Upload scripts/MDNClean.py with huggingface_hub

scripts/MDNClean.py

@@ -0,0 +1,111 @@
+import asyncio
+import pathlib
+import re
+import markdownify
+from markdownify import chomp
+import urllib.parse
+import orjson
+import tqdm
+import typer
+import httpx
+from bs4 import BeautifulSoup, Tag
+
+app = typer.Typer()
+
+class MDNConverter(markdownify.MarkdownConverter):
+    def convert_a(self, el, text, convert_as_inline):
+        prefix, suffix, text = chomp(text)
+        if not text:
+            return ""
+        return "%s%s%s" % (prefix, text, suffix)
+
+    def convert_pre(self, el:Tag, text, convert_as_inline):
+        if not text:
+            return ''
+        
+        code_language = self.options['code_language']
+        # 
+        if el.previous_sibling and el.previous_sibling.name and el.previous_sibling.name == "div":
+            if "example-header" in el.previous_sibling.get("class",[]):
+                code_language = el.previous_sibling.find('span').get_text()
+                el.previous_sibling.decompose()
+
+        # if self.options['code_language_callback']:
+        #     code_language = self.options['code_language_callback'](el) or code_language
+
+        return '\n```%s\n%s\n```\n' % (code_language, text)
+
+inst = MDNConverter()
+
+rgx = re.compile(r"(.+)\n```\1")
+
+def quote_fx(matchobj):
+    return f"\n```{matchobj.group(1)}"
+
+def soup_processor(soup:BeautifulSoup):
+    base_article = soup.select_one("main#content > article")
+    [script.decompose() for script in base_article.find_all("script")]
+    if not base_article:
+        return None
+    mod_time = base_article.find("aside",attrs={"class":"metadata"})
+    if mod_time:
+        mod_time.decompose()
+    hydro = base_article.select_one("#hydration")
+    if hydro:
+        hydro.decompose()
+    
+    r = inst.convert_soup(base_article).replace("\n\n\n\n","\n\n").replace("\n\n\n","\n\n")
+    r = rgx.sub(quote_fx, r).replace("\n \n","\n\n").replace("\n\n```\n","\n```\n\n").rstrip().replace("\t","    ")
+    
+    data = {"title":soup.find("title").get_text().split("|")[0].strip(),"text":r}
+    return data
+    
+
+
+
+@app.command()
+def clean(root:pathlib.Path, output_jsonl:pathlib.Path):
+    with open(output_jsonl.resolve(),"wb") as f:
+        for file in root.rglob("*.html"):
+            if file.is_file() and "docs/MDN/" not in str(file):
+                soup = BeautifulSoup(file.read_text(encoding="utf-8"),"lxml")
+                print(file)
+                r = soup_processor(soup)
+                f.write(orjson.dumps(r)+b"\n")
+            
+            
+
+async def async_executor(urls:list[str], root:pathlib.Path):
+    sem = asyncio.Semaphore(64)
+    client = httpx.AsyncClient(timeout=None)
+    client.headers["user-agent"] = "curl/7.54.1 (compatible; kinda like curl) " + client.headers["user-agent"] + " mdncrawl/1.0.0 (SiteMap Crawling; +shinon@recursal.ai; Contact if bot errors.)"
+    pbar = tqdm.tqdm()
+    async def url_executor(url:str):
+        pth = urllib.parse.urlparse(url).path[1:]
+        url_write = (root / pth).resolve()
+        # print(url_write)
+        url_write.parent.mkdir(exist_ok=True,parents=True)
+        async with sem:
+            response = await client.get(url)
+            if response.status_code == 200:
+                if url_write.suffix != ".html":
+                    url_write = url_write.with_suffix(f"{url_write.suffix}.html")
+                    url_write.write_bytes(response.content)
+                    pbar.update(1)
+            else:
+                print(response.status_code)
+    loop = asyncio.get_running_loop()
+    workers = [loop.create_task(url_executor(url)) for url in urls]
+    await asyncio.gather(*workers)
+    pass
+
+@app.command()
+def sitemap(xml_file:pathlib.Path, root:pathlib.Path):
+    root = root.resolve()
+    soup = BeautifulSoup(xml_file.read_text(encoding="utf-8"), "xml")
+    urls = [i.get_text() for i in soup.find_all("loc")]
+    # print(urls)
+    asyncio.run(async_executor(urls, root))
+    
+if __name__ == "__main__":
+    app()