fix: prevent concurrent state crashes (#21568)

This commit is contained in:
buua436 authored and GitHub committed 2026-10-28 09:44:43 +08:00
1 parent 99ae9ca5ad
commit 752dc02157
33 files changed
+492 -155

No files matched your search

+99 -64
View File
@@ -29,7 +29,11 @@ void RAGAnalyzer_Destroy(RAGAnalyzerHandle handle) {
int RAGAnalyzer_Load(RAGAnalyzerHandle handle) {
if (!handle) return -1;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
return analyzer->Load();
try {
return analyzer->Load();
} catch (...) {
return -1;
}
}
void RAGAnalyzer_SetFineGrained(RAGAnalyzerHandle handle, bool fine_grained) {
@@ -46,47 +50,58 @@ void RAGAnalyzer_SetEnablePosition(RAGAnalyzerHandle handle, bool enable_positio
void RAGAnalyzer_SetLanguage(RAGAnalyzerHandle handle, const char* language) {
if (!handle || !language) return;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
analyzer->SetLanguage(std::string(language));
try {
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
analyzer->SetLanguage(std::string(language));
} catch (...) {
}
}
int RAGAnalyzer_Analyze(RAGAnalyzerHandle handle, const char* text, RAGTokenCallback callback) {
if (!handle || !text || !callback) return -1;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
try {
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
Term input;
input.text_ = std::string(text);
Term input;
input.text_ = std::string(text);
TermList output;
// Use the analyzer's internal state for fine_grained and enable_position
int ret = analyzer->Analyze(input, output, analyzer->fine_grained_, analyzer->enable_position_);
TermList output;
// Use the analyzer's internal state for fine_grained and enable_position
int ret = analyzer->Analyze(input, output, analyzer->fine_grained_, analyzer->enable_position_);
if (ret != 0) {
return ret;
if (ret != 0) {
return ret;
}
// Call callback for each token
for (const auto& term : output) {
callback(term.text_.c_str(), term.text_.length(), term.word_offset_, term.end_offset_);
}
return 0;
} catch (...) {
return -1;
}
// Call callback for each token
for (const auto& term : output) {
callback(term.text_.c_str(), term.text_.length(), term.word_offset_, term.end_offset_);
}
return 0;
}
char* RAGAnalyzer_Tokenize(RAGAnalyzerHandle handle, const char* text) {
if (!handle || !text) return nullptr;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
try {
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
std::string result = analyzer->Tokenize(std::string(text));
std::string result = analyzer->Tokenize(std::string(text));
// Allocate memory for C string
char* c_result = static_cast<char*>(malloc(result.size() + 1));
if (c_result) {
std::memcpy(c_result, result.c_str(), result.size() + 1);
// Allocate memory for C string
char* c_result = static_cast<char*>(malloc(result.size() + 1));
if (c_result) {
std::memcpy(c_result, result.c_str(), result.size() + 1);
}
return c_result;
} catch (...) {
return nullptr;
}
return c_result;
}
RAGTokenList* RAGAnalyzer_TokenizeWithPosition(RAGAnalyzerHandle handle, const char* text) {
@@ -94,19 +109,27 @@ RAGTokenList* RAGAnalyzer_TokenizeWithPosition(RAGAnalyzerHandle handle, const c
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
auto [tokens, positions] = analyzer->TokenizeWithPosition(std::string(text));
if (analyzer->fine_grained_) {
std::string joined_tokens;
for (size_t i = 0; i < tokens.size(); ++i) {
if (i > 0) joined_tokens += " ";
joined_tokens += tokens[i];
}
std::vector<std::string> tokens;
std::vector<std::pair<unsigned, unsigned>> positions;
try {
auto result = analyzer->TokenizeWithPosition(std::string(text));
tokens = std::move(result.first);
positions = std::move(result.second);
if (analyzer->fine_grained_) {
std::string joined_tokens;
for (size_t i = 0; i < tokens.size(); ++i) {
if (i > 0) joined_tokens += " ";
joined_tokens += tokens[i];
}
std::vector<std::string> fine_tokens;
std::vector<std::pair<unsigned, unsigned>> fine_positions;
analyzer->FineGrainedTokenizeWithPosition(joined_tokens, positions, fine_tokens, fine_positions);
tokens = std::move(fine_tokens);
positions = std::move(fine_positions);
std::vector<std::string> fine_tokens;
std::vector<std::pair<unsigned, unsigned>> fine_positions;
analyzer->FineGrainedTokenizeWithPosition(joined_tokens, positions, fine_tokens, fine_positions);
tokens = std::move(fine_tokens);
positions = std::move(fine_positions);
}
} catch (...) {
return nullptr;
}
RAGTokenList* token_list = static_cast<RAGTokenList*>(malloc(sizeof(RAGTokenList)));
@@ -176,49 +199,61 @@ uint32_t RAGToken_GetEndOffset(void* token) {
char* RAGAnalyzer_FineGrainedTokenize(RAGAnalyzerHandle handle, const char* tokens) {
if (!handle || !tokens) return nullptr;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
try {
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
std::vector<std::string> result;
analyzer->FineGrainedTokenize(std::string(tokens), result);
std::vector<std::string> result;
analyzer->FineGrainedTokenize(std::string(tokens), result);
// Join results with space
std::string result_str;
for (size_t i = 0; i < result.size(); ++i) {
if (i > 0) result_str += " ";
result_str += result[i];
// Join results with space
std::string result_str;
for (size_t i = 0; i < result.size(); ++i) {
if (i > 0) result_str += " ";
result_str += result[i];
}
// Allocate memory for C string
char* c_result = static_cast<char*>(malloc(result_str.size() + 1));
if (c_result) {
std::memcpy(c_result, result_str.c_str(), result_str.size() + 1);
}
return c_result;
} catch (...) {
return nullptr;
}
// Allocate memory for C string
char* c_result = static_cast<char*>(malloc(result_str.size() + 1));
if (c_result) {
std::memcpy(c_result, result_str.c_str(), result_str.size() + 1);
}
return c_result;
}
int32_t RAGAnalyzer_GetTermFreq(RAGAnalyzerHandle handle, const char* term) {
if (!handle || !term) return 0;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
return analyzer->Freq(term);
try {
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
return analyzer->Freq(term);
} catch (...) {
return 0;
}
}
char* RAGAnalyzer_GetTermTag(RAGAnalyzerHandle handle, const char* term) {
if (!handle || !term) return nullptr;
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
std::string tag_result = analyzer->Tag(term);
try {
RAGAnalyzer* analyzer = static_cast<RAGAnalyzer*>(handle);
std::string tag_result = analyzer->Tag(term);
if (tag_result.empty()) {
if (tag_result.empty()) {
return nullptr;
}
// Allocate memory for C string
char* c_result = static_cast<char*>(malloc(tag_result.size() + 1));
if (c_result) {
std::memcpy(c_result, tag_result.c_str(), tag_result.size() + 1);
}
return c_result;
} catch (...) {
return nullptr;
}
// Allocate memory for C string
char* c_result = static_cast<char*>(malloc(tag_result.size() + 1));
if (c_result) {
std::memcpy(c_result, tag_result.c_str(), tag_result.size() + 1);
}
return c_result;
}
RAGAnalyzerHandle RAGAnalyzer_Copy(RAGAnalyzerHandle handle) {