diff options
author | Peng Wu <alexepico@gmail.com> | 2013-04-10 14:47:15 +0800 |
---|---|---|
committer | Peng Wu <alexepico@gmail.com> | 2013-04-10 14:47:15 +0800 |
commit | a762acbcb5c16fe8bbcbc67ff05d0882ae9faa56 (patch) | |
tree | 645888c14ef893f0b755bd12de88053540158ac1 /utils/training/gen_unigram.cpp | |
parent | 9ce3b3437d4e7164475aaf8efd00935f82ee3801 (diff) | |
download | libpinyin-a762acbcb5c16fe8bbcbc67ff05d0882ae9faa56.tar.gz libpinyin-a762acbcb5c16fe8bbcbc67ff05d0882ae9faa56.tar.xz libpinyin-a762acbcb5c16fe8bbcbc67ff05d0882ae9faa56.zip |
update gen_unigram.cpp
Diffstat (limited to 'utils/training/gen_unigram.cpp')
-rw-r--r-- | utils/training/gen_unigram.cpp | 17 |
1 files changed, 14 insertions, 3 deletions
diff --git a/utils/training/gen_unigram.cpp b/utils/training/gen_unigram.cpp index 62584a9..47aee91 100644 --- a/utils/training/gen_unigram.cpp +++ b/utils/training/gen_unigram.cpp @@ -27,13 +27,24 @@ int main(int argc, char * argv[]){ + SystemTableInfo system_table_info; + + bool retval = system_table_info.load("table.conf"); + 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_table_info(); + /* Note: please increase the value when corpus size becomes larger. * To avoid zero value when computing unigram frequency in float format. */ for (size_t i = 0; i < PHRASE_INDEX_LIBRARY_COUNT; ++i) { - const pinyin_table_info_t * table_info = pinyin_phrase_files + i; + const pinyin_table_info_t * table_info = phrase_files + i; assert(table_info->m_dict_index == i); if (SYSTEM_FILE != table_info->m_file_type && @@ -68,10 +79,10 @@ int main(int argc, char * argv[]){ } } - if (!save_phrase_index(&phrase_index)) + if (!save_phrase_index(phrase_files, &phrase_index)) exit(ENOENT); - if (!save_dictionary(&phrase_index)) + if (!save_dictionary(phrase_files, &phrase_index)) exit(ENOENT); return 0; |