datamarket 0.9.26__tar.gz → 0.9.27__tar.gz

{datamarket-0.9.26 → datamarket-0.9.27}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: datamarket
-Version: 0.9.26
+Version: 0.9.27
 Summary: Utilities that integrate advanced scraping knowledge into just one library.
 License: GPL-3.0-or-later
 Author: DataMarket

{datamarket-0.9.26 → datamarket-0.9.27}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "datamarket"
-version = "0.9.26"
+version = "0.9.27"
 description = "Utilities that integrate advanced scraping knowledge into just one library."
 authors = ["DataMarket <techsupport@datamarket.es>"]
 license = "GPL-3.0-or-later"

{datamarket-0.9.26 → datamarket-0.9.27}/src/datamarket/utils/strings.py

@@ -37,8 +37,14 @@ class NamingConvention(Enum):
 
 def transliterate_symbols(s: str) -> str:
     """
-    Translates symbols (category S*) to lowercase Unicode names,
-    with spaces→underscores. The rest of the text remains the same.
+    Translates Unicode symbols (category S*) in the input string to their lowercase Unicode names,
+    with spaces replaced by underscores. Other characters remain unchanged.
+
+    Args:
+        s: The input string.
+
+    Returns:
+        The string with symbols transliterated.
     """
     out: list[str] = []
     for c in s:
@@ -55,16 +61,32 @@ def normalize(
     s: Any, mode: NormalizationMode = NormalizationMode.BASIC, naming: NamingConvention = NamingConvention.NONE
 ) -> str:
     """
-    1. Normalizes the string according to `mode`:
-       - NONE: returns the original input as an unprocessed string.
-       - BASIC: removes accents, converts punctuation to spaces, preserves alphanumeric characters.
-       - SYMBOLS: translates only symbols to Unicode name.
-       - FULL: combines BASIC + SYMBOLS.
-    2. Applies naming convention according to `naming`:
-       - NONE: returns the normalized text.
-       - PARAM: parameterize (hyphens).
-       - SNAKE: snake_case (underscore, lowercase).
-       - CONSTANT: CONSTANT_CASE (underscore, uppercase).
+    Normalizes and applies a naming convention to the input.
+
+    Handles None and NaN values by returning an empty string. Converts non-string inputs to strings.
+
+    Normalization is applied according to `mode`:
+    - NONE: Returns the input as a string without any normalization.
+    - BASIC: Removes accents, converts punctuation and spaces to single spaces, and preserves alphanumeric characters.
+    - SYMBOLS: Translates only Unicode symbols (category S*) to their lowercase Unicode names with underscores.
+    - FULL: Applies both BASIC and SYMBOLS normalization.
+
+    After normalization, a naming convention is applied according to `naming`:
+    - NONE: Returns the normalized text.
+    - CONSTANT: Converts to CONSTANT_CASE (uppercase with underscores).
+    - SNAKE: Converts to snake_case (lowercase with underscores).
+    - CAMEL: Converts to camelCase (lowercase first word, capitalize subsequent words, no spaces).
+    - PASCAL: Converts to PascalCase (capitalize all words, no spaces).
+    - PARAM: Converts to parameterize (lowercase with hyphens).
+    - TITLE: Converts to Title Case (capitalize each word).
+
+    Args:
+        s: The input value to normalize and format. Can be any type.
+        mode: The normalization mode to apply. Defaults to NormalizationMode.BASIC.
+        naming: The naming convention to apply. Defaults to NamingConvention.NONE.
+
+    Returns:
+        The normalized and formatted string.
     """
     # Parameter mapping
     if isinstance(mode, str):