🐛 修复bilibili推送的一些格式错误 (#263)

* 🎈 perf(platform/bilibili): 增加了相似度计算前文本的预处理

将动态和简介文本中较长的一段按照较短的一段进行截取(分了从前截和从后截的两种情况)

* 🐞 fix(bilibili): 修复视频简介多余空格的bug

* 🦄 refactor(bilibili): 更改文本相似度比较函数
This commit is contained in:
UKM
2023-07-18 15:06:46 +08:00
committed by GitHub
parent 5922a7827f
commit ff3c0ffe65
3 changed files with 5 additions and 19 deletions
+3 -8
View File
@@ -12,7 +12,7 @@ from typing_extensions import Self
from ..post import Post
from ..types import ApiError, Category, RawPost, Tag, Target
from ..utils import SchedulerConfig, jaccard_text_similarity
from ..utils import SchedulerConfig, text_similarity
from .platform import CategoryNotRecognize, CategoryNotSupport, NewMessage, StatusChange
@@ -151,7 +151,7 @@ class Bilibili(NewMessage):
title = card["title"]
desc = card.get("desc", "")
if jaccard_text_similarity(desc, dynamic) > 0.8:
if text_similarity(desc, dynamic) > 0.8:
# 如果视频简介和动态内容相似,就只保留长的那个
if len(dynamic) > len(desc):
text = f"{dynamic}\n=================\n{title}"
@@ -159,12 +159,7 @@ class Bilibili(NewMessage):
text = f"{title}\n\n{desc}"
else:
# 否则就把两个拼起来
text = f"""
{dynamic}
\n=================\n
{title}\n\n
{desc}
"""
text = f"{dynamic}\n=================\n{title}\n\n{desc}"
pic = [card["pic"]]
elif post_type == 4: