Add C API for punctuation (#768)

sherpa-onnx/c-api/c-api.cc

@@ -15,6 +15,7 @@
 #include "sherpa-onnx/csrc/display.h"
 #include "sherpa-onnx/csrc/keyword-spotter.h"
 #include "sherpa-onnx/csrc/macros.h"
+#include "sherpa-onnx/csrc/offline-punctuation.h"
 #include "sherpa-onnx/csrc/offline-recognizer.h"
 #include "sherpa-onnx/csrc/online-recognizer.h"
 #include "sherpa-onnx/csrc/speaker-embedding-extractor.h"
@@ -1299,3 +1300,48 @@ void SherpaOnnxAudioTaggingFreeResults(
 
   delete[] events;
 }
+
+struct SherpaOnnxOfflinePunctuation {
+  std::unique_ptr<sherpa_onnx::OfflinePunctuation> impl;
+};
+
+const SherpaOnnxOfflinePunctuation *SherpaOnnxCreateOfflinePunctuation(
+    const SherpaOnnxOfflinePunctuationConfig *config) {
+  sherpa_onnx::OfflinePunctuationConfig c;
+  c.model.ct_transformer = SHERPA_ONNX_OR(config->model.ct_transformer, "");
+  c.model.num_threads = SHERPA_ONNX_OR(config->model.num_threads, 1);
+  c.model.debug = config->model.debug;
+  c.model.provider = SHERPA_ONNX_OR(config->model.provider, "cpu");
+
+  if (c.model.debug) {
+    SHERPA_ONNX_LOGE("%s\n", c.ToString().c_str());
+  }
+
+  if (!c.Validate()) {
+    SHERPA_ONNX_LOGE("Errors in config");
+    return nullptr;
+  }
+
+  SherpaOnnxOfflinePunctuation *punct = new SherpaOnnxOfflinePunctuation;
+  punct->impl = std::make_unique<sherpa_onnx::OfflinePunctuation>(c);
+
+  return punct;
+}
+
+void SherpaOnnxDestroyOfflinePunctuation(
+    const SherpaOnnxOfflinePunctuation *punct) {
+  delete punct;
+}
+
+const char *SherpaOfflinePunctuationAddPunct(
+    const SherpaOnnxOfflinePunctuation *punct, const char *text) {
+  std::string text_with_punct = punct->impl->AddPunctuation(text);
+
+  char *ans = new char[text_with_punct.size() + 1];
+  std::copy(text_with_punct.begin(), text_with_punct.end(), ans);
+  ans[text_with_punct.size()] = 0;
+
+  return ans;
+}
+
+void SherpaOfflinePunctuationFreeText(const char *text) { delete[] text; }

sherpa-onnx/c-api/c-api.h

@@ -1149,6 +1149,41 @@ SherpaOnnxAudioTaggingCompute(const SherpaOnnxAudioTagging *tagger,
 SHERPA_ONNX_API void SherpaOnnxAudioTaggingFreeResults(
     const SherpaOnnxAudioEvent *const *p);
 
+// ============================================================
+// For punctuation
+// ============================================================
+
+SHERPA_ONNX_API typedef struct SherpaOnnxOfflinePunctuationModelConfig {
+  const char *ct_transformer;
+  int32_t num_threads;
+  int32_t debug;  // true to print debug information of the model
+  const char *provider;
+} SherpaOnnxOfflinePunctuationModelConfig;
+
+SHERPA_ONNX_API typedef struct SherpaOnnxOfflinePunctuationConfig {
+  SherpaOnnxOfflinePunctuationModelConfig model;
+} SherpaOnnxOfflinePunctuationConfig;
+
+SHERPA_ONNX_API typedef struct SherpaOnnxOfflinePunctuation
+    SherpaOnnxOfflinePunctuation;
+
+// The user has to invoke SherpaOnnxDestroyOfflinePunctuation()
+// to free the returned pointer to avoid memory leak
+SHERPA_ONNX_API const SherpaOnnxOfflinePunctuation *
+SherpaOnnxCreateOfflinePunctuation(
+    const SherpaOnnxOfflinePunctuationConfig *config);
+
+SHERPA_ONNX_API void SherpaOnnxDestroyOfflinePunctuation(
+    const SherpaOnnxOfflinePunctuation *punct);
+
+// Add punctuations to the input text.
+// The user has to invoke SherpaOfflinePunctuationFreeText()
+// to free the returned pointer to avoid memory leak
+SHERPA_ONNX_API const char *SherpaOfflinePunctuationAddPunct(
+    const SherpaOnnxOfflinePunctuation *punct, const char *text);
+
+SHERPA_ONNX_API void SherpaOfflinePunctuationFreeText(const char *text);
+
 #if defined(__GNUC__)
 #pragma GCC diagnostic pop
 #endif

sherpa-onnx/csrc/offline-punctuation-ct-transformer-impl.h

@@ -134,25 +134,40 @@ class OfflinePunctuationCtTransformerImpl : public OfflinePunctuationImpl {
       }
     }  // for (int32_t i = 0; i != num_segments; ++i)
 
-    std::string ans;
+    if (punctuations.empty()) {
+      return text + meta_data.id2punct[meta_data.dot_id];
+    }
+    std::vector<std::string> words_punct;
 
     for (int32_t i = 0; i != static_cast<int32_t>(punctuations.size()); ++i) {
-      if (i > tokens.size()) {
+      if (i >= tokens.size()) {
         break;
       }
-      const std::string &w = tokens[i];
-      if (i > 0 && !(ans.back() & 0x80) && !(w[0] & 0x80)) {
-        ans.push_back(' ');
+      std::string &w = tokens[i];
+      if (i > 0 && !(words_punct.back()[0] & 0x80) && !(w[0] & 0x80)) {
+        words_punct.push_back(" ");
       }
-      ans.append(w);
+      words_punct.push_back(std::move(w));
+
       if (punctuations[i] != meta_data.underline_id) {
-        ans.append(meta_data.id2punct[punctuations[i]]);
+        words_punct.push_back(meta_data.id2punct[punctuations[i]]);
       }
     }
-    if (ans.back() != meta_data.dot_id && ans.back() != meta_data.quest_id) {
-      ans.push_back(meta_data.dot_id);
-    }
 
+    if (words_punct.back() == meta_data.id2punct[meta_data.comma_id] ||
+        words_punct.back() == meta_data.id2punct[meta_data.pause_id]) {
+      words_punct.back() = meta_data.id2punct[meta_data.dot_id];
+    }
+
+    if (words_punct.back() != meta_data.id2punct[meta_data.dot_id] &&
+        words_punct.back() != meta_data.id2punct[meta_data.quest_id]) {
+      words_punct.push_back(meta_data.id2punct[meta_data.dot_id]);
+    }
+
+    std::string ans;
+    for (const auto &w : words_punct) {
+      ans.append(w);
+    }
     return ans;
   }
 

sherpa-onnx/python/csrc/offline-punctuation.cc

@@ -4,6 +4,8 @@
 
 #include "sherpa-onnx/python/csrc/offline-punctuation.h"
 
+#include <string>
+
 #include "sherpa-onnx/csrc/offline-punctuation.h"
 
 namespace sherpa_onnx {