2024.5.16 测试整体构建

CMakeLists.txt

@@ -33,7 +33,7 @@ include(openfst)
 include_directories(
         ${OPENFST_INCLUDE_DIR}
         ${CMAKE_CURRENT_SOURCE_DIR}
-#        ${CMAKE_CURRENT_SOURCE_DIR}/kaldi
+        ${CMAKE_CURRENT_SOURCE_DIR}/kaldi
 )
 # Build all libraries
 add_subdirectory(utils)

decoder/asr_decoder.cc

@@ -15,10 +15,7 @@
 
 #include "decoder/asr_decoder.h"
 
-#include <ctype.h>
-
 #include <algorithm>
-#include <limits>
 #include <utility>
 
 #include "utils/timer.h"

frontend/fbank.h

@@ -27,7 +27,7 @@
 #endif
 
 namespace wenet {
-
+using namespace fst;
 // This code is based on kaldi Fbank implementation, please see
 // https://github.com/kaldi-asr/kaldi/blob/master/src/feat/feature-fbank.cc
 

frontend/feature_pipeline.cc

@@ -18,8 +18,8 @@
 #include <utility>
 
 namespace wenet {
-
-FeaturePipeline::FeaturePipeline(const FeaturePipelineConfig& config)
+    using namespace fst;
+    FeaturePipeline::FeaturePipeline(const FeaturePipelineConfig& config)
     : config_(config),
       feature_dim_(config.num_bins),
       fbank_(config.num_bins, config.sample_rate, config.frame_length,

kaldi/fstext/lattice-weight.h

@@ -401,8 +401,8 @@ template
 inline std::ostream& operator<<(std::ostream& strm,
                                 const LatticeWeightTpl<FloatType>& w) {
   LatticeWeightTpl<FloatType>::WriteFloatType(strm, w.Value1());
-  CHECK(FST_FLAGS_fst_weight_separator.size() == 1);  // NOLINT
-  strm << FST_FLAGS_fst_weight_separator[0];          // comma by default;
+  CHECK(FLAGS_fst_weight_separator.size() == 1);  // NOLINT
+  strm << FLAGS_fst_weight_separator[0];          // comma by default;
   // may or may not be settable from Kaldi programs.
   LatticeWeightTpl<FloatType>::WriteFloatType(strm, w.Value2());
   return strm;
@@ -411,9 +411,9 @@ inline std::ostream& operator<<(std::ostream& strm,
 template <class FloatType>
 inline std::istream& operator>>(std::istream& strm,
                                 LatticeWeightTpl<FloatType>& w1) {
-  CHECK(FST_FLAGS_fst_weight_separator.size() == 1);  // NOLINT
+  CHECK(FLAGS_fst_weight_separator.size() == 1);  // NOLINT
   // separator defaults to ','
-  return w1.ReadNoParen(strm, FST_FLAGS_fst_weight_separator[0]);
+  return w1.ReadNoParen(strm, FLAGS_fst_weight_separator[0]);
 }
 
 // CompactLattice will be an acceptor (accepting the words/output-symbols),
@@ -752,8 +752,8 @@ template
 inline std::ostream& operator<<(
     std::ostream& strm, const CompactLatticeWeightTpl<WeightType, IntType>& w) {
   strm << w.Weight();
-  CHECK(FST_FLAGS_fst_weight_separator.size() == 1);  // NOLINT
-  strm << FST_FLAGS_fst_weight_separator[0];          // comma by default.
+  CHECK(FLAGS_fst_weight_separator.size() == 1);  // NOLINT
+  strm << FLAGS_fst_weight_separator[0];          // comma by default.
   for (size_t i = 0; i < w.String().size(); i++) {
     strm << w.String()[i];
     if (i + 1 < w.String().size())
@@ -771,8 +771,8 @@ inline std::istream& operator>>(
   if (strm.fail()) {
     return strm;
   }
-  CHECK(FST_FLAGS_fst_weight_separator.size() == 1);            // NOLINT
-  size_t pos = s.find_last_of(FST_FLAGS_fst_weight_separator);  // normally ","
+  CHECK(FLAGS_fst_weight_separator.size() == 1);            // NOLINT
+  size_t pos = s.find_last_of(FLAGS_fst_weight_separator);  // normally ","
   if (pos == std::string::npos) {
     strm.clear(std::ios::badbit);
     return strm;

kaldi/lat/determinize-lattice-pruned.cc

@@ -22,7 +22,6 @@
 #include <climits>
 #include <vector>
 #include "fstext/determinize-lattice.h"  // for LatticeStringRepository
-#include "fstext/fstext-utils.h"
 #include "lat/lattice-functions.h"  // for PruneLattice
 // #include "lat/minimize-lattice.h"   // for minimization
 // #include "lat/push-lattice.h"       // for minimization

post_processor/processor/wetext_processor.cc

@@ -15,21 +15,22 @@
 #include "wetext_processor.h"
 #include "fst/string.h"
 namespace wetext {
-Processor::Processor(const std::string& tagger_path,
+    using namespace fst;
+    Processor::Processor(const std::string& tagger_path,
                      const std::string& verbalizer_path) {
-  tagger_.reset(StdVectorFst::Read(tagger_path));
-  verbalizer_.reset(StdVectorFst::Read(verbalizer_path));
-  compiler_ = std::make_shared<StringCompiler<StdArc>>(fst::StringTokenType::BYTE);
-  printer_ = std::make_shared<StringPrinter<StdArc>>(fst::StringTokenType::BYTE);
+          tagger_.reset(StdVectorFst::Read(tagger_path));
+          verbalizer_.reset(StdVectorFst::Read(verbalizer_path));
+          compiler_ = std::make_shared<StringCompiler<StdArc>>(fst::StringTokenType::BYTE);
+          printer_ = std::make_shared<StringPrinter<StdArc>>(fst::StringTokenType::BYTE);
 
-  if (tagger_path.find("_tn_") != tagger_path.npos) {
-    parse_type_ = ParseType::kTN;
-  } else if (tagger_path.find("_itn_") != tagger_path.npos) {
-    parse_type_ = ParseType::kITN;
-  } else {
-    LOG(FATAL) << "Invalid fst prefix, prefix should contain"
-               << " either \"_tn_\" or \"_itn_\".";
-  }
+    if (tagger_path.find("_tn_") != tagger_path.npos) {
+        parse_type_ = ParseType::kTN;
+    } else if (tagger_path.find("_itn_") != tagger_path.npos) {
+        parse_type_ = ParseType::kITN;
+    } else {
+        LOG(FATAL) << "Invalid fst prefix, prefix should contain"
+                   << " either \"_tn_\" or \"_itn_\".";
+    }
 }
 
 std::string Processor::ShortestPath(const StdVectorFst& lattice) {

post_processor/processor/wetext_token_parser.cc

@@ -18,6 +18,7 @@
 #include "../utils/wetext_string.h"
 
 namespace wetext {
+using namespace fst;
 const char EOS[] = "<EOS>";
 const std::set<std::string> UTF8_WHITESPACE = {" ", "\t", "\n", "\r",
                                                "\x0b\x0c"};

post_processor/utils/wetext_string.cc

@@ -17,7 +17,8 @@
 #include "wetext_log.h"
 
 namespace wetext {
-const char* WHITESPACE = " \n\r\t\f\v";
+    using namespace fst;
+    const char* WHITESPACE = " \n\r\t\f\v";
 
 int UTF8CharLength(char ch) {
   int num_bytes = 1;

utils/wn_string.cc

@@ -21,7 +21,7 @@
 #include "wn_utils.h"
 
 namespace wenet {
-
+using namespace fst;
 void SplitString(const std::string& str, std::vector<std::string>* strs) {
   SplitStringToVector(Trim(str), " \t", true, strs);
 }