Merge pull request #65 from IndicoDataSolutions/development

Development

CHANGES.txt

@@ -24,3 +24,4 @@ v0.5.2, Tue March 7 -- Required API keys, configuration settings
 v0.5.3, Wed Apr 15 -- Added scipy to requirements, edited Readme to not break pypi page
 v0.6.0, Thu May 29 -- Remove numpy / scipy dependency in favor of Pillow
 v0.7.0, Tue Jun 9 -- Added support for calling multiple APIs in a single function and accepting filenames as image API inputs
+v0.7.1 Thu Jun 11 -- High quality sentiment API for private beta, fix for multi API support

indicoio/__init__.py

@@ -1,6 +1,6 @@
 from functools import partial
 
-Version, version, __version__, VERSION = ('0.7.0',) * 4
+Version, version, __version__, VERSION = ('0.7.1',) * 4
 
 JSON_HEADERS = {
     'Content-type': 'application/json',
@@ -9,7 +9,7 @@ JSON_HEADERS = {
     'version-number': VERSION
 }
 
-from indicoio.text.sentiment import political, posneg
+from indicoio.text.sentiment import political, posneg, sentiment_hq
 from indicoio.text.sentiment import posneg as sentiment
 from indicoio.text.lang import language
 from indicoio.text.tagging import text_tags

indicoio/config.py

@@ -49,7 +49,8 @@ TEXT_APIS = [
     'text_tags',
     'political',
     'sentiment',
-    'language'
+    'language',
+    'sentiment_hq'
 ]
 
 IMAGE_APIS = [

indicoio/text/sentiment.py

@@ -50,3 +50,26 @@ def posneg(text, cloud=None, batch=False, api_key=None, **kwargs):
     """
 
     return api_handler(text, cloud=cloud, api="sentiment", url_params={"batch":batch, "api_key":api_key}, **kwargs)
+
+def sentiment_hq(text, cloud=None, batch=False, api_key=None, **kwargs):
+    """
+    Given input text, returns a scalar estimate of the sentiment of that text.
+    Values are roughly in the range 0 to 1 with 0.5 indicating neutral sentiment.
+    For reference, 0 suggests very negative sentiment and 1 suggests very positive sentiment.
+
+    Example usage:
+
+    .. code-block:: python
+
+       >>> from indicoio import sentimenthq
+       >>> text = 'Thanks everyone for the birthday wishes!! It was a crazy few days ><'
+       >>> sentiment = sentimenthq(text)
+       >>> sentiment
+       0.6210052967071533
+
+    :param text: The text to be analyzed.
+    :type text: str or unicode
+    :rtype: Float
+    """
+
+    return api_handler(text, cloud=cloud, api="sentimenthq", url_params={"batch":batch, "api_key":api_key}, **kwargs)

indicoio/utils/api.py

@@ -17,6 +17,7 @@ def api_handler(arg, cloud, api, url_params = {"batch":False, "api_key":None}, *
 
     if cloud:
         host = "%s.indico.domains" % cloud
+
     else:
         # default to indico public cloud
         host = config.PUBLIC_API_HOST
@@ -24,8 +25,14 @@ def api_handler(arg, cloud, api, url_params = {"batch":False, "api_key":None}, *
     url = config.url_protocol + "//%s/%s" % (host, api)
     url = url + "/batch" if url_params.get("batch", False) else url
     url += "?key=%s" % (url_params.get("api_key", None) or config.api_key)
-    if "apis" in url_params:
+    apis = url_params.get("apis", [])
-        url += "&apis=%s" % ",".join(url_params["apis"])
+    if apis:
+        url += "&apis=%s" % ",".join(apis)
+
+    # private beta
+    if host == config.PUBLIC_API_HOST:
+        if (api == 'sentimenthq') or ('sentimenthq' in apis):
+            raise IndicoError("The high quality sentiment API is currently in private beta.")
 
     response = requests.post(url, data=json_data, headers=JSON_HEADERS)
     if response.status_code == 503 and cloud != None:

indicoio/utils/multi.py

@@ -6,8 +6,12 @@ from indicoio.utils.errors import IndicoError
 
 CLIENT_SERVER_MAP = dict((api, api.strip().replace("_", "").lower()) for api in API_NAMES)
 SERVER_CLIENT_MAP = dict((v, k) for k, v in CLIENT_SERVER_MAP.iteritems())
+AVAILABLE_APIS = {
+    'text': TEXT_APIS,
+    'image': IMAGE_APIS
+}
 
-def multi(data, type, apis, available, batch=False, **kwargs):
+def multi(data, datatype, apis, batch=False, **kwargs):
     """
     Helper to make multi requests of different types.
 
@@ -22,23 +26,39 @@ def multi(data, type, apis, available, batch=False, **kwargs):
     :rtype: Dictionary of api responses
     """
     # Client side api name checking - strictly only accept func name api
+    available = AVAILABLE_APIS.get(datatype)
     invalid_apis = [api for api in apis if api not in available]
     if invalid_apis:
-        raise IndicoError("%s are not valid %s APIs. Please reference the available APIs below:\n%s"
+        raise IndicoError(
-                            % (", ".join(invalid_apis), type, ", ".join(available))
+            "%s are not valid %s APIs. Please reference the available APIs below:\n%s"
-                        )
+            % (", ".join(invalid_apis), datatype, ", ".join(available))
+        )
+
     # Convert client api names to server names before sending request
     apis = map(CLIENT_SERVER_MAP.get, apis)
-    result = api_handler(data, url_params = {"apis":apis, "batch":batch}, **kwargs)
+    cloud = kwargs.pop("cloud", None)
+    api_key = kwargs.pop('api_key', None)
+    result = api_handler(
+        data,
+        cloud=cloud,
+        api='apis',
+        url_params={
+            "apis":apis,
+            "batch":batch,
+            "api_key":api_key
+        },
+        **kwargs
+    )
     return handle_response(result)
 
+
 def handle_response(result):
     # Parse out the results to a dicionary of api: result
     return dict((SERVER_CLIENT_MAP[api], parsed_response(api, res))
         for api, res in result.iteritems())
 
 
-def predict_text(input_text, apis=TEXT_APIS, cloud=None, batch=False, api_key=None, **kwargs):
+def predict_text(input_text, apis=TEXT_APIS, **kwargs):
     """
     Given input text, returns the results of specified text apis. Possible apis
     include: [ 'text_tags', 'political', 'sentiment', 'language' ]
@@ -60,19 +80,22 @@ def predict_text(input_text, apis=TEXT_APIS, cloud=None, batch=False, api_key=No
     :rtype: Dictionary of api responses
     """
 
+    cloud = kwargs.pop('cloud', None)
+    batch = kwargs.pop('batch', False)
+    api_key = kwargs.pop('api_key', None)
+
     return multi(
-        api="apis",
         data=input_text,
-        type="text",
+        datatype="text",
-        available = TEXT_APIS,
         cloud=cloud,
         batch=batch,
         api_key=api_key,
         apis=apis,
-        **kwargs)
+        **kwargs
+    )
 
 
-def predict_image(image, apis=IMAGE_APIS, cloud=None, batch=False, api_key=None, **kwargs):
+def predict_image(image, apis=IMAGE_APIS, **kwargs):
     """
     Given input image, returns the results of specified image apis. Possible apis
     include: ['fer', 'facial_features', 'image_features']
@@ -95,16 +118,19 @@ def predict_image(image, apis=IMAGE_APIS, cloud=None, batch=False, api_key=None,
     :rtype: Dictionary of api responses
     """
 
+    cloud = kwargs.pop('cloud', None)
+    batch = kwargs.pop('batch', False)
+    api_key = kwargs.pop('api_key', None)
+
     return multi(
-        api="apis",
         data=image_preprocess(image, batch=batch),
-        type="image",
+        datatype="image",
-        available=IMAGE_APIS,
         cloud=cloud,
         batch=batch,
         api_key=api_key,
         apis=apis,
-        **kwargs)
+        **kwargs
+    )
 
 def parsed_response(api, response):
     result = response.get('results', False)

tests/test_remote.py

@@ -9,9 +9,13 @@ from indicoio import config
 from indicoio import political, sentiment, fer, facial_features, language, image_features, text_tags
 from indicoio import batch_political, batch_sentiment, batch_fer, batch_facial_features
 from indicoio import batch_language, batch_image_features, batch_text_tags
+from indicoio import sentiment_hq, batch_sentiment_hq
 from indicoio import predict_image, predict_text, batch_predict_image, batch_predict_text
 from indicoio.utils.errors import IndicoError
 
+# TODO: remove once sentiment_hq is added to the public API
+config.TEXT_APIS.remove("sentiment_hq")
+
 DIR = os.path.dirname(os.path.realpath(__file__))
 
 class BatchAPIRun(unittest.TestCase):
@@ -37,6 +41,13 @@ class BatchAPIRun(unittest.TestCase):
         self.assertTrue(isinstance(response, list))
         self.assertTrue(response[0] < 0.5)
 
+    # TODO: uncomment once the high quality sentiment API is publicly released
+    # def test_batch_sentiment_hq(self):
+    #     test_data = ['Worst song ever', 'Best song ever']
+    #     response = batch_sentiment_hq(test_data, api_key=self.api_key)
+    #     self.assertTrue(isinstance(response, list))
+    #     self.assertTrue(response[0] < 0.5)
+
     def test_batch_political(self):
         test_data = ["Guns don't kill people, people kill people."]
         response = batch_political(test_data, api_key=self.api_key)
@@ -220,6 +231,19 @@ class FullAPIRun(unittest.TestCase):
         self.assertTrue(isinstance(response, float))
         self.assertTrue(response > 0.5)
 
+    # TODO: uncomment when the high quality sentiment API is publicly released
+    # def test_sentiment_hq(self):
+    #     test_string = "Worst song ever."
+    #     response = sentiment_hq(test_string)
+
+    #     self.assertTrue(isinstance(response, float))
+    #     self.assertTrue(response < 0.5)
+
+    #     test_string = "Best song ever."
+    #     response = sentiment_hq(test_string)
+    #     self.assertTrue(isinstance(response, float))
+    #     self.assertTrue(response > 0.5)
+
     def test_good_fer(self):
         fer_set = set(['Angry', 'Sad', 'Neutral', 'Surprise', 'Fear', 'Happy'])
         test_face = generate_array((48,48))