ReXploreBackend / main.py
raannakasturi's picture
Refactor title and citation encoding in paper_data and test functions for improved handling of special characters
72c7a0a
raw
history blame
3.19 kB
import json
import os
import dotenv
import html
from summarize_paper import summarize_paper
from fetch_data import fetch_paper_data_with_category
from post_blog import post_blog
from send_mail import send_email
dotenv.load_dotenv()
access_key = os.getenv("ACCESS_KEY")
def paper_data(paper_data, wait_time=5):
data = {"status": "success"}
data['data'] = {}
paper_data = json.loads(paper_data)
for category, papers in paper_data.items():
print(f"Processing category: {category}")
data['data'][category] = {}
for paper_id, details in papers.items():
doi = details.get("doi")
pdf_url = details.get("pdf_url")
title = details.get("title")
citation = details.get("citation")
if not all([paper_id, doi, pdf_url, title, citation]):
print(f"Skipping paper with ID: {paper_id} (missing details)")
continue
summary, mindmap = summarize_paper(pdf_url, paper_id, access_key)
if not summary or not mindmap:
print(f"Skipping paper with ID: {paper_id} (summary/mindmap not found)")
continue
try:
try:
escaped_title = repr(title.encode('latin1').decode('unicode-escape')).strip()
except:
escaped_title = repr(title).strip()
title = html.escape(str(escaped_title).strip()[1:-1])
try:
encoded_bytes = citation.encode('latin1').decode('utf-8', errors='replace')
except:
encoded_bytes = repr(citation)
citation = html.unescape(encoded_bytes)
post_blog(title, category, summary, mindmap, citation, access_key, wait_time)
except Exception as e:
print(f"Error posting blog '{title}': {e}")
continue
data['data'][category][paper_id] = {
"id": paper_id,
"doi": doi,
"title": title,
"category": category,
"citation": citation,
"summary": summary,
"mindmap": mindmap,
}
data = json.dumps(data, indent=4, ensure_ascii=False)
return data
def post_blogpost(uaccess_key, wait_time=5):
if uaccess_key != access_key:
return False
data = fetch_paper_data_with_category(uaccess_key)
pdata = paper_data(data, wait_time)
try:
send_email(pdata)
print("\n-------------------------------------------------------\nMail Sent\n-------------------------------------------------------\n")
except Exception as e:
print(f"\n-------------------------------------------------------\nError sending mail: {e}\n-------------------------------------------------------\n")
finally:
print("\n-------------------------------------------------------\nProcess Completed\n-------------------------------------------------------\n")
return pdata