Merge pull request #18 from Clarifai/revert-17-fixlang

Revert "refer to the server for languages"

clarifai/client/__init__.py

@@ -1,3 +1,3 @@
 from __future__ import absolute_import
 
-from .client import ClarifaiApi, ApiError, ApiThrottledError, ApiBadRequestError, ApiClientError
+from .client import ClarifaiApi, ApiError, ApiThrottledError, ApiBadRequestError

clarifai/client/client.py

@@ -27,6 +27,7 @@ def iteritems(d):
   def iteritems(d):
     return d.iteritems()
 
+#logging.basicConfig(level=logging.DEBUG)
 logger = logging.getLogger(__name__)
 
 
@@ -42,11 +43,11 @@ def __str__(self):
   def __repr__(self):
     return "Error: '%s'" % str(self.msg)
 
+
 class ApiClientError(ApiError):
   """Thrown when client side validation fails"""
   pass
 
-
 class ApiThrottledError(Exception):
   """The usage limit throttle was hit.  Client should wait for wait_seconds before retrying."""
 
@@ -76,14 +77,37 @@ class ClarifaiApi(object):
     app_id: the client_id for an application you've created in your Clarifai account.
     app_secret: the client_secret for the same application.
     base_url: Base URL of the API endpoints.
-    model: Name of the recognition model to query. Defaults to None so that server side defaults
-  in your app settings are used.
+    model: Name of the recognition model to query. Defaults to None so that server side defaults 
+  in your app settings are used. 
     wait_on_throttle: When the API returns a 429 throttled error, sleep for the amount of time
         reported in the X-Throttle-Wait-Seconds HTTP response header.
     language: set the default language using it's two letter (with options -XX variant) ISO 639-1
   code to use for all requests. Defaults to None so that server side defaults in your app settings
   are used.
   """
+  _SUPPORTED_LANGUAGES = {
+    'ar': 'Arabic',
+    'bn': 'Bengali',
+    'da': 'Danish',
+    'de': 'German',
+    'en': 'English',
+    'es': 'Spanish',
+    'fi': 'Finnish',
+    'fr': 'French',
+    'hi': 'Hindi',
+    'it': 'Italian',
+    'ja': 'Japanese',
+    'nl': 'Dutch',
+    'no': 'Norwegian',
+    'pa': 'Punjabi',
+    'pl': 'Polish',
+    'pt': 'Portuguese',
+    'ru': 'Russian',
+    'sv': 'Swedish',
+    'tr': 'Turkish',
+    'zh': 'Chinese (Simplified)',
+    'zh-TW': 'Chinese (Traditional)'
+  }
 
   def __init__(self, app_id=None, app_secret=None, base_url='https://api.clarifai.com',
                model=None, wait_on_throttle=True, language=None):
@@ -106,8 +130,7 @@ def __init__(self, app_id=None, app_secret=None, base_url='https://api.clarifai.
       'multiop': "/".join([self._base_url, '%s/multiop/' % API_VERSION]),
       'feedback': "/".join([self._base_url, '%s/feedback/' % API_VERSION]),
       'token': "/".join([self._base_url, '%s/token/' % API_VERSION]),
-      'info': "/".join([self._base_url, '%s/info/' % API_VERSION]),
-      'languages': "/".join([self._base_url, '%s/info/languages' % API_VERSION])
+      'info': "/".join([self._base_url, '%s/info/' % API_VERSION])
       }
     self.access_token = None
     self.api_info = None
@@ -121,15 +144,25 @@ def language(self):
 
   @language.setter
   def language(self, lang_code):
-    self._language = self._sanitize_param(lang_code, default=None)
+    self._language = self._parse_language(lang_code)
 
-  def get_languages(self):
-    response = self._parse_response(self._get_raw_response(
-      self._get_json_headers,
-      self._get_json_response,
-      self._url_for_op('languages'),
-      {}))
-    return response['languages']
+  def _parse_language(self, lang_code):
+    """
+    Checks to see if the language code is supported and sanitizes it.
+
+    Args:
+      lang_code: language code
+
+    Returns:
+      lang_code: validated and sanitized language code
+
+    Raises:
+      ApiClientError: if the language code that was provided is not supported
+    """
+    if lang_code is not None and lang_code not in self._SUPPORTED_LANGUAGES:
+      raise ApiClientError('Invalid language code {code}. Should be one of {supported}'
+                           .format(code=lang_code, supported=self._SUPPORTED_LANGUAGES.items()))
+    return self._sanitize_param(lang_code, default=None)
 
   def get_access_token(self, renew=False):
     """ Get an access token using your app_id and app_secret.
@@ -176,6 +209,9 @@ def get_info(self):
     self.api_info = response['results']
     return self.api_info
 
+  def get_languages(self):
+    return self._SUPPORTED_LANGUAGES
+
   def _url_for_op(self, ops):
     if not isinstance(ops, list):
       ops = [ops]
@@ -202,7 +238,7 @@ def tag(self, files, model=None, local_ids=None, meta=None, select_classes=None,
     back in order).
       meta: a string of any extra information to accompany the request. This has to be a string, so
     if passing structured data, pass a json.dumps(meta) string.
-      select_classes: to select only a subset of all possible classes, enter a comma separated list
+      select_classes: to select only a subset of all possible classes, enter a comma separated list 
     of classes you want to predict. Ex: "dog,cat,tree,car,boat"
       language: set the default language using it's two letter (with options -XX variant) ISO 639-1
     code to use for all requests.
@@ -217,7 +253,7 @@ def tag(self, files, model=None, local_ids=None, meta=None, select_classes=None,
       clarifai_api.tag([open('/path/to/local/image.jpeg'),
                         open('/path/to/local/image2.jpeg')])
     """
-    return self._multi_data_op(files, ['tag'], model=model, local_ids=local_ids, meta=meta,
+    return self._multi_data_op(files, ['tag'], model=model, local_ids=local_ids, meta=meta, 
                                select_classes=select_classes,
                                language=language)
 
@@ -273,7 +309,7 @@ def tag_and_embed(self, files, model=None, local_ids=None, meta=None, select_cla
     back in order).
       meta: a string of any extra information to accompany the request. This has to be a string, so
     if passing structured data, pass a json.dumps(meta) string.
-      select_classes: to select only a subset of all possible classes, enter a comma separated list
+      select_classes: to select only a subset of all possible classes, enter a comma separated list 
     of classes you want to predict. Ex: "dog,cat,tree,car,boat"
       language: set the default language using it's two letter (with options -XX variant) ISO 639-1
     code to use for all requests.
@@ -307,7 +343,7 @@ def tag_urls(self, urls, model=None, local_ids=None, meta=None, select_classes=N
     back in order).
       meta: a string of any extra information to accompany the request. This has to be a string, so
     if passing structured data, pass a json.dumps(meta) string.
-      select_classes: to select only a subset of all possible classes, enter a comma separated list
+      select_classes: to select only a subset of all possible classes, enter a comma separated list 
     of classes you want to predict. Ex: "dog,cat,tree,car,boat"
       language: set the default language using it's two letter (with options -XX variant) ISO 639-1
     code to use for all requests.
@@ -370,7 +406,7 @@ def tag_and_embed_urls(self, urls, model=None, local_ids=None, meta=None, select
     back in order).
       meta: a string of any extra information to accompany the request. This has to be a string, so
     if passing structured data, pass a json.dumps(meta) string.
-      select_classes: to select only a subset of all possible classes, enter a comma separated list
+      select_classes: to select only a subset of all possible classes, enter a comma separated list 
     of classes you want to predict. Ex: "dog,cat,tree,car,boat"
       language: set the default language using it's two letter (with options -XX variant) ISO 639-1
     code to use for all requests.
@@ -560,7 +596,7 @@ def _setup_multi_data(self, ops, num_cases, model=None, local_ids=None, meta=Non
     elif self._model:  # use the variable passed into __init__
       data['model'] = self._model
     if language:  # use the variable passed into method
-      data['language'] = self._sanitize_param(language, default=None)
+      data['language'] = self._parse_language(language)
     elif self.language:  # use the variable passed into __init__
       data['language'] = self.language
     if local_ids:

tests/test_client.py

@@ -8,24 +8,19 @@
 import os
 import hashlib
 import unittest
-from clarifai.client import ClarifaiApi, ApiClientError
+from clarifai.client import ClarifaiApi
 
 class TestClarifaiApi(unittest.TestCase):
   """
   test the Clarifai API Python client with all supported features
   """
-  image_url = 'http://clarifai-img.s3.amazonaws.com/test/toddler-flowers.jpeg'
-  video_url = 'http://techslides.com/demos/sample-videos/small.mp4'
-
-  def get_client(self, *args, **kwargs):
-    return ClarifaiApi()
 
   def test_api_connection(self):
-    api = self.get_client()
+    api = ClarifaiApi()
     self.assertTrue(api)
 
   def test_get_info(self):
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.get_info()
     self.assertTrue(response.get('api_version'))
     self.assertTrue(len(response) > 0)
@@ -34,20 +29,14 @@ def test_tag_one_image(self):
     """ tag one image, from url and disk """
     # tag image from online URL
     image_url = 'http://clarifai-img.s3.amazonaws.com/test/toddler-flowers.jpeg'
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.tag_image_urls(image_url)
     self.assertTrue(response)
-    result = response['results'][0]
-    self.assertTrue(result['url'] == image_url)
-
-    tag = result['result']['tag']
-    self.assertTrue('probs' in tag, 'probs not included in response')
-    self.assertTrue('classes' in tag, 'classes not included in response')
-    self.assertTrue(len(tag['probs']) == len(tag['classes']), 'lengths of tag fields did not match')
+    self.assertTrue(response['results'][0]['url'] == image_url)
 
     # tag image from local fs
     image_file = 'tests/data/toddler-flowers.jpeg'
-    api = self.get_client()
+    api = ClarifaiApi()
     if os.path.exists(image_file):
       with open(image_file, 'rb') as fb:
         response = api.tag_images(fb)
@@ -60,15 +49,15 @@ def test_tag_images(self):
     image_files = ['metro-north.jpg', 'octopus.jpg', 'tahoe.jpg', 'thai-market.jpg']
     image_urls = [os.path.join(image_url_base, one_file) for one_file in image_files]
 
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.tag_image_urls(image_urls)
     self.assertTrue(response)
 
     # tag images frmo local fs
     image_dir = 'tests/data'
     image_files = ['metro-north.jpg', 'octopus.jpg', 'tahoe.jpg', 'thai-market.jpg']
 
-    api = self.get_client()
+    api = ClarifaiApi()
     if os.path.exists(image_dir):
       image_files = [open(os.path.join(image_dir, one_file), 'rb') for one_file in image_files]
       response = api.tag_images(image_files)
@@ -79,7 +68,7 @@ def test_tag_images(self):
   def test_unicode_urls(self):
     image_url = u'http://www.alvaronoboa.com/wp-content/uploads/2013/02/Álvaro-Noboa-y-Annabella-Azín-Votaciones-41-1024x682.jpg'
 
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.tag_image_urls(image_url)
     self.assertTrue(response)
     self.assertTrue(response['results'][0]['url'] == image_url)
@@ -89,13 +78,13 @@ def test_tag_gif(self):
     # source: http://media.giphy.com/media/fRZn2vraBGiA0/giphy.gif
     image_url = 'http://media.giphy.com/media/fRZn2vraBGiA0/giphy.gif'
 
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.tag_image_urls(image_url)
     self.assertTrue(response)
     self.assertTrue(response['results'][0]['url'] == image_url)
 
     image_file = 'tests/data/water-ocean-turtle.gif'
-    api = self.get_client()
+    api = ClarifaiApi()
     if os.path.exists(image_file):
       with open(image_file, 'rb') as fb:
         response = api.tag_images(fb)
@@ -105,30 +94,30 @@ def test_tag_one_video(self):
     # video source: http://techslides.com/demos/sample-videos/small.mp4
     video_url = 'http://techslides.com/demos/sample-videos/small.mp4'
 
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.tag_image_urls(video_url)
     self.assertTrue(response)
     self.assertTrue(response['results'][0]['url'] == video_url)
 
   def test_tag_one_video_from_localfs(self):
     # video source: http://techslides.com/demos/sample-videos/small.mp4
     video_file = 'tests/data/small.mp4'
-    api = self.get_client()
+    api = ClarifaiApi()
     if os.path.exists(video_file):
       with open(video_file, 'rb') as fb:
         response = api.tag_images(fb)
         self.assertTrue(response)
 
   def test_embed_one_image(self):
     image_url = 'http://clarifai-img.s3.amazonaws.com/test/toddler-flowers.jpeg'
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.embed_image_urls(image_url)
     self.assertTrue(response)
     self.assertTrue(response['results'][0]['url'] == image_url)
 
   def test_embed_one_image_from_localfs(self):
     image_file = 'tests/data/toddler-flowers.jpeg'
-    api = self.get_client()
+    api = ClarifaiApi()
     if os.path.exists(image_file):
       with open(image_file, 'rb') as fb:
         response = api.embed_images(fb)
@@ -139,15 +128,15 @@ def test_tag_n_embed_one_image(self):
     image_files = ['metro-north.jpg', 'octopus.jpg', 'tahoe.jpg', 'thai-market.jpg']
     image_urls = [os.path.join(image_url_base, one_file) for one_file in image_files]
 
-    api = self.get_client()
+    api = ClarifaiApi()
     response = api.tag_and_embed_image_urls(image_urls)
     self.assertTrue(response)
 
   def test_tag_n_embed_from_localfs(self):
     image_dir = 'tests/data'
     image_files = ['metro-north.jpg', 'octopus.jpg', 'tahoe.jpg', 'thai-market.jpg']
 
-    api = self.get_client()
+    api = ClarifaiApi()
     if os.path.exists(image_dir):
       image_files = [open(os.path.join(image_dir, one_file), 'rb') for one_file in image_files]
       response = api.tag_and_embed_images(image_files)
@@ -162,7 +151,7 @@ def test_send_feedback(self):
             'http://clarifai-img.s3.amazonaws.com/test/metro-north.jpg', \
             'http://clarifai-img.s3.amazonaws.com/test/octopus.jpg']
 
-    api = self.get_client()
+    api = ClarifaiApi()
 
     response = api.feedback(urls=urls[0], add_tags='train')
     self.assertTrue(response)
@@ -184,42 +173,6 @@ def test_send_feedback(self):
     response = api.feedback(urls=urls, similar_docids=docids[:2], dissimilar_docids=docids[1:])
     self.assertTrue(response)
 
-  def test_i18n(self):
-    with self.assertRaises(ApiClientError):
-      api = ClarifaiApi(language='aaa')
-
-    api = self.get_client()
-    request_body = api._setup_multi_data([], ['urls'], language='en')
-    self.assertEqual(request_body['language'], 'en', 'language field was not set')
-
-    languages = api.get_languages()
-    self.assertTrue(len(languages), 'did not return any languages')
-    self.assertTrue('en' in languages, 'english code not included in languages')
-
-  def test_concept_ids(self):
-    """new models should return concept_ids"""
-    api = self.get_client()
-
-    api.set_model('general-v1.3')
-    response = api.tag_image_urls(self.image_url)
-    tag = response['results'][0]['result']['tag']
-    self.assertTrue('concept_ids' in tag, 'concept_ids not included in new model')
-    self.assertTrue(tag['concept_ids'][0].startswith('ai_'), 'concept id doesn\'t start with ai_')
-
-    response = api.tag_image_urls(self.video_url)
-    self.assertTrue(response['results'][0]['result']['tag']['concept_ids'][0][0].startswith('ai_'),
-                    "video concept_ids didn't start wit ai_")
-
-
-    api.set_model('general-v1.1')
-    response = api.tag_image_urls(self.image_url)
-    tag = response['results'][0]['result']['tag']
-    self.assertTrue('concept_ids' not in tag, 'concept_ids included in old model')
-
-    response = api.tag_image_urls(self.video_url)
-    tag = response['results'][0]['result']['tag']
-    self.assertTrue('concept_ids' not in tag, 'concept_ids included in old model')
-
-
 if __name__ == '__main__':
   unittest.main()
+