Compare commits

..

10 Commits

11 changed files with 279 additions and 93 deletions

9
opencode.jsonc Normal file
View File

@@ -0,0 +1,9 @@
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"marketplace-scrape": {
"type": "remote",
"url": "http://localhost:4006/mcp"
}
}
}

View File

@@ -2,7 +2,13 @@
"name": "marketplace-scrapers-monorepo",
"version": "1.0.0",
"scripts": {
"ci": "biome ci"
"ci": "biome ci",
"clean": "rm -rf dist",
"build:api": "bun build ./packages/api-server/src/index.ts --target=bun --outdir=./dist/api --minify",
"build:mcp": "bun build ./packages/mcp-server/src/index.ts --target=bun --outdir=./dist/mcp --minify",
"build:all": "bun run build:api && bun run build:mcp",
"build": "bun run clean && bun run build:all",
"start": "./scripts/start.sh"
},
"private": true,
"type": "module",

View File

@@ -46,7 +46,7 @@ export async function kijijiRoute(req: Request): Promise<Response> {
try {
const items = await fetchKijijiItems(
SEARCH_QUERY,
1,
4, // 4 requests per second for faster scraping
"https://www.kijiji.ca",
searchOptions,
{},

View File

@@ -1,4 +1,5 @@
/* eslint-disable @typescript-eslint/no-explicit-any */
import { parseHTML } from "linkedom";
import { delay } from "../utils/delay";
// ----------------------------- Types -----------------------------

View File

@@ -1,5 +1,3 @@
/* eslint-disable @typescript-eslint/no-explicit-any */
import cliProgress from "cli-progress";
import { parseHTML } from "linkedom";
import type { HTMLString } from "../types/common";
@@ -842,8 +840,7 @@ export function parseFacebookAds(
title,
listingPrice: {
amountFormatted:
priceObj.formatted_amount ||
formatCentsToCurrency(cents / 100, "en-CA"),
priceObj.formatted_amount || formatCentsToCurrency(cents, "en-CA"),
cents,
currency: priceObj.currency || "CAD", // Facebook marketplace often uses CAD
},
@@ -890,8 +887,7 @@ export function parseFacebookItem(
if (!Number.isNaN(amount)) {
cents = Math.round(amount * 100);
amountFormatted =
item.formatted_price?.text ||
formatCentsToCurrency(cents / 100, "en-CA");
item.formatted_price?.text || formatCentsToCurrency(cents, "en-CA");
}
}
}

View File

@@ -1,5 +1,3 @@
/* eslint-disable @typescript-eslint/no-explicit-any */
import cliProgress from "cli-progress";
import { parseHTML } from "linkedom";
import unidecode from "unidecode";
@@ -496,7 +494,7 @@ function _parseListing(
const cents = price?.amount != null ? Number(price.amount) : undefined;
const amountFormatted =
cents != null ? formatCentsToCurrency(cents / 100, "en-CA") : undefined;
cents != null ? formatCentsToCurrency(cents, "en-CA") : undefined;
const numberOfViews =
metrics?.views != null ? Number(metrics.views) : undefined;
@@ -575,7 +573,7 @@ export async function parseDetailedListing(
const cents = price?.amount != null ? Number(price.amount) : undefined;
const amountFormatted =
cents != null ? formatCentsToCurrency(cents / 100, "en-CA") : undefined;
cents != null ? formatCentsToCurrency(cents, "en-CA") : undefined;
const numberOfViews =
metrics?.views != null ? Number(metrics.views) : undefined;
@@ -758,51 +756,75 @@ export default async function fetchKijijiItems(
`\nFound ${newListingLinks.length} new listing links on page ${page}. Total unique: ${seenUrls.size}`,
);
// Fetch details for this page's listings
const progressBar = new cliProgress.SingleBar(
{},
cliProgress.Presets.shades_classic,
);
// Fetch details for this page's listings with controlled concurrency
const isTTY = process.stdout?.isTTY ?? false;
const progressBar = isTTY
? new cliProgress.SingleBar({}, cliProgress.Presets.shades_classic)
: null;
const totalProgress = newListingLinks.length;
let currentProgress = 0;
progressBar.start(totalProgress, currentProgress);
progressBar?.start(totalProgress, currentProgress);
for (const link of newListingLinks) {
try {
const html = await fetchHtml(link, DELAY_MS, {
onRateInfo: (remaining, reset) => {
if (remaining && reset) {
console.log(
`\nItem - Rate limit remaining: ${remaining}, reset in: ${reset}s`,
);
}
},
});
const parsed = await parseDetailedListing(
html,
BASE_URL,
finalListingOptions,
// Process in batches for controlled concurrency
const CONCURRENT_REQUESTS = REQUESTS_PER_SECOND * 2; // 2x rate for faster processing
const results: (DetailedListing | null)[] = [];
for (let i = 0; i < newListingLinks.length; i += CONCURRENT_REQUESTS) {
const batch = newListingLinks.slice(i, i + CONCURRENT_REQUESTS);
const batchPromises = batch.map(async (link) => {
try {
const html = await fetchHtml(link, 0, {
// No per-request delay, batch handles rate limit
onRateInfo: (remaining, reset) => {
if (remaining && reset) {
console.log(
`\nItem - Rate limit remaining: ${remaining}, reset in: ${reset}s`,
);
}
},
});
const parsed = await parseDetailedListing(
html,
BASE_URL,
finalListingOptions,
);
return parsed;
} catch (err) {
if (err instanceof HttpError) {
console.error(
`\nFailed to fetch ${link}\n - ${err.statusCode} ${err.message}`,
);
} else {
console.error(
`\nFailed to fetch ${link}\n - ${String((err as Error)?.message || err)}`,
);
}
return null;
} finally {
currentProgress++;
progressBar?.update(currentProgress);
if (!progressBar) {
console.log(`Progress: ${currentProgress}/${totalProgress}`);
}
}
});
const batchResults = await Promise.all(batchPromises);
results.push(...batchResults);
// Wait between batches to respect rate limit
if (i + CONCURRENT_REQUESTS < newListingLinks.length) {
await new Promise((resolve) =>
setTimeout(resolve, DELAY_MS * batch.length),
);
if (parsed) {
allListings.push(parsed);
}
} catch (err) {
if (err instanceof HttpError) {
console.error(
`\nFailed to fetch ${link}\n - ${err.statusCode} ${err.message}`,
);
} else {
console.error(
`\nFailed to fetch ${link}\n - ${String((err as Error)?.message || err)}`,
);
}
} finally {
currentProgress++;
progressBar.update(currentProgress);
}
}
progressBar.stop();
allListings.push(
...results.filter((r): r is DetailedListing => r !== null),
);
progressBar?.stop();
// If we got fewer results than expected (40 per page), we've reached the end
if (searchResults.length < 40) {

View File

@@ -5,7 +5,7 @@ const PORT = process.env.MCP_PORT || 4006;
const server = Bun.serve({
port: PORT as number | string,
idleTimeout: 0,
idleTimeout: 255, // 255 seconds (max allowed)
routes: {
// MCP metadata discovery endpoint
"/.well-known/mcp/server-card.json": new Response(

View File

@@ -1,10 +1,8 @@
import {
fetchEbayItems,
fetchFacebookItems,
fetchKijijiItems,
} from "@marketplace-scrapers/core";
import { tools } from "./tools";
const API_BASE_URL = process.env.API_BASE_URL || "http://localhost:4005/api";
const API_TIMEOUT = Number(process.env.API_TIMEOUT) || 180000; // 3 minutes default
/**
* Handle MCP JSON-RPC 2.0 protocol requests
*/
@@ -105,24 +103,44 @@ export async function handleMcpRequest(req: Request): Promise<Response> {
error: { code: -32602, message: "query parameter is required" },
});
}
const searchOptions = {
location: args.location,
category: args.category,
keywords: args.keywords,
sortBy: args.sortBy,
sortOrder: args.sortOrder,
maxPages: args.maxPages || 5,
priceMin: args.priceMin,
priceMax: args.priceMax,
};
const items = await fetchKijijiItems(
query,
1,
"https://www.kijiji.ca",
searchOptions,
{},
const params = new URLSearchParams({ q: query });
if (args.location) params.append("location", args.location);
if (args.category) params.append("category", args.category);
if (args.keywords) params.append("keywords", args.keywords);
if (args.sortBy) params.append("sortBy", args.sortBy);
if (args.sortOrder) params.append("sortOrder", args.sortOrder);
if (args.maxPages)
params.append("maxPages", args.maxPages.toString());
if (args.priceMin)
params.append("priceMin", args.priceMin.toString());
if (args.priceMax)
params.append("priceMax", args.priceMax.toString());
console.log(
`[MCP] Calling Kijiji API: ${API_BASE_URL}/kijiji?${params.toString()}`,
);
const response = await Promise.race([
fetch(`${API_BASE_URL}/kijiji?${params.toString()}`),
new Promise<Response>((_, reject) =>
setTimeout(
() =>
reject(new Error(`Request timed out after ${API_TIMEOUT}ms`)),
API_TIMEOUT,
),
),
]);
if (!response.ok) {
const errorText = await response.text();
console.error(
`[MCP] Kijiji API error ${response.status}: ${errorText}`,
);
throw new Error(`API returned ${response.status}: ${errorText}`);
}
result = await response.json();
console.log(
`[MCP] Kijiji returned ${Array.isArray(result) ? result.length : 0} items`,
);
result = items || [];
} else if (name === "search_facebook") {
const query = args.query;
if (!query) {
@@ -132,15 +150,37 @@ export async function handleMcpRequest(req: Request): Promise<Response> {
error: { code: -32602, message: "query parameter is required" },
});
}
const items = await fetchFacebookItems(
query,
1,
args.location || "toronto",
args.maxItems || 25,
args.cookiesSource,
undefined,
const params = new URLSearchParams({ q: query });
if (args.location) params.append("location", args.location);
if (args.maxItems)
params.append("maxItems", args.maxItems.toString());
if (args.cookiesSource) params.append("cookies", args.cookiesSource);
console.log(
`[MCP] Calling Facebook API: ${API_BASE_URL}/facebook?${params.toString()}`,
);
const response = await Promise.race([
fetch(`${API_BASE_URL}/facebook?${params.toString()}`),
new Promise<Response>((_, reject) =>
setTimeout(
() =>
reject(new Error(`Request timed out after ${API_TIMEOUT}ms`)),
API_TIMEOUT,
),
),
]);
if (!response.ok) {
const errorText = await response.text();
console.error(
`[MCP] Facebook API error ${response.status}: ${errorText}`,
);
throw new Error(`API returned ${response.status}: ${errorText}`);
}
result = await response.json();
console.log(
`[MCP] Facebook returned ${Array.isArray(result) ? result.length : 0} items`,
);
result = items || [];
} else if (name === "search_ebay") {
const query = args.query;
if (!query) {
@@ -150,18 +190,49 @@ export async function handleMcpRequest(req: Request): Promise<Response> {
error: { code: -32602, message: "query parameter is required" },
});
}
const items = await fetchEbayItems(query, 1, {
minPrice: args.minPrice,
maxPrice: args.maxPrice,
strictMode: args.strictMode || false,
exclusions: args.exclusions || [],
keywords: args.keywords || [query],
buyItNowOnly: args.buyItNowOnly !== false,
canadaOnly: args.canadaOnly !== false,
});
const params = new URLSearchParams({ q: query });
if (args.minPrice)
params.append("minPrice", args.minPrice.toString());
if (args.maxPrice)
params.append("maxPrice", args.maxPrice.toString());
if (args.strictMode !== undefined)
params.append("strictMode", args.strictMode.toString());
if (args.exclusions?.length)
params.append("exclusions", args.exclusions.join(","));
if (args.keywords?.length)
params.append("keywords", args.keywords.join(","));
if (args.buyItNowOnly !== undefined)
params.append("buyItNowOnly", args.buyItNowOnly.toString());
if (args.canadaOnly !== undefined)
params.append("canadaOnly", args.canadaOnly.toString());
if (args.maxItems)
params.append("maxItems", args.maxItems.toString());
const results = args.maxItems ? items.slice(0, args.maxItems) : items;
result = results || [];
console.log(
`[MCP] Calling eBay API: ${API_BASE_URL}/ebay?${params.toString()}`,
);
const response = await Promise.race([
fetch(`${API_BASE_URL}/ebay?${params.toString()}`),
new Promise<Response>((_, reject) =>
setTimeout(
() =>
reject(new Error(`Request timed out after ${API_TIMEOUT}ms`)),
API_TIMEOUT,
),
),
]);
if (!response.ok) {
const errorText = await response.text();
console.error(
`[MCP] eBay API error ${response.status}: ${errorText}`,
);
throw new Error(`API returned ${response.status}: ${errorText}`);
}
result = await response.json();
console.log(
`[MCP] eBay returned ${Array.isArray(result) ? result.length : 0} items`,
);
} else {
return Response.json({
jsonrpc: "2.0",

26
scripts/biome-symlink.sh Executable file
View File

@@ -0,0 +1,26 @@
#!/usr/bin/env bash
# Get the path to the system biome executable
BIOME_PATH=$(which biome)
if [ -z "$BIOME_PATH" ]; then
echo "Error: biome executable not found in PATH"
exit 1
fi
# Find all biome executables in node_modules
files=$(fd biome node_modules --type executable --no-ignore --follow)
if [ -z "$files" ]; then
echo "No biome executables found in node_modules"
exit 0
fi
# Replace each with a symlink to the system biome
for file in $files; do
echo "Replacing $file with symlink to $BIOME_PATH"
rm "$file"
ln -s "$BIOME_PATH" "$file"
done
echo "Done."

30
scripts/remove-eslint.sh Executable file
View File

@@ -0,0 +1,30 @@
#!/usr/bin/env bash
PATTERN="eslint"
FILES="$(fd .)" # Or use 'find .' to search recursively
for file in $FILES; do
if [[ -f "$file" ]]; then
# 1. Use rg with line numbers (-n) and only the matched line (-o)
# 2. Use awk to print ONLY the line number (field 1)
# 3. Use xargs to pass multiple line numbers to a single sed command
LINE_NUMBERS=$(rg --line-number --no-filename "$PATTERN" "$file" | awk -F':' '{print $1}' | tr '\n' ',')
# Remove trailing comma if any
LINE_NUMBERS=${LINE_NUMBERS%,}
if [[ -n "$LINE_NUMBERS" ]]; then
echo "Deleting lines $LINE_NUMBERS from $file..."
# Use sed to delete the specified comma-separated line numbers in-place (-i)
# NOTE: The syntax for -i might vary slightly between GNU sed (Linux) and BSD sed (macOS).
sed -i.bak "${LINE_NUMBERS}d" "$file"
# Optional: Remove the backup file created by sed -i.bak
# rm "${file}.bak"
else
echo "$file: No lines matching pattern found."
fi
fi
done

25
scripts/start.sh Executable file
View File

@@ -0,0 +1,25 @@
#!/usr/bin/env bash
set -e
# Trap SIGTERM and SIGINT for graceful shutdown
trap 'echo "Received shutdown signal, stopping services..."; kill -TERM $API_PID $MCP_PID 2>/dev/null; wait' TERM INT
# Start API Server in background
echo "Starting API Server on port ${API_PORT:-4005}..."
bun dist/api/index.js &
API_PID=$!
# Give API server a moment to initialize
sleep 1
# Start MCP Server in background
echo "Starting MCP Server on port ${API_PORT:-4006}..."
bun dist/mcp/index.js &
MCP_PID=$!
echo "Both services started successfully"
echo "API Server PID: $API_PID"
echo "MCP Server PID: $MCP_PID"
# Wait for both processes
wait $API_PID $MCP_PID