gptwritelang.py

import openai
import pywikibot
import re
import random
seed = random.randint(100000, 999999)
import requests
import traceback
from bs4 import BeautifulSoup
import datetime
import time
import threading
NUM_THREADS = 10
ERROR_SLEEP = 5 
openai.api_key = ""
openai.base_url = ""
model=None

from cstitle import *

def main():
    topic = get_random_cs_topic()
    topic_prompt = (
        f"The phrase \"{topic}\" was extracted from Stack Overflow and Reddit etc. "
        "It may come from a question or descriptive phrase.\n\n"
        "Your task is to rewrite this phrase as the name of a **programming language, software development tool, or operating system only**.\n\n"
        "Rules:\n"
        "- Only return **1 name**. Don't too common\n"
        "- The result must be a **real, widely used** language (e.g., Python, C++, JavaScript), tool (e.g., Git, Docker, GCC, Vim), or OS (e.g., Linux, Windows), or framework (e.g. Flask, Django).\n"
        "- Do NOT include niche libraries, company names, course titles, or made-up terms.\n"
        "- Only return the name, with **no punctuation, no quotes, and no explanation**.\n"
        "- Keep correct casing (e.g., C++, JavaScript, TypeScript, macOS).\n\n"
        f"Random seed: {seed}"
    )


    topic = openai.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": topic_prompt}],
        timeout=60
    ).choices[0].message.content.strip()

    print(f"🎯 Selected topic: {topic}")

    site = pywikibot.Site("en", "EdwardWiki")
    page = pywikibot.Page(site, topic)

    if page.exists():
        print(f"⚠️ Skipping existing topic: {topic}")
        return
    
    article_prompt = f"""
    Write a long and detailed Wikipedia-style article about '''{topic}''' in pure MediaWiki markup.

    == REQUIRED FORMAT ==
    - Use only MediaWiki formatting (== Section headings ==, '''bold''', [[internal links]], etc.).
    - The article must contain **at least 6 main sections**, each with its own == heading ==. Subsections must use === subheadings ===.
    - Do not include bullet lists as substitutes for content.
    - Never output a flat article without headings.
    - Do not use markdown, bullet points, or explanatory assistant messages.
    - Content must be at least 3000 words, well-organized and fully structured like a real Wikipedia article.

    == EXAMPLE HEADINGS ==
    - Background or History
    - Architecture or Design
    - Implementation or Applications
    - Real-world Examples
    - Criticism or Limitations
    - See also
    - References (only official websites or trusted docs, no academic papers)

    == STYLE ==
    - Use formal, encyclopedic tone.
    - Explain clearly for general readers.
    - Avoid generic filler or vague summaries.
    - Avoid duplicating the topic name excessively.
    - The article must contain at least 6 sections, each starting with == Heading == and every sections use the == Heading ==. Do not skip this. Articles without enough section headers will be rejected. Do not include an '== Introduction ==' section – instead, place the introduction before any headers.
    Start with the following exact format:
    '''{topic}''' is ...
    == Other Heading... ==
    ...
    Then continue with at least 4 more major sections.

    == OUTPUT ==
    Return only the article in MediaWiki format. No additional instructions or notes.
    """


    article_text = openai.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": article_prompt}],
        timeout=60
    ).choices[0].message.content

    article_text = re.sub(r"```.*?\n", "", article_text)
    article_text = re.sub(r"\n```", "", article_text)

    article_text = re.sub(r'^[\"\']{1,2}(.*?)[\"\']{1,2}$', r"\1", article_text, flags=re.M)  
    article_text = re.sub(r'^[-•]\s+', '', article_text, flags=re.M)  
    article_text = re.sub(r'^\s*[*]+\s+', '* ', article_text, flags=re.M)  
    article_text = article_text.strip()


    cat_prompt = (
        f"Given the MediaWiki article titled '{topic}', list up to three appropriate Wikipedia-style "
        "categories (broadest first). Respond with category names only, comma-separated, no extra text."
    )
    raw_cats = openai.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": cat_prompt}],
    ).choices[0].message.content

    cats = [c.strip() for c in raw_cats.split(",") if c.strip()]
    if cats:
        article_text += "\n\n" + "\n".join(f"[[Category:{c}]]" for c in cats)

    site = pywikibot.Site("en", "EdwardWiki")
    page = pywikibot.Page(site, topic)

    page.text = article_text
    page.save(summary=f"Created article '{topic}' with auto-categories 🏷️")
    print(f"✅ Article '{topic}' published, categories: {', '.join(cats) if cats else 'none'}")

def log_error(e):
    with open("error.log", "a", encoding="utf-8") as f:
        f.write(f"\n[{datetime.datetime.now().isoformat()}] ERROR:\n")
        f.write(traceback.format_exc())
        f.write("\n" + "-"*80 + "\n")


def worker(tid):
    run_count = 0
    while True:
        try:
            start = time.time()
            print(f"\n🧵[T{tid}] ⏱️  Run #{run_count+1}  "
                  f"{datetime.datetime.now().isoformat()}")
            main()      
            run_count += 1
            print(f"🧵[T{tid}] ✅  Done #{run_count}   "
                  f"{time.time()-start:.2f}s")
        except KeyboardInterrupt:
            print(f"🧵[T{tid}] 🛑  User interrupt, exit.")
            break
        except Exception as e:
            print(f"🧵[T{tid}] ⚠️  Error, logged & retry in "
                  f"{ERROR_SLEEP}s.")
            log_error(e, tid)
            time.sleep(ERROR_SLEEP)

def loop_run_multithread():
    threads = []
    for tid in range(NUM_THREADS):
        t = threading.Thread(target=worker, args=(tid,), daemon=True)
        t.start()
        threads.append(t)
    try:
        while True:
            time.sleep(1)
    except KeyboardInterrupt:
        print("🛑  Master interrupt, exiting…")

if __name__ == "__main__":
    loop_run_multithread()