/ups/app.js';
-const policies = new Map([dpd, indiaPost, ups].map((policy) => [policy.sourceCarrierId, policy]));
+import { sameInstantIdentityPolicy as swissPost } from '../../carriers/swiss-post/app.js';
+import { sameInstantIdentityPolicy as universal } from '../../providers/shared/app.js';
+const policies = new Map([dpd, indiaPost, ups, swissPost, universal].map((policy) => [policy.sourceCarrierId, policy]));
 /** Unlisted sources cannot identify a reworded scan by its instant alone. */
 export function sameInstantIdentityPolicy(sourceCarrierId, options = {}) {
     const policy = policies.get(sourceCarrierId);

package/dist/core/catalog/locationIdentity.d.ts

@@ -0,0 +1,3 @@
+import type { SameInstantScan } from './eventIdentity.js';
+/** Adding a place must keep the scan's wording and known stage; conflicting places do not match. */
+export declare function matchesLocationEnrichment(incoming: SameInstantScan, stored: SameInstantScan): boolean;

package/dist/core/catalog/locationIdentity.js

@@ -0,0 +1,10 @@
+const normalized = (value) => value.replace(/\s+/g, ' ').trim().toLocaleLowerCase('en-US');
+/** Adding a place must keep the scan's wording and known stage; conflicting places do not match. */
+export function matchesLocationEnrichment(incoming, stored) {
+    const wording = normalized(incoming.description);
+    const location = normalized(incoming.location);
+    const savedLocation = normalized(stored.location);
+    return Boolean(location && wording) && !['', 'unknown'].includes(incoming.stage)
+        && incoming.stage === stored.stage && wording === normalized(stored.description)
+        && (!savedLocation || location === savedLocation);
+}

package/dist/providers/parcelsapp/adapter.js

@@ -221,6 +221,10 @@ function parseHistory(payload, trackingNumber, timezone = null) {
             continue;
         }
         if (parsed) {
+            const location = typeof raw.location === 'string' ? raw.location.replace(/\s+/g, ' ').trim().slice(0, 200) : '';
+            // Some Chronopost rows put the service label in the location field.
+            if (location && !/^Type de livraison\s*:/i.test(location))
+                parsed.location = location;
             events.push(parsed);
             scanCarriers.add(name);
         }

package/dist/providers/shared/app.d.ts

@@ -0,0 +1,2 @@
+import type { SameInstantIdentityPolicy } from '../../core/catalog/eventIdentity.js';
+export declare const sameInstantIdentityPolicy: SameInstantIdentityPolicy;

package/dist/providers/shared/app.js

@@ -0,0 +1,6 @@
+import { matchesLocationEnrichment } from '../../core/catalog/locationIdentity.js';
+export const sameInstantIdentityPolicy = {
+    sourceCarrierId: 'unknown', storedSources: ['unknown'], requireProviderCode: false,
+    matchEachScan: true,
+    matches: matchesLocationEnrichment,
+};

package/dist/providers/shared/result.js

@@ -163,7 +163,14 @@ function moment(value) {
 export function result(events, source, preserveOrder = false) {
     // Wall times sort alongside UTC instants only approximately; providers that
     // omit offsets everywhere pass preserveOrder instead.
-    const unique = [...new Map(events.map((e) => [`${moment(e)}|${e.description}`, e])).values()]
+    const scans = new Map();
+    for (const event of events) {
+        const key = `${moment(event)}|${event.description}`;
+        const earlier = scans.get(key);
+        // A carrier's copy can lack the place another copy of the same scan reported.
+        scans.set(key, !event.location && earlier?.location ? { ...event, location: earlier.location } : event);
+    }
+    const unique = [...scans.values()]
         .sort((a, b) => preserveOrder ? 0 : moment(b).localeCompare(moment(a))).slice(0, 100);
     // A well-formed reply with no scan left after notices and private details proves nothing about the parcel.
     if (!unique.length)

package/dist/server/openapi.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "Universal Parcel Scraper",
-    "version": "0.7.0"
+    "version": "0.7.2"
   },
   "paths": {
     "/health": {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "universal-parcel-scraper",
-  "version": "0.7.0-main.326",
+  "version": "0.7.2-main.328",
   "description": "Self-hosted parcel tracking: carrier detection, dedicated scrapers and optional universal providers.",
   "type": "module",
   "license": "Apache-2.0",