Gender detaction for SSML

inbox.py

@@ -14,6 +14,7 @@ import html
 import urllib.parse
 from linked_data_sig import verifyJsonSignature
 from utils import getDisplayName
+from utils import getGenderFromBio
 from utils import removeHtml
 from utils import getConfigParam
 from utils import hasUsersPath
@@ -2194,6 +2195,7 @@ def _updateSpeaker(baseDir: str, nickname: str, domain: str,
 
     speakerName = \
         getDisplayName(baseDir, postJsonObject['actor'], personCache)
+    gender = getGenderFromBio(baseDir, postJsonObject['actor'], personCache)
     if not speakerName:
         return
     if announcingActor:
@@ -2204,7 +2206,7 @@ def _updateSpeaker(baseDir: str, nickname: str, domain: str,
             translate['announces'] + ' ' + announcedHandle + '. ' + content
     speakerJson = speakerEndpointJson(speakerName, summary,
                                       content, imageDescription,
-                                      detectedLinks)
+                                      detectedLinks, gender)
     saveJson(speakerJson, speakerFilename)
 
 

speaker.py

@@ -194,16 +194,19 @@ def getSpeakerFromServer(baseDir: str, session,
 
 def speakerEndpointJson(displayName: str, summary: str,
                         content: str, imageDescription: str,
-                        links: []) -> {}:
+                        links: [], gender: str) -> {}:
     """Returns a json endpoint for the TTS speaker
     """
-    return {
+    speakerJson = {
         "name": displayName,
         "summary": summary,
         "say": content,
         "imageDescription": imageDescription,
         "detectedLinks": links
     }
+    if gender:
+        speakerJson['gender'] = gender
+    return speakerJson
 
 
 def _speakerEndpointSSML(displayName: str, summary: str,
@@ -223,13 +226,10 @@ def _speakerEndpointSSML(displayName: str, summary: str,
     else:
         if langShort == 'en':
             gender = gender.lower()
-            if 'him' in gender or 'male' in gender:
+            if 'he/him' in gender:
                 gender = 'male'
-            elif 'her' in gender or 'she' in gender or \
+            elif 'she/her' in gender:
-                 'fem' in gender or 'woman' in gender:
                 gender = 'female'
-            elif 'man' in gender:
-                gender = 'male'
             else:
                 gender = 'neutral'
 

utils.py

@@ -669,6 +669,41 @@ def getDisplayName(baseDir: str, actor: str, personCache: {}) -> str:
     return nameFound
 
 
+def getGenderFromBio(baseDir: str, actor: str, personCache: {}) -> str:
+    """Tries to ascertain gender from bio description
+    """
+    if '/statuses/' in actor:
+        actor = actor.split('/statuses/')[0]
+    if not personCache.get(actor):
+        return None
+    bioFound = None
+    if personCache[actor].get('actor'):
+        if personCache[actor]['actor'].get('summary'):
+            bioFound = personCache[actor]['actor']['summary']
+    else:
+        # Try to obtain from the cached actors
+        cachedActorFilename = \
+            baseDir + '/cache/actors/' + (actor.replace('/', '#')) + '.json'
+        if os.path.isfile(cachedActorFilename):
+            actorJson = loadJson(cachedActorFilename, 1)
+            if actorJson:
+                if actorJson.get('summary'):
+                    bioFound = actorJson['summary']
+    if not bioFound:
+        return None
+    gender = 'They/Them'
+    bioFoundOrig = bioFound
+    bioFound = bioFound.lower()
+    if 'him' in bioFound or 'male' in bioFound:
+        gender = 'He/Him'
+    elif 'her' in bioFound or 'she' in bioFound or \
+         'fem' in bioFound or 'woman' in bioFound:
+        gender = 'She/Her'
+    elif 'man' in bioFound or 'He' in bioFoundOrig:
+        gender = 'He/Him'
+    return gender
+
+
 def getNicknameFromActor(actor: str) -> str:
     """Returns the nickname from an actor url
     """