PaddlePaddle/PaddleSpeech

[TTS] C++ 前端崩溃测试用例

Open

#3.035 aberto em 14 de mar. de 2023

Ver no GitHub
 (2 comments) (0 reactions) (1 assignee)Python (1.702 forks)batch import
BugT2Sfeature requestgood first issue

Métricas do repositório

Stars
 (9.453 stars)
Métricas de merge de PR
 (Nenhuma PRs mesclada em 30d)

Description

记录一些导致C++前端崩溃的示例文本：

  1. 包含字母，断言异常。
./run_front_demo.sh --sentence 'TTS语音合成服务'

tts_front_demo: /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:358: int ppspeech::FrontEngineInterface::GetInitialsFinals(const string&, std::vector<std::__cxx11::basic_string<char> >&, std::vector<std::__cxx11::basic_string<char> >&): Assertion `word_finals.size() == ppspeech::utf8string2wstring(word).length() && word_finals.size() == word_initials.size()' failed.
  1. 字母在结尾，内存越界。
./run_front_demo.sh --sentence '语音合成服务a'

# 调试
gdb ./build/tts_front_demo -ex "set args --sentence '语音合成服务a'"

Program received signal SIGSEGV, Segmentation fault.
(gdb) bt
#0  __memcpy_generic () at ../sysdeps/aarch64/multiarch/../memcpy.S:100
#1  0x0000007ff7fa8db0 in void std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::_M_construct<char*>(char*, char*, std::forward_iterator_tag) () from /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/third-party/build/lib/libgflags.so.2.2
#2  0x000000555558c500 in __gnu_cxx::new_allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >::construct<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&> (this=0x7fffffeb18, __p=0x555ac91430) at /usr/include/c++/9/ext/new_allocator.h:146                                                                                                                                                                            
#3  0x0000005555584ecc in std::allocator_traits<std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::construct<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&> (__a=..., __p=0x555ac91430) at /usr/include/c++/9/bits/alloc_traits.h:483                                                                                                                                                                        
#4  0x000000555557f010 in std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::push_back (this=0x7fffffeb18, 
    __x=<error: Cannot access memory at address 0xff1f0000ff1b>) at /usr/include/c++/9/bits/stl_vector.h:1189
#5  0x000000555556dd48 in ppspeech::FrontEngineInterface::GetInitialsFinals (this=0x5555638350, word="a", word_initials=std::vector of length 1, capacity 1 = {...}, word_finals=std::vector of length 0, capacity 2)
    at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:349
#6  0x000000555556df58 in ppspeech::FrontEngineInterface::GetFinals (this=0x5555638350, word="a", word_finals=std::vector of length 0, capacity 2) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:366
#7  0x000000555556ef30 in ppspeech::FrontEngineInterface::MergeThreeTones (this=0x5555638350, seg_result=std::vector of length 4, capacity 4 = {...}) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:512
#8  0x000000555556ff34 in ppspeech::FrontEngineInterface::MergeforModify (this=0x5555638350, seg_word_type=std::vector of length 4, capacity 4 = {...}, modify_seg_word_type=std::vector of length 4, capacity 4 = {...})
    at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:634
#9  0x000000555556d4f8 in ppspeech::FrontEngineInterface::Cut (this=0x5555638350, sentence="语音合成服务a", cut_result=std::vector of length 4, capacity 4 = {...}) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:269
#10 0x000000555556cc70 in ppspeech::FrontEngineInterface::GetSentenceIds (this=0x5555638350, sentence="语音合成服务a", phoneids=std::vector of length 0, capacity 0, toneids=std::vector of length 0, capacity 0)
    at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:183
#11 0x00000055555603b4 in main (argc=1, argv=0x7ffffff288) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/front_demo/front_demo.cpp:53
  1. 包含空格，内存越界。
./run_front_demo.sh --sentence '语音合成 服务'
gdb ./build/tts_front_demo -ex "set args --sentence '语音合成 服务'"

Program received signal SIGSEGV, Segmentation fault.
__memcpy_generic () at ../sysdeps/aarch64/multiarch/../memcpy.S:100
(gdb) bt
#0  __memcpy_generic () at ../sysdeps/aarch64/multiarch/../memcpy.S:100
#1  0x0000007ff7fa8db0 in void std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::_M_construct<char*>(char*, char*, std::forward_iterator_tag) () from /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/third-party/build/lib/libgflags.so.2.2
#2  0x000000555558c500 in __gnu_cxx::new_allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >::construct<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&> (this=0x7fffffeb18, __p=0x555ac91430) at /usr/include/c++/9/ext/new_allocator.h:146                                                                                                                                                                            
#3  0x0000005555584ecc in std::allocator_traits<std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::construct<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&> (__a=..., __p=0x555ac91430) at /usr/include/c++/9/bits/alloc_traits.h:483                                                                                                                                                                        
#4  0x000000555557f010 in std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::push_back (this=0x7fffffeb18, 
    __x=<error: Cannot access memory at address 0xff1f0000ff1b>) at /usr/include/c++/9/bits/stl_vector.h:1189
#5  0x000000555556dd48 in ppspeech::FrontEngineInterface::GetInitialsFinals (this=0x5555638350, word=" ", word_initials=std::vector of length 1, capacity 1 = {...}, word_finals=std::vector of length 0, capacity 2)
    at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:349
#6  0x000000555556df58 in ppspeech::FrontEngineInterface::GetFinals (this=0x5555638350, word=" ", word_finals=std::vector of length 0, capacity 2) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:366
#7  0x000000555556ef30 in ppspeech::FrontEngineInterface::MergeThreeTones (this=0x5555638350, seg_result=std::vector of length 4, capacity 4 = {...}) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:512
#8  0x000000555556ff34 in ppspeech::FrontEngineInterface::MergeforModify (this=0x5555638350, seg_word_type=std::vector of length 4, capacity 4 = {...}, modify_seg_word_type=std::vector of length 4, capacity 4 = {...})
    at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:634
#9  0x000000555556d4f8 in ppspeech::FrontEngineInterface::Cut (this=0x5555638350, sentence="语音合成 服务", cut_result=std::vector of length 4, capacity 4 = {...}) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:269
#10 0x000000555556cc70 in ppspeech::FrontEngineInterface::GetSentenceIds (this=0x5555638350, sentence="语音合成 服务", phoneids=std::vector of length 0, capacity 0, toneids=std::vector of length 0, capacity 0)
    at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/src/front/front_interface.cpp:183
#11 0x00000055555603b4 in main (argc=1, argv=0x7ffffff288) at /home/firefly/work/tts/PaddleSpeech/demos/TTSCppFrontend/front_demo/front_demo.cpp:53

遇到其他情况时会继续追加。

Guia do colaborador

Direção de pesquisa
O problema documenta falhas no frontend C++ quando a entrada contém letras inglesas ou espaços. As falhas ocorrem em funções como GetInitialsFinals e MergeThreeTones em front interface.cpp. Para corrigir, é necessário validar a entrada para rejeitar ou tratar caracteres não chineses e espaços. Comece adicionando sanitização de entrada na função principal de front demo.cpp ou dentro de FrontEngineInterface::GetSentenceIds. Em seguida, certifique se de que o mecanismo não prossiga com entrada inválida que cause falhas de asserção ou corrupção de memória.
Pilha de tecnologia
cpp
Domain
backend
Tipo Issue
Bug
DifficultyDificuldade de implementação estimada para um novo contribuidor, de 1 para alterações muito pequenas a 5 para trabalho de nível especializado.
3
Tempo estimadoUm intervalo de tempo aproximado para um colaborador experiente investigar, implementar, testar e preparar um pull request.
1-2 dias
Status da atividadeQuão disponível o issue aparece agora: novo, ativo, obsoleto, bloqueado ou aguardando entrada do mantenedor.
Antigo
ClarityCom que clareza o issue explica a mudança esperada, os critérios de aceitação e a próxima etapa.
Claro
Prerequisites
C++ basicsunderstanding of memory managementfamiliarity with front interface.cpp
Simpatia para novatosUma pontuação de 1 a 100 que estima o quão acessível este issue é para colaboradores iniciantes.
50