Álvaro Valenzuela Valdes commited on
Commit ·
7259e84
1
Parent(s): 4ceb9d2
feat: Live Compra Ágil web scraper integration for real-time discovery
Browse files- backend/app/routers/tenders.py +5 -0
- backend/app/services/scraper.py +76 -0
- frontend/app/page.tsx +13 -3
- frontend/components/TenderSearch.tsx +3 -2
- frontend/lib/api.ts +8 -0
backend/app/routers/tenders.py
CHANGED
|
@@ -59,3 +59,8 @@ def get_tenders_count(db: Session = Depends(get_db)):
|
|
| 59 |
@router.post("/tenders/sync")
|
| 60 |
async def manual_sync(keyword: Optional[str] = None, db: Session = Depends(get_db)):
|
| 61 |
return await sync_tenders_to_db(db, keyword=keyword)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
@router.post("/tenders/sync")
|
| 60 |
async def manual_sync(keyword: Optional[str] = None, db: Session = Depends(get_db)):
|
| 61 |
return await sync_tenders_to_db(db, keyword=keyword)
|
| 62 |
+
|
| 63 |
+
@router.get("/tenders/scrape", response_model=List[Tender])
|
| 64 |
+
async def live_scrape(keyword: str):
|
| 65 |
+
from app.services.scraper import scrape_compra_agil
|
| 66 |
+
return await scrape_compra_agil(keyword)
|
backend/app/services/scraper.py
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import httpx
|
| 2 |
+
from bs4 import BeautifulSoup
|
| 3 |
+
from typing import List
|
| 4 |
+
from app.schemas.tender import Tender
|
| 5 |
+
from datetime import datetime
|
| 6 |
+
import re
|
| 7 |
+
|
| 8 |
+
async def scrape_compra_agil(keywords: str) -> List[Tender]:
|
| 9 |
+
"""
|
| 10 |
+
Scrapes the Mercado Público Compra Ágil search results page.
|
| 11 |
+
"""
|
| 12 |
+
# Base URL for Compra Ágil search
|
| 13 |
+
url = f"https://buscador.mercadopublico.cl/compra-agil?keywords={keywords}&status=2&order_by=recent"
|
| 14 |
+
|
| 15 |
+
headers = {
|
| 16 |
+
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
|
| 17 |
+
}
|
| 18 |
+
|
| 19 |
+
try:
|
| 20 |
+
async with httpx.AsyncClient(timeout=30.0, follow_redirects=True) as client:
|
| 21 |
+
print(f"[Scraper] Navigating to: {url}")
|
| 22 |
+
response = await client.get(url, headers=headers)
|
| 23 |
+
response.raise_for_status()
|
| 24 |
+
|
| 25 |
+
soup = BeautifulSoup(response.text, 'html.parser')
|
| 26 |
+
|
| 27 |
+
# Find tender cards (this is based on standard Mercado Público searcher structure)
|
| 28 |
+
# Note: The classes might change, but usually they are consistent
|
| 29 |
+
tenders = []
|
| 30 |
+
|
| 31 |
+
# Find elements that look like tender containers
|
| 32 |
+
cards = soup.select('.card-tender, .item-busqueda, .search-result-item')
|
| 33 |
+
|
| 34 |
+
if not cards:
|
| 35 |
+
# Fallback: try to find any link with a COT26 pattern
|
| 36 |
+
all_text = soup.get_text()
|
| 37 |
+
codes = re.findall(r'[0-9]+-[0-9]+-COT26', all_text)
|
| 38 |
+
print(f"[Scraper] Found {len(codes)} codes via regex fallback.")
|
| 39 |
+
# We would need more info to build a full Tender object from regex
|
| 40 |
+
|
| 41 |
+
for card in cards[:20]: # Limit for performance
|
| 42 |
+
try:
|
| 43 |
+
name_elem = card.select_one('.title-tender, h3, .name')
|
| 44 |
+
code_elem = card.select_one('.code-tender, .code, span:contains("COT26")')
|
| 45 |
+
buyer_elem = card.select_one('.buyer-name, .organismo')
|
| 46 |
+
|
| 47 |
+
if not name_elem or not code_elem:
|
| 48 |
+
continue
|
| 49 |
+
|
| 50 |
+
name = name_elem.get_text(strip=True)
|
| 51 |
+
code = code_elem.get_text(strip=True)
|
| 52 |
+
buyer = buyer_elem.get_text(strip=True) if buyer_elem else "Unknown"
|
| 53 |
+
|
| 54 |
+
tenders.append(Tender(
|
| 55 |
+
code=code,
|
| 56 |
+
name=name,
|
| 57 |
+
description=name,
|
| 58 |
+
buyer=buyer,
|
| 59 |
+
status="Publicada",
|
| 60 |
+
closing_date=datetime.now().strftime("%Y-%m-%d"),
|
| 61 |
+
estimated_amount=0,
|
| 62 |
+
source="Mercado Público (Scraped)",
|
| 63 |
+
region="Nacional",
|
| 64 |
+
sector="Compra Ágil",
|
| 65 |
+
items=[],
|
| 66 |
+
attachments=[]
|
| 67 |
+
))
|
| 68 |
+
except Exception as e:
|
| 69 |
+
print(f"Error parsing card: {e}")
|
| 70 |
+
|
| 71 |
+
print(f"[Scraper] Successfully scraped {len(tenders)} tenders.")
|
| 72 |
+
return tenders
|
| 73 |
+
|
| 74 |
+
except Exception as e:
|
| 75 |
+
print(f"❌ Scraper error: {e}")
|
| 76 |
+
return []
|
frontend/app/page.tsx
CHANGED
|
@@ -114,9 +114,19 @@ export default function HomePage() {
|
|
| 114 |
window.history.pushState({}, '', `?tab=tender_search&q=${encodeURIComponent(value)}`);
|
| 115 |
};
|
| 116 |
|
| 117 |
-
const handleSearch = async (params: { keyword?: string; buyer?: string; provider_code?: string; date?: string; skip?: number; limit?: number }) => {
|
| 118 |
-
|
| 119 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 120 |
};
|
| 121 |
|
| 122 |
const handleProfileSave = async (profile: CompanyProfileType) => {
|
|
|
|
| 114 |
window.history.pushState({}, '', `?tab=tender_search&q=${encodeURIComponent(value)}`);
|
| 115 |
};
|
| 116 |
|
| 117 |
+
const handleSearch = async (params: { keyword?: string; buyer?: string; provider_code?: string; date?: string; skip?: number; limit?: number; isAgile?: boolean }) => {
|
| 118 |
+
try {
|
| 119 |
+
let results: Tender[];
|
| 120 |
+
if (params.isAgile && params.keyword) {
|
| 121 |
+
const { scrapeTenders } = await import("../lib/api");
|
| 122 |
+
results = await scrapeTenders(params.keyword);
|
| 123 |
+
} else {
|
| 124 |
+
results = await searchTenders(params);
|
| 125 |
+
}
|
| 126 |
+
setTenders(results);
|
| 127 |
+
} catch (error) {
|
| 128 |
+
console.error("Search error:", error);
|
| 129 |
+
}
|
| 130 |
};
|
| 131 |
|
| 132 |
const handleProfileSave = async (profile: CompanyProfileType) => {
|
frontend/components/TenderSearch.tsx
CHANGED
|
@@ -7,7 +7,7 @@ import { Language, translations } from "../lib/translations";
|
|
| 7 |
|
| 8 |
type Props = {
|
| 9 |
tenders: Tender[];
|
| 10 |
-
onSearch: (params: { keyword?: string; buyer?: string; provider_code?: string; date?: string; skip?: number; limit?: number }) => void;
|
| 11 |
onAnalyze: (tender: Tender) => void;
|
| 12 |
forceShowFollowed?: boolean;
|
| 13 |
initialKeyword?: string;
|
|
@@ -83,7 +83,8 @@ export default function TenderSearch({ tenders, onSearch, onAnalyze, forceShowFo
|
|
| 83 |
buyer: buyerCode,
|
| 84 |
date,
|
| 85 |
skip: 0,
|
| 86 |
-
limit: itemsPerPage
|
|
|
|
| 87 |
});
|
| 88 |
} catch (error) {
|
| 89 |
console.error(error);
|
|
|
|
| 7 |
|
| 8 |
type Props = {
|
| 9 |
tenders: Tender[];
|
| 10 |
+
onSearch: (params: { keyword?: string; buyer?: string; provider_code?: string; date?: string; skip?: number; limit?: number; isAgile?: boolean }) => void;
|
| 11 |
onAnalyze: (tender: Tender) => void;
|
| 12 |
forceShowFollowed?: boolean;
|
| 13 |
initialKeyword?: string;
|
|
|
|
| 83 |
buyer: buyerCode,
|
| 84 |
date,
|
| 85 |
skip: 0,
|
| 86 |
+
limit: itemsPerPage,
|
| 87 |
+
isAgile: isAgileMode
|
| 88 |
});
|
| 89 |
} catch (error) {
|
| 90 |
console.error(error);
|
frontend/lib/api.ts
CHANGED
|
@@ -114,3 +114,11 @@ export async function syncDatabase() {
|
|
| 114 |
}
|
| 115 |
return res.json();
|
| 116 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 114 |
}
|
| 115 |
return res.json();
|
| 116 |
}
|
| 117 |
+
|
| 118 |
+
export async function scrapeTenders(keyword: string): Promise<Tender[]> {
|
| 119 |
+
const res = await fetch(`${API_BASE}/api/tenders/scrape?keyword=${encodeURIComponent(keyword)}`);
|
| 120 |
+
if (!res.ok) {
|
| 121 |
+
throw new Error("Error scraping tenders");
|
| 122 |
+
}
|
| 123 |
+
return res.json();
|
| 124 |
+
}
|