From e6b0e55c086bc4e3d1f0fcf248cf46b67dd0aef7 Mon Sep 17 00:00:00 2001
From: Michael Hansen <mike@rhasspy.org>
Date: Thu, 27 Jul 2023 15:51:40 -0500
Subject: [PATCH] Add sentence splitting tests

---
 src/python_test.py | 11 ++++++++++-
 src/test.cpp       | 16 +++++++++++++++-
 2 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/src/python_test.py b/src/python_test.py
index 38385d0..25f0038 100644
--- a/src/python_test.py
+++ b/src/python_test.py
@@ -22,7 +22,7 @@
 de_phonemes = phonemize_espeak("licht!", "de")
 
 # "lˈɪçt!" where "ç" is decomposed into two codepoints
-assert de_phonemes == [["l", "ˈ", "ɪ", "c", "̧", "t", "!"]]
+assert de_phonemes == [["l", "ˈ", "ɪ", "c", "̧", "t", "!"]], de_phonemes
 
 # phoneme -> [id, ...]
 espeak_map = get_espeak_map()
@@ -44,6 +44,15 @@
 
 # -----------------------------------------------------------------------------
 
+# Capitalization is required to get espeak to split the sentences.
+en_phonemes = phonemize_espeak("Test 1. Test2.", "en-us")
+assert en_phonemes == [
+    ["t", "ˈ", "ɛ", "s", "t", " ", "w", "ˈ", "ʌ", "n", "."],
+    ["t", "ˈ", "ɛ", "s", "t", " ", "t", "ˈ", "u", "ː", "."],
+], en_phonemes
+
+# -----------------------------------------------------------------------------
+
 codepoints_map = get_codepoints_map()
 assert "uk" in codepoints_map, "uk not supported"
 uk_phonemes = phonemize_codepoints("ВЕСЕ́ЛКА")
diff --git a/src/test.cpp b/src/test.cpp
index 55177e6..f580b60 100644
--- a/src/test.cpp
+++ b/src/test.cpp
@@ -37,6 +37,8 @@ phonemeString(const std::vector<std::vector<piper::Phoneme>> &phonemes) {
 
       phonemeStr << una::utf32to8(phonemeU32Str);
     }
+
+    phonemeStr << "\n";
   }
 
   return phonemeStr.str();
@@ -82,11 +84,23 @@ int main(int argc, char *argv[]) {
   piper::phonemize_eSpeak("this, is: a; test.", phonemeConfig, phonemes);
 
   std::string phonemeStr = phonemeString(phonemes);
-  if (phonemeStr != "ðˈɪs, ɪz: ˈeɪ; tˈɛst.") {
+  if (phonemeStr != "ðˈɪs, ɪz: ˈeɪ; tˈɛst.\n") {
     std::cerr << "punctuation test: " << phonemeStr << std::endl;
     return 1;
   }
 
+  // Check sentence splitting.
+  phonemes.clear();
+
+  // Capitalization is required to get espeak to split the sentences.
+  piper::phonemize_eSpeak("Test 1. Test 2.", phonemeConfig, phonemes);
+
+   phonemeStr = phonemeString(phonemes);
+  if (phonemeStr != "tˈɛst wˈʌn.\ntˈɛst tˈuː.\n") {
+    std::cerr << "sentence split: " << phonemeStr << std::endl;
+    return 1;
+  }
+
   // Check "ВЕСЕ́ЛКА" in Ukrainian
   piper::CodepointsPhonemeConfig codepointsConfig;
   phonemes.clear();