Rhaarm validators

benchmark/benchmark.py

@@ -5,17 +5,7 @@
 from fastavro import writer, reader
 from fastavro._timezone import utc
 
-try:
-    from fastavro.validate import validate, validate_many
-except ImportError:
-    try:
-        from fastavro._write import validate
-    except ImportError:
-        from fastavro._write_py import validate
-
-
-    def validate_many(records, schema):
-        return all([validate(record, schema) for record in records])
+from fastavro.validate import validate, validate_many
 
 
 def write(schema, records, runs=1):

fastavro/__init__.py

@@ -47,7 +47,7 @@
 import fastavro.read
 import fastavro.write
 import fastavro.schema
-import fastavro.validate
+import fastavro.validation
 
 
 def _acquaint_schema(schema):
@@ -71,7 +71,7 @@ def _acquaint_schema(schema):
 acquaint_schema = _acquaint_schema
 fastavro.schema.acquaint_schema = _acquaint_schema
 is_avro = fastavro.read.is_avro
-validator = fastavro.validate.validate
+validate = fastavro.validation.validate
 
 __all__ = [
     n for n in locals().keys() if not n.startswith('_')

fastavro/_schema.pyx

@@ -28,7 +28,7 @@ cpdef inline str extract_logical_type(schema):
     return None
 
 
-def schema_name(object schema, parent_ns):
+def schema_name(schema, parent_ns):
     name = schema.get('name')
     if not name:
         return parent_ns, None

fastavro/_validate_common.py

@@ -9,10 +9,10 @@ def __str__(self):
             self.field = ''
 
         if self.datum is None:
-            return 'Field({field}) is null' \
+            return 'Field({field}) is None' \
                    ' expected {schema}'.format(field=self.field,
                                                schema=self.schema)
-        return '{field} is {datum} of type ' \
+        return '{field} is <{datum}> of type ' \
                '{given_type} expected {schema}'. \
             format(datum=self.datum, given_type=type(self.datum),
                    schema=self.schema, field=self.field)

fastavro/_validate.pyx → fastavro/_validation.pyx

@@ -19,8 +19,8 @@ cdef int32 INT_MAX_VALUE = const.INT_MAX_VALUE
 cdef long64 LONG_MIN_VALUE = const.LONG_MIN_VALUE
 cdef long64 LONG_MAX_VALUE = const.LONG_MAX_VALUE
 
-cpdef bint validate_null(datum, schema=None,
-                         str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_null(datum, schema=None,
+                               str parent_ns='', bint raise_errors=False):
     """
     Checks that the data value is None.
 
@@ -32,8 +32,8 @@ cpdef bint validate_null(datum, schema=None,
     """
     return datum is None
 
-cpdef bint validate_boolean(datum, schema=None,
-                            str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_boolean(datum, schema=None,
+                                  str parent_ns='', bint raise_errors=False):
     """Check that the data value is bool instance
 
     :param datum: data to validate as boolean
@@ -44,8 +44,8 @@ cpdef bint validate_boolean(datum, schema=None,
     """
     return isinstance(datum, bool)
 
-cpdef bint validate_string(datum, schema=None,
-                           str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_string(datum, schema=None,
+                                 str parent_ns='', bint raise_errors=False):
     """Check that the data value is string type, uses
     six for Python version compatibility.
 
@@ -57,8 +57,8 @@ cpdef bint validate_string(datum, schema=None,
     """
     return is_str(datum)
 
-cpdef bint validate_bytes(datum, schema=None,
-                          str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_bytes(datum, schema=None,
+                                str parent_ns='', bint raise_errors=False):
     """
     Check that the data value is
     (bytes or decimal.Decimal)
@@ -71,8 +71,8 @@ cpdef bint validate_bytes(datum, schema=None,
     """
     return isinstance(datum, (bytes, decimal.Decimal))
 
-cpdef bint validate_int(datum, schema=None,
-                        str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_int(datum, schema=None,
+                              str parent_ns='', bint raise_errors=False):
     """
     Check that the data value is a non floating
     point number with size less that Int32.
@@ -95,8 +95,8 @@ cpdef bint validate_int(datum, schema=None,
             datetime.time, datetime.datetime, datetime.date))
     )
 
-cpdef bint validate_long(datum, schema=None,
-                         str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_long(datum, schema=None,
+                               str parent_ns='', bint raise_errors=False):
     """
     Check that the data value is a non floating
     point number with size less that Int32.
@@ -118,8 +118,8 @@ cpdef bint validate_long(datum, schema=None,
             datetime.time, datetime.datetime, datetime.date))
     )
 
-cpdef bint validate_float(datum, schema=None,
-                          str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_float(datum, schema=None,
+                                str parent_ns='', bint raise_errors=False):
     """
     Check that the data value is a floating
     point number or double precision.
@@ -132,8 +132,8 @@ cpdef bint validate_float(datum, schema=None,
     """
     return isinstance(datum, (int, long, float, numbers.Real))
 
-cpdef bint validate_fixed(datum, dict schema,
-                          str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_fixed(datum, dict schema,
+                                str parent_ns='', bint raise_errors=False):
     """
     Check that the data value is fixed width bytes,
     matching the schema['size'] exactly!
@@ -148,8 +148,8 @@ cpdef bint validate_fixed(datum, dict schema,
             len(datum) == schema['size']) or \
            (isinstance(datum, decimal.Decimal))
 
-cpdef bint validate_enum(datum, dict schema,
-                         str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_enum(datum, dict schema,
+                               str parent_ns='', bint raise_errors=False):
     """
     Check that the data value matches one of the enum symbols.
 
@@ -163,8 +163,8 @@ cpdef bint validate_enum(datum, dict schema,
     """
     return datum in schema['symbols']
 
-cpdef bint validate_array(datum, dict schema,
-                          str parent_ns='', bint raise_errors=False):
+cdef inline bint validate_array(datum, dict schema,
+                                str parent_ns='', bint raise_errors=False):
     """
     Check that the data list values all match schema['items'].
 
@@ -189,8 +189,8 @@ cpdef bint validate_array(datum, dict schema,
             return False
     return True
 
-cpdef bint validate_map(object datum, dict schema, str parent_ns='',
-                        bint raise_errors=False):
+cdef inline bint validate_map(object datum, dict schema, str parent_ns='',
+                              bint raise_errors=False):
     """
     Check that the data is a Map(k,v)
     matching values to schema['values'] type.
@@ -220,8 +220,8 @@ cpdef bint validate_map(object datum, dict schema, str parent_ns='',
             return False
     return True
 
-cpdef bint validate_record(object datum, dict schema, str parent_ns='',
-                           bint raise_errors=False):
+cdef inline bint validate_record(object datum, dict schema, str parent_ns='',
+                                 bint raise_errors=False):
     """
     Check that the data is a Mapping type with all schema defined fields
     validated as True.
@@ -247,8 +247,8 @@ cpdef bint validate_record(object datum, dict schema, str parent_ns='',
             return False
     return True
 
-cpdef bint validate_union(object datum, list schema, str parent_ns=None,
-                          bint raise_errors=False):
+cdef inline bint validate_union(object datum, list schema, str parent_ns=None,
+                                bint raise_errors=False):
     """
     Check that the data is a list type with possible options to
     validate as True.
@@ -287,36 +287,6 @@ cpdef bint validate_union(object datum, list schema, str parent_ns=None,
         raise ValidationError(*errors)
     return False
 
-cpdef BASE_VALIDATORS = {
-    'null': validate_null,
-    'boolean': validate_boolean,
-    'string': validate_string,
-    'int': validate_int,
-    'long': validate_long,
-    'float': validate_float,
-    'double': validate_float,
-    'bytes': validate_bytes,
-    'fixed': validate_fixed,
-    'enum': validate_enum,
-    'array': validate_array,
-    'map': validate_map,
-    'union': validate_union,
-    'error_union': validate_union,
-    'record': validate_record,
-    'error': validate_record,
-    'request': validate_record
-}
-
-cpdef VALIDATORS = BASE_VALIDATORS.copy()
-
-cpdef void register_validator(record_type, validator):
-    if record_type in BASE_VALIDATORS:
-        raise ValueError("Not allowed to override Base Validators.")
-    VALIDATORS[record_type] = validator
-
-cpdef get_validator(record_type):
-    return VALIDATORS.get(record_type)
-
 cpdef validate(object datum, object schema, str field='',
                bint raise_errors=False):
     """Determine if a python datum is an instance of a schema."""
@@ -369,24 +339,17 @@ cpdef validate(object datum, object schema, str field='',
     elif record_type in ('record', 'error', 'request'):
         result = validate_record(datum, schema=schema, parent_ns=ns_field,
                                  raise_errors=raise_errors)
-    else:
-        validator = get_validator(record_type)
-        if validator:
-            result = validator(datum, schema=schema, parent_ns=ns_field,
-                               raise_errors=raise_errors)
-
-    if record_type in SCHEMA_DEFS and result is None:
+    elif record_type in SCHEMA_DEFS:
         result = validate(datum,
                           schema=SCHEMA_DEFS[record_type],
                           field=ns_field,
                           raise_errors=raise_errors)
+    else:
+        raise UnknownType(record_type)
 
     if raise_errors and result is False:
         raise ValidationError(ValidationErrorData(datum, schema, ns_field))
 
-    if result is None:
-        raise UnknownType(record_type)
-
     return bool(result)
 
 

fastavro/_validate_py.py → fastavro/_validation_py.py

@@ -3,8 +3,9 @@
 import numbers
 from collections import Iterable, Mapping
 
-from fastavro.const import INT_MAX_VALUE, INT_MIN_VALUE, \
-    LONG_MAX_VALUE, LONG_MIN_VALUE
+from fastavro.const import (
+    INT_MAX_VALUE, INT_MIN_VALUE, LONG_MAX_VALUE, LONG_MIN_VALUE
+)
 from ._validate_common import ValidationError, ValidationErrorData
 from .schema import extract_record_type, schema_name, UnknownType
 from .six import long, is_str, iterkeys, itervalues
@@ -161,7 +162,7 @@ def validate_array(datum, schema, parent_ns=None, raise_errors=False):
     :except: ValidationError
     """
     if raise_errors:
-        namespace, name = schema_name(schema, parent_ns)
+        _, name = schema_name(schema, parent_ns)
     else:
         name = parent_ns
     return (
@@ -186,7 +187,7 @@ def validate_map(datum, schema, parent_ns=None, raise_errors=False):
     :except: ValidationError
     """
     if raise_errors:
-        namespace, name = schema_name(schema, parent_ns)
+        _, name = schema_name(schema, parent_ns)
     else:
         name = parent_ns
     return (
@@ -265,7 +266,7 @@ def validate_union(datum, schema, parent_ns=None, raise_errors=False):
     return False
 
 
-BASE_VALIDATORS = {
+VALIDATORS = {
     'null': validate_null,
     'boolean': validate_boolean,
     'string': validate_string,
@@ -285,18 +286,6 @@ def validate_union(datum, schema, parent_ns=None, raise_errors=False):
     'request': validate_record
 }
 
-VALIDATORS = BASE_VALIDATORS.copy()
-
-
-def register_validator(record_type, validator):
-    if record_type in BASE_VALIDATORS:
-        raise ValueError("Not allowed to override Base Validators.")
-    VALIDATORS[record_type] = validator
-
-
-def get_validator(record_type):
-    return VALIDATORS.get(record_type)
-
 
 def validate(datum, schema, field=None, raise_errors=False):
     """Determine if a python datum is an instance of a schema."""
@@ -309,7 +298,7 @@ def validate(datum, schema, field=None, raise_errors=False):
     elif field:
         ns_field = field
 
-    validator = get_validator(record_type)
+    validator = VALIDATORS.get(record_type)
     if validator:
         result = validator(datum, schema=schema,
                            parent_ns=ns_field,

fastavro/_write.pyx

@@ -16,7 +16,7 @@ from os import urandom
 from zlib import compress
 
 from fastavro import const
-from ._validate import validate
+from ._validation import validate
 from ._six import utob, long, iteritems, mk_bits
 from ._read import HEADER_SCHEMA, SYNC_SIZE, MAGIC
 from ._schema import (

fastavro/_write_py.py

@@ -17,7 +17,7 @@
 from struct import pack
 from zlib import compress
 
-from .validate import validate
+from .validation import validate
 from .const import (
     MCS_PER_HOUR, MCS_PER_MINUTE, MCS_PER_SECOND, MLS_PER_HOUR, MLS_PER_MINUTE,
     MLS_PER_SECOND, DAYS_SHIFT