feat(encoding): enhance encoding detection with confidence threshold

This commit is contained in:
InfinityPacer
2024-11-27 12:33:57 +08:00
parent 2086651dbe
commit 83fc474dbe
4 changed files with 117 additions and 27 deletions

View File

@@ -219,6 +219,10 @@ class ConfigModel(BaseModel):
BIG_MEMORY_MODE: bool = False
# 全局图片缓存,将媒体图片缓存到本地
GLOBAL_IMAGE_CACHE: bool = False
# 是否启用编码探测的兼容模式
ENCODING_DETECTION_COMPATIBLE_MODE: bool = True
# 编码探测的最低置信度阈值
ENCODING_DETECTION_MIN_CONFIDENCE: float = 0.8
# 允许的图片缓存域名
SECURITY_IMAGE_DOMAINS: List[str] = Field(
default_factory=lambda: ["image.tmdb.org",

View File

@@ -344,11 +344,9 @@ class SiteParserBase(metaclass=ABCMeta):
logger.warn(
f"{self._site_name} 检测到Cloudflare请更新Cookie和UA")
return ""
if re.search(r"charset=\"?utf-8\"?", res.text, re.IGNORECASE):
res.encoding = "utf-8"
else:
res.encoding = res.apparent_encoding
return res.text
return RequestUtils.get_decoded_html_content(res,
settings.ENCODING_DETECTION_COMPATIBLE_MODE,
settings.ENCODING_DETECTION_MIN_CONFIDENCE)
return ""

View File

@@ -5,7 +5,6 @@ import traceback
from typing import List
from urllib.parse import quote, urlencode, urlparse, parse_qs
import chardet
from jinja2 import Template
from pyquery import PyQuery
from ruamel.yaml import CommentedMap
@@ -250,27 +249,9 @@ class TorrentSpider:
referer=self.referer,
proxies=self.proxies
).get_res(searchurl, allow_redirects=True)
if ret is not None:
# 使用chardet检测字符编码
raw_data = ret.content
if raw_data:
try:
result = chardet.detect(raw_data)
encoding = result['encoding']
# 解码为字符串
page_source = raw_data.decode(encoding)
except Exception as e:
logger.debug(f"chardet解码失败{str(e)}")
# 探测utf-8解码
if re.search(r"charset=\"?utf-8\"?", ret.text, re.IGNORECASE):
ret.encoding = "utf-8"
else:
ret.encoding = ret.apparent_encoding
page_source = ret.text
else:
page_source = ret.text
else:
page_source = ""
page_source = RequestUtils.get_decoded_html_content(ret,
settings.ENCODING_DETECTION_COMPATIBLE_MODE,
settings.ENCODING_DETECTION_MIN_CONFIDENCE)
# 解析
return self.parse(page_source)

View File

@@ -1,5 +1,7 @@
import re
from typing import Any, Optional, Union
import chardet
import requests
import urllib3
from requests import Response, Session
@@ -273,3 +275,108 @@ class RequestUtils:
cache_headers["Cache-Control"] = f"max-age={max_age}"
return cache_headers
@staticmethod
def detect_encoding_from_html_response(response: Response,
compatible_mode: bool = False, confidence_threshold: float = 0.8):
"""
根据HTML响应内容探测编码信息
:param response: HTTP 响应对象
:param compatible_mode: 是否使用兼容模式,默认为 False (性能模式)
:param confidence_threshold: chardet 检测置信度阈值,默认为 0.8
:return: 解析得到的字符编码
"""
fallback_encoding = None
try:
if compatible_mode:
# 兼容模式使用chardet分析后再处理 BOM 和 meta 信息
# 1. 使用 chardet 库进一步分析内容
detection = chardet.detect(response.content)
if detection["confidence"] > confidence_threshold:
return detection.get("encoding")
# 保存 chardet 的结果备用
fallback_encoding = detection.get("encoding")
# 2. 检查响应体中的 BOM 标记(例如 UTF-8 BOM
if response.content[:3] == b"\xef\xbb\xbf": # UTF-8 BOM
return "utf-8"
# 3. 如果是 HTML 响应体,检查其中的 <meta charset="..."> 标签
if re.search(r"charset=[\"']?utf-8[\"']?", response.text, re.IGNORECASE):
return "utf-8"
# 4. 尝试从 response headers 中获取编码信息
content_type = response.headers.get("Content-Type", "")
if re.search(r"charset=[\"']?utf-8[\"']?", content_type, re.IGNORECASE):
return "utf-8"
else:
# 性能模式:优先从 headers 和 BOM 标记获取,最后使用 chardet 分析
# 1. 尝试从 response headers 中获取编码信息
content_type = response.headers.get("Content-Type", "")
if re.search(r"charset=[\"']?utf-8[\"']?", content_type, re.IGNORECASE):
return "utf-8"
# 暂不支持直接提取字符集仅提取UTF8
# match = re.search(r"charset=[\"']?([^\"';\s]+)", content_type, re.IGNORECASE)
# if match:
# return match.group(1)
# 2. 检查响应体中的 BOM 标记(例如 UTF-8 BOM
if response.content[:3] == b"\xef\xbb\xbf":
return "utf-8"
# 3. 如果是 HTML 响应体,检查其中的 <meta charset="..."> 标签
if re.search(r"charset=[\"']?utf-8[\"']?", response.text, re.IGNORECASE):
return "utf-8"
# 暂不支持直接提取字符集仅提取UTF8
# match = re.search(r"<meta[^>]+charset=[\"']?([^\"'>\s]+)", response.text, re.IGNORECASE)
# if match:
# return match.group(1)
# 4. 使用 chardet 库进一步分析内容
detection = chardet.detect(response.content)
if detection.get("confidence", 0) > confidence_threshold:
return detection.get("encoding")
# 保存 chardet 的结果备用
fallback_encoding = detection.get("encoding")
# 5. 如果上述方法都无法确定,信任 chardet 的结果(即使置信度较低),否则返回默认字符集
return fallback_encoding or "utf-8"
except Exception as e:
logger.debug(f"Error when detect_encoding_from_response: {str(e)}")
return fallback_encoding or "utf-8"
@staticmethod
def get_decoded_html_content(response: Response,
compatible_mode: bool = False, confidence_threshold: float = 0.8) -> str:
"""
获取HTML响应的解码文本内容
:param response: HTTP 响应对象
:param compatible_mode: 是否使用兼容模式,默认为 False (性能模式)
:param confidence_threshold: chardet 检测置信度阈值,默认为 0.8
:return: 解码后的响应文本内容
"""
try:
if not response:
return ""
if response.content:
# 1. 获取编码信息
encoding = (RequestUtils.detect_encoding_from_html_response(response, compatible_mode,
confidence_threshold)
or response.apparent_encoding)
# 2. 根据解析得到的编码进行解码
try:
# 尝试用推测的编码解码
return response.content.decode(encoding)
except Exception as e:
logger.debug(f"Decoding failed, error message: {str(e)}")
# 如果解码失败,尝试 fallback 使用 apparent_encoding
response.encoding = response.apparent_encoding
return response.text
else:
return response.text
except Exception as e:
logger.debug(f"Error when getting decoded content: {str(e)}")
return response.text