feat: add sources to retrieve news and divide the IA reflexions in 2 steps to limit the number of news
This commit is contained in:
@ -1,206 +1,94 @@
|
||||
package bloomberg
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"context"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
"github.com/chromedp/chromedp"
|
||||
"github.com/tradarr/backend/internal/scraper"
|
||||
)
|
||||
|
||||
type Bloomberg struct {
|
||||
username string
|
||||
password string
|
||||
chromePath string
|
||||
scraperURL string
|
||||
client *http.Client
|
||||
}
|
||||
|
||||
func New(username, password, chromePath string) *Bloomberg {
|
||||
return &Bloomberg{username: username, password: password, chromePath: chromePath}
|
||||
func New(scraperURL string) *Bloomberg {
|
||||
if scraperURL == "" {
|
||||
scraperURL = "http://scraper:3001"
|
||||
}
|
||||
return &Bloomberg{
|
||||
scraperURL: scraperURL,
|
||||
client: &http.Client{Timeout: 10 * time.Minute},
|
||||
}
|
||||
}
|
||||
|
||||
func (b *Bloomberg) Name() string { return "bloomberg" }
|
||||
|
||||
func (b *Bloomberg) Scrape(ctx context.Context, symbols []string) ([]scraper.Article, error) {
|
||||
if b.username == "" || b.password == "" {
|
||||
return nil, fmt.Errorf("bloomberg credentials not configured")
|
||||
}
|
||||
|
||||
opts := []chromedp.ExecAllocatorOption{
|
||||
chromedp.NoFirstRun,
|
||||
chromedp.NoDefaultBrowserCheck,
|
||||
chromedp.Headless,
|
||||
chromedp.DisableGPU,
|
||||
chromedp.Flag("no-sandbox", true),
|
||||
chromedp.Flag("disable-setuid-sandbox", true),
|
||||
chromedp.Flag("disable-dev-shm-usage", true),
|
||||
chromedp.Flag("disable-blink-features", "AutomationControlled"),
|
||||
chromedp.Flag("disable-infobars", true),
|
||||
chromedp.Flag("window-size", "1920,1080"),
|
||||
chromedp.Flag("ignore-certificate-errors", true),
|
||||
chromedp.UserAgent("Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36"),
|
||||
}
|
||||
if b.chromePath != "" {
|
||||
opts = append(opts, chromedp.ExecPath(b.chromePath))
|
||||
}
|
||||
|
||||
allocCtx, cancelAlloc := chromedp.NewExecAllocator(ctx, opts...)
|
||||
defer cancelAlloc()
|
||||
|
||||
chromeCtx, cancelChrome := chromedp.NewContext(allocCtx)
|
||||
defer cancelChrome()
|
||||
|
||||
timeoutCtx, cancelTimeout := context.WithTimeout(chromeCtx, 5*time.Minute)
|
||||
defer cancelTimeout()
|
||||
|
||||
if err := b.login(timeoutCtx); err != nil {
|
||||
return nil, fmt.Errorf("bloomberg login: %w", err)
|
||||
}
|
||||
|
||||
var articles []scraper.Article
|
||||
pages := []string{
|
||||
"https://www.bloomberg.com/markets",
|
||||
"https://www.bloomberg.com/technology",
|
||||
"https://www.bloomberg.com/economics",
|
||||
}
|
||||
for _, u := range pages {
|
||||
pageArticles, err := b.scrapePage(timeoutCtx, u, symbols)
|
||||
if err != nil {
|
||||
fmt.Printf("bloomberg scrape %s: %v\n", u, err)
|
||||
continue
|
||||
}
|
||||
articles = append(articles, pageArticles...)
|
||||
}
|
||||
fmt.Printf("bloomberg: %d articles fetched total\n", len(articles))
|
||||
return articles, nil
|
||||
type scraperRequest struct {
|
||||
Username string `json:"username"`
|
||||
Password string `json:"password"`
|
||||
}
|
||||
|
||||
func (b *Bloomberg) login(ctx context.Context) error {
|
||||
loginCtx, cancel := context.WithTimeout(ctx, 2*time.Minute)
|
||||
defer cancel()
|
||||
|
||||
// Masquer la détection d'automation via JS
|
||||
if err := chromedp.Run(loginCtx,
|
||||
chromedp.ActionFunc(func(ctx context.Context) error {
|
||||
return chromedp.Evaluate(`
|
||||
Object.defineProperty(navigator, 'webdriver', {get: () => undefined});
|
||||
window.chrome = { runtime: {} };
|
||||
`, nil).Do(ctx)
|
||||
}),
|
||||
); err != nil {
|
||||
fmt.Printf("bloomberg: could not inject stealth JS: %v\n", err)
|
||||
}
|
||||
|
||||
err := chromedp.Run(loginCtx,
|
||||
chromedp.Navigate("https://www.bloomberg.com/account/signin"),
|
||||
chromedp.Sleep(2*time.Second),
|
||||
// Essayer plusieurs sélecteurs pour l'email
|
||||
chromedp.ActionFunc(func(ctx context.Context) error {
|
||||
selectors := []string{
|
||||
`input[name="email"]`,
|
||||
`input[type="email"]`,
|
||||
`input[data-type="email"]`,
|
||||
`input[placeholder*="email" i]`,
|
||||
`input[placeholder*="mail" i]`,
|
||||
}
|
||||
for _, sel := range selectors {
|
||||
var count int
|
||||
if err := chromedp.Evaluate(fmt.Sprintf(`document.querySelectorAll('%s').length`, sel), &count).Do(ctx); err == nil && count > 0 {
|
||||
fmt.Printf("bloomberg: using email selector: %s\n", sel)
|
||||
return chromedp.SendKeys(sel, b.username, chromedp.ByQuery).Do(ctx)
|
||||
}
|
||||
}
|
||||
return fmt.Errorf("could not find email input — Bloomberg login page structure may have changed")
|
||||
}),
|
||||
chromedp.Sleep(500*time.Millisecond),
|
||||
// Submit email
|
||||
chromedp.ActionFunc(func(ctx context.Context) error {
|
||||
selectors := []string{`button[type="submit"]`, `input[type="submit"]`, `button[data-testid*="submit"]`}
|
||||
for _, sel := range selectors {
|
||||
var count int
|
||||
if err := chromedp.Evaluate(fmt.Sprintf(`document.querySelectorAll('%s').length`, sel), &count).Do(ctx); err == nil && count > 0 {
|
||||
return chromedp.Click(sel, chromedp.ByQuery).Do(ctx)
|
||||
}
|
||||
}
|
||||
// Fallback: press Enter
|
||||
return chromedp.KeyEvent("\r").Do(ctx)
|
||||
}),
|
||||
chromedp.Sleep(2*time.Second),
|
||||
// Password
|
||||
chromedp.ActionFunc(func(ctx context.Context) error {
|
||||
selectors := []string{`input[type="password"]`, `input[name="password"]`}
|
||||
for _, sel := range selectors {
|
||||
var count int
|
||||
if err := chromedp.Evaluate(fmt.Sprintf(`document.querySelectorAll('%s').length`, sel), &count).Do(ctx); err == nil && count > 0 {
|
||||
fmt.Printf("bloomberg: using password selector: %s\n", sel)
|
||||
return chromedp.SendKeys(sel, b.password, chromedp.ByQuery).Do(ctx)
|
||||
}
|
||||
}
|
||||
return fmt.Errorf("could not find password input")
|
||||
}),
|
||||
chromedp.Sleep(500*time.Millisecond),
|
||||
chromedp.ActionFunc(func(ctx context.Context) error {
|
||||
selectors := []string{`button[type="submit"]`, `input[type="submit"]`}
|
||||
for _, sel := range selectors {
|
||||
var count int
|
||||
if err := chromedp.Evaluate(fmt.Sprintf(`document.querySelectorAll('%s').length`, sel), &count).Do(ctx); err == nil && count > 0 {
|
||||
return chromedp.Click(sel, chromedp.ByQuery).Do(ctx)
|
||||
}
|
||||
}
|
||||
return chromedp.KeyEvent("\r").Do(ctx)
|
||||
}),
|
||||
chromedp.Sleep(3*time.Second),
|
||||
)
|
||||
return err
|
||||
type scraperArticle struct {
|
||||
Title string `json:"title"`
|
||||
URL string `json:"url"`
|
||||
}
|
||||
|
||||
func (b *Bloomberg) scrapePage(ctx context.Context, pageURL string, symbols []string) ([]scraper.Article, error) {
|
||||
pageCtx, cancel := context.WithTimeout(ctx, 60*time.Second)
|
||||
defer cancel()
|
||||
type scraperResponse struct {
|
||||
Articles []scraperArticle `json:"articles"`
|
||||
Error string `json:"error,omitempty"`
|
||||
}
|
||||
|
||||
var articleNodes []map[string]string
|
||||
err := chromedp.Run(pageCtx,
|
||||
chromedp.Navigate(pageURL),
|
||||
chromedp.Sleep(3*time.Second),
|
||||
chromedp.Evaluate(`
|
||||
(function() {
|
||||
var items = [];
|
||||
var seen = new Set();
|
||||
var links = document.querySelectorAll('a[href*="/news/articles"], a[href*="/opinion/"], a[href*="/markets/"]');
|
||||
links.forEach(function(a) {
|
||||
if (seen.has(a.href)) return;
|
||||
seen.add(a.href);
|
||||
var title = a.querySelector('h1,h2,h3,h4,[class*="headline"],[class*="title"]');
|
||||
var text = title ? title.innerText.trim() : a.innerText.trim();
|
||||
if (text.length > 20 && a.href.includes('bloomberg.com')) {
|
||||
items.push({title: text, url: a.href});
|
||||
}
|
||||
});
|
||||
return items.slice(0, 25);
|
||||
})()
|
||||
`, &articleNodes),
|
||||
)
|
||||
func (b *Bloomberg) ScrapeWithCredentials(ctx context.Context, username, password string, symbols []string) ([]scraper.Article, error) {
|
||||
payload, _ := json.Marshal(scraperRequest{Username: username, Password: password})
|
||||
req, err := http.NewRequestWithContext(ctx, http.MethodPost, b.scraperURL+"/bloomberg/scrape", bytes.NewReader(payload))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("navigate %s: %w", pageURL, err)
|
||||
return nil, err
|
||||
}
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
|
||||
resp, err := b.client.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("scraper service unreachable: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, _ := io.ReadAll(resp.Body)
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return nil, fmt.Errorf("scraper service HTTP %d: %s", resp.StatusCode, body)
|
||||
}
|
||||
|
||||
var result scraperResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse scraper response: %w", err)
|
||||
}
|
||||
if result.Error != "" {
|
||||
return nil, fmt.Errorf("bloomberg: %s", result.Error)
|
||||
}
|
||||
|
||||
var articles []scraper.Article
|
||||
now := time.Now()
|
||||
for _, node := range articleNodes {
|
||||
title := strings.TrimSpace(node["title"])
|
||||
url := node["url"]
|
||||
if title == "" || url == "" || !strings.Contains(url, "bloomberg.com") {
|
||||
var articles []scraper.Article
|
||||
for _, a := range result.Articles {
|
||||
title := strings.TrimSpace(a.Title)
|
||||
url := a.URL
|
||||
if title == "" || url == "" {
|
||||
continue
|
||||
}
|
||||
syms := scraper.DetectSymbols(title, symbols)
|
||||
articles = append(articles, scraper.Article{
|
||||
Title: title,
|
||||
Content: title, // contenu minimal — l'article complet nécessite un accès payant
|
||||
Content: title,
|
||||
URL: url,
|
||||
PublishedAt: &now,
|
||||
Symbols: syms,
|
||||
})
|
||||
}
|
||||
fmt.Printf("bloomberg: %d articles fetched\n", len(articles))
|
||||
return articles, nil
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user