add relevance to cp ninjs output (#239)

SDCP-714
superdesk · May 14, 2024 · 09fa463 · 09fa463
1 parent 0cbf813
commit 09fa463
Show file tree

Hide file tree

Showing 9 changed files with 89 additions and 14 deletions.
diff --git a/client/extensions/auto-tagger/src/adapter.ts b/client/extensions/auto-tagger/src/adapter.ts
@@ -12,6 +12,8 @@ export interface ITagBase {
     altids: {[key: string]: string};
     aliases?: Array<string>;
     original_source?: string;
+    creator?: string;
+    relevance?: number;
 }
 
 export interface ISubjectTag extends ITagBase {
@@ -44,11 +46,11 @@ export function toClientFormat(response: IServerResponse): OrderedMap<string, IT
     let tags = OrderedMap<string, ITagUi>();
 
     response.subject?.forEach((item) => {
-        const {name, description, qcode, source, altids, aliases, original_source, parent, scheme} = item;
+        const {name, description, qcode, source, altids, aliases, original_source, parent, scheme, relevance, creator} = item;
 
         // Checking if the item has original_source to filter auto tagger tags
         if (original_source != null) {
-            if(scheme === 'http://cv.iptc.org/newscodes/mediatopic/' || scheme === 'subject') {
+            if (scheme === 'http://cv.iptc.org/newscodes/mediatopic/' || scheme === 'subject') {
                 const tag: ITagUi = {
                     name,
                     description,
@@ -62,6 +64,8 @@ export function toClientFormat(response: IServerResponse): OrderedMap<string, IT
                         kind: 'scheme',
                         value: 'subject' || '',
                     },
+                    relevance,
+                    creator,
                 };
 
                 tags = tags.set(tag.qcode, tag);
@@ -79,6 +83,8 @@ export function toClientFormat(response: IServerResponse): OrderedMap<string, IT
                         kind: 'scheme',
                         value: item.scheme || '',
                     },
+                    relevance,
+                    creator,
                 };
 
                 tags = tags.set(tag.qcode, tag);
@@ -110,7 +116,7 @@ export function toClientFormat(response: IServerResponse): OrderedMap<string, IT
 
     others.forEach(({group, items}) => {
         items.forEach((item) => {
-            const {name, description, qcode, source, altids, aliases, original_source, scheme} = item;
+            const {name, description, qcode, source, altids, aliases, original_source, scheme, relevance, creator} = item;
 
             const tag: ITagUi = {
                 name,
@@ -125,6 +131,8 @@ export function toClientFormat(response: IServerResponse): OrderedMap<string, IT
                     kind: 'visual',
                     value: group,
                 },
+                relevance,
+                creator,
             };
 
             if (!tags.has(tag.name)) {
@@ -149,7 +157,7 @@ export function toServerFormat(items: OrderedMap<string, ITagUi>, superdesk: ISu
                 result.subject = [];
             }
 
-            const {name, description, qcode, source, altids, aliases, original_source, parent} = item;
+            const {name, description, qcode, source, altids, aliases, original_source, parent, relevance, creator} = item;
 
             const subjectTag: ISubjectTag = {
                 name,
@@ -161,6 +169,8 @@ export function toServerFormat(items: OrderedMap<string, ITagUi>, superdesk: ISu
                 scheme: item.group.value,
                 aliases,
                 original_source,
+                relevance,
+                creator,
             };
 
             result.subject.push(subjectTag);
@@ -171,7 +181,7 @@ export function toServerFormat(items: OrderedMap<string, ITagUi>, superdesk: ISu
                 result[groupValue] = [];
             }
 
-            const {name, description, qcode, source, altids, aliases, original_source, scheme} = item;
+            const {name, description, qcode, source, altids, aliases, original_source, scheme, relevance, creator} = item;
 
             const tagBase: ITagBase = {
                 name,
@@ -182,6 +192,8 @@ export function toServerFormat(items: OrderedMap<string, ITagUi>, superdesk: ISu
                 aliases,
                 original_source,
                 scheme,
+                relevance,
+                creator,
             };
 
             result[groupValue]!.push(tagBase);

diff --git a/client/extensions/auto-tagger/src/types.ts b/client/extensions/auto-tagger/src/types.ts
@@ -10,4 +10,6 @@ export interface ITagUi {
     /** The identifier of a controlled vocabulary which includes a code for the tag. */
     scheme?: string;
     group: {kind: 'scheme' | 'visual'; value: string};
+    creator?: string;
+    relevance?: number;
 }
diff --git a/server/cp/ai/semaphore.py b/server/cp/ai/semaphore.py
@@ -23,6 +23,12 @@
 TIMEOUT = (5, 30)
 
 
+def format_relevance(value: str) -> int:
+    if value:
+        return int(float(value) * 100)
+    return 100
+
+
 ResponseType = Mapping[str, Union[str, List[str]]]
 
 
@@ -121,7 +127,7 @@ def fetch_parent_info(self, qcode):
                             {
                                 "name": field.get("NAME"),
                                 "qcode": field.get("ID"),
-                                "relevance": score,
+                                "relevance": format_relevance(score),
                                 "parent": None,  # Set to None initially
                             }
                         )
@@ -195,7 +201,7 @@ def transform_xml_response(api_response):
                         "qcode": item["id"],
                         "source": "Semaphore",
                         "creator": "Human",
-                        "relevance": score,
+                        "relevance": int(score),
                         "altids": {"source_name": "source_id"},
                         "original_source": "original_source_value",
                         "scheme": scheme_url,
@@ -238,7 +244,7 @@ def transform_xml_response(api_response):
                                 ),
                                 "creator": "Human",
                                 "source": "Semaphore",
-                                "relevance": "100",
+                                "relevance": format_relevance("100"),
                                 "altids": {"source_name": "source_id"},
                                 "original_source": "original_source_value",
                                 "scheme": "http://cv.iptc.org/newscodes/mediatopic/",
@@ -553,7 +559,7 @@ def adjust_score(score, existing_scores):
                                     "qcode": meta_id if meta_id else "",
                                     "creator": "Machine",
                                     "source": "Semaphore",
-                                    "relevance": meta_score,
+                                    "relevance": format_relevance(meta_score),
                                     "altids": f'{{"{meta_value}": "{meta_id}"}}',
                                     "original_source": "original_source_value",
                                     "scheme": scheme_url,
@@ -574,7 +580,7 @@ def adjust_score(score, existing_scores):
                             "parent": parent_qcode,
                             "source": "Semaphore",
                             "creator": "Machine",
-                            "relevance": score,
+                            "relevance": format_relevance(score),
                             "altids": {"source_name": "source_id"},
                             "original_source": "original_source_value",
                             "scheme": "http://cv.iptc.org/newscodes/mediatopic/",

diff --git a/server/cp/output/formatter/cp_ninjs_formatter.py b/server/cp/output/formatter/cp_ninjs_formatter.py
@@ -13,6 +13,8 @@ def format_cv_item(item, language):
                 "code": item.get("qcode"),
                 "name": get_locale_name(item, language),
                 "scheme": "http://cv.iptc.org/newscodes/mediatopic/",
+                "creator": item.get("creator", ""),
+                "relevance": item.get("relevance", 100),
             }
         )
     else:
@@ -21,6 +23,8 @@ def format_cv_item(item, language):
                 "code": item.get("qcode"),
                 "name": get_locale_name(item, language),
                 "scheme": item.get("scheme"),
+                "creator": item.get("creator", ""),
+                "relevance": item.get("relevance", 100),
             }
         )
 

diff --git a/server/tests/conftest.py b/server/tests/conftest.py
@@ -0,0 +1,17 @@
+import flask
+import pytest
+
+
+@pytest.fixture(autouse=True)
+def app():
+    app = flask.Flask(__name__)
+    app.config.update(
+        {
+            "VERSION": "version",
+            "DEFAULT_LANGUAGE": "en",
+        }
+    )
+    ctx = app.app_context()
+    ctx.push()
+    yield app
+    ctx.pop()
diff --git a/server/tests/orangelogic_test.py b/server/tests/orangelogic_test.py
@@ -68,8 +68,7 @@ def setUp(self):
         self.ctx.push()
 
     def tearDown(self):
-        if hasattr(self, "ctx"):
-            self.ctx.pop()
+        self.ctx.pop()
 
     def test_instance(self):
         OrangelogicSearchProvider(self.provider)

diff --git a/server/tests/output/formatter/__init__.py b/server/tests/output/formatter/__init__.py
@@ -20,7 +20,11 @@ def setUp(self):
                 "DEFAULT_LANGUAGE": "en",
             }
         )
-        self.app.app_context().push()
+        self.ctx = self.app.app_context()
+        self.ctx.push()
+
+    def tearDown(self):
+        self.ctx.pop()
 
     def format(self, updates=None, _all=False):
         article = self.article.copy()

diff --git a/server/tests/output/formatter/cp_ninjs_formatter_test.py b/server/tests/output/formatter/cp_ninjs_formatter_test.py
@@ -0,0 +1,26 @@
+from unittest.mock import patch
+
+from cp.output.formatter.cp_ninjs_formatter import CPNINJSFormatter
+
+
+@patch("superdesk.get_resource_service")
+def test_subject_relevance(mock):
+    item = {
+        "type": "text",
+        "subject": [
+            {
+                "name": "Society",
+                "qcode": "eef4a135-e188-4d1f-93f1-cf7af1f594a6",
+                "source": "Semaphore",
+                "altids": {"source_name": "source_id"},
+                "scheme": "subject",
+                "original_source": "original_source_value",
+                "relevance": 54,
+                "creator": "Machine",
+            },
+        ],
+    }
+    formatter = CPNINJSFormatter()
+    ninjs = formatter._transform_to_ninjs(item, {})
+    assert ninjs["subject"][0]["relevance"] == 54
+    assert ninjs["subject"][0]["creator"] == "Machine"
diff --git a/server/tests/set_provice_on_publish_test.py b/server/tests/set_provice_on_publish_test.py
@@ -16,7 +16,12 @@ def setUp(self) -> None:
                 "DEFAULT_LANGUAGE": "en",
             }
         )
-        self.app.app_context().push()
+        self.ctx = self.app.app_context()
+        self.ctx.push()
+
+    def tearDown(self) -> None:
+        super().tearDown()
+        self.ctx.pop()
 
     def test_publish_signal(self):
         with patch.dict(superdesk.resources, resources):