feat: include the partial RAS XML with the 422 error when g2p fails in assemble (#292)

Squashes these commits:

* feat: include the partial RAS XML with the 422 error when g2p fails in assemble

* fix: make the g2p error words easier to see with spacing

* feat: with g2p 422, send partially processed RAS file

* test: add assertions about g2p errors in assemble call

Author: joanise

readalongs/web_api.py

@@ -36,7 +36,7 @@
 
 from fastapi import Body, FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import FileResponse
+from fastapi.responses import FileResponse, JSONResponse
 from lxml import etree
 from pydantic import BaseModel, Field
 from starlette.background import BackgroundTask
@@ -272,16 +272,20 @@ async def assemble(
         if not valid:
             if non_convertible_words:
                 logs = (
-                    "These words could not be converted from text to phonemes by g2p: '"
-                    + "', '".join(non_convertible_words)
-                    + "'."
+                    "These words could not be converted from text to phonemes by g2p: ' "
+                    + " ', ' ".join(non_convertible_words)
+                    + " '."
                 )
             else:
                 logs = "Logs: " + captured_logs.getvalue()
-            raise HTTPException(
+            return JSONResponse(
                 status_code=422,
-                detail="g2p could not be performed, please check your text or your language code. "
-                + logs,
+                content={
+                    "detail": "g2p could not be performed, please check your text or your language code. "
+                    + logs,
+                    "g2p_error_words": non_convertible_words,
+                    "partial_ras": xml_to_string(g2ped),
+                },
             )
         # create grammar
         dict_data, text_input = create_grammar(g2ped)

tests/test_web_api.py

@@ -219,10 +219,14 @@ def test_empty_g2p(self):
         with redirect_stderr(StringIO()):
             response = self.API_CLIENT.post("/api/v1/assemble", json=request)
         self.assertEqual(response.status_code, 422)
-        content_log = response.json()["detail"]
+        content = response.json()
+        content_log = content["detail"]
         for message_part in ["These words could not", "24", "23"]:
             self.assertIn(message_part, content_log)
 
+        self.assertEqual(content["g2p_error_words"], ["24", "23", "99", "1234"])
+        self.assertIn("partial_ras", content)
+
     def test_langs(self):
         # Test the langs endpoint
         with redirect_stderr(StringIO()):