From 3b2a433d2b8f5e3e043f1b8ad1eafc16aa041802 Mon Sep 17 00:00:00 2001
From: Eric Joanis <eric.joanis@nrc-cnrc.gc.ca>
Date: Tue, 28 Feb 2023 13:46:44 -0500
Subject: [PATCH] fix: do not fail when the lang code is invalid

plus, adjust the test suite to the updated code
---
 readalongs/text/convert_xml.py |  4 ++--
 test/test_g2p_cli.py           |  7 ++++---
 test/test_web_api.py           | 11 ++---------
 3 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/readalongs/text/convert_xml.py b/readalongs/text/convert_xml.py
index 115caf54..18bac10f 100644
--- a/readalongs/text/convert_xml.py
+++ b/readalongs/text/convert_xml.py
@@ -179,8 +179,8 @@ def convert_word(word: str, lang: str):
                         if g2p_fallback_warning_count < 2 or verbose_warnings:
                             g2p_fallback_warning_count += 1
                             LOGGER.warning(
-                                f'Could not g2p "{text_to_g2p}" as {langs[g2p_lang]} ({g2p_lang}). '
-                                f"Trying fallback: {langs[lang]} ({lang})."
+                                f'Could not g2p "{text_to_g2p}" as {langs.get(g2p_lang, "")} ({g2p_lang}). '
+                                f"Trying fallback: {langs.get(lang, '')} ({lang})."
                             )
                         g2p_lang = lang.strip()
                         g2p_text, valid = convert_word(text_to_g2p, g2p_lang)
diff --git a/test/test_g2p_cli.py b/test/test_g2p_cli.py
index 57a052b0..e60a502d 100755
--- a/test/test_g2p_cli.py
+++ b/test/test_g2p_cli.py
@@ -218,7 +218,7 @@ def test_three_way_fallback(self):
         )
         results = self.runner.invoke(g2p, [tok_file2, "-"])
         self.assertEqual(results.exit_code, 0)
-        self.assertIn("Trying fallback: und", results.output)
+        self.assertIn("Trying fallback: Und", results.output)
 
     def test_align_with_error(self):
         """handling g2p errors in readalongs align with --g2p-fallback"""
@@ -251,6 +251,7 @@ def test_align_with_error(self):
                     "eng",  # lang 1 is eng
                     "-f",
                     "--language=fra:iku",  # fallback langs are fra and iku
+                    "--debug-g2p",
                     text_file,
                     os.path.join(self.data_dir, "noise.mp3"),
                     output_dir,
@@ -262,8 +263,8 @@ def test_align_with_error(self):
                 f"results.output='{results.output}' "
                 f"results.exception={results.exception!r}"
             )
-        self.assertIn("Trying fallback: fra", results.output)
-        self.assertIn("Trying fallback: iku", results.output)
+        self.assertIn("Trying fallback: French", results.output)
+        self.assertIn("Trying fallback: Inuktitut", results.output)
         self.assertNotIn("could not be g2p", results.output)
         self.assertIn("Number of aligned segments", results.output)
 
diff --git a/test/test_web_api.py b/test/test_web_api.py
index 9553912f..9333f0a8 100755
--- a/test/test_web_api.py
+++ b/test/test_web_api.py
@@ -135,14 +135,7 @@ def test_empty_g2p(self):
         response = API_CLIENT.post("/api/v1/assemble", json=request)
         self.assertEqual(response.status_code, 200)
         content_log = response.json()["log"]
-        for message_part in [
-            "The g2p output for",
-            "24",
-            "23",
-            "99",
-            "1234",
-            "is empty",
-        ]:
+        for message_part in ["The output of the g2p process", "24", "23", "is empty"]:
             self.assertIn(message_part, content_log)
 
     def test_langs(self):
@@ -164,7 +157,7 @@ def test_logs(self):
         response = API_CLIENT.post("/api/v1/assemble", json=request)
         content = response.json()
         # print("Content", content)
-        self.assertIn('Could not g2p "ña" as fra', content["log"])
+        self.assertIn('Could not g2p "ña" as French', content["log"])
 
     def test_debug(self):
         # Test the assemble endpoint with debug mode on