v1.20.11 Performance improvements

tableschema/VERSION

@@ -1 +1 @@
-1.20.10
+1.20.11

tableschema/field.py

@@ -46,6 +46,7 @@ def __init__(self, descriptor, missing_values=config.DEFAULT_MISSING_VALUES,
         self.__schema = schema
         self.__cast_function = self.__get_cast_function()
         self.__check_functions = self.__get_check_functions()
+        self.__preserve_missing_values = os.environ.get('TABLESCHEMA_PRESERVE_MISSING_VALUES')
 
     @cached_property
     def schema(self):
@@ -155,7 +156,7 @@ def cast_value(self, value, constraints=True):
         # Null value
         if value in self.__missing_values:
             # Whether missing_values should be preserved without being cast
-            if os.environ.get('TABLESCHEMA_PRESERVE_MISSING_VALUES'):
+            if self.__preserve_missing_values:
                 return value
             value = None
 

tableschema/types/integer.py

@@ -21,7 +21,7 @@ def cast_integer(format, value, **options):
 
     elif isinstance(value, six.string_types):
         if not options.get('bareNumber', _DEFAULT_BARE_NUMBER):
-            value = re.sub(r'((^\D*)|(\D*$))', '', value)
+            value = _RE_BARE_NUMBER.sub('', value)
 
         try:
             value = int(value)
@@ -41,5 +41,5 @@ def cast_integer(format, value, **options):
 
 
 # Internal
-
+_RE_BARE_NUMBER = re.compile(r'((^\D*)|(\D*$))')
 _DEFAULT_BARE_NUMBER = True

tableschema/types/number.py

@@ -13,31 +13,42 @@
 # Module API
 
 def cast_number(format, value, **options):
-    group_char = options.get('groupChar', _DEFAULT_GROUP_CHAR)
-    decimal_char = options.get('decimalChar', _DEFAULT_DECIMAL_CHAR)
-    if not isinstance(value, Decimal):
-        if isinstance(value, six.string_types):
-            value = re.sub(r'\s', '', value)
-            value = value.replace(decimal_char, '__decimal_char__')
+    if isinstance(value, six.string_types):
+        group_char = options.get('groupChar', _DEFAULT_GROUP_CHAR)
+        decimal_char = options.get('decimalChar', _DEFAULT_DECIMAL_CHAR)
+        value = _RE_WHITESPACE.sub('', value)
+        if decimal_char != '.':
+            if group_char:
+                value = value.replace(decimal_char, '__decimal_char__')
+                value = value.replace(group_char, '')
+                value = value.replace('__decimal_char__', '.')
+            else:
+                value = value.replace(decimal_char, '__decimal_char__')
+                value = value.replace('__decimal_char__', '.')
+        elif group_char:
             value = value.replace(group_char, '')
-            value = value.replace('__decimal_char__', '.')
-            if not options.get('bareNumber', _DEFAULT_BARE_NUMBER):
-                value = re.sub(r'((^\D*)|(\D*$))', '', value)
-        elif not isinstance(value, six.integer_types + (float,)):
-            return ERROR
-        elif value is True or value is False:
-            return ERROR
-        try:
-            if isinstance(value, float):
-                value = str(value)
-            value = Decimal(value)
-        except Exception:
-            return ERROR
+
+        if not options.get('bareNumber', _DEFAULT_BARE_NUMBER):
+            value = _RE_BARE_NUMBER.sub('', value)
+    elif isinstance(value, Decimal):
+        return value
+    elif not isinstance(value, six.integer_types + (float,)):
+        return ERROR
+    elif value is True or value is False:
+        return ERROR
+    else:
+        value = str(value)
+    try:
+        value = Decimal(value)
+    except Exception:
+        return ERROR
     return value
 
 
 # Internal
 
+_RE_WHITESPACE = re.compile(r'\s')
+_RE_BARE_NUMBER = re.compile(r'((^\D*)|(\D*$))')
 _DEFAULT_GROUP_CHAR = ''
 _DEFAULT_DECIMAL_CHAR = '.'
 _DEFAULT_BARE_NUMBER = True

tableschema/types/string.py

@@ -19,6 +19,8 @@
 def cast_string(format, value, **options):
     if not isinstance(value, six.string_types):
         return ERROR
+    if format in _SIMPLE_FORMATS:
+        return value
     if format == 'uri':
         uri = _uri_from_string(value)
         try:
@@ -43,6 +45,7 @@ def cast_string(format, value, **options):
 
 # Internal
 
+_SIMPLE_FORMATS = {'default', None}
 _EMAIL_PATTERN = re.compile(r'[^@]+@[^@]+\.[^@]+')
 _uri_from_string = rfc3986.uri.URIReference.from_string
 _uri_validator = rfc3986.validators.Validator().require_presence_of('scheme')