int main(int argc, char * argv[]){ const char * evals_text = "evals.text"; pinyin_option_t options = USE_TONE; FacadeChewingTable largetable; MemoryChunk * chunk = new MemoryChunk; chunk->load("pinyin_index.bin"); largetable.load(options, chunk, NULL); FacadePhraseTable2 phrase_table; chunk = new MemoryChunk; chunk->load("phrase_index.bin"); phrase_table.load(chunk, NULL); FacadePhraseIndex phrase_index; if (!load_phrase_index(&phrase_index)) exit(ENOENT); Bigram system_bigram; system_bigram.attach("bigram.db", ATTACH_READONLY); Bigram user_bigram; user_bigram.attach(NULL, ATTACH_CREATE|ATTACH_READWRITE); PinyinLookup2 pinyin_lookup(options, &largetable, &phrase_index, &system_bigram, &user_bigram); /* open evals.text. */ FILE * evals_file = fopen(evals_text, "r"); if ( NULL == evals_file ) { fprintf(stderr, "Can't open file:%s\n", evals_text); exit(ENOENT); } PhraseTokens phrase_tokens; memset(phrase_tokens, 0, sizeof(PhraseTokens)); phrase_index.prepare_tokens(phrase_tokens); /* Evaluates the correction rate of test text documents. */ size_t tested_count = 0; size_t passed_count = 0; char* linebuf = NULL; size_t size = 0; TokenVector tokens = g_array_new(FALSE, TRUE, sizeof(phrase_token_t)); phrase_token_t token = null_token; while( getline(&linebuf, &size, evals_file) ) { if ( feof(evals_file) ) break; if ( '\n' == linebuf[strlen(linebuf)-1] ) linebuf[strlen(linebuf)-1] = '\0'; glong phrase_len = 0; ucs4_t * phrase = g_utf8_to_ucs4(linebuf, -1, NULL, &phrase_len, NULL); token = null_token; if ( 0 != phrase_len ) { int result = phrase_table.search(phrase_len, phrase, phrase_tokens); int num = get_first_token(phrase_tokens, token); if ( !(result & SEARCH_OK) ) token = null_token; g_free(phrase); phrase = NULL; } if ( null_token == token ) { if ( tokens->len ) { /* one test. */ if ( do_one_test(&pinyin_lookup, &phrase_index, tokens) ) { tested_count ++; passed_count ++; } else { tested_count ++; } g_array_set_size(tokens, 0); } } else { g_array_append_val(tokens, token); } } if ( tokens->len ) { /* one test. */ if ( do_one_test(&pinyin_lookup, &phrase_index, tokens) ) { tested_count ++; passed_count ++; } else { tested_count ++; } } parameter_t rate = passed_count / (parameter_t) tested_count; printf("correction rate:%f\n", rate); g_array_free(tokens, TRUE); fclose(evals_file); free(linebuf); phrase_index.destroy_tokens(phrase_tokens); return 0; }
int main(int argc, char * argv[]){ int i = 1; bool train_pi_gram = true; const char * bigram_filename = "deleted_bigram.db"; setlocale(LC_ALL, ""); while ( i < argc ){ if ( strcmp("--help", argv[i]) == 0){ print_help(); exit(0); } else if ( strcmp("--skip-pi-gram-training", argv[i]) == 0 ){ train_pi_gram = false; } else if ( strcmp("--deleted-bigram-file", argv[i]) == 0){ if ( ++i >= argc ) { print_help(); exit(EINVAL); } bigram_filename = argv[i]; } else { print_help(); exit(EINVAL); } ++i; } /* load phrase table. */ PhraseLargeTable2 phrase_table; MemoryChunk * new_chunk = new MemoryChunk; new_chunk->load("phrase_index.bin"); phrase_table.load(new_chunk); FacadePhraseIndex phrase_index; if (!load_phrase_index(&phrase_index)) exit(ENODATA); Bigram bigram; bigram.attach(bigram_filename, ATTACH_CREATE|ATTACH_READWRITE); char* linebuf = NULL; size_t size = 0; phrase_token_t last_token, cur_token = last_token = 0; while( getline(&linebuf, &size, stdin) ){ if ( feof(stdin) ) break; if ( '\n' == linebuf[strlen(linebuf) - 1] ) { linebuf[strlen(linebuf) - 1] = '\0'; } TAGLIB_PARSE_SEGMENTED_LINE(&phrase_index, token, linebuf); last_token = cur_token; cur_token = token; /* skip null_token in second word. */ if ( null_token == cur_token ) continue; /* skip pi-gram training. */ if ( null_token == last_token ){ if ( !train_pi_gram ) continue; last_token = sentence_start; } /* train bi-gram */ SingleGram * single_gram = NULL; bigram.load(last_token, single_gram); if ( NULL == single_gram ){ single_gram = new SingleGram; } guint32 freq, total_freq; //increase freq if (single_gram->get_freq(cur_token, freq)) assert(single_gram->set_freq(cur_token, freq + 1)); else assert(single_gram->insert_freq(cur_token, 1)); //increase total freq single_gram->get_total_freq(total_freq); single_gram->set_total_freq(total_freq + 1); bigram.store(last_token, single_gram); delete single_gram; } free(linebuf); return 0; }
int main(int argc, char * argv[]){ FILE * input = stdin; FILE * output = stdout; setlocale(LC_ALL, ""); GError * error = NULL; GOptionContext * context; context = g_option_context_new("- shortest path segment"); g_option_context_add_main_entries(context, entries, NULL); if (!g_option_context_parse(context, &argc, &argv, &error)) { g_print("option parsing failed:%s\n", error->message); exit(EINVAL); } if (outputfile) { output = fopen(outputfile, "w"); if (NULL == output) { perror("open file failed"); exit(EINVAL); } } if (argc > 2) { fprintf(stderr, "too many arguments.\n"); exit(EINVAL); } if (2 == argc) { input = fopen(argv[1], "r"); if (NULL == input) { perror("open file failed"); exit(EINVAL); } } SystemTableInfo2 system_table_info; bool retval = system_table_info.load(SYSTEM_TABLE_INFO); if (!retval) { fprintf(stderr, "load table.conf failed.\n"); exit(ENOENT); } /* init phrase table */ FacadePhraseTable3 phrase_table; phrase_table.load(SYSTEM_PHRASE_INDEX, NULL); /* init phrase index */ FacadePhraseIndex phrase_index; const pinyin_table_info_t * phrase_files = system_table_info.get_default_tables(); if (!load_phrase_index(phrase_files, &phrase_index)) exit(ENOENT); CONTEXT_STATE state, next_state; GArray * current_ucs4 = g_array_new(TRUE, TRUE, sizeof(ucs4_t)); PhraseTokens tokens; memset(tokens, 0, sizeof(PhraseTokens)); phrase_index.prepare_tokens(tokens); char * linebuf = NULL; size_t size = 0; ssize_t read; while( (read = getline(&linebuf, &size, input)) != -1 ){ if ( '\n' == linebuf[strlen(linebuf) - 1] ) { linebuf[strlen(linebuf) - 1] = '\0'; } /* check non-ucs4 characters. */ const glong num_of_chars = g_utf8_strlen(linebuf, -1); glong len = 0; ucs4_t * sentence = g_utf8_to_ucs4(linebuf, -1, NULL, &len, NULL); if ( len != num_of_chars ) { fprintf(stderr, "non-ucs4 characters encountered:%s.\n", linebuf); fprintf(output, "%d \n", null_token); continue; } /* only new-line persists. */ if ( 0 == num_of_chars ) { fprintf(output, "%d \n", null_token); continue; } state = CONTEXT_INIT; int result = phrase_table.search( 1, sentence, tokens); g_array_append_val( current_ucs4, sentence[0]); if ( result & SEARCH_OK ) state = CONTEXT_SEGMENTABLE; else state = CONTEXT_UNKNOWN; for ( int i = 1; i < num_of_chars; ++i) { int result = phrase_table.search( 1, sentence + i, tokens); if ( result & SEARCH_OK ) next_state = CONTEXT_SEGMENTABLE; else next_state = CONTEXT_UNKNOWN; if ( state == next_state ){ g_array_append_val(current_ucs4, sentence[i]); continue; } assert ( state != next_state ); if ( state == CONTEXT_SEGMENTABLE ) deal_with_segmentable(&phrase_table, &phrase_index, current_ucs4, output); if ( state == CONTEXT_UNKNOWN ) deal_with_unknown(current_ucs4, output); /* save the current character */ g_array_set_size(current_ucs4, 0); g_array_append_val(current_ucs4, sentence[i]); state = next_state; } if ( current_ucs4->len ) { /* this seems always true. */ if ( state == CONTEXT_SEGMENTABLE ) deal_with_segmentable(&phrase_table, &phrase_index, current_ucs4, output); if ( state == CONTEXT_UNKNOWN ) deal_with_unknown(current_ucs4, output); g_array_set_size(current_ucs4, 0); } /* print extra enter */ if ( gen_extra_enter ) fprintf(output, "%d \n", null_token); g_free(sentence); } phrase_index.destroy_tokens(tokens); /* print enter at file tail */ fprintf(output, "%d \n", null_token); g_array_free(current_ucs4, TRUE); free(linebuf); fclose(input); fclose(output); return 0; }
int main(int argc, char * argv[]){ setlocale(LC_ALL, ""); SystemTableInfo system_table_info; bool retval = system_table_info.load("../../data/table.conf"); if (!retval) { fprintf(stderr, "load table.conf failed.\n"); exit(ENOENT); } /* init phrase table */ FacadePhraseTable2 phrase_table; MemoryChunk * chunk = new MemoryChunk; chunk->load("../../data/phrase_index.bin"); phrase_table.load(chunk, NULL); const pinyin_table_info_t * phrase_files = system_table_info.get_table_info(); /* init phrase index */ FacadePhraseIndex phrase_index; if (!load_phrase_index(phrase_files, &phrase_index)) exit(ENOENT); /* init bi-gram */ Bigram system_bigram; system_bigram.attach("../../data/bigram.db", ATTACH_READONLY); Bigram user_bigram; gfloat lambda = system_table_info.get_lambda(); /* init phrase lookup */ PhraseLookup phrase_lookup(lambda, &phrase_table, &phrase_index, &system_bigram, &user_bigram); /* try one sentence */ char * linebuf = NULL; size_t size = 0; ssize_t read; while( (read = getline(&linebuf, &size, stdin)) != -1 ){ if ( '\n' == linebuf[strlen(linebuf) - 1] ) { linebuf[strlen(linebuf) - 1] = '\0'; } if ( strcmp ( linebuf, "quit" ) == 0) break; /* check non-ucs4 characters */ const glong num_of_chars = g_utf8_strlen(linebuf, -1); glong len = 0; ucs4_t * sentence = g_utf8_to_ucs4(linebuf, -1, NULL, &len, NULL); if ( len != num_of_chars ) { fprintf(stderr, "non-ucs4 characters are not accepted.\n"); g_free(sentence); continue; } try_phrase_lookup(&phrase_lookup, sentence, len); g_free(sentence); } free(linebuf); return 0; }
int main(int argc, char * argv[]){ int i = 1; bool gen_extra_enter = false; setlocale(LC_ALL, ""); /* deal with options */ while ( i < argc ){ if ( strcmp ("--help", argv[i]) == 0 ){ print_help(); exit(0); } else if ( strcmp("--generate-extra-enter", argv[i]) == 0 ){ gen_extra_enter = true; } else { print_help(); exit(EINVAL); } ++i; } /* init phrase table */ FacadePhraseTable2 phrase_table; MemoryChunk * chunk = new MemoryChunk; chunk->load("phrase_index.bin"); phrase_table.load(chunk, NULL); /* init phrase index */ FacadePhraseIndex phrase_index; if (!load_phrase_index(&phrase_index)) exit(ENOENT); /* init bi-gram */ Bigram system_bigram; system_bigram.attach("bigram.db", ATTACH_READONLY); Bigram user_bigram; /* init phrase lookup */ PhraseLookup phrase_lookup(&phrase_table, &phrase_index, &system_bigram, &user_bigram); CONTEXT_STATE state, next_state; GArray * current_ucs4 = g_array_new(TRUE, TRUE, sizeof(ucs4_t)); PhraseTokens tokens; memset(tokens, 0, sizeof(PhraseTokens)); phrase_index.prepare_tokens(tokens); /* split the sentence */ char * linebuf = NULL; size_t size = 0; ssize_t read; while( (read = getline(&linebuf, &size, stdin)) != -1 ){ if ( '\n' == linebuf[strlen(linebuf) - 1] ) { linebuf[strlen(linebuf) - 1] = '\0'; } /* check non-ucs4 characters */ const glong num_of_chars = g_utf8_strlen(linebuf, -1); glong len = 0; ucs4_t * sentence = g_utf8_to_ucs4(linebuf, -1, NULL, &len, NULL); if ( len != num_of_chars ) { fprintf(stderr, "non-ucs4 characters encountered:%s.\n", linebuf); printf("\n"); continue; } /* only new-line persists. */ if ( 0 == num_of_chars ) { printf("\n"); continue; } state = CONTEXT_INIT; int result = phrase_table.search( 1, sentence, tokens); g_array_append_val( current_ucs4, sentence[0]); if ( result & SEARCH_OK ) state = CONTEXT_SEGMENTABLE; else state = CONTEXT_UNKNOWN; for ( int i = 1; i < num_of_chars; ++i) { int result = phrase_table.search( 1, sentence + i, tokens); if ( result & SEARCH_OK ) next_state = CONTEXT_SEGMENTABLE; else next_state = CONTEXT_UNKNOWN; if ( state == next_state ){ g_array_append_val(current_ucs4, sentence[i]); continue; } assert ( state != next_state ); if ( state == CONTEXT_SEGMENTABLE ) deal_with_segmentable(&phrase_lookup, current_ucs4); if ( state == CONTEXT_UNKNOWN ) deal_with_unknown(current_ucs4); /* save the current character */ g_array_set_size(current_ucs4, 0); g_array_append_val(current_ucs4, sentence[i]); state = next_state; } if ( current_ucs4->len ) { /* this seems always true. */ if ( state == CONTEXT_SEGMENTABLE ) deal_with_segmentable(&phrase_lookup, current_ucs4); if ( state == CONTEXT_UNKNOWN ) deal_with_unknown(current_ucs4); g_array_set_size(current_ucs4, 0); } /* print extra enter */ if ( gen_extra_enter ) printf("\n"); } phrase_index.destroy_tokens(tokens); /* print enter at file tail */ printf("\n"); g_array_free(current_ucs4, TRUE); free(linebuf); return 0; }
int main( int argc, char * argv[]){ pinyin_option_t options = USE_TONE | USE_RESPLIT_TABLE | PINYIN_CORRECT_ALL | PINYIN_AMB_ALL; FacadeChewingTable largetable; MemoryChunk * chunk = new MemoryChunk; chunk->load("../../data/pinyin_index.bin"); largetable.load(options, chunk, NULL); FacadePhraseIndex phrase_index; if (!load_phrase_index(&phrase_index)) exit(ENOENT); Bigram system_bigram; system_bigram.attach("../../data/bigram.db", ATTACH_READONLY); Bigram user_bigram; user_bigram.attach(NULL, ATTACH_CREATE|ATTACH_READWRITE); PinyinLookup2 pinyin_lookup(options, &largetable, &phrase_index, &system_bigram, &user_bigram); /* prepare the prefixes for get_best_match. */ TokenVector prefixes = g_array_new (FALSE, FALSE, sizeof(phrase_token_t)); g_array_append_val(prefixes, sentence_start); CandidateConstraints constraints = g_array_new (TRUE, FALSE, sizeof(lookup_constraint_t)); MatchResults results = g_array_new(FALSE, FALSE, sizeof(phrase_token_t)); char* linebuf = NULL; size_t size = 0; ssize_t read; while( (read = getline(&linebuf, &size, stdin)) != -1 ){ if ( '\n' == linebuf[strlen(linebuf) - 1] ) { linebuf[strlen(linebuf) - 1] = '\0'; } if ( strcmp ( linebuf, "quit" ) == 0) break; FullPinyinParser2 parser; ChewingKeyVector keys = g_array_new(FALSE, FALSE, sizeof(ChewingKey)); ChewingKeyRestVector key_rests = g_array_new(FALSE, FALSE, sizeof(ChewingKeyRest)); parser.parse(options, keys, key_rests, linebuf, strlen(linebuf)); if ( 0 == keys->len ) /* invalid pinyin */ continue; /* initialize constraints. */ g_array_set_size(constraints, keys->len); for ( size_t i = 0; i < constraints->len; ++i){ lookup_constraint_t * constraint = &g_array_index(constraints, lookup_constraint_t, i); constraint->m_type = NO_CONSTRAINT; } guint32 start_time = record_time(); for ( size_t i = 0; i < bench_times; ++i) pinyin_lookup.get_best_match(prefixes, keys, constraints, results); print_time(start_time, bench_times); for ( size_t i = 0; i < results->len; ++i){ phrase_token_t * token = &g_array_index(results, phrase_token_t, i); if ( null_token == *token) continue; printf("pos:%ld,token:%d\t", i, *token); } printf("\n"); char * sentence = NULL; pinyin_lookup.convert_to_utf8(results, sentence); printf("%s\n", sentence); g_array_free(keys, TRUE); g_array_free(key_rests, TRUE); g_free(sentence); } g_array_free(prefixes, TRUE); g_array_free(constraints, TRUE); g_array_free(results, TRUE); free(linebuf); return 0; }
int main(int argc, char * argv[]){ setlocale(LC_ALL, ""); GError * error = NULL; GOptionContext * context; context = g_option_context_new("- generate deleted n-gram"); g_option_context_add_main_entries(context, entries, NULL); if (!g_option_context_parse(context, &argc, &argv, &error)) { g_print("option parsing failed:%s\n", error->message); exit(EINVAL); } SystemTableInfo2 system_table_info; bool retval = system_table_info.load(SYSTEM_TABLE_INFO); if (!retval) { fprintf(stderr, "load table.conf failed.\n"); exit(ENOENT); } FacadePhraseIndex phrase_index; const pinyin_table_info_t * phrase_files = system_table_info.get_default_tables(); if (!load_phrase_index(phrase_files, &phrase_index)) exit(ENODATA); Bigram bigram; bigram.attach(bigram_filename, ATTACH_CREATE|ATTACH_READWRITE); char* linebuf = NULL; size_t size = 0; phrase_token_t last_token, cur_token = last_token = 0; while( getline(&linebuf, &size, stdin) ){ if ( feof(stdin) ) break; if ( '\n' == linebuf[strlen(linebuf) - 1] ) { linebuf[strlen(linebuf) - 1] = '\0'; } TAGLIB_PARSE_SEGMENTED_LINE(&phrase_index, token, linebuf); last_token = cur_token; cur_token = token; /* skip null_token in second word. */ if ( null_token == cur_token ) continue; /* skip pi-gram training. */ if ( null_token == last_token ){ if ( !train_pi_gram ) continue; last_token = sentence_start; } /* train bi-gram */ SingleGram * single_gram = NULL; bigram.load(last_token, single_gram); if ( NULL == single_gram ){ single_gram = new SingleGram; } guint32 freq, total_freq; //increase freq if (single_gram->get_freq(cur_token, freq)) assert(single_gram->set_freq(cur_token, freq + 1)); else assert(single_gram->insert_freq(cur_token, 1)); //increase total freq single_gram->get_total_freq(total_freq); single_gram->set_total_freq(total_freq + 1); bigram.store(last_token, single_gram); delete single_gram; } free(linebuf); return 0; }