CP Auto tagger

superdesk/publish/formatters/imatrics.py

@@ -1,6 +1,9 @@
 from superdesk.text_utils import get_text
 from .ninjs_formatter import NINJSFormatter
+import logging
 
+logging.basicConfig(level=logging.DEBUG) 
+logger = logging.getLogger(__name__)
 
 def format_datetime(date):
     return date.isoformat()
@@ -11,6 +14,10 @@ def can_format(self, format_type, article):
         return format_type.lower() == "imatrics" and article.get("type") == "text"
 
     def _transform_to_ninjs(self, article, subscriber, recursive=True):
+
+        logger.warning('In Formatter IMatrics. lets log.')
+        logging.debug("Transforming article to NINJS: %s", article)
+
         return {
             "uuid": article["guid"],
             "createdTimestamp": format_datetime(article["firstcreated"]),

superdesk/publish/formatters/ninjs_formatter.py

@@ -75,9 +75,16 @@ def get_locale_name(item, language):
 
 def format_cv_item(item, language):
     """Format item from controlled vocabulary for output."""
-    return filter_empty_vals(
-        {"code": item.get("qcode"), "name": get_locale_name(item, language), "scheme": item.get("scheme")}
+    if item.get("scheme") == "subject":
+
+        return filter_empty_vals(
+        {"code": item.get("qcode"), "name": get_locale_name(item, language), "scheme": "http://cv.iptc.org/newscodes/mediatopic/"}
     )
+    else:
+
+        return filter_empty_vals(
+            {"code": item.get("qcode"), "name": get_locale_name(item, language), "scheme": item.get("scheme")}
+        )
 
 
 class NINJSFormatter(Formatter):
@@ -213,8 +220,18 @@ def _transform_to_ninjs(self, article, subscriber, recursive=True):
         else:
             ninjs["priority"] = 5
 
-        if article.get("subject"):
-            ninjs["subject"] = self._get_subject(article)
+        # Merging Various Entities into Subjects for ninjs Response
+        # ---------------------------------------------------------
+        # This section of the code is responsible for aggregating different entity types 
+        # like 'organisation', 'place', 'event', and 'person' along with 'subject' into 
+        # a single list.
+
+
+        if article.get("subject") or article.get("organisation") or article.get("place") or article.get("event") or article.get("person"):
+            combined_subjects = (self._get_subject(article) + self._get_organisation(article) + 
+                                self._get_place(article) + self._get_event(article) + 
+                                self._get_person(article))
+            ninjs["subject"] = combined_subjects
 
         if article.get("anpa_category"):
             ninjs["service"] = self._get_service(article)
@@ -414,10 +431,32 @@ def _get_genre(self, article):
         lang = article.get("language", "")
         return [format_cv_item(item, lang) for item in article["genre"]]
 
+
+
+
     def _get_subject(self, article):
         """Get subject list for article."""
         return [format_cv_item(item, article.get("language", "")) for item in article.get("subject", [])]
 
+    #  Updated Code here to fetch Organisations from Article
+    def _get_organisation(self, article):
+        return [format_cv_item(item, article.get("language", "")) for item in article.get("organisation", [])]
+
+    #  Updated Code here to fetch Places from Article
+    def _get_place(self, article):
+        """Get place list for article."""
+        return [format_cv_item(item, article.get("language", "")) for item in article.get("place", [])]
+
+    #  Updated Code here to fetch Events from Article
+    def _get_event(self, article):
+        """Get event list for article."""
+        return [format_cv_item(item, article.get("language", "")) for item in article.get("event", [])]
+
+    #  Updated Code here to fetch Person from Article
+    def _get_person(self, article):
+        """Get person list for article."""
+        return [format_cv_item(item, article.get("language", "")) for item in article.get("person", [])]
+
     def _get_service(self, article):
         """Get service list for article.
 

superdesk/publish/formatters/semaphore.py

@@ -0,0 +1,26 @@
+import logging
+from superdesk.text_utils import get_text
+from .ninjs_formatter import NINJSFormatter
+from superdesk.text_checkers.ai.semaphore import Semaphore  # Import the Semaphore integration class
+
+logger = logging.getLogger(__name__)
+
+class SemaphoreFormatter(NINJSFormatter):
+    def can_format(self, format_type, article):
+        return format_type.lower() == "semaphore" and article.get("type") == "text"
+
+    def _transform_to_ninjs(self, article, subscriber, recursive=True):
+        semaphore = Semaphore()  # Initialize the Semaphore integration
+        formatted_data = {}  # Define how you want to format the data for Semaphore
+
+        try:
+            # Example: format the data
+            formatted_data["uuid"] = article["guid"]
+            formatted_data["headline"] = get_text(article["headline"])
+            # Add more formatting logic here
+
+        except Exception as e:
+            logger.error(f"Error formatting data for Semaphore: {str(e)}")
+            formatted_data = {}  # Return an empty dictionary in case of an error
+
+        return formatted_data

superdesk/publish/transmitters/imatrics.py

@@ -3,12 +3,19 @@
 from superdesk.publish import register_transmitter
 from superdesk.publish.publish_service import PublishService
 from superdesk.text_checkers.ai.imatrics import IMatrics
+import logging
 
+logging.basicConfig(level=logging.DEBUG) 
+logger = logging.getLogger(__name__)
 
 class IMatricsTransmitter(PublishService):
     def _transmit(self, queue_item, subscriber):
         imatrics = IMatrics(current_app)
         item = json.loads(queue_item["formatted_item"])
+
+        logger.warning("Logging in Transmitter for IMAtrics")
+        logging.debug("Transmitting item: %s", item)
+
         imatrics.publish(item)
 
 

superdesk/publish/transmitters/semaphore.py

@@ -0,0 +1,15 @@
+from flask import current_app, json
+
+from superdesk.publish import register_transmitter
+from superdesk.publish.publish_service import PublishService
+from superdesk.text_checkers.ai.semaphore import Semaphore  # Import the Semaphore integration class
+
+class SemaphoreTransmitter(PublishService):
+    def _transmit(self, queue_item, subscriber):
+        semaphore = Semaphore(current_app)  # Initialize the Semaphore integration
+        item = json.loads(queue_item["formatted_item"])
+        # Modify this part to transmit the item using the Semaphore integration
+        semaphore.transmit(item)
+
+# Register the Semaphore transmitter
+register_transmitter("semaphore", SemaphoreTransmitter(), [])

superdesk/text_checkers/ai/__init__.py

@@ -33,11 +33,13 @@ class AIResource(Resource):
             "type": "dict",
             "required": True,
             "schema": {
-                "guid": {"type": "string", "required": True},
+                "guid": {"type": "string", "required": False},
                 "abstract": {"type": "string", "required": False},
-                "language": {"type": "string", "required": True},
-                "headline": {"type": "string", "nullable": True},
-                "body_html": {"type": "string", "required": True},
+                "language": {"type": "string", "required": False},
+                "headline": {"type": "string", "nullable": False},
+                "slugline": {"type": "string", "required": False},
+                "searchString": {"type": "string", "required": False},
+                "body_html": {"type": "string", "required": False},
             },
         },
         "tags": {
@@ -88,7 +90,8 @@ def create(self, docs, **kwargs):
         except KeyError:
             raise SuperdeskApiError.notFoundError("{service} service can't be found".format(service=service))
 
-        analyzed_data = service.analyze(item, doc.get("tags"))
+        # analyzed_data = service.analyze(item, doc.get("tags"))
+        analyzed_data = service.analyze(item)
         docs[0].update({"analysis": analyzed_data})
         return [0]
 

superdesk/text_checkers/ai/imatrics.py

@@ -226,9 +226,10 @@ def search_images(self, items: list) -> list:
         data = items
         try:
             r_data = self._search_images(data)
-        except Exception:
+        except Exception as err:
+            logger.exception(err)
             return []
-        return [image for image in r_data if type(image["imageUrl"]) == str and image["imageUrl"] != ""]
+        return [image for image in r_data if isinstance(image["imageUrl"], str) and image["imageUrl"] != ""]
 
     def _search_images(self, data, **params):
         return self._request_images(