scrapper hai hai
This commit is contained in:
parent
02f6ed5b9b
commit
4583e08ca5
56
scraper.py
56
scraper.py
@ -13,7 +13,7 @@ def scrape_and_store(db: Session, req: ScrapeRequest) -> List[ScrapedPost]:
|
||||
|
||||
try:
|
||||
user_id = client.user_id_from_username(req.username)
|
||||
posts = client.user_medias_v1(user_id, amount=30) # fetch more to skip duplicates
|
||||
posts = client.user_medias_v1(user_id, amount=30)
|
||||
except Exception as e:
|
||||
print(f"❌ Failed to fetch posts for @{req.username}: {e}")
|
||||
return []
|
||||
@ -37,50 +37,64 @@ def scrape_and_store(db: Session, req: ScrapeRequest) -> List[ScrapedPost]:
|
||||
for post in posts:
|
||||
post_id = str(post.id)
|
||||
|
||||
# Skip if already in DB
|
||||
if db.query(InstagramPost).filter_by(ig_post_id=post_id).first():
|
||||
continue
|
||||
|
||||
media_items = []
|
||||
|
||||
try:
|
||||
# Handle carousel
|
||||
# Carousel
|
||||
if post.media_type == 8 and hasattr(post, "resources"):
|
||||
for res in post.resources:
|
||||
media_url = res.video_url if res.media_type == 2 else res.thumbnail_url
|
||||
if not media_url:
|
||||
continue
|
||||
local_url = upload_media_from_url(media_url, str(req.seller_id), post_id)
|
||||
if local_url:
|
||||
media_items.append(MediaItem(
|
||||
media_type="video" if res.media_type == 2 else "image",
|
||||
media_url=media_url,
|
||||
local_media=local_url
|
||||
))
|
||||
try:
|
||||
local_url = upload_media_from_url(media_url, str(req.seller_id), f"{post_id}/")
|
||||
if local_url:
|
||||
media_items.append(MediaItem(
|
||||
media_type="video" if res.media_type == 2 else "image",
|
||||
media_url=media_url,
|
||||
local_media=local_url
|
||||
))
|
||||
except Exception as e:
|
||||
print(f"⚠️ Error uploading media: {e}")
|
||||
continue
|
||||
else:
|
||||
media_url = post.video_url if post.media_type == 2 else post.thumbnail_url
|
||||
if not media_url:
|
||||
continue
|
||||
local_url = upload_media_from_url(media_url, str(req.seller_id), post_id)
|
||||
if local_url:
|
||||
media_items.append(MediaItem(
|
||||
media_type="video" if post.media_type == 2 else "image",
|
||||
media_url=media_url,
|
||||
local_media=local_url
|
||||
))
|
||||
try:
|
||||
local_url = upload_media_from_url(media_url, str(req.seller_id), f"{post_id}/")
|
||||
if local_url:
|
||||
media_items.append(MediaItem(
|
||||
media_type="video" if post.media_type == 2 else "image",
|
||||
media_url=media_url,
|
||||
local_media=local_url
|
||||
))
|
||||
except Exception as e:
|
||||
print(f"⚠️ Error uploading media: {e}")
|
||||
continue
|
||||
|
||||
except LoginRequired:
|
||||
print("⛔ Login expired or failed mid-request.")
|
||||
break
|
||||
except Exception as e:
|
||||
print(f"⚠️ Error downloading media: {e}")
|
||||
print(f"⚠️ Unexpected error while handling media: {e}")
|
||||
continue
|
||||
|
||||
if not media_items:
|
||||
continue
|
||||
|
||||
# ✅ Upload thumbnail separately and store local version
|
||||
uploaded_thumb = upload_media_from_url(post.thumbnail_url, str(req.seller_id), f"{post_id}/thumbnail")
|
||||
final_thumbnail = uploaded_thumb or post.thumbnail_url # fallback if upload fails
|
||||
# ✅ Upload thumbnail safely
|
||||
final_thumbnail = post.thumbnail_url or ""
|
||||
if post.thumbnail_url:
|
||||
try:
|
||||
uploaded_thumb = upload_media_from_url(post.thumbnail_url, str(req.seller_id), f"{post_id}/thumbnail")
|
||||
if uploaded_thumb:
|
||||
final_thumbnail = uploaded_thumb
|
||||
except Exception as e:
|
||||
print(f"⚠️ Failed to upload thumbnail: {e}")
|
||||
|
||||
# Save to DB
|
||||
post_entry = InstagramPost(
|
||||
|
||||
Loading…
Reference in New Issue
Block a user