🐞 fix(text-similarity): 修复除0报错 (#302)

* 🐞 fix(text-similarity): 修复除0报错

没有考虑到bilibili的推送出现动态或者视频简介长度为零的情况,出现文本相似度除0Error

* 🧪 test(bilibili): 添加视频动态内容为空的情况的测试

* 🧪 test(text-similarity): 增加文本相似度函数的测试

* Update test_rss.py

* 📃 docs(text_similarity): 添加文本相似度函数的注释

* 🦄 refactor(text_similarity): 重构文本相似度的比较方法

* 🎈 perf(similar_text): 将比较函数的return改成raise

* 🦄 refactor(text_similarity): 重构文本相似度比较方法

* Update nonebot_bison/platform/bilibili.py

Co-authored-by: felinae98 <731499577@qq.com>

* Update nonebot_bison/platform/rss.py

Co-authored-by: felinae98 <731499577@qq.com>

---------

Co-authored-by: felinae98 <731499577@qq.com>
This commit is contained in:
UKM 2023-08-27 16:28:26 +08:00 committed by GitHub
parent d8dd0cdd39
commit cfa91df5e4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 69 additions and 20 deletions

View File

@ -125,6 +125,16 @@ class Bilibili(NewMessage):
def get_tags(self, raw_post: RawPost) -> list[Tag]:
return [*(tp["topic_name"] for tp in raw_post["display"]["topic_info"]["topic_details"])]
def _text_process(self, dynamic: str, desc: str, title: str) -> str:
similarity = 1.0 if len(dynamic) == 0 or len(desc) == 0 else text_similarity(dynamic, desc)
if len(dynamic) == 0 and len(desc) == 0:
text = title
elif similarity > 0.8:
text = title + "\n\n" + desc if len(dynamic) < len(desc) else dynamic + "\n=================\n" + title
else:
text = dynamic + "\n=================\n" + title + "\n\n" + desc
return text
def _get_info(self, post_type: Category, card) -> tuple[str, list]:
if post_type == 1:
# 一般动态
@ -139,17 +149,7 @@ class Bilibili(NewMessage):
dynamic = card.get("dynamic", "")
title = card["title"]
desc = card.get("desc", "")
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}"
else:
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 = self._text_process(dynamic, desc, title)
pic = [card["pic"]]
elif post_type == 4:
# 纯文字

View File

@ -53,18 +53,19 @@ class Rss(NewMessage):
entry["_target_name"] = feed.feed.title
return feed.entries
def _text_process(self, title: str, desc: str) -> str:
similarity = 1.0 if len(title) == 0 or len(desc) == 0 else text_similarity(title, desc)
if similarity > 0.8:
text = title if len(title) > len(desc) else desc
else:
text = title + "\n\n" + desc
return text
async def parse(self, raw_post: RawPost) -> Post:
title = raw_post.get("title", "")
soup = bs(raw_post.description, "html.parser")
desc = soup.text.strip()
if not title or not desc:
text = title or desc
else:
if text_similarity(desc, title) > 0.8:
text = desc if len(desc) > len(title) else title
else:
text = f"{title}\n\n{desc}"
text = self._text_process(title, desc)
pics = [x.attrs["src"] for x in soup("img")]
if raw_post.get("media_content"):
for media in raw_post["media_content"]:

View File

@ -95,6 +95,9 @@ if plugin_config.bison_filter_log:
def text_similarity(str1, str2) -> float:
"""利用最长公共子序列的算法判断两个字符串是否相似并返回0到1.0的相似度"""
if len(str1) == 0 or len(str2) == 0:
raise ValueError("The length of string can not be 0")
matcher = difflib.SequenceMatcher(None, str1, str2)
t = sum(temp.size for temp in matcher.get_matching_blocks())
return t / min(len(str1), len(str2))

View File

@ -425,7 +425,7 @@
"rid_str": "975400699",
"bvid": "BV1K44y1h7Xg"
},
"card": "{\"aid\":975400699,\"attribute\":0,\"cid\":406644689,\"copyright\":1,\"ctime\":1631408447,\"desc\":\"本系列视频为饼组成员的有趣直播录播,主要内容为方舟相关,未来可能系列其他视频会包含部分饼组团建日常等。仅为娱乐性视频,内容与常规饼学预测无关。视频仅为当期主播主观观点,不代表饼组观点。仅供娱乐。\\n\\n直播主播:@寒蝉慕夏 \\n后期剪辑:@Melodiesviel \\n\\n本群视频为9.11组员慕夏直播录播包含慕夏对新PV的个人解读风笛厨力疯狂放出CP言论输出9.16轮换池预测视频分析和理智规划杂谈内容。\\n注意:内含大量个人性质对风笛的厨力观点与多CP混乱发言不适者请及时点击退出或跳到下一片段。\",\"dimension\":{\"height\":1080,\"rotate\":0,\"width\":1920},\"duration\":4318,\"dynamic\":\"昨天慕夏直播的录播剪辑版关于新PV慕夏对风笛的看法新一期轮换池预测视频的分析以及理智规划。错过直播且有兴趣的朋友可以看啦。\",\"first_frame\":\"https:\\/\\/i1.hdslb.com\\/bfs\\/storyff\\/n210911a297vzlaeyhb8g26etg86gci5_firsti.jpg\",\"jump_url\":\"bilibili:\\/\\/video\\/975400699\\/?page=1&player_preload=null&player_width=1920&player_height=1080&player_rotate=0\",\"owner\":{\"face\":\"https:\\/\\/i0.hdslb.com\\/bfs\\/face\\/00776b6ddde4874af87b8bc2870da86ed39c2c80.jpg\",\"mid\":8412516,\"name\":\"罗德岛蜜饼工坊\"},\"pic\":\"https:\\/\\/i0.hdslb.com\\/bfs\\/archive\\/c8cb0073819a0c8171db5009002eec19a80c85f6.jpg\",\"player_info\":null,\"pubdate\":1631408446,\"rights\":{\"autoplay\":1,\"bp\":0,\"download\":0,\"elec\":0,\"hd5\":0,\"is_cooperation\":0,\"movie\":0,\"no_background\":0,\"no_reprint\":1,\"pay\":0,\"ugc_pay\":0,\"ugc_pay_preview\":0},\"short_link\":\"https:\\/\\/b23.tv\\/BV1K44y1h7Xg\",\"short_link_v2\":\"https:\\/\\/b23.tv\\/BV1K44y1h7Xg\",\"stat\":{\"aid\":975400699,\"coin\":46,\"danmaku\":156,\"dislike\":0,\"favorite\":45,\"his_rank\":0,\"like\":495,\"now_rank\":0,\"reply\":45,\"share\":6,\"view\":3293},\"state\":0,\"tid\":172,\"title\":\"阿消的罗德岛闲谈直播#01:《女人最喜欢的女人,就是在战场上熠熠生辉的女人》\",\"tname\":\"手机游戏\",\"up_from_v2\":35,\"videos\":1}",
"card": "{\"aid\":975400699,\"attribute\":0,\"cid\":406644689,\"copyright\":1,\"ctime\":1631408447,\"desc\":\"本系列视频为饼组成员的有趣直播录播,主要内容为方舟相关,未来可能系列其他视频会包含部分饼组团建日常等。仅为娱乐性视频,内容与常规饼学预测无关。视频仅为当期主播主观观点,不代表饼组观点。仅供娱乐。\\n\\n直播主播:@寒蝉慕夏 \\n后期剪辑:@Melodiesviel \\n\\n本群视频为9.11组员慕夏直播录播包含慕夏对新PV的个人解读风笛厨力疯狂放出CP言论输出9.16轮换池预测视频分析和理智规划杂谈内容。\\n注意:内含大量个人性质对风笛的厨力观点与多CP混乱发言不适者请及时点击退出或跳到下一片段。\",\"dimension\":{\"height\":1080,\"rotate\":0,\"width\":1920},\"duration\":4318,\"dynamic\":\"\",\"first_frame\":\"https:\\/\\/i1.hdslb.com\\/bfs\\/storyff\\/n210911a297vzlaeyhb8g26etg86gci5_firsti.jpg\",\"jump_url\":\"bilibili:\\/\\/video\\/975400699\\/?page=1&player_preload=null&player_width=1920&player_height=1080&player_rotate=0\",\"owner\":{\"face\":\"https:\\/\\/i0.hdslb.com\\/bfs\\/face\\/00776b6ddde4874af87b8bc2870da86ed39c2c80.jpg\",\"mid\":8412516,\"name\":\"罗德岛蜜饼工坊\"},\"pic\":\"https:\\/\\/i0.hdslb.com\\/bfs\\/archive\\/c8cb0073819a0c8171db5009002eec19a80c85f6.jpg\",\"player_info\":null,\"pubdate\":1631408446,\"rights\":{\"autoplay\":1,\"bp\":0,\"download\":0,\"elec\":0,\"hd5\":0,\"is_cooperation\":0,\"movie\":0,\"no_background\":0,\"no_reprint\":1,\"pay\":0,\"ugc_pay\":0,\"ugc_pay_preview\":0},\"short_link\":\"https:\\/\\/b23.tv\\/BV1K44y1h7Xg\",\"short_link_v2\":\"https:\\/\\/b23.tv\\/BV1K44y1h7Xg\",\"stat\":{\"aid\":975400699,\"coin\":46,\"danmaku\":156,\"dislike\":0,\"favorite\":45,\"his_rank\":0,\"like\":495,\"now_rank\":0,\"reply\":45,\"share\":6,\"view\":3293},\"state\":0,\"tid\":172,\"title\":\"阿消的罗德岛闲谈直播#01:《女人最喜欢的女人,就是在战场上熠熠生辉的女人》\",\"tname\":\"手机游戏\",\"up_from_v2\":35,\"videos\":1}",
"extend_json": "{\"\":{\"ogv\":{\"ogv_id\":0}},\"dispute\":{\"content\":\"\"},\"from\":{\"from\":\"\",\"verify\":{}},\"like_icon\":{\"action\":\"\",\"action_url\":\"\",\"end\":\"\",\"end_url\":\"\",\"start\":\"\",\"start_url\":\"\"},\"topic\":{\"is_attach_topic\":1}}",
"extra": {
"is_space_top": 0

View File

@ -52,6 +52,22 @@ async def test_video_forward(bilibili, bing_dy_list):
)
@pytest.mark.asyncio
async def test_video_forward_without_dynamic(bilibili, bing_dy_list):
# 视频简介和动态文本其中一方为空的情况
post = await bilibili.parse(bing_dy_list[2])
assert (
post.text
== "阿消的罗德岛闲谈直播#01:《女人最喜欢的女人,就是在战场上熠熠生辉的女人》"
+ "\n\n"
+ "本系列视频为饼组成员的有趣直播录播,主要内容为方舟相关,未来可能系列其他视频会包含部分饼组团建日常等。"
"仅为娱乐性视频,内容与常规饼学预测无关。视频仅为当期主播主观观点,不代表饼组观点。仅供娱乐。"
"\n\n直播主播:@寒蝉慕夏 \n后期剪辑:@Melodiesviel \n\n本群视频为9.11组员慕夏直播录播,"
"包含慕夏对新PV的个人解读风笛厨力疯狂放出CP言论输出9.16轮换池预测视频分析和理智规划杂谈内容。"
"\n注意:内含大量个人性质对风笛的厨力观点与多CP混乱发言不适者请及时点击退出或跳到下一片段。"
)
@pytest.mark.asyncio
async def test_article_forward(bilibili, bing_dy_list):
post = await bilibili.parse(bing_dy_list[4])

View File

@ -163,3 +163,32 @@ async def test_fetch_new_4(
post1 = res2[0][1][0]
assert post1.url == "https://wallhaven.cc/w/85rjej"
assert post1.text == "85rjej.jpg"
def test_similar_text_process():
from nonebot_bison.utils import text_similarity
str1 = ""
str2 = "xxxx"
with pytest.raises(ValueError, match="The length of string can not be 0"):
text_similarity(str1, str2)
str1 = "xxxx"
str2 = ""
with pytest.raises(ValueError, match="The length of string can not be 0"):
text_similarity(str1, str2)
str1 = (
"天使九局下被追平,米基-莫尼亚克(Mickey Moniak)超前安打拒绝剧本天使7-6老虎阿莱克-博姆Alec"
" Bohm再见安打费城人4-3金莺..."
)
str2 = (
"天使九局下被追平,米基-莫尼亚克(Mickey Moniak)超前安打拒绝剧本天使7-6老虎阿莱克-博姆Alec"
" Bohm再见安打费城人4-3金莺布兰登-洛维Brandon Lowe阳春炮光芒4-1马林鱼皮特-阿隆索(Pete Alonso)、"
"丹尼尔-沃格尔巴克(Daniel Vogelbach)背靠背本垒打大都会9-3扬基吉田正尚Masataka Yoshida3安2打点"
"红袜7-1勇士凯尔-塔克Kyle Tucker阳春炮太空人4-3连胜游骑兵。"
)
res = text_similarity(str1, str2)
assert res > 0.8
str1 = "我爱你"
str2 = "你爱我"
res = text_similarity(str1, str2)
assert res <= 0.8