scripts/coling/coling.py

import json
import requests
from bs4 import BeautifulSoup

SOURCE_URL = "https://aclanthology.org/events/coling-{year}/"
OUTPUT_FILE = "data/coling/coling{year}_papers.jsonl"
BASE_URL = "https://aclanthology.org"


def fetch_page(url):
    headers = {"User-Agent": "Mozilla/5.0 (ACL-parser/1.0; research)"}
    print(f"Fetching {url} ...", flush=True)
    resp = requests.get(url, headers=headers, timeout=120)
    resp.raise_for_status()
    return resp.text


def abs_url(href):
    if not href:
        return ""
    return href if href.startswith("http") else BASE_URL + href


def parse_papers(html):
    soup = BeautifulSoup(html, "html.parser")
    papers = []
    
    # Find all paper divs - each paper is in a div with class "d-sm-flex align-items-stretch mb-3"
    paper_divs = soup.find_all("div", class_="d-sm-flex align-items-stretch mb-3")
    
    for paper_div in paper_divs:
        paper = parse_paper_div(paper_div)
        if paper:
            papers.append(paper)
    
    return papers


def parse_paper_div(paper_div):
    # Extract the paper title and URL
    title_element = paper_div.find("strong")
    if not title_element:
        return None
    
    title_link = title_element.find("a")
    if not title_link:
        return None
    
    title = title_link.get_text(strip=False)
    href = title_link.get("href", "")
    paper_id = href.strip("/").split("/")[-1] if href else ""
    url = abs_url(href)
    
    # Skip if no paper_id found
    if not paper_id:
        return None
    
    # PDF / BIB links from the button-row div
    pdf_url = bib_url = ""
    btn_row = paper_div.find("div", class_="list-button-row")
    if btn_row:
        for a in btn_row.find_all("a", href=True):
            text = a.get_text(strip=True).lower()
            h = a["href"]
            if text == "pdf" or h.endswith(".pdf"):
                pdf_url = abs_url(h)
            elif text == "bib" or h.endswith(".bib"):
                bib_url = abs_url(h)
    
    # Authors: <a href="/people/..."> links inside the d-block span
    authors = []
    span = paper_div.find("span", class_="d-block")
    if span:
        for a in span.find_all("a", href=True):
            if "/people/" in a["href"]:
                authors.append(a.get_text(strip=True))
    
    # Abstract: look for the collapse div with abstract content
    abstract = ""
    # Find the abstract ID from the button
    abs_button = paper_div.find("a", class_="badge text-bg-info")
    if abs_button and "data-bs-target" in abs_button.attrs:
        abs_id = abs_button["data-bs-target"].lstrip("#")
        # Find the abstract div using this ID
        abs_div = soup.find("div", id=abs_id)
        if abs_div:
            body = abs_div.find("div", class_="card-body")
            if body:
                abstract = body.get_text(strip=True)
    
    # Determine volume from paper_id
    volume = ""
    if paper_id:
        parts = paper_id.split(".")
        if len(parts) >= 2:
            volume = parts[1]  # e.g., "acl-long" from "2023.acl-long.19"
    
    return {
        "paper_id": paper_id,
        "title": title,
        "authors": authors,
        "abstract": abstract,
        "url": url,
        "pdf_url": pdf_url,
        "bib_url": bib_url,
    }


def main():
    for year in range(0, 25, 2):
        year = f"20{year:02d}"
        try:
            html = fetch_page(SOURCE_URL.format(year=year))
            print("Parsing ...", flush=True)
            papers = parse_papers(html)
            
            print(f"Found {len(papers)} papers. Writing to {OUTPUT_FILE.format(year=year)} ...", flush=True)
            with open(OUTPUT_FILE.format(year=year), "w", encoding="utf-8") as f:
                for p in papers:
                    f.write(json.dumps(p, ensure_ascii=False) + "\n")
            
            print("Done")
        except Exception as e:
            print(f"Error: {e}")


if __name__ == "__main__":
    main()