site-web-targo/supabase/functions/process-import-job/index.ts
louispaulb 88dc3714a1 Initial deploy: gigafibre.ca website with self-hosted address search
React/Vite/shadcn-ui site for Gigafibre ISP.
Address qualification via PostgreSQL (5.2M AQ addresses, pg_trgm fuzzy search).
No Supabase dependency for address search.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-27 14:37:50 -04:00

616 lines
24 KiB
TypeScript

import { createClient } from "https://esm.sh/@supabase/supabase-js@2";
const corsHeaders = {
"Access-Control-Allow-Origin": "*",
"Access-Control-Allow-Headers":
"authorization, x-client-info, apikey, content-type, x-supabase-client-platform, x-supabase-client-platform-version, x-supabase-client-runtime, x-supabase-client-runtime-version",
};
const INITIAL_BATCH_SIZE = 1000;
const MIN_BATCH_SIZE = 100;
const MAX_BATCH_SIZE = 2000;
const BATCH_GROW_FACTOR = 1.25; // Increase by 25% on success
const BATCH_SHRINK_FACTOR = 0.5; // Halve on error
const MAX_ROWS_PER_CALL = 50000;
const MAX_RETRIES = 3;
const RETRY_DELAY_MS = 2000;
const TIME_LIMIT_MS = 110000;
const LOG_INTERVAL = 5000;
// ─── GIN Index Definitions (to drop/recreate) ──────────────────────────────
const GIN_INDEXES = [
{ name: "idx_addr_formatee_trgm", def: `CREATE INDEX idx_addr_formatee_trgm ON public.addresses USING gin (adresse_formatee gin_trgm_ops)` },
{ name: "idx_addr_municipalite_trgm", def: `CREATE INDEX idx_addr_municipalite_trgm ON public.addresses USING gin (nom_municipalite gin_trgm_ops)` },
{ name: "idx_addr_odonyme_court_trgm", def: `CREATE INDEX idx_addr_odonyme_court_trgm ON public.addresses USING gin (odonyme_recompose_court gin_trgm_ops)` },
{ name: "idx_addr_odonyme_long_trgm", def: `CREATE INDEX idx_addr_odonyme_long_trgm ON public.addresses USING gin (odonyme_recompose_long gin_trgm_ops)` },
{ name: "idx_addr_odonyme_normal_trgm", def: `CREATE INDEX idx_addr_odonyme_normal_trgm ON public.addresses USING gin (odonyme_recompose_normal gin_trgm_ops)` },
{ name: "idx_addresses_adresse_formatee_trgm", def: `CREATE INDEX idx_addresses_adresse_formatee_trgm ON public.addresses USING gin (adresse_formatee gin_trgm_ops)` },
{ name: "idx_addresses_num_odonyme_trgm", def: `CREATE INDEX idx_addresses_num_odonyme_trgm ON public.addresses USING gin (((COALESCE(numero_municipal, ''::text) || ' '::text) || COALESCE(odonyme_recompose_normal, ''::text))) gin_trgm_ops)` },
];
// ─── CSV Parsing helpers ─────────────────────────────────────────────────────
function parseLine(line: string, delimiter: string): string[] {
if (delimiter === "\t" || delimiter === ";") {
return line.split(delimiter).map((v) => v.trim().replace(/^"|"$/g, ""));
}
const values: string[] = [];
let current = "";
let inQuotes = false;
for (let i = 0; i < line.length; i++) {
const char = line[i];
if (char === '"') {
if (inQuotes && i + 1 < line.length && line[i + 1] === '"') {
current += '"';
i++;
} else {
inQuotes = !inQuotes;
}
} else if (char === delimiter && !inQuotes) {
values.push(current.trim());
current = "";
} else {
current += char;
}
}
values.push(current.trim());
return values;
}
function normalizeHeader(name: string): string {
return name
.toLowerCase()
.normalize("NFD")
.replace(/[\u0300-\u036f]/g, "")
.replace(/[_\s]+/g, "_")
.replace(/[^a-z0-9_]/g, "")
.trim();
}
const FIBER_HEADER_MAP: Record<string, string> = {
id_placemark: "id_placemark", idplacemark: "id_placemark",
id_appart: "id_appart", idappart: "id_appart",
uuidadresse: "uuidadresse", idgouvqc: "uuidadresse", id_gouv_qc: "uuidadresse",
nom: "nom", civique: "civique",
appartement: "appartement", appt: "appartement", apt: "appartement",
rue: "rue", ville: "ville",
code_postal: "code_postal", codepostal: "code_postal", cp: "code_postal",
lien_googlemap: "lien_googlemap", liengooglemap: "lien_googlemap", googlemap: "lien_googlemap",
zone_tarifaire: "zone_tarifaire", zonetarifaire: "zone_tarifaire", zone: "zone_tarifaire",
max_speed: "max_speed", maxspeed: "max_speed", vitesse: "max_speed", vitesse_max: "max_speed",
commentaire: "commentaire", commentaires: "commentaire", comment: "commentaire",
lien_map_targo: "_skip", lienmaptargo: "_skip",
};
// ─── Row Mapping ─────────────────────────────────────────────────────────────
function mapAddressRow(r: Record<string, string>) {
return {
identifiant_unique_adresse: r.identifiant_unique_adresse,
date_diffusion_version: r.date_diffusion_version || null,
date_creation: r.date_creation || null,
date_modification: r.date_modification || null,
date_fin: r.date_fin || null,
numero_municipal: r.numero_municipal || null,
numero_municipal_suffixe: r.numero_municipal_suffixe || null,
numero_unite: r.numero_unite || null,
type_unite: r.type_unite || null,
code_postal: r.code_postal || null,
seqodo: r.seqodo || null,
generique_odonyme: r.generique_odonyme || null,
particule_odonyme: r.particule_odonyme || null,
specifique_odonyme: r.specifique_odonyme || null,
point_cardinal_odonyme: r.point_cardinal_odonyme || null,
odonyme_recompose_normal: r.odonyme_recompose_normal || null,
odonyme_recompose_court: r.odonyme_recompose_court || null,
odonyme_recompose_long: r.odonyme_recompose_long || null,
discriminant_odonyme: r.discriminant_odonyme || null,
odonyme_officiel_ctop: r.odonyme_officiel_ctop || null,
adresse_formatee: r.adresse_formatee || null,
qualite_positionnement_geometrique: r.qualite_positionnement_geometrique || null,
etat: r.etat || null,
code_utilisation: r.code_utilisation || null,
categorie: r.categorie || null,
nombre_unite: r.nombre_unite ? parseInt(r.nombre_unite) : null,
numero_lot: r.numero_lot || null,
matricule_unite_evaluation: r.matricule_unite_evaluation || null,
nosqnocivq: r.nosqnocivq || null,
caracteristique_adresse: r.caracteristique_adresse || null,
code_region_administrative: r.code_region_administrative || null,
nom_region_administrative: r.nom_region_administrative || null,
code_mrc: r.code_mrc || null,
nom_mrc: r.nom_mrc || null,
code_municipalite: r.code_municipalite || null,
nom_municipalite: r.nom_municipalite || null,
nom_municipalite_complet: r.nom_municipalite_complet || null,
code_arrondissement: r.code_arrondissement || null,
nom_arrondissement: r.nom_arrondissement || null,
code_communaute_metropolitaine: r.code_communaute_metropolitaine || null,
nom_communaute_metropolitaine: r.nom_communaute_metropolitaine || null,
identifiant_reseau_routier: r.identifiant_reseau_routier || null,
cote_route: r.cote_route || null,
longitude: r.longitude ? parseFloat(r.longitude) : null,
latitude: r.latitude ? parseFloat(r.latitude) : null,
};
}
function mapFiberRow(r: Record<string, string>) {
if (!r.uuidadresse) return null;
return {
id_placemark: r.id_placemark || null,
id_appart: r.id_appart || null,
uuidadresse: r.uuidadresse,
nom: r.nom || null,
civique: r.civique || null,
appartement: r.appartement || null,
rue: r.rue || null,
ville: r.ville || null,
code_postal: r.code_postal || null,
lien_googlemap: r.lien_googlemap || null,
zone_tarifaire: r.zone_tarifaire ? parseInt(r.zone_tarifaire) : 1,
max_speed: r.max_speed ? parseInt(r.max_speed) : 0,
commentaire: r.commentaire || null,
};
}
// ─── Upsert with retry ──────────────────────────────────────────────────────
async function insertWithRetry(
supabase: any,
table: string,
batch: any[],
): Promise<{ error: any; timedOut: boolean }> {
for (let attempt = 0; attempt < MAX_RETRIES; attempt++) {
const { error } = await supabase.from(table).insert(batch);
if (!error) return { error: null, timedOut: false };
// On duplicate key, skip silently (data already exists)
if (error.code === "23505") return { error: null, timedOut: false };
const isTimeout = !!error.message?.includes("statement timeout");
if (isTimeout && attempt < MAX_RETRIES - 1) {
await new Promise((r) => setTimeout(r, RETRY_DELAY_MS * (attempt + 1)));
continue;
}
return { error, timedOut: isTimeout };
}
return { error: { message: "Max retries exceeded" }, timedOut: true };
}
// ─── Index Management ───────────────────────────────────────────────────────
async function dropGinIndexes(supabase: any, log: (msg: string) => void) {
log("🗑️ Dropping GIN trigram indexes for faster import…");
for (const idx of GIN_INDEXES) {
const { error } = await supabase.rpc("exec_sql", { query: `DROP INDEX IF EXISTS public.${idx.name}` });
if (error) {
// Try via direct SQL if rpc doesn't exist
log(`⚠️ Could not drop ${idx.name} via rpc: ${error.message}`);
} else {
log(` ✅ Dropped ${idx.name}`);
}
}
}
async function createGinIndexes(supabase: any, log: (msg: string) => void) {
log("🔨 Recreating GIN trigram indexes (this may take a while)…");
for (const idx of GIN_INDEXES) {
const { error } = await supabase.rpc("exec_sql", { query: idx.def });
if (error) {
log(`⚠️ Could not create ${idx.name}: ${error.message}`);
} else {
log(` ✅ Created ${idx.name}`);
}
}
}
// ─── Main Handler ────────────────────────────────────────────────────────────
Deno.serve(async (req) => {
if (req.method === "OPTIONS") {
return new Response(null, { headers: corsHeaders });
}
const serviceClient = createClient(
Deno.env.get("SUPABASE_URL")!,
Deno.env.get("SUPABASE_SERVICE_ROLE_KEY")!,
{ db: { schema: "public" }, global: { headers: { "x-statement-timeout": "120000" } } }
);
// Find the oldest active import job
const { data: jobs, error: jobErr } = await serviceClient
.from("import_jobs")
.select("*")
.in("status", ["running", "queued"])
.order("created_at", { ascending: true })
.limit(1);
if (jobErr || !jobs || jobs.length === 0) {
return new Response(JSON.stringify({ message: "No active jobs" }), {
headers: { ...corsHeaders, "Content-Type": "application/json" },
});
}
const job = jobs[0];
const logLines: string[] = [...(job.logs || [])].slice(-300);
function log(msg: string) {
const ts = new Date().toISOString().slice(11, 19);
logLines.push(`[${ts}] ${msg}`);
console.log(`[process-import-job] ${msg}`);
}
const updateJob = async (updates: Record<string, any>) => {
await serviceClient
.from("import_jobs")
.update({ ...updates, logs: logLines.slice(-500), updated_at: new Date().toISOString() })
.eq("id", job.id);
};
await updateJob({ status: "running" });
const url = job.source_url;
if (!url) {
log("❌ No source URL");
await updateJob({ status: "error", error_messages: ["No source URL"] });
return new Response(JSON.stringify({ error: "No source URL" }), {
headers: { ...corsHeaders, "Content-Type": "application/json" },
});
}
const type = job.type as "addresses" | "fiber";
const byteOffset: number = job.byte_offset || 0;
let totalInserted = job.total_inserted || 0;
let totalErrors = job.total_errors || 0;
const errorMessages: string[] = [...(job.error_messages || [])];
const startTime = Date.now();
// Detect if this is a direct CSV (not ZIP) — only CSV supports Range
const isZip = url.toLowerCase().includes(".zip");
const useRange = !isZip && byteOffset > 0;
// Drop GIN indexes on addresses table for faster upserts (only on first chunk or when explicitly needed)
const indexesDropped = job.indexes_dropped === true;
if (type === "addresses" && !indexesDropped) {
await dropGinIndexes(serviceClient, log);
await updateJob({ indexes_dropped: true });
}
try {
log(`🚀 Job ${job.id.slice(0, 8)}… type=${type} byteOffset=${byteOffset.toLocaleString()} totalInserted=${totalInserted.toLocaleString()}`);
// Download the source file, using Range header for CSV resume
const fetchHeaders: Record<string, string> = {};
if (useRange) {
fetchHeaders["Range"] = `bytes=${byteOffset}-`;
log(`📥 Resuming from byte ${byteOffset.toLocaleString()}`);
} else {
log("📥 Downloading source file…");
}
const resp = await fetch(url, { headers: fetchHeaders });
if (!resp.ok || !resp.body) {
throw new Error(`Download failed: ${resp.status} ${resp.statusText}`);
}
const rangeSupported = useRange && resp.status === 206;
if (useRange && !rangeSupported) {
log(`⚠️ Server returned ${resp.status} instead of 206 — Range not supported, will skip rows`);
}
log(`📥 Download started (status ${resp.status})`);
let csvStream: ReadableStream<Uint8Array>;
if (isZip) {
csvStream = await decompressZipStream(resp.body, log);
} else {
csvStream = resp.body;
}
// Process CSV from stream
const streamReader = csvStream.getReader();
const decoder = new TextDecoder("utf-8", { fatal: false });
const knownHeaders = job.headers_cache as string[] | null;
let headers: string[] = knownHeaders || [];
let delimiter = ",";
let leftover = "";
let totalRowsRead = 0;
let processedRows = 0;
let inserted = 0;
let errors = 0;
let pendingRows: any[] = [];
let done = false;
let stoppedByLimit = false;
let bytesConsumed = 0;
let currentBatchSize = job.current_batch_size || INITIAL_BATCH_SIZE;
let consecutiveSuccesses = 0;
const needsHeaderDetection = headers.length === 0;
if (!needsHeaderDetection) {
log(`📋 Using cached headers (${headers.length} cols) | batch=${currentBatchSize}`);
}
const flushBatch = async (batch: any[]) => {
if (batch.length === 0) return;
const table = type === "addresses" ? "addresses" : "fiber_availability";
const { error, timedOut } = await insertWithRetry(serviceClient, table, batch);
if (error) {
const msg = error.message || "Unknown error";
if (timedOut) {
// Shrink batch size on timeout — re-queue these rows by NOT counting them as errors
const oldSize = currentBatchSize;
currentBatchSize = Math.max(MIN_BATCH_SIZE, Math.floor(currentBatchSize * BATCH_SHRINK_FACTOR));
consecutiveSuccesses = 0;
log(`⚠️ Timeout on batch of ${batch.length} — shrinking batch ${oldSize}${currentBatchSize}`);
// Split and retry the failed batch in smaller chunks
for (let i = 0; i < batch.length; i += currentBatchSize) {
const subBatch = batch.slice(i, i + currentBatchSize);
const retry = await insertWithRetry(serviceClient, table, subBatch);
if (retry.error) {
errors += subBatch.length;
if (errorMessages.length < 20) errorMessages.push(retry.error.message || msg);
} else {
inserted += subBatch.length;
}
}
} else {
log(`❌ Batch error (${batch.length} rows): ${msg}`);
errors += batch.length;
if (errorMessages.length < 20) errorMessages.push(msg);
}
} else {
inserted += batch.length;
consecutiveSuccesses++;
// Grow batch size after 3 consecutive successes
if (consecutiveSuccesses >= 3 && currentBatchSize < MAX_BATCH_SIZE) {
const oldSize = currentBatchSize;
currentBatchSize = Math.min(MAX_BATCH_SIZE, Math.floor(currentBatchSize * BATCH_GROW_FACTOR));
if (currentBatchSize !== oldSize) {
log(`🚀 3 successes — growing batch ${oldSize}${currentBatchSize}`);
}
consecutiveSuccesses = 0;
}
}
};
const skipRows = (!rangeSupported && !isZip) ? (job.skip_rows || 0) : 0;
if (skipRows > 0) {
log(`⏩ Will skip ${skipRows.toLocaleString()} rows (no Range support)`);
}
while (true) {
if (Date.now() - startTime > TIME_LIMIT_MS) { stoppedByLimit = true; break; }
if (processedRows >= MAX_ROWS_PER_CALL) { stoppedByLimit = true; break; }
const { done: streamDone, value } = await streamReader.read();
const chunk = value ? decoder.decode(value, { stream: !streamDone }) : "";
if (value) bytesConsumed += value.byteLength;
const text = leftover + chunk;
const lines = text.split(/\r?\n/);
leftover = streamDone ? "" : (lines.pop() || "");
for (const line of lines) {
const trimmed = line.trim();
if (!trimmed) continue;
if (headers.length === 0) {
const cleanLine = trimmed.charCodeAt(0) === 0xfeff ? trimmed.slice(1) : trimmed;
const tabCount = (cleanLine.match(/\t/g) || []).length;
const semiCount = (cleanLine.match(/;/g) || []).length;
const commaCount = (cleanLine.match(/,/g) || []).length;
delimiter = tabCount >= semiCount && tabCount >= commaCount ? "\t"
: semiCount >= commaCount ? ";" : ",";
const candidateHeaders = parseLine(cleanLine, delimiter);
if (candidateHeaders.length < 3) continue;
headers = candidateHeaders;
log(`📋 Headers (${headers.length}): ${headers.slice(0, 5).join(", ")}`);
continue;
}
// Auto-detect delimiter for Range resume
if (needsHeaderDetection === false && totalRowsRead === 0 && processedRows === 0) {
const tabCount = (trimmed.match(/\t/g) || []).length;
const semiCount = (trimmed.match(/;/g) || []).length;
const commaCount = (trimmed.match(/,/g) || []).length;
delimiter = tabCount >= semiCount && tabCount >= commaCount ? "\t"
: semiCount >= commaCount ? ";" : ",";
}
totalRowsRead++;
if (skipRows > 0 && totalRowsRead <= skipRows) {
if (totalRowsRead % 500000 === 0) {
log(`⏩ Skipping… ${totalRowsRead.toLocaleString()} / ${skipRows.toLocaleString()}`);
}
continue;
}
const values = parseLine(trimmed, delimiter);
const row: Record<string, string> = {};
if (type === "fiber") {
headers.forEach((h, idx) => {
const norm = normalizeHeader(h);
const dbCol = FIBER_HEADER_MAP[norm];
if (dbCol && dbCol !== "_skip") {
row[dbCol] = values[idx] || "";
}
});
} else {
headers.forEach((h, idx) => {
row[h] = values[idx] || "";
});
}
const mapped = type === "addresses" ? mapAddressRow(row) : mapFiberRow(row);
if (mapped) {
pendingRows.push(mapped);
processedRows++;
}
if (pendingRows.length >= currentBatchSize) {
await flushBatch(pendingRows);
pendingRows = [];
}
// Log progress at regular intervals
if (processedRows > 0 && processedRows % LOG_INTERVAL === 0) {
const elapsed = ((Date.now() - startTime) / 1000).toFixed(0);
const rate = (processedRows / ((Date.now() - startTime) / 1000)).toFixed(0);
log(`📊 ${processedRows.toLocaleString()} rows (+${inserted.toLocaleString()} ok, ${errors} err) | ${rate} rows/s | batch=${currentBatchSize} | ${elapsed}s`);
// Persist progress periodically
await updateJob({
total_inserted: totalInserted + inserted,
total_errors: totalErrors + errors,
byte_offset: byteOffset + bytesConsumed,
});
}
if (processedRows >= MAX_ROWS_PER_CALL || Date.now() - startTime > TIME_LIMIT_MS) {
stoppedByLimit = true;
break;
}
}
if (stoppedByLimit) break;
if (streamDone) { done = true; break; }
}
// Flush remaining
if (pendingRows.length > 0) {
await flushBatch(pendingRows);
}
if (!done) {
try { streamReader.cancel(); } catch {}
}
const newByteOffset = byteOffset + bytesConsumed;
const nextSkipRows = (job.skip_rows || 0) + processedRows;
totalInserted += inserted;
totalErrors += errors;
const isComplete = done && !stoppedByLimit;
const elapsed = ((Date.now() - startTime) / 1000).toFixed(1);
const rate = (processedRows / (parseFloat(elapsed) || 1)).toFixed(0);
if (isComplete) {
log(`✅ COMPLETE: ${processedRows.toLocaleString()} rows this run (${rate}/s), total: ${totalInserted.toLocaleString()}, errors: ${totalErrors} (${elapsed}s)`);
// Recreate GIN indexes now that import is done
if (type === "addresses") {
await createGinIndexes(serviceClient, log);
}
await updateJob({
status: "done",
byte_offset: newByteOffset,
skip_rows: nextSkipRows,
total_inserted: totalInserted,
total_errors: totalErrors,
error_messages: errorMessages.slice(-20),
indexes_dropped: false,
});
} else {
log(`⏸️ Chunk: +${processedRows.toLocaleString()} rows (${rate}/s), inserted=${inserted.toLocaleString()}, batch=${currentBatchSize}, byteOffset=${newByteOffset.toLocaleString()} (${elapsed}s)`);
await updateJob({
status: "running",
byte_offset: newByteOffset,
skip_rows: nextSkipRows,
total_inserted: totalInserted,
total_errors: totalErrors,
error_messages: errorMessages.slice(-20),
headers_cache: headers,
current_batch_size: currentBatchSize,
});
}
return new Response(
JSON.stringify({
success: true,
jobId: job.id,
processed: processedRows,
inserted,
totalInserted,
done: isComplete,
byteOffset: newByteOffset,
}),
{ headers: { ...corsHeaders, "Content-Type": "application/json" } }
);
} catch (err: any) {
log(`💥 Fatal: ${err.message}`);
console.error("[process-import-job] Error:", err);
await updateJob({
status: "paused",
error_messages: [...errorMessages, err.message].slice(-20),
});
return new Response(JSON.stringify({ error: err.message }), {
status: 500,
headers: { ...corsHeaders, "Content-Type": "application/json" },
});
}
});
// ─── Streaming ZIP Decompression (legacy) ───────────────────────────────────
async function decompressZipStream(
zipStream: ReadableStream<Uint8Array>,
log: (msg: string) => void
): Promise<ReadableStream<Uint8Array>> {
const reader = zipStream.getReader();
let buffer = new Uint8Array(0);
async function readBytes(n: number): Promise<Uint8Array> {
while (buffer.length < n) {
const { done, value } = await reader.read();
if (done) throw new Error("Unexpected end of ZIP stream");
const newBuf = new Uint8Array(buffer.length + value.length);
newBuf.set(buffer);
newBuf.set(value, buffer.length);
buffer = newBuf;
}
const result = buffer.slice(0, n);
buffer = buffer.slice(n);
return result;
}
const header = await readBytes(30);
const view = new DataView(header.buffer, header.byteOffset, header.byteLength);
const sig = view.getUint32(0, true);
if (sig !== 0x04034b50) {
throw new Error(`Invalid ZIP local file header signature: 0x${sig.toString(16)}`);
}
const method = view.getUint16(8, true);
const compressedSize = view.getUint32(18, true);
const filenameLen = view.getUint16(26, true);
const extraLen = view.getUint16(28, true);
const nameExtra = await readBytes(filenameLen + extraLen);
const filename = new TextDecoder().decode(nameExtra.slice(0, filenameLen));
log(`📄 ZIP entry: "${filename}" method=${method === 8 ? "DEFLATE" : method === 0 ? "STORED" : method} size=${compressedSize > 0 ? (compressedSize / 1024 / 1024).toFixed(1) + "MB" : "streaming"}`);
const remainingStream = new ReadableStream<Uint8Array>({
async pull(controller) {
if (buffer.length > 0) {
controller.enqueue(buffer);
buffer = new Uint8Array(0);
return;
}
const { done, value } = await reader.read();
if (done) { controller.close(); return; }
controller.enqueue(value);
},
cancel() { reader.cancel(); },
});
if (method === 0) {
return remainingStream;
} else if (method === 8) {
log("🔓 Streaming DEFLATE decompression...");
return remainingStream.pipeThrough(new DecompressionStream("deflate-raw"));
} else {
throw new Error(`Unsupported ZIP compression method: ${method}`);
}
}