refactored jsonconfig, included jsonschema validation

and type casting.

6 files changed, 565 insertions, 88 deletions

diff --git a/src/leap/base/config.py b/src/leap/base/config.py
index 7f69a41c..dc047f80 100644
--- a/src/leap/base/config.py
+++ b/src/leap/base/config.py
@@ -9,14 +9,12 @@ import tempfile
 import os
 
 logger = logging.getLogger(name=__name__)
-logger.setLevel('DEBUG')
 
-import configuration
-import jsonschema
 import requests
 
 from leap.base import exceptions
 from leap.base import constants
+from leap.base.pluggableconfig import PluggableConfig
 from leap.util.fileutil import (mkdir_p)
 
 # move to base!
@@ -47,20 +45,6 @@ class BaseLeapConfig(object):
         raise NotImplementedError("abstract base class")
 
 
-class SchemaEncoder(json.JSONEncoder):
-    def default(self, obj):
-        if obj is str:
-            return 'string'
-        if obj is unicode:
-            return 'string'
-        if obj is int:
-            return 'int'
-        if obj is list:
-            return 'array'
-        if obj is dict:
-            return 'object'
-
-
 class MetaConfigWithSpec(type):
     """
     metaclass for JSONLeapConfig classes.
@@ -73,63 +57,43 @@ class MetaConfigWithSpec(type):
     # place where we want to enforce
     # singletons, read-only and similar stuff.
 
-    # TODO:
-    # - add a error handler for missing options that
-    #   we can act easily upon (sys.exit is ugly, for $deity's sake)
-
     def __new__(meta, classname, bases, classDict):
         schema_obj = classDict.get('spec', None)
-        if schema_obj:
-            spec_options = schema_obj.get('properties', None)
-            schema_json = SchemaEncoder().encode(schema_obj)
-            schema = json.loads(schema_json)
-        else:
-            spec_options = None
-            schema = None
+
         # not quite happy with this workaround.
         # I want to raise if missing spec dict, but only
         # for grand-children of this metaclass.
         # maybe should use abc module for this.
         abcderived = ("JSONLeapConfig",)
-        if spec_options is None and classname not in abcderived:
-            if not schema_obj:
-                raise exceptions.ImproperlyConfigured(
-                    "missing spec dict on your derived class (%s)" % classname)
-            if schema_obj and not spec_options:
-                raise exceptions.ImproperlyConfigured(
-                    "missing properties attr in spec dict "
-                    "on your derived class (%s)" % classname)
-
-        # we create a configuration spec attribute from the spec dict
+        if schema_obj is None and classname not in abcderived:
+            raise exceptions.ImproperlyConfigured(
+                "missing spec dict on your derived class (%s)" % classname)
+
+        # we create a configuration spec attribute
+        # from the spec dict
         config_class = type(
             classname + "Spec",
-            (configuration.Configuration, object),
-            {'options': spec_options})
+            (PluggableConfig, object),
+            {'options': schema_obj})
         classDict['spec'] = config_class
-        # A shipped json-schema for validation
-        classDict['schema'] = schema
 
         return type.__new__(meta, classname, bases, classDict)
 
 ##########################################################
-# hacking in progress:
+# some hacking still in progress:
 
 # Configs have:
+
 # - a slug (from where a filename/folder is derived)
 # - a spec (for validation and defaults).
-#   this spec is basically a dict that will be used
+#   this spec is conformant to the json-schema.
+#   basically a dict that will be used
 #   for type casting and validation, and defaults settings.
 
 # all config objects, since they are derived from  BaseConfig, implement basic
 # useful methods:
 # - save
 # - load
-# - get_config (returns a optparse.OptionParser object)
-
-# TODO:
-# [done] raise validation errors
-# - have a good type cast repertory (uris, version, hashes...)
-# - multilingual objects
 
 ##########################################################
 
@@ -152,10 +116,10 @@ class JSONLeapConfig(BaseLeapConfig):
             raise exceptions.ImproperlyConfigured(
                 "missing spec on JSONLeapConfig"
                 " derived class")
-        assert issubclass(self.spec, configuration.Configuration)
+        assert issubclass(self.spec, PluggableConfig)
 
-        self._config = self.spec()
-        self._config.parse_args(list(args))
+        self._config = self.spec(format="json")
+        self._config.load()
         self.fetcher = kwargs.pop('fetcher', requests)
 
     # mandatory baseconfig interface
@@ -166,13 +130,6 @@ class JSONLeapConfig(BaseLeapConfig):
         folder, filename = os.path.split(to)
         if folder and not os.path.isdir(folder):
             mkdir_p(folder)
-        # lazy evaluation until first level of nesting
-        # to allow lambdas with context-dependant info
-        # like os.path.expanduser
-        config = self.get_config()
-        for k, v in config.iteritems():
-            if callable(v):
-                config[k] = v()
         self._config.serialize(to)
 
     def load(self, fromfile=None, from_uri=None, fetcher=None, verify=False):
@@ -183,10 +140,7 @@ class JSONLeapConfig(BaseLeapConfig):
         if fromfile is None:
             fromfile = self.filename
         if os.path.isfile(fromfile):
-            newconfig = self._config.deserialize(fromfile)
-            # XXX check for no errors, etc
-            # XXX could validate here!
-            self._config.config = newconfig
+            self._config.load(fromfile=fromfile)
         else:
             logger.error('tried to load config from non-existent path')
             logger.error('Not Found: %s', fromfile)
@@ -196,19 +150,25 @@ class JSONLeapConfig(BaseLeapConfig):
             fetcher = self.fetcher
         logger.debug('verify: %s', verify)
         request = fetcher.get(uri, verify=verify)
+        # XXX should send a if-modified-since header
 
         # XXX get 404, ...
         # and raise a UnableToFetch...
         request.raise_for_status()
         fd, fname = tempfile.mkstemp(suffix=".json")
-        if not request.json:
+
+        if request.json:
+            self._config.load(json.dumps(request.json))
+
+        else:
+            # not request.json
+            # might be server did not announce content properly,
+            # let's try deserializing all the same.
             try:
-                json.loads(request.content)
+                self._config.load(request.content)
             except ValueError:
                 raise eipexceptions.LeapBadConfigFetchedError
-        with open(fname, 'w') as tmp:
-            tmp.write(json.dumps(request.json))
-        self._loadtemp(fname)
+
         return True
 
     def get_config(self):
@@ -223,20 +183,16 @@ class JSONLeapConfig(BaseLeapConfig):
     def filename(self):
         return self.get_filename()
 
-    def jsonvalidate(self, data):
-        jsonschema.validate(data, self.schema)
+    def validate(self, data):
+        logger.debug('validating schema')
+        self._config.validate(data)
         return True
 
     # private
 
-    def _loadtemp(self, filename):
-        self.load(fromfile=filename)
-        os.remove(filename)
-
     def _slug_to_filename(self):
         # is this going to work in winland if slug is "foo/bar" ?
         folder, filename = os.path.split(self.slug)
-        # XXX fix import
         config_file = get_config_file(filename, folder)
         return config_file
 
diff --git a/src/leap/base/constants.py b/src/leap/base/constants.py
index 7a1415fb..48a18dc3 100644
--- a/src/leap/base/constants.py
+++ b/src/leap/base/constants.py
@@ -16,7 +16,7 @@ DEFINITION_EXPECTED_PATH = "provider.json"
 DEFAULT_PROVIDER_DEFINITION = {
     u'api_uri': u'https://api.%s/' % DEFAULT_PROVIDER,
     u'api_version': u'0.1.0',
-    u'ca_cert': u'8aab80ae4326fd30721689db813733783fe0bd7e',
+    u'ca_cert_fingerprint': u'8aab80ae4326fd30721689db813733783fe0bd7e',
     u'ca_cert_uri': u'https://%s/cacert.pem' % DEFAULT_PROVIDER,
     u'description': {u'en': u'This is a test provider'},
     u'display_name': {u'en': u'Test Provider'},
diff --git a/src/leap/base/pluggableconfig.py b/src/leap/base/pluggableconfig.py
new file mode 100644
index 00000000..b8615ad8
--- /dev/null
+++ b/src/leap/base/pluggableconfig.py
@@ -0,0 +1,421 @@
+"""
+generic configuration handlers
+"""
+import copy
+import json
+import logging
+import os
+import time
+import urlparse
+
+import jsonschema
+
+logger = logging.getLogger(__name__)
+
+
+__all__ = ['PluggableConfig',
+           'adaptors',
+           'types',
+           'UnknownOptionException',
+           'MissingValueException',
+           'ConfigurationProviderException',
+           'TypeCastException']
+
+# exceptions
+
+
+class UnknownOptionException(Exception):
+    """exception raised when a non-configuration
+    value is present in the configuration"""
+
+
+class MissingValueException(Exception):
+    """exception raised when a required value is missing"""
+
+
+class ConfigurationProviderException(Exception):
+    """exception raised when a configuration provider is missing, etc"""
+
+
+class TypeCastException(Exception):
+    """exception raised when a
+    configuration item cannot be coerced to a type"""
+
+
+class ConfigAdaptor(object):
+    """
+    abstract base class for config adaotors for
+    serialization/deserialization and custom validation
+    and type casting.
+    """
+    def read(self, filename):
+        raise NotImplementedError("abstract base class")
+
+    def write(self, config, filename):
+        with open(filename, 'w') as f:
+            self._write(f, config)
+
+    def _write(self, fp, config):
+        raise NotImplementedError("abstract base class")
+
+    def validate(self, config, schema):
+        raise NotImplementedError("abstract base class")
+
+
+adaptors = {}
+
+
+class JSONSchemaEncoder(json.JSONEncoder):
+    """
+    custom default encoder that
+    casts python objects to json objects for
+    the schema validation
+    """
+    def default(self, obj):
+        if obj is str:
+            return 'string'
+        if obj is unicode:
+            return 'string'
+        if obj is int:
+            return 'integer'
+        if obj is list:
+            return 'array'
+        if obj is dict:
+            return 'object'
+        if obj is bool:
+            return 'boolean'
+
+
+class JSONAdaptor(ConfigAdaptor):
+    indent = 2
+    extensions = ['json']
+
+    def read(self, _from):
+        if isinstance(_from, file):
+            _from_string = _from.read()
+        if isinstance(_from, str):
+            _from_string = _from
+        return json.loads(_from_string)
+
+    def _write(self, fp, config):
+        fp.write(json.dumps(config,
+                 indent=self.indent,
+                 sort_keys=True))
+
+    def validate(self, config, schema_obj):
+        schema_json = JSONSchemaEncoder().encode(schema_obj)
+        schema = json.loads(schema_json)
+        jsonschema.validate(config, schema)
+
+
+adaptors['json'] = JSONAdaptor()
+
+#
+# Adaptors
+#
+# Allow to apply a predefined set of types to the
+# specs, so it checks the validity of formats and cast it
+# to proper python types.
+
+# TODO:
+# - multilingual object.
+# - HTTPS uri
+
+
+class DateType(object):
+    fmt = '%Y-%m-%d'
+
+    def to_python(self, data):
+        return time.strptime(data, self.fmt)
+
+    def get_prep_value(self, data):
+        return time.strftime(self.fmt, data)
+
+
+class URIType(object):
+
+    def to_python(self, data):
+        parsed = urlparse.urlparse(data)
+        if not parsed.scheme:
+            raise TypeCastException("uri %s has no schema" % data)
+        return parsed
+
+    def get_prep_value(self, data):
+        return data.geturl()
+
+
+class HTTPSURIType(object):
+
+    def to_python(self, data):
+        parsed = urlparse.urlparse(data)
+        if not parsed.scheme:
+            raise TypeCastException("uri %s has no schema" % data)
+        if parsed.scheme != "https":
+            raise TypeCastException(
+                "uri %s does not has "
+                "https schema" % data)
+        return parsed
+
+    def get_prep_value(self, data):
+        return data.geturl()
+
+
+types = {
+    'date': DateType(),
+    'uri': URIType(),
+    'https-uri': HTTPSURIType(),
+}
+
+
+class PluggableConfig(object):
+
+    options = {}
+
+    def __init__(self,
+                 adaptors=adaptors,
+                 types=types,
+                 format=None):
+
+        self.config = {}
+        self.adaptors = adaptors
+        self.types = types
+        self._format = format
+
+    @property
+    def option_dict(self):
+        if hasattr(self, 'options') and isinstance(self.options, dict):
+            return self.options.get('properties', None)
+
+    def items(self):
+        """
+        act like an iterator
+        """
+        if isinstance(self.option_dict, dict):
+            return self.option_dict.items()
+        return self.options
+
+    def validate(self, config, format=None):
+        """
+        validate config
+        """
+        schema = self.options
+        if format is None:
+            format = self._format
+
+        if format:
+            adaptor = self.get_adaptor(self._format)
+            adaptor.validate(config, schema)
+        else:
+            # we really should make format mandatory...
+            logger.error('no format passed to validate')
+
+        # first round of validation is ok.
+        # now we proceed to cast types if any specified.
+        self.to_python(config)
+
+    def to_python(self, config):
+        """
+        cast types following first type and then format indications.
+        """
+        unseen_options = [i for i in config if i not in self.option_dict]
+        if unseen_options:
+            raise UnknownOptionException(
+                "Unknown options: %s" % ', '.join(unseen_options))
+
+        for key, value in config.items():
+            _type = self.option_dict[key].get('type')
+            if _type is None and 'default' in self.option_dict[key]:
+                _type = type(self.option_dict[key]['default'])
+            if _type is not None:
+                tocast = True
+                if not callable(_type) and isinstance(value, _type):
+                    tocast = False
+                if tocast:
+                    try:
+                        config[key] = _type(value)
+                    except BaseException, e:
+                        raise TypeCastException(
+                            "Could not coerce %s, %s, "
+                            "to type %s: %s" % (key, value, _type.__name__, e))
+            _format = self.option_dict[key].get('format', None)
+            _ftype = self.types.get(_format, None)
+            if _ftype:
+                try:
+                    config[key] = _ftype.to_python(value)
+                except BaseException, e:
+                    raise TypeCastException(
+                        "Could not coerce %s, %s, "
+                        "to format %s: %s" % (key, value,
+                        _ftype.__class__.__name__,
+                        e))
+
+        return config
+
+    def prep_value(self, config):
+        """
+        the inverse of to_python method,
+        called just before serialization
+        """
+        for key, value in config.items():
+            _format = self.option_dict[key].get('format', None)
+            _ftype = self.types.get(_format, None)
+            if _ftype and hasattr(_ftype, 'get_prep_value'):
+                try:
+                    config[key] = _ftype.get_prep_value(value)
+                except BaseException, e:
+                    raise TypeCastException(
+                        "Could not serialize %s, %s, "
+                        "by format %s: %s" % (key, value,
+                        _ftype.__class__.__name__,
+                        e))
+            else:
+                config[key] = value
+        return config
+
+    # methods for adding configuration
+
+    def get_default_values(self):
+        """
+        return a config options from configuration defaults
+        """
+        defaults = {}
+        for key, value in self.items():
+            if 'default' in value:
+                defaults[key] = value['default']
+        return copy.deepcopy(defaults)
+
+    def get_adaptor(self, format):
+        """
+        get specified format adaptor or
+        guess for a given filename
+        """
+        adaptor = self.adaptors.get(format, None)
+        if adaptor:
+            return adaptor
+
+        # not registered in adaptors dict, let's try all
+        for adaptor in self.adaptors.values():
+            if format in adaptor.extensions:
+                return adaptor
+
+    def filename2format(self, filename):
+        extension = os.path.splitext(filename)[-1]
+        return extension.lstrip('.') or None
+
+    def serialize(self, filename, format=None, full=False):
+        if not format:
+            format = self._format
+        if not format:
+            format = self.filename2format(filename)
+        if not format:
+            raise Exception('Please specify a format')
+            # TODO: more specific exception type
+
+        adaptor = self.get_adaptor(format)
+        if not adaptor:
+            raise Exception("Adaptor not found for format: %s" % format)
+
+        config = copy.deepcopy(self.config)
+        serializable = self.prep_value(config)
+        adaptor.write(serializable, filename)
+
+    def deserialize(self, string=None, fromfile=None, format=None):
+        """
+        load configuration from a file or string
+        """
+
+        def _try_deserialize():
+            if fromfile:
+                with open(fromfile, 'r') as f:
+                    content = adaptor.read(f)
+            elif string:
+                content = adaptor.read(string)
+            return content
+
+        # XXX cleanup this!
+
+        if fromfile:
+            assert os.path.exists(fromfile)
+            if not format:
+                format = self.filename2format(fromfile)
+
+        if not format:
+            format = self._format
+        if format:
+            adaptor = self.get_adaptor(format)
+        else:
+            adaptor = None
+
+        if adaptor:
+            content = _try_deserialize()
+            return content
+
+        # no adaptor, let's try rest of adaptors
+
+        adaptors = self.adaptors[:]
+
+        if format:
+            adaptors.sort(
+                key=lambda x: int(
+                    format in x.extensions),
+                reverse=True)
+
+        for adaptor in adaptors:
+            content = _try_deserialize()
+        return content
+
+    def load(self, *args, **kwargs):
+        """
+        load from string or file
+        if no string of fromfile option is given,
+        it will attempt to load from defaults
+        defined in the schema.
+        """
+        string = args[0] if args else None
+        fromfile = kwargs.get("fromfile", None)
+        content = None
+
+        # start with defaults, so we can
+        # have partial values applied.
+        content = self.get_default_values()
+        if string and isinstance(string, str):
+            content = self.deserialize(string)
+
+        if not string and fromfile is not None:
+            #import ipdb;ipdb.set_trace()
+            content = self.deserialize(fromfile=fromfile)
+
+        if not content:
+            logger.error('no content could be loaded')
+            # XXX raise!
+            return
+
+        # lazy evaluation until first level of nesting
+        # to allow lambdas with context-dependant info
+        # like os.path.expanduser
+        for k, v in content.iteritems():
+            if callable(v):
+                content[k] = v()
+
+        self.validate(content)
+        self.config = content
+        return True
+
+
+def testmain():
+    from tests import test_validation as t
+    import pprint
+
+    config = PluggableConfig(_format="json")
+    properties = copy.deepcopy(t.sample_spec)
+
+    config.options = properties
+    config.load(fromfile='data.json')
+
+    print 'config'
+    pprint.pprint(config.config)
+
+    config.serialize('/tmp/testserial.json')
+
+if __name__ == "__main__":
+    testmain()
diff --git a/src/leap/base/specs.py b/src/leap/base/specs.py
index 641e795a..b4bb8dcf 100644
--- a/src/leap/base/specs.py
+++ b/src/leap/base/specs.py
@@ -44,11 +44,16 @@ leap_provider_spec = {
         'public_key': {
             'type': unicode  # fingerprint
         },
-        'ca_cert': {
-            'type': unicode
+        'ca_cert_fingerprint': {
+            'type': unicode,
         },
         'ca_cert_uri': {
-            'type': unicode
+            'type': unicode,
+            'format': 'https-uri'
+        },
+        'languages': {
+            'type': list,
+            'default': ['en']
         }
     }
 }
diff --git a/src/leap/base/tests/test_providers.py b/src/leap/base/tests/test_providers.py
index d667a7e0..8d3b8847 100644
--- a/src/leap/base/tests/test_providers.py
+++ b/src/leap/base/tests/test_providers.py
@@ -1,15 +1,13 @@
+import copy
 import json
 try:
     import unittest2 as unittest
 except ImportError:
     import unittest
-
-# XXX FIXME
-import logging
-logging.basicConfig()
-
 import os
 
+import jsonschema
+
 from leap import __branding as BRANDING
 from leap.testing.basetest import BaseLeapTest
 from leap.base import providers
@@ -25,6 +23,7 @@ EXPECTED_DEFAULT_CONFIG = {
     u"services": [
         u"eip"
     ],
+    u"languages": [u"en"],
     u"version": u"0.1.0"
 }
 
@@ -84,6 +83,13 @@ class TestLeapProviderDefinition(BaseLeapTest):
         self.assertDictEqual(self.config,
                              EXPECTED_DEFAULT_CONFIG)
 
+    def test_provider_validation(self):
+        self.definition.validate(self.config)
+        _config = copy.deepcopy(self.config)
+        _config['serial'] = 'aaa'
+        with self.assertRaises(jsonschema.ValidationError):
+            self.definition.validate(_config)
+
     @unittest.skip
     def test_load_malformed_json_definition(self):
         raise NotImplementedError
@@ -94,9 +100,6 @@ class TestLeapProviderDefinition(BaseLeapTest):
         # type cast
         raise NotImplementedError
 
-    def test_provider_validation(self):
-        self.definition.jsonvalidate(self.config)
-
 
 class TestLeapProviderSet(BaseLeapTest):
 
diff --git a/src/leap/base/tests/test_validation.py b/src/leap/base/tests/test_validation.py
new file mode 100644
index 00000000..87e99648
--- /dev/null
+++ b/src/leap/base/tests/test_validation.py
@@ -0,0 +1,92 @@
+import copy
+import datetime
+#import json
+try:
+    import unittest2 as unittest
+except ImportError:
+    import unittest
+import os
+
+import jsonschema
+
+from leap.base.config import JSONLeapConfig
+from leap.base import pluggableconfig
+from leap.testing.basetest import BaseLeapTest
+
+SAMPLE_CONFIG_DICT = {
+    'prop_one': 1,
+    'prop_uri': "http://example.org",
+    'prop_date': '2012-12-12',
+}
+
+EXPECTED_CONFIG = {
+    'prop_one': 1,
+    'prop_uri': "http://example.org",
+    'prop_date': datetime.datetime(2012, 12, 12)
+}
+
+sample_spec = {
+    'description': 'sample schema definition',
+    'type': 'object',
+    'properties': {
+        'prop_one': {
+            'type': int,
+            'default': 1,
+            'required': True
+        },
+        'prop_uri': {
+            'type': str,
+            'default': 'http://example.org',
+            'required': True,
+            'format': 'uri'
+        },
+        'prop_date': {
+            'type': str,
+            'default': '2012-12-12',
+            'format': 'date'
+        }
+    }
+}
+
+
+class SampleConfig(JSONLeapConfig):
+    spec = sample_spec
+
+    @property
+    def slug(self):
+        return os.path.expanduser('~/sampleconfig.json')
+
+
+class TestJSONLeapConfigValidation(BaseLeapTest):
+    def setUp(self):
+        self.sampleconfig = SampleConfig()
+        self.sampleconfig.save()
+        self.sampleconfig.load()
+        self.config = self.sampleconfig.config
+
+    def tearDown(self):
+        if hasattr(self, 'testfile') and os.path.isfile(self.testfile):
+            os.remove(self.testfile)
+
+    # tests
+
+    def test_good_validation(self):
+        self.sampleconfig.validate(SAMPLE_CONFIG_DICT)
+
+    def test_broken_int(self):
+        _config = copy.deepcopy(SAMPLE_CONFIG_DICT)
+        _config['prop_one'] = '1'
+        with self.assertRaises(jsonschema.ValidationError):
+            self.sampleconfig.validate(_config)
+
+    def test_format_property(self):
+        # JsonSchema Validator does not check the format property.
+        # We should have to extend the Configuration class
+        blah = copy.deepcopy(SAMPLE_CONFIG_DICT)
+        blah['prop_uri'] = 'xxx'
+        with self.assertRaises(pluggableconfig.TypeCastException):
+            self.sampleconfig.validate(blah)
+
+
+if __name__ == "__main__":
+    unittest.main()