FastApi-ISS/scraper.py

94 lines
2.9 KiB
Python

from sqlalchemy.orm import Session
from database import InstagramAccount, InstagramPost
from utils.boxapi_client import fetch_boxapi_posts
from utils.s3_uploader import upload_media_from_url
from schemas import ScrapeRequest, ScrapedPost
from typing import List
from datetime import datetime
def scrape_and_store(db: Session, req: ScrapeRequest) -> List[ScrapedPost]:
posts = fetch_boxapi_posts(req.username, req.max_count)
if not posts:
return []
# Get or create InstagramAccount
account = db.query(InstagramAccount).filter_by(username=req.username).first()
if not account:
account = InstagramAccount(
username=req.username,
seller_id=req.seller_id,
is_active=True,
last_synced=None
)
db.add(account)
db.commit()
db.refresh(account)
result: List[ScrapedPost] = []
new_post_count = 0
for post in posts:
post_id = str(post.get("id") or post.get("pk"))
caption = post.get("caption", "")
thumbnail_url = post.get("thumbnail_url")
taken_at = post.get("taken_at") or datetime.utcnow()
media_type = post.get("media_type") or "image"
if db.query(InstagramPost).filter_by(ig_post_id=post_id).first():
continue
media_urls: List[str] = []
for media in post.get("media", []):
media_url = media.get("media_url")
if not media_url:
continue
try:
local_url = upload_media_from_url(media_url, str(req.seller_id), f"{post_id}/")
if local_url:
media_urls.append(local_url)
except Exception as e:
print(f"⚠️ Error uploading media: {e}")
continue
if not media_urls:
continue
# Upload thumbnail (optional)
final_thumbnail = thumbnail_url
if thumbnail_url:
try:
thumb_uploaded = upload_media_from_url(thumbnail_url, str(req.seller_id), f"{post_id}/thumbnail")
if thumb_uploaded:
final_thumbnail = thumb_uploaded
except Exception as e:
print(f"⚠️ Failed to upload thumbnail: {e}")
# Save to DB
post_entry = InstagramPost(
ig_post_id=post_id,
account_id=account.id,
media_type=media_type,
caption=caption,
thumbnail_url=final_thumbnail,
local_media=media_urls,
created_at=taken_at,
)
db.add(post_entry)
db.commit()
result.append(ScrapedPost(
ig_post_id=post_id,
media_type=media_type,
caption=caption,
thumbnail_url=final_thumbnail,
local_media=media_urls,
created_at=taken_at
))
new_post_count += 1
if new_post_count >= req.max_count:
break
return result