def process_post(store: Minio, post: CleanPost, data_dir: str) -> CleanPost:
likes = post.likes
comments = post.comments
date = post.date
_id = post.id
rel = relevance(likes, comments)
print(f"~ [{_id}]: L: {likes}, C: {comments}, D: {date}, R: {rel}")
filepath = path.join(data_dir, post.id)
bucket = os.getenv("S3_BUCKET", "")
endpoint = os.getenv("S3_ENDPOINT", "")
folder_name = os.getenv("S3_DATA_DIR_NAME", "laciudadinvisible")
destination = folder_name + "/" + post.id + ".jpg"
full_filepath = filepath + ".jpg"
download_pic(full_filepath, post.image_uri)
post_processed_img = crop_image(full_filepath)
store.fput_object(bucket,
destination,
post_processed_img,
content_type="image/jpg",
metadata={"x-amz-acl": "public-read"})
link = f"https://{bucket}.{endpoint}/{destination}"
full_comments = post.comments_content
p = CleanPost(
id=post.id,
date=post.date,
likes=post.likes,
comments=post.comments,
hashtags=post.hashtags,
mentions=post.mentions,
relevance=rel,
image_uri=link,
description=post.description,
comments_content=full_comments,
)
return p
def process_post(loader: Instaloader, post: Post, store: Minio,
data_dir: str) -> CleanPost:
likes = post.likes
comments = post.comments
date = post.date_local
_id = post.shortcode
rel = relevance(likes, comments)
print(f"\n~ [{_id}]: L: {likes}, C: {comments}, D: {date}, R: {rel}")
filepath = path.join(data_dir, post.shortcode)
loader.download_pic(filepath, post.url, post.date)
print("")
bucket = os.getenv("S3_BUCKET", "")
endpoint = os.getenv("S3_ENDPOINT", "")
folder_name = os.getenv("S3_DATA_DIR_NAME", "laciudadinvisible")
destination = folder_name + "/" + post.shortcode + ".jpg"
full_filepath = filepath + ".jpg"
store.fput_object(bucket,
destination,
full_filepath,
content_type="image/jpg",
metadata={"x-amz-acl": "public-read"})
link = f"https://{bucket}.{endpoint}/{destination}"
full_comments: List[str] = []
for comment in post.get_comments():
full_comments.append(comment.text)
p = CleanPost(
id=post.shortcode,
date=str(post.date_local),
likes=post.likes,
comments=post.comments,
hashtags=post.caption_hashtags,
mentions=post.caption_mentions,
relevance=rel,
image_uri=link,
description=post.caption,
comments_content=full_comments,
)
# coll_name = "Posts"
# try:
# db.create_collection(coll_name)
# except:
# print("error at execute create collection")
# posts = db.collection(coll_name)
# posts.insert(asdict(p))
# if path.exists(full_filepath):
# os.remove(full_filepath)
return p
