| 203 | return ret |
| 204 | |
| 205 | def get_html(self, url) -> str: |
| 206 | cache_key = os.path.basename(url) |
| 207 | if cache := self.cache.get(cache_key, CacheType.HTMLDocument, "html"): |
| 208 | return cache |
| 209 | logger.debug(f"getting law html file {url}") |
| 210 | response = requests.get('https://wb.flk.npc.gov.cn' + url, |
| 211 | headers=REQUEST_HEADER) |
| 212 | sleep(1) |
| 213 | response.encoding = "utf8" |
| 214 | ret = response.text |
| 215 | self.cache.set(cache_key, CacheType.HTMLDocument, ret, "html") |
| 216 | return ret |
| 217 | |
| 218 | def get_word(self, url) -> Document: |
| 219 | filename = os.path.basename(url) |