datamarket 0.7.98__tar.gz → 0.7.99__tar.gz

{datamarket-0.7.98 → datamarket-0.7.99}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: datamarket
-Version: 0.7.98
+Version: 0.7.99
 Summary: Utilities that integrate advanced scraping knowledge into just one library.
 License: GPL-3.0-or-later
 Author: DataMarket

{datamarket-0.7.98 → datamarket-0.7.99}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "datamarket"
-version = "0.7.98"
+version = "0.7.99"
 description = "Utilities that integrate advanced scraping knowledge into just one library."
 authors = ["DataMarket <techsupport@datamarket.es>"]
 license = "GPL-3.0-or-later"

{datamarket-0.7.98 → datamarket-0.7.99}/src/datamarket/params/nominatim.py

@@ -412,7 +412,7 @@ COUNTRY_PARSING_RULES = {
         
         "zip_search_pattern": re.compile(r"\b\d{5}\b"), 
         
-        "phone_validate_pattern": re.compile(r"^(\+?34)?[6|7]\d{8}$")
+        "phone_validate_pattern": re.compile(r"^(\+?34)?[67]\d{8}$")
     },
     "pt": {
         "zip_validate_pattern": re.compile(r"^\d{4}[- ]{0,1}\d{3}$|^\d{4}$"),

datamarket-0.7.99/src/datamarket/utils/strings/standardization.py

@@ -0,0 +1,38 @@
+########################################################################################################################
+# IMPORTS
+
+import re
+from typing import Literal
+from ...params.nominatim import COUNTRY_PARSING_RULES
+
+########################################################################################################################
+# FUNCTIONS
+
+def parse_phone_number(number: str, country_code: Literal["es", "pt"]) -> str | None:
+    """Clean and standardize phone number from a certain country_code
+
+    Args:
+        number (str): phone number
+        country_code (Literal["es", "pt"]): country code of the phone number to parse
+
+    Raises:
+        ValueError: when parsing is not supported for a certain country
+
+    Returns:
+        str | None: standardized phone number
+    """
+    clean_number = re.sub(r"\D", "", number)
+    if country_code in {"es", "pt"}:
+        # Get the validation regex from params
+        pattern = COUNTRY_PARSING_RULES[country_code]["phone_validate_pattern"]
+
+        # Validate and extract in one step
+        if len(clean_number) >= 9: # Check if the cleaned number has at least 9 digits
+            match = pattern.match(clean_number)
+
+            # Return the captured group (the 9-digit number)
+            return match.group(0)[-9:] if match else None
+        else:
+            return None # Or handle the case where the number is too short
+    else:
+        raise ValueError(f"Country code ({country_code}) is not currently supported")

datamarket-0.7.98/src/datamarket/utils/strings/standardization.py

@@ -1,69 +0,0 @@
-########################################################################################################################
-# IMPORTS
-
-import re
-from typing import Literal
-from ...params.nominatim import COUNTRY_PARSING_RULES
-
-########################################################################################################################
-# FUNCTIONS
-
-
-def _standardize_es_phone_number(number: str) -> str | None:
-    """Standardize phone numbers from Spain using regex validation.
-
-    Args:
-        number (str): cleaned, digits-only phone number
-
-    Returns:
-        str | None: standardized 9-digit phone number
-    """
-    # Get the validation regex from params
-    pattern = COUNTRY_PARSING_RULES["es"]["phone_validate_pattern"]
-
-    # Validate and extract in one step
-    match = pattern.match(number)
-
-    # Return the captured group (the 9-digit number)
-    return match.group(1) if match else None
-
-
-def _standardize_pt_phone_number(number: str) -> str | None:
-    """Standardize phone numbers from Portugal using regex validation.
-
-    Args:
-        number (str): cleaned, digits-only phone number
-
-    Returns:
-        str | None: standardized 9-digit phone number
-    """
-    # Get the validation regex from params
-    pattern = COUNTRY_PARSING_RULES["pt"]["phone_validate_pattern"]
-
-    # Validate and extract in one step
-    match = pattern.match(number)
-
-    # Return the captured group (the 9-digit number)
-    return match.group(1) if match else None
-
-
-def parse_phone_number(number: str, country_code: Literal["es", "pt"]) -> str | None:
-    """Clean and standardize phone number from a certain country_code
-
-    Args:
-        number (str): phone number
-        country_code (Literal["es", "pt"]): country code of the phone number to parse
-
-    Raises:
-        ValueError: when parsing is not supported for a certain country
-
-    Returns:
-        str | None: standardized phone number
-    """
-    clean_number = re.sub(r"\D", "", number)
-    if country_code == "es":
-        return _standardize_es_phone_number(clean_number)
-    elif country_code == "pt":
-        return _standardize_pt_phone_number(clean_number)
-    else:
-        raise ValueError(f"Country code ({country_code}) is not currently supported")