re-common 10.0.8__tar.gz → 10.0.9__tar.gz

{re_common-10.0.8 → re_common-10.0.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: re_common
-Version: 10.0.8
+Version: 10.0.9
 Summary: a library about all python projects
 Home-page: https://gitee.com/xujiangios/re-common
 Author: vic

{re_common-10.0.8 → re_common-10.0.9}/re_common/v2/baselibrary/tools/dict_tools.py

@@ -2,6 +2,7 @@ class DotDict(dict):
     """
     让字典成为对象 既可以用字典方式访问 也可以用点访问key
     """
+
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
         # 递归地将嵌套字典转换为 DotDict
@@ -21,4 +22,16 @@ class DotDict(dict):
     def __setattr__(self, key, value):
         if isinstance(value, dict):  # 如果值是字典，转换为 DotDict
             value = DotDict(value)
-        self[key] = value
+        self[key] = value
+
+    def to_dict(self):
+        """
+        将 DotDict 实例转换为普通字典
+        """
+        result = {}
+        for key, value in self.items():
+            if isinstance(value, DotDict):
+                result[key] = value.to_dict()
+            else:
+                result[key] = value
+        return result

{re_common-10.0.8 → re_common-10.0.9}/re_common/v2/baselibrary/tools/search_hash_tools.py

@@ -1,7 +1,7 @@
 from typing import List
 
 import jieba
-from datasketch import MinHash
+from datasketch import MinHash, minhash
 
 
 def tokenize(text: str, stopwords=None) -> List[str]:
@@ -31,3 +31,13 @@ def create_minhash(words: List[str], num_perm=128) -> MinHash:
     for word in words:
         minhash.update(word.encode("utf-8"))
     return minhash
+
+
+def get_str_minhash(title):
+    from re_common.v2.baselibrary.utils.string_clear import rel_clear
+    rel_title = rel_clear(title)
+    if not rel_title:
+        return ""
+    words = tokenize(rel_title)
+    minhash = create_minhash(words)
+    return minhash

re_common-10.0.9/re_common/v2/baselibrary/utils/string_bool.py

@@ -0,0 +1,46 @@
+import re
+
+
+def is_all_english_chars(s):
+    return bool(re.match(r'^[A-Za-z]+$', s))
+
+
+def contains_chinese_chars(s):
+    return bool(re.search(r'[\u3400-\u9fff]', s))
+
+
+def is_empty(value):
+    """
+    判断一个值是否为空。
+
+    支持的类型：
+    - None
+    - 空字符串（去除空白后）
+    - pandas 的 NaN
+    - 其他可迭代类型（如列表、字典等）的长度为 0
+    - 其他情况返回 False
+    """
+    # 如果是 None，直接返回 True
+    if value is None:
+        return True
+
+    # 尝试处理 pandas 的 NaN
+    try:
+        import pandas as pd
+        if pd.isna(value):
+            return True
+    except ImportError:
+        pass  # 如果没有安装 pandas，跳过
+
+    # 如果是字符串，检查去除空白后是否为空
+    if isinstance(value, str):
+        return value.strip() == ""
+
+
+
+    # 处理其他可迭代类型（如列表、字典等）
+    if hasattr(value, "__len__"):
+        return len(value) == 0
+
+    # 默认情况下，非 None、非空类型返回 False
+    return False

{re_common-10.0.8 → re_common-10.0.9}/re_common/v2/baselibrary/utils/string_clear.py

@@ -165,10 +165,11 @@ def ref_clear(str_obj):
 def clear_obj(str_obj):
     # 为对象化定制的清理
     str_obj = clear_au_organ(str_obj)
-    str_obj = str_obj.replace("ß", "SS")
-    return (StringClear(str_obj)
-            .remove_diacritics()  # 清理音标
-            .upper()
-            .get_str()  # 获取str
-            .strip()  # 去掉空格
-            )
+    # str_obj = str_obj.replace("ß", "SS") # "ß" 的 大写就是 "SS"
+    result = (StringClear(str_obj)
+              .remove_diacritics()  # 清理音标
+              .upper()
+              .get_str()  # 获取str
+              .strip()  # 去掉空格
+              )
+    return result

{re_common-10.0.8 → re_common-10.0.9}/re_common.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: re_common
-Version: 10.0.8
+Version: 10.0.9
 Summary: a library about all python projects
 Home-page: https://gitee.com/xujiangios/re-common
 Author: vic

{re_common-10.0.8 → re_common-10.0.9}/setup.py

@@ -34,7 +34,7 @@ long_description = """
 """
 setuptools.setup(
     name="re_common",
-    version="10.0.8",
+    version="10.0.9",
     author="vic",
     author_email="xujiang5@163.com",
     description="a library about all python projects",

re_common-10.0.8/re_common/v2/baselibrary/utils/string_bool.py

@@ -1,26 +0,0 @@
-import re
-
-
-def is_all_english_chars(s):
-    return bool(re.match(r'^[A-Za-z]+$', s))
-
-
-def contains_chinese_chars(s):
-    return bool(re.search(r'[\u3400-\u9fff]', s))
-
-
-def is_empty(value):
-    # 如果是 None，直接返回 True
-    if value is None:
-        return True
-
-    # 如果是字符串，检查去除空白后是否为空
-    if isinstance(value, str):
-        return value.strip() == ""
-
-    # 可选：处理其他可迭代类型（如列表、字典等）
-    if hasattr(value, "__len__"):
-        return len(value) == 0
-
-    # 默认情况下，非 None、非空类型返回 False
-    return False