feat: EMEX VIN decode → catalog DB matching via pathData/SSD
Some checks failed
CI / Lint, Typecheck, Test & Build (push) Has been cancelled

Parse path_data from Vehicle.aspx URLs (Base64-decoded vehicle name),
persist vehicleLabel/vid/pathData in rawData, and add 4-step matching
cascade for category resolution:
1. pathData name → matchByName (DB-only, ~50ms)
2. emexVehicleName → matchByName (DB-only, ~50ms)
3. ssd + catalogCode → matchBySsd (wizard HTTP + DB, ~2-3s)
4. Fallback → on-demand EMEX scrape (unchanged, ~6-13s)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-03-06 14:46:00 +00:00
parent adfba6c543
commit 53a6296a91
5 changed files with 207 additions and 30 deletions

View File

@@ -393,6 +393,108 @@ export class EmexCatalogService {
return rows;
}
/**
* Match a VIN-decoded vehicle to DB using wizard API flow.
* Fetches Vehicles.aspx?ft=findByWizard2 to get motor/variant-level names,
* then matches them against emex_vehicles in DB.
*/
async matchBySsd(catalogCode: string, ssd: string, pathData?: string): Promise<EmexVehicleMatch | null> {
const cacheKey = `emex:matchssd:${catalogCode}:${Buffer.from(ssd).toString("base64url").slice(0, 32)}`;
const cached = await this.redis.getJson<EmexVehicleMatch>(cacheKey);
if (cached) return cached;
try {
// Fetch vehicle list from wizard SSD via Vehicles.aspx
const params = new URLSearchParams({ ft: "findByWizard2", c: catalogCode, ssd });
const url = `${EmexCatalogService.EMEX_BASE}/Vehicles.aspx?${params}`;
const res = await fetch(url, {
headers: { ...EmexCatalogService.EMEX_HDR, Accept: "text/html" },
signal: AbortSignal.timeout(15000),
});
if (!res.ok) {
this.logger.warn(`matchBySsd: HTTP ${res.status} for ${url}`);
return null;
}
const html = await res.text();
// Parse vehicle names from the HTML (same link pattern as VIN decode)
const linkRx = /href="Vehicle\.aspx\?[^"]*">([^<]+)<\/a>/g;
const vehicleNames: string[] = [];
let m: RegExpExecArray | null;
while ((m = linkRx.exec(html)) !== null) {
const name = m[1].trim();
if (name && !vehicleNames.includes(name)) vehicleNames.push(name);
}
if (vehicleNames.length === 0) {
this.logger.log(`matchBySsd: no vehicles found for c=${catalogCode} ssd=${ssd.slice(0, 30)}...`);
return null;
}
this.logger.log(`matchBySsd: found ${vehicleNames.length} vehicle names from wizard, trying DB match`);
// Find catalog in DB
const [catalog] = await this.db
.select({ id: emexCatalogs.id })
.from(emexCatalogs)
.where(eq(emexCatalogs.catalogId, catalogCode))
.limit(1);
if (!catalog) return null;
const vehicleCols = {
id: emexVehicles.id,
vehicleId: emexVehicles.vehicleId,
name: emexVehicles.name,
engine: emexVehicles.engine,
engineCode: emexVehicles.engineCode,
bodyType: emexVehicles.bodyType,
transmission: emexVehicles.transmission,
driveType: emexVehicles.driveType,
fuelType: emexVehicles.fuelType,
yearFrom: emexVehicles.yearFrom,
yearTo: emexVehicles.yearTo,
optionsRaw: emexVehicles.optionsRaw,
};
// Try each vehicle name against DB
for (const vName of vehicleNames) {
// Extract model part before brackets: "M 103 [123.981] (1985-1992)" → "M 103"
const nameMatch = vName.match(/^([^\[]+)/);
const cleanName = nameMatch ? nameMatch[1].trim() : vName;
const candidates = await this.db
.select(vehicleCols)
.from(emexVehicles)
.where(and(eq(emexVehicles.catalogId, catalog.id), ilike(emexVehicles.name, `${cleanName}%`)))
.orderBy(emexVehicles.optionsRaw)
.limit(50);
if (candidates.length > 0) {
const displayName = pathData
? pathData.replace(/^Name:\s*/i, "").trim()
: vName;
const match: EmexVehicleMatch = {
vehicleId: candidates[0].id,
catalogCode,
vehicleName: displayName,
candidates,
};
await this.redis.setJson(cacheKey, match, CACHE_TTL.match);
this.logger.log(`matchBySsd: DB match found — "${cleanName}" → ${candidates.length} candidates`);
return match;
}
}
this.logger.log(`matchBySsd: no DB match for any of ${vehicleNames.length} wizard vehicles`);
return null;
} catch (err) {
this.logger.warn(`matchBySsd failed: ${(err as Error).message}`);
return null;
}
}
async searchByOem(query: string): Promise<EmexSearchResult[]> {
if (!query || query.length < 3) return [];