static PyObject *constituents(PyObject *self, PyObject *args) {
Dictionary dict;
Parse_Options opts;
Sentence sent;
Linkage linkage;
CNode * cn;
/// Link counts
int num_linkages;
const char *text;
PyObject *output_list;
if (!PyArg_ParseTuple(args, "s", &text))
return NULL;
opts = parse_options_create();
parse_options_set_verbosity(opts, -1);
setlocale(LC_ALL, "");
dict = dictionary_create_default_lang();
if (!dict) {
PyErr_SetString(PyExc_RuntimeError, "Fatal error: Unable to open the dictionary");
Py_INCREF(Py_None);
return Py_None;
}
sent = sentence_create(text, dict);
sentence_split(sent, opts);
num_linkages = sentence_parse(sent, opts);
if (num_linkages > 0) {
linkage = linkage_create(0, sent, opts);
cn = linkage_constituent_tree(linkage);
output_list = build_tree(cn, linkage);
if(output_list == Py_None) {
Py_INCREF(output_list);
return output_list;
}
linkage_free_constituent_tree(cn);
linkage_delete(linkage);
} else {
sentence_delete(sent);
dictionary_delete(dict);
parse_options_delete(opts);
Py_INCREF(Py_None);
return Py_None;
}
sentence_delete(sent);
dictionary_delete(dict);
parse_options_delete(opts);
return Py_BuildValue("S", output_list);
}
int sentence_parse(Sentence sent, Parse_Options opts)
{
int rc;
sent->num_valid_linkages = 0;
/* If the sentence has not yet been split, do so now.
* This is for backwards compatibility, for existing programs
* that do not explicitly call the splitter.
*/
if (0 == sent->length)
{
rc = sentence_split(sent, opts);
if (rc) return -1;
}
else
{
/* During a panic parse, we enter here a second time, with leftover
* garbage. Free it. We really should make the code that is panicking
* do this free, but right now, they have no API for it, so we do it
* as a favor. XXX FIXME someday. */
free_sentence_disjuncts(sent);
}
/* Check for bad sentence length */
if (MAX_SENTENCE <= sent->length)
{
prt_error("Error: sentence too long, contains more than %d words\n",
MAX_SENTENCE);
return -2;
}
resources_reset(opts->resources);
/* Expressions were set up during the tokenize stage.
* Prune them, and then parse.
*/
expression_prune(sent, opts);
print_time(opts, "Finished expression pruning");
if (opts->use_sat_solver)
{
sat_parse(sent, opts);
}
else
{
classic_parse(sent, opts);
}
print_time(opts, "Finished parse");
if ((verbosity > 0) &&
(PARSE_NUM_OVERFLOW < sent->num_linkages_found))
{
prt_error("Warning: Combinatorial explosion! nulls=%zu cnt=%d\n"
"Consider retrying the parse with the max allowed disjunct cost set lower.\n"
"At the command line, use !cost-max\n",
sent->null_count, sent->num_linkages_found);
}
return sent->num_valid_linkages;
}
int sentence_parse(Sentence sent, Parse_Options opts)
{
int rc;
verbosity = opts->verbosity;
debug = opts->debug;
test = opts->test;
sent->num_valid_linkages = 0;
/* If the sentence has not yet been split, do so now.
* This is for backwards compatibility, for existing programs
* that do not explicitly call the splitter.
*/
if (0 == sent->length)
{
rc = sentence_split(sent, opts);
if (rc) return -1;
}
/* Check for bad sentence length */
if (MAX_SENTENCE <= sent->length)
{
prt_error("Error: sentence too long, contains more than %d words\n",
MAX_SENTENCE);
return -2;
}
/* Initialize/free any leftover garbage */
free_sentence_disjuncts(sent); /* Is this really needed ??? */
resources_reset_space(opts->resources);
if (resources_exhausted(opts->resources))
return 0;
/* Expressions were previously set up during the tokenize stage. */
expression_prune(sent);
print_time(opts, "Finished expression pruning");
if (opts->use_sat_solver)
{
sat_parse(sent, opts);
}
else
{
chart_parse(sent, opts);
}
print_time(opts, "Finished parse");
if ((verbosity > 0) &&
(PARSE_NUM_OVERFLOW < sent->num_linkages_found))
{
prt_error("WARNING: Combinatorial explosion! nulls=%zu cnt=%d\n"
"Consider retrying the parse with the max allowed disjunct cost set lower.\n"
"At the command line, use !cost-max\n",
sent->null_count, sent->num_linkages_found);
}
return sent->num_valid_linkages;
}
// **************************************************************************
// HTTPヘッダを受信して解析する。
//
// 処理するのはGETのみ。GET以外のメソッドが来たらエラー
// 今のところ、URIとuser_agent、Range、Hostを解析。
// URIは、URIデコードもやる。
//
// return: 0 正常終了
// return: 0以外 エラー
// **************************************************************************
static int http_header_receive(int accept_socket, HTTP_RECV_INFO *http_recv_info_p)
{
int result = 0;
int recv_len;
unsigned char line_buf[1024]; // 大きめに。
unsigned char work_buf[1024];
unsigned char work_buf2[1024];
unsigned char split1[1024];
unsigned char split2[1024];
int ret;
int i;
int j;
// ================================
// １行づつ HTTPヘッダを受信
// ================================
for (i=0;;i++)
{
// 1行受信 実行。
memset(line_buf, '\0', sizeof(line_buf));
recv_len = line_receive(accept_socket, line_buf, sizeof(line_buf));
// debug. 受信したヘッダ表示
debug_log_output("'%s'(%d byte)\n", line_buf, recv_len );
// 受信した内容をチェック。
if ( i != 0 && recv_len == 0 ) // 空行検知。ヘッダ受信終了。
{
break;
}
else if ( recv_len < 0 ) // 受信失敗
{
return ( -1 );
}
// --------------------------
// GETメッセージチェック
// --------------------------
if ( i == 0 ) // １行目のみチェック
{
debug_log_output("URI Check start.'%s'\n", line_buf);
// GET/POSTある？
if (strncmp(line_buf, "GET ", 4) && strncmp(line_buf, "POST ", 5))
{
debug_log_output("'GET' or 'POST' not found. error.");
return ( -1 );
}
strncpy(http_recv_info_p->request_method, line_buf, sizeof(http_recv_info_p->request_method));
cut_after_character(http_recv_info_p->request_method, ' ');
// 最初のスペースまでを削除。
cut_before_character(line_buf, ' ');
// 次にスペースが出てきたところの後ろまでを削除。
cut_after_character(line_buf, ' ');
// ===========================
// GETオプション部解析
// ===========================
// REQUEST_URI用・Proxy用に値を保存
strncpy(http_recv_info_p->request_uri, line_buf, sizeof(http_recv_info_p->request_uri));
// '?'が存在するかチェック。
if ( strchr(line_buf, '?') != NULL )
{
strncpy(work_buf, line_buf, sizeof(work_buf));
// '?'より前をカット
cut_before_character(work_buf, '?' );
debug_log_output("work_buf = '%s'", work_buf );
while ( 1 )
{
memset(split1, '\0', sizeof(split1));
memset(split2, '\0', sizeof(split2));
uri_decode(split1, sizeof(split1), work_buf, sizeof(work_buf) );
strcpy(work_buf, split1);
// 最初に登場する'&'で分割
ret = sentence_split(work_buf, '&', split1, split2 );
if ( ret == 0 ) // 分割成功
{
strncpy(work_buf, split2, sizeof(work_buf));
}
else if (strlen(work_buf) > 0) // まだwork_bufに中身ある？
{
strncpy( split1, work_buf, sizeof(split1));
strncpy( work_buf, "", sizeof(work_buf));
}
else // 処理終了
{
break;
}
// -------------------------------------
// GETした内容 解析開始
// 超安直。いいのかこんな比較で。
// -------------------------------------
// URIデコード
// uri_decode(work_buf2, sizeof(work_buf2), split1, sizeof(split1) );
strcpy(work_buf2, split1);
// "page="あるか調査。
if (strncasecmp( work_buf2, "page=", strlen("page=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
if ( strlen(work_buf2) > 0 )
{
// 構造体に値を保存。
http_recv_info_p->page = atoi(work_buf2);
}
continue;
}
// "menupage="あるか調査。
if (strncasecmp( work_buf2, "menupage=", strlen("menupage=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
if ( strlen(work_buf2) > 0 )
{
// 構造体に値を保存。
http_recv_info_p->menupage = atoi(work_buf2);
}
continue;
}
// "title=" DVD title selection
if (strncasecmp( work_buf2, "title=", strlen("title=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
if ( strlen(work_buf2) > 0 )
{
// 構造体に値を保存。
http_recv_info_p->title = atoi(work_buf2);
}
continue;
}
if (strncasecmp( work_buf2, "width=", strlen("width=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
if ( strlen(work_buf2) > 0 )
{
// 構造体に値を保存。
global_param.target_jpeg_width = atoi(work_buf2);
}
continue;
}
if (strncasecmp( work_buf2, "height=", strlen("height=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
if ( strlen(work_buf2) > 0 )
{
// 構造体に値を保存。
global_param.target_jpeg_height = atoi(work_buf2);
}
continue;
}
// "action="あるか調査。
if (strncasecmp( work_buf2, "action=", strlen("action=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->action, work_buf2, sizeof(http_recv_info_p->action));
continue;
}
// "type=" allplay list type
if (strncasecmp( work_buf2, "type=movie", strlen("type=movie") ) == 0 )
{
http_recv_info_p->default_file_type = TYPE_MOVIE;
continue;
}
if (strncasecmp( work_buf2, "type=music", strlen("type=music") ) == 0 )
{
http_recv_info_p->default_file_type = TYPE_MUSIC;
continue;
}
if (strncasecmp( work_buf2, "type=photo", strlen("type=photo") ) == 0 )
{
http_recv_info_p->default_file_type = TYPE_JPEG;
continue;
}
if (strncasecmp( work_buf2, "type=soundtrack", strlen("type=soundtrack") ) == 0 )
{
http_recv_info_p->default_file_type = TYPE_MUSICLIST;
continue;
}
if (strncasecmp( work_buf2, "type=slideshow", strlen("type=slideshow") ) == 0 )
{
http_recv_info_p->default_file_type = TYPE_PLAYLIST;
continue;
}
// "option="あるか調査
if (strncasecmp( work_buf2, "option=", strlen("option=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->option, work_buf2, sizeof(http_recv_info_p->option));
continue;
}
// "alias="あるか調査
if (strncasecmp( work_buf2, "alias=", strlen("alias=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->alias, work_buf2, sizeof(http_recv_info_p->alias));
if (strncasecmp(http_recv_info_p->alias, "movie", strlen("movie")) == 0)
http_recv_info_p->default_file_type = TYPE_MOVIE;
else if (strncasecmp(http_recv_info_p->alias, "music", strlen("music")) == 0)
http_recv_info_p->default_file_type = TYPE_MUSIC;
else if (strncasecmp(http_recv_info_p->alias, "photo", strlen("photo")) == 0)
http_recv_info_p->default_file_type = TYPE_JPEG;
else
http_recv_info_p->default_file_type = TYPE_UNKNOWN;
continue;
}
if (strncasecmp( work_buf2, "lsearch=", strlen("lsearch=") ) == 0 )
{
cut_before_character(work_buf2, '=');
strcpy(http_recv_info_p->lsearch, work_buf2);
strcpy(http_recv_info_p->recv_uri, work_buf2);
continue;
}
// "search=
if (strncasecmp( work_buf2, "search", strlen("search") ) == 0 )
{
if (strncasecmp(work_buf2, "search_movie", strlen("search_movie")) == 0) {
http_recv_info_p->search_type = TYPE_MOVIE;
strcpy(http_recv_info_p->search_str, "_movie");
http_recv_info_p->default_file_type = TYPE_MOVIE;
// printf("search movie for ");
} else if (strncasecmp(work_buf2, "search_music", strlen("search_music")) == 0) {
http_recv_info_p->search_type = TYPE_MUSIC;
strcpy(http_recv_info_p->search_str, "_music");
http_recv_info_p->default_file_type = TYPE_MUSIC;
// printf("search music for ");
} else if (strncasecmp(work_buf2, "search_photo", strlen("search_photo")) == 0) {
http_recv_info_p->search_type = TYPE_JPEG;
strcpy(http_recv_info_p->search_str, "_photo");
http_recv_info_p->default_file_type = TYPE_JPEG;
// printf("search photo for ");
} else if (strncasecmp(work_buf2, "search_all", strlen("search_all")) == 0) {
http_recv_info_p->search_type = TYPE_UNKNOWN;
strcpy(http_recv_info_p->search_str, "_all");
http_recv_info_p->default_file_type = TYPE_UNKNOWN;
// printf("search all for ");
} else
continue;
cut_before_character(work_buf2, '=');
strncpy(http_recv_info_p->search, work_buf2, sizeof(http_recv_info_p->search));
if (http_recv_info_p->search[0] == '\0')
// everything qualifies
strcpy(http_recv_info_p->search, ".*");
// printf("%s\n", http_recv_info_p->search);
continue;
}
// "sort="あるか調査
if (strncasecmp( work_buf2, "sort=", strlen("sort=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->sort, work_buf2, sizeof(http_recv_info_p->sort));
continue;
}
// "dvdopt="あるか調査
if (strncasecmp( work_buf2, "dvdopt=", strlen("dvdopt=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->dvdopt, work_buf2, sizeof(http_recv_info_p->dvdopt));
continue;
}
// "focus="あるか調査
if (strncasecmp( work_buf2, "focus=", strlen("focus=") ) == 0 )
{
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->focus, work_buf2, sizeof(http_recv_info_p->focus));
continue;
}
}
}
debug_log_output("http_recv_info_p->page = '%d'", http_recv_info_p->page);
debug_log_output("http_recv_info_p->title = '%d'", http_recv_info_p->title);
debug_log_output("http_recv_info_p->action = '%s'", http_recv_info_p->action);
debug_log_output("http_recv_info_p->option = '%s'", http_recv_info_p->option);
debug_log_output("http_recv_info_p->dvdopt = '%s'", http_recv_info_p->dvdopt);
// URIデコード
cut_after_character(line_buf, '?');
uri_decode(work_buf, sizeof(work_buf), line_buf, sizeof(line_buf) );
strncpy(line_buf, work_buf, sizeof(line_buf));
debug_log_output("URI(decoded):'%s'\n", line_buf);
convert_language_code(line_buf, work_buf, sizeof(work_buf), CODE_AUTO, CODE_EUC);
debug_log_output("URI(decoded,euc,FYI):'%s'\n", work_buf);
// 構造体に保存
if (http_recv_info_p->lsearch[0] == '\0')
strncpy(http_recv_info_p->recv_uri, line_buf, sizeof(http_recv_info_p->recv_uri));
continue;
}
// User-agent切り出し
if ( strncasecmp(line_buf, HTTP_USER_AGENT, strlen(HTTP_USER_AGENT) ) == 0 )
{
// ':'より前を切る
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
// 構造体に保存
strncpy( http_recv_info_p->user_agent, line_buf, sizeof(http_recv_info_p->user_agent));
// Set the skin name based on user agent, if desired
for(j=0; j<global_param.alternate_skin_count; j++) {
debug_log_output("Checking for '%s'", global_param.alternate_skin_match[j]);
if(strstr(line_buf, global_param.alternate_skin_match[j]) != NULL) {
strcpy(global_param.skin_name, global_param.alternate_skin_name[j]);
debug_log_output("User agent matches alternate skin '%s'", global_param.skin_name);
break;
}
}
continue;
}
// Rangeあるかチェック
if ( strncasecmp(line_buf, HTTP_RANGE, strlen(HTTP_RANGE) ) == 0 )
{
debug_log_output("%s Detect.\n", HTTP_RANGE);
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
// recv_range にRangeの中身保存
strncpy(http_recv_info_p->recv_range, line_buf, sizeof(http_recv_info_p->recv_range));
// '=' より前を切る
cut_before_character(line_buf, '=');
// '-'で前後に分割。
sentence_split(line_buf, '-', work_buf, work_buf2);
debug_log_output("work_buf='%s'\n", work_buf);
debug_log_output("work_buf2='%s'\n", work_buf2);
// 値を文字列→数値変換
http_recv_info_p->range_start_pos = strtoull(work_buf, NULL, 10);
if ( strlen(work_buf2) > 0 )
{
http_recv_info_p->range_end_pos = strtoull(work_buf2, NULL, 10);
}
debug_log_output("range_start_pos=%d\n", http_recv_info_p->range_start_pos);
debug_log_output("range_end_pos=%d\n", http_recv_info_p->range_end_pos);
continue;
}
// Hostあるかチェック
if ( strncasecmp(line_buf, HTTP_HOST, strlen(HTTP_HOST) ) == 0 )
{
debug_log_output("%s Detect.\n", HTTP_HOST);
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
strncpy(http_recv_info_p->recv_host, line_buf, sizeof(http_recv_info_p->recv_host));
if(NULL == strchr(http_recv_info_p->recv_host, ':')) {
debug_log_output("CLIENT BUG: Host header field was missing port number - fixing");
snprintf(http_recv_info_p->recv_host + strlen(http_recv_info_p->recv_host), sizeof(http_recv_info_p->recv_host)-1, ":%d", global_param.server_port);
debug_log_output("%s '%s'", HTTP_HOST, http_recv_info_p->recv_host);
}
continue;
}
if (strncasecmp(line_buf, HTTP_AUTHORIZATION, strlen(HTTP_AUTHORIZATION)) == 0) {
debug_log_output("%s Detect.\n", HTTP_AUTHORIZATION);
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
if (strncmp(line_buf, "Basic ", 6)) {
debug_log_output("received '%s', is not supported.", line_buf);
continue;
}
strncpy(http_recv_info_p->passwd, line_buf + 6, sizeof(http_recv_info_p->passwd));
continue;
}
// Content-Lengthあるかチェック
if ( strncasecmp(line_buf, HTTP_CONTENT_LENGTH_STR, strlen(HTTP_CONTENT_LENGTH_STR) ) == 0 )
{
debug_log_output("%s Detect.\n", HTTP_CONTENT_LENGTH_STR);
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
/* recv_content_length は わざと long です. */
http_recv_info_p->recv_content_length = strtol(line_buf, NULL, 10);
debug_log_output("Content-Length: %ld", http_recv_info_p->recv_content_length);
continue;
}
}
return result;
}
static PyObject *domains(PyObject *self, PyObject *args) {
Dictionary dict;
Parse_Options opts;
Sentence sent;
Linkage linkage;
//CNode * cn;
/// Link counts
int num_linkages;
int links;
int i;
int j = 0;
int num_domains;
const char *text;
PyObject *output_list;
PyObject *temp;
output_list = PyList_New(0);
if (!PyArg_ParseTuple(args, "s", &text))
return NULL;
opts = parse_options_create();
parse_options_set_verbosity(opts, -1);
setlocale(LC_ALL, "");
dict = dictionary_create_default_lang();
if (!dict) {
PyErr_SetString(PyExc_RuntimeError, "Fatal error: Unable to open the dictionary");
Py_INCREF(Py_None);
return Py_None;
}
sent = sentence_create(text, dict);
sentence_split(sent, opts);
num_linkages = sentence_parse(sent, opts);
if (num_linkages > 0) {
linkage = linkage_create(0, sent, opts);
links = linkage_get_num_sublinkages(linkage);
for(i=0; i<=links; i++) {
num_domains = linkage_get_link_num_domains(linkage, i);
const char **temp1 = linkage_get_link_domain_names(linkage, i);
//for(j=0; j<=num_domains; j++){
while(num_domains < j) {
temp = PyString_FromString(temp1[j]);
PyList_Append(output_list, temp);
j++;
}
j = 0;
}
linkage_delete(linkage);
} else {
sentence_delete(sent);
dictionary_delete(dict);
parse_options_delete(opts);
Py_INCREF(Py_None);
return Py_None;
}
sentence_delete(sent);
dictionary_delete(dict);
parse_options_delete(opts);
return Py_BuildValue("Si", output_list, num_domains);
}
/// This is the basic sentence dissection
static PyObject *sentence(PyObject *self, PyObject *args) {
Dictionary dict;
Parse_Options opts;
Sentence sent;
Linkage linkage;
Linkage sub_linkage;
char * diagram;
/// Link counts
int num_linkages;
int links;
/// Index's for the iterators
int link_idx;
int word_idx;
int num_words;
long span;
long sub_linkages;
const char *text;
const char *d_output;
PyObject *output_list;
PyObject *word_list;
PyObject *word2_list;
PyObject *span_list;
PyObject *temp;
PyObject *sublinkage_list;
PyObject *_diagram;
output_list = PyList_New(0);
word_list = PyList_New(0);
word2_list = PyList_New(0);
sublinkage_list = PyList_New(0);
span_list = PyList_New(0);
if (!PyArg_ParseTuple(args, "s", &text))
return NULL;
opts = parse_options_create();
parse_options_set_verbosity(opts, -1);
parse_options_set_screen_width(opts, 50);
setlocale(LC_ALL, "");
dict = dictionary_create_default_lang();
if (!dict) {
PyErr_SetString(PyExc_RuntimeError, "Fatal error: Unable to open the dictionary");
Py_INCREF(Py_None);
return Py_None;
}
sent = sentence_create(text, dict);
sentence_split(sent, opts);
num_linkages = sentence_parse(sent, opts);
if (num_linkages > 0) {
linkage = linkage_create(0, sent, opts);
/// Get the lengths of everything
num_words = linkage_get_num_words(linkage);
links = linkage_get_num_links(linkage);
for(link_idx=0; link_idx < links; link_idx++) {
PyObject *temp_subLen;
diagram = linkage_print_diagram(linkage);
_diagram = PyString_FromString(diagram);
sub_linkage = linkage_create(link_idx, sent, opts);
sub_linkages = linkage_get_num_sublinkages(linkage);
temp_subLen = PyLong_FromLong(sub_linkages);
linkage_delete(sub_linkage);
PyList_Append(sublinkage_list, temp_subLen);
span = linkage_get_link_length(linkage, link_idx);
PyList_Append(span_list, PyInt_FromLong(span));
PyObject *temp_list;
temp_list = PyList_New(0);
/// Sub Group these (left and right labels)
const char *t1 = linkage_get_link_llabel(linkage, link_idx);
temp = PyString_FromString(t1);
PyList_Append(temp_list, temp);
const char *t2 = linkage_get_link_rlabel(linkage, link_idx);
temp = PyString_FromString(t2);
PyList_Append(temp_list, temp);
/// Then add to the main list
PyList_Append(output_list, temp_list);
/// Just the label
const char *t3 = linkage_get_link_label(linkage, link_idx);
temp = PyString_FromString(t3);
PyList_Append(word2_list, temp);
}
for(word_idx=0; word_idx < num_words; word_idx++) {
d_output = linkage_get_word(linkage, word_idx);
PyObject *word;
word = PyString_FromString(d_output);
PyList_Append(word_list, word);
}
linkage_free_diagram(diagram);
linkage_delete(linkage);
} else {
sentence_delete(sent);
dictionary_delete(dict);
parse_options_delete(opts);
Py_INCREF(Py_None);
return Py_None;
}
sentence_delete(sent);
dictionary_delete(dict);
parse_options_delete(opts);
return Py_BuildValue("SSSSSS", word_list, span_list, output_list, word2_list, sublinkage_list, _diagram);
}
int main(int argc, char * argv[])
{
FILE *input_fh = stdin;
Dictionary dict;
const char *language = NULL;
int num_linkages;
Label label = NO_LABEL;
Command_Options *copts;
Parse_Options opts;
bool batch_in_progress = false;
isatty_stdin = isatty(fileno(stdin));
isatty_stdout = isatty(fileno(stdout));
#ifdef _WIN32
/* If compiled with MSVC/MinGW, we still support running under Cygwin.
* This is done by checking running_under_cygwin to resolve
* incompatibilities. */
const char *ostype = getenv("OSTYPE");
if ((NULL != ostype) && (0 == strcmp(ostype, "cygwin")))
running_under_cygwin = true;
/* argv encoding is in the current locale. */
argv = argv2utf8(argc);
if (NULL == argv)
{
prt_error("Fatal error: Unable to parse command line\n");
exit(-1);
}
#ifdef _MSC_VER
_set_printf_count_output(1); /* enable %n support for display_1line_help()*/
#endif /* _MSC_VER */
win32_set_utf8_output();
#endif /* _WIN32 */
#if LATER
/* Try to catch the SIGWINCH ... except this is not working. */
struct sigaction winch_act;
winch_act.sa_handler = winch_handler;
winch_act.sa_sigaction = NULL;
sigemptyset (&winch_act.sa_mask);
winch_act.sa_flags = 0;
sigaction (SIGWINCH, &winch_act, NULL);
#endif
copts = command_options_create();
if (copts == NULL || copts->panic_opts == NULL)
{
prt_error("Fatal error: unable to create parse options\n");
exit(-1);
}
opts = copts->popts;
setup_panic_parse_options(copts->panic_opts);
copts->panic_mode = true;
parse_options_set_max_parse_time(opts, 30);
parse_options_set_linkage_limit(opts, 1000);
parse_options_set_min_null_count(opts, 0);
parse_options_set_max_null_count(opts, 0);
parse_options_set_short_length(opts, 16);
parse_options_set_islands_ok(opts, false);
parse_options_set_display_morphology(opts, false);
save_default_opts(copts); /* Options so far are the defaults */
if ((argc > 1) && (argv[1][0] != '-')) {
/* The dictionary is the first argument if it doesn't begin with "-" */
language = argv[1];
}
for (int i = 1; i < argc; i++)
{
if (strcmp("--help", argv[i]) == 0)
{
print_usage(stdout, argv[0], copts, 0);
}
if (strcmp("--version", argv[i]) == 0)
{
printf("Version: %s\n", linkgrammar_get_version());
printf("%s\n", linkgrammar_get_configuration());
exit(0);
}
}
/* Process command line variable-setting commands (only). */
for (int i = 1; i < argc; i++)
{
if (argv[i][0] == '-')
{
const char *var = argv[i] + ((argv[i][1] != '-') ? 1 : 2);
if ((var[0] != '!') && (0 > issue_special_command(var, copts, NULL)))
print_usage(stderr, argv[0], copts, -1);
}
else if (i != 1)
{
prt_error("Fatal error: Unknown argument '%s'.\n", argv[i]);
print_usage(stderr, argv[0], copts, -1);
}
}
if (language && *language)
{
dict = dictionary_create_lang(language);
if (dict == NULL)
{
prt_error("Fatal error: Unable to open dictionary.\n");
exit(-1);
}
}
else
{
dict = dictionary_create_default_lang();
if (dict == NULL)
{
prt_error("Fatal error: Unable to open default dictionary.\n");
exit(-1);
}
}
/* Process the command line '!' commands */
for (int i = 1; i < argc; i++)
{
if ((argv[i][0] == '-') && (argv[i][1] == '!'))
{
if (0 > issue_special_command(argv[i]+1, copts, dict))
print_usage(stderr, argv[0], copts, -1);
}
}
check_winsize(copts);
prt_error("Info: Dictionary version %s, locale %s\n",
linkgrammar_get_dict_version(dict),
linkgrammar_get_dict_locale(dict));
prt_error("Info: Library version %s. Enter \"!help\" for help.\n",
linkgrammar_get_version());
/* Main input loop */
while (true)
{
char *input_string;
Sentence sent = NULL;
/* Make sure stderr is shown even when MSVC binary runs under
* Cygwin/MSYS pty (in that case it is fully buffered(!)). */
fflush(stderr);
verbosity = parse_options_get_verbosity(opts);
debug = parse_options_get_debug(opts);
test = parse_options_get_test(opts);
input_string = fget_input_string(input_fh, stdout, /*check_return*/false);
check_winsize(copts);
if (NULL == input_string)
{
if (ferror(input_fh))
prt_error("Error: Read: %s\n", strerror(errno));
if (input_fh == stdin) break;
fclose (input_fh);
input_fh = stdin;
continue;
}
/* Discard whitespace characters from end of string. */
for (char *p = &input_string[strlen(input_string)-1];
(p > input_string) && strchr(WHITESPACE, *p) ; p--)
{
*p = '\0';
}
/* If the input string is just whitespace, then ignore it. */
if (strspn(input_string, WHITESPACE) == strlen(input_string))
continue;
char command = special_command(input_string, copts, dict);
if ('e' == command) break; /* It was an exit command */
if ('c' == command) continue; /* It was another command */
if (-1 == command) continue; /* It was a bad command */
/* We have to handle the !file command inline; it's too hairy
* otherwise ... */
if ('f' == command)
{
char * filename = &input_string[strcspn(input_string, WHITESPACE)] + 1;
int fnlen = strlen(filename);
if (0 == fnlen)
{
prt_error("Error: Missing file name argument\n");
continue;
}
if ('\n' == filename[fnlen-1]) filename[fnlen-1] = '\0';
struct stat statbuf;
if ((0 == stat(filename, &statbuf)) && statbuf.st_mode & S_IFDIR)
{
prt_error("Error: Cannot open %s: %s\n",
filename, strerror(EISDIR));
continue;
}
input_fh = fopen(filename, "r");
if (NULL == input_fh)
{
prt_error("Error: Cannot open %s: %s\n", filename, strerror(errno));
input_fh = stdin;
continue;
}
continue;
}
if (!copts->batch_mode) batch_in_progress = false;
if ('\0' != test[0])
{
/* In batch mode warn only once.
* In auto-next-linkage mode don't warn at all. */
if (!batch_in_progress && !auto_next_linkage_test(test))
{
fflush(stdout);
/* Remind the developer this is a test mode. */
prt_error("Warning: Tests enabled: %s\n", test);
if (copts->batch_mode) batch_in_progress = true;
}
}
if (copts->echo_on)
{
printf("%s\n", input_string);
}
if (copts->batch_mode || auto_next_linkage_test(test))
{
label = strip_off_label(input_string);
}
// Post-processing-based pruning will clip away connectors
// that we might otherwise want to examine. So disable PP
// pruning in this situation.
if (copts->display_bad)
parse_options_set_perform_pp_prune(opts, false);
else
parse_options_set_perform_pp_prune(opts, true);
sent = sentence_create(input_string, dict);
if (sentence_split(sent, opts) < 0)
{
sentence_delete(sent);
sent = NULL;
continue;
}
if (0 != copts->display_wordgraph)
{
const char *wg_display_flags = ""; /* default flags */
switch (copts->display_wordgraph)
{
case 1: /* default flags */
break;
case 2: /* subgraphs with a legend */
wg_display_flags = "sl";
break;
case 3:
{
/* Use esoteric flags from the test user variable. */
const char *s = test_enabled(test, "wg");
if ((NULL != s) && (':' == s[0])) wg_display_flags = s;
}
break;
default:
prt_error("Warning: wordgraph=%d: Unknown value, using 1\n",
copts->display_wordgraph);
copts->display_wordgraph = 1;
}
sentence_display_wordgraph(sent, wg_display_flags);
}
/* First parse with the default disjunct_cost as set by the library
* (currently 2.7). Usually parse here with no null links.
* However, if "-test=one-step-parse" is used and we are said to
* parse with null links, allow parsing here with null links too. */
bool one_step_parse = !copts->batch_mode && copts->allow_null &&
test_enabled(test, "one-step-parse");
int max_null_count = one_step_parse ? sentence_length(sent) : 0;
parse_options_set_min_null_count(opts, 0);
parse_options_set_max_null_count(opts, max_null_count);
parse_options_reset_resources(opts);
num_linkages = sentence_parse(sent, opts);
/* num_linkages is negative only on a hard-error;
* typically, due to a zero-length sentence. */
if (num_linkages < 0)
{
sentence_delete(sent);
sent = NULL;
continue;
}
#if 0
/* Try again, this time omitting the requirement for
* definite articles, etc. This should allow for the parsing
* of newspaper headlines and other clipped speech.
*
* XXX Unfortunately, this also allows for the parsing of
* all sorts of ungrammatical sentences which should not
* parse, and leads to bad parses of many other unparsable
* but otherwise grammatical sentences. Thus, this trick
* pretty much fails; we leave it here to document the
* experiment.
*/
if (num_linkages == 0)
{
parse_options_set_disjunct_cost(opts, 4.5);
num_linkages = sentence_parse(sent, opts);
if (num_linkages < 0) continue;
}
#endif /* 0 */
/* Try using a larger list of disjuncts */
/* XXX FIXME: the lg_expand_disjunct_list() routine is not
* currently a part of the public API; it should be made so,
* or this expansion idea should be abandoned... not sure which.
*/
if ((num_linkages == 0) && parse_options_get_use_cluster_disjuncts(opts))
{
int expanded;
if (verbosity > 0) fprintf(stdout, "No standard linkages, expanding disjunct set.\n");
parse_options_set_disjunct_cost(opts, 3.9);
expanded = lg_expand_disjunct_list(sent);
if (expanded)
{
num_linkages = sentence_parse(sent, opts);
}
if (0 < num_linkages) printf("Got One !!!!!!!!!!!!!!!!!\n");
}
/* If asked to show bad linkages, then show them. */
if ((num_linkages == 0) && (!copts->batch_mode))
{
if (copts->display_bad)
{
num_linkages = sentence_num_linkages_found(sent);
}
}
/* Now parse with null links */
if (!one_step_parse && num_linkages == 0 && !copts->batch_mode)
{
if (verbosity > 0) fprintf(stdout, "No complete linkages found.\n");
if (copts->allow_null)
{
/* XXX should use expanded disjunct list here too */
parse_options_set_min_null_count(opts, 1);
parse_options_set_max_null_count(opts, sentence_length(sent));
num_linkages = sentence_parse(sent, opts);
}
}
if (verbosity > 0)
{
if (parse_options_timer_expired(opts))
fprintf(stdout, "Timer is expired!\n");
if (parse_options_memory_exhausted(opts))
fprintf(stdout, "Memory is exhausted!\n");
}
if ((num_linkages == 0) &&
copts->panic_mode &&
parse_options_resources_exhausted(opts))
{
/* print_total_time(opts); */
batch_errors++;
if (verbosity > 0) fprintf(stdout, "Entering \"panic\" mode...\n");
/* If the parser used was the SAT solver, set the panic parser to
* it too.
* FIXME? Currently, the SAT solver code is not too useful in
* panic mode since it doesn't handle parsing with null words, so
* using the regular parser in that case could be beneficial.
* However, this currently causes a crash due to a memory
* management mess. */
parse_options_set_use_sat_parser(copts->panic_opts,
parse_options_get_use_sat_parser(opts));
parse_options_reset_resources(copts->panic_opts);
parse_options_set_verbosity(copts->panic_opts, verbosity);
(void)sentence_parse(sent, copts->panic_opts);
if (verbosity > 0)
{
if (parse_options_timer_expired(copts->panic_opts))
fprintf(stdout, "Panic timer is expired!\n");
}
}
if (verbosity > 1) parse_options_print_total_time(opts);
const char *rc = "";
if (copts->batch_mode)
{
batch_process_some_linkages(label, sent, copts);
}
else
{
rc = process_some_linkages(input_fh, sent, copts);
}
fflush(stdout);
sentence_delete(sent);
sent = NULL;
if ((NULL == rc) && (input_fh == stdin)) break;
}
if (copts->batch_mode)
{
/* print_time(opts, "Total"); */
fprintf(stderr,
"%d error%s.\n", batch_errors, (batch_errors==1) ? "" : "s");
}
/* Free stuff, so that mem-leak detectors don't complain. */
command_options_delete(copts);
dictionary_delete(dict);
printf ("Bye.\n");
return 0;
}
SPOTriplets NLP::sentence2triplets ( const char* sentence )
{
// vector of triplets
SPOTriplets triplets;
#ifdef DEBUG
std::cout << "The sentence: " << sentence << std::endl;
#endif
// creates a Sentence from the input char*
Sentence sent = sentence_create ( sentence, dict_ );
#ifdef DEBUG
std::cout << "Sentence created" << std::endl;
#endif
// tokenizes the sentence
sentence_split ( sent, parse_opts_ );
#ifdef DEBUG
std::cout << "Sentence splitted" << std::endl;
#endif
// searches for all possible linkages
int num_linkages = sentence_parse ( sent, parse_opts_ );
#ifdef DEBUG
std::cout << "Sentence parsed" << std::endl;
std::cout << "Number of linkages: " << num_linkages << std::endl;
#endif
// just one triplet
SPOTriplet triplet;
// if there is any linkage in the sentence
if( num_linkages > 0 )
{
// create the linkage
Linkage linkage = linkage_create ( 0, sent, parse_opts_ );
#ifdef DEBUG
// prints the sentence's diagram
std::cout << "The diagram: " << std::endl;
char *diagram = linkage_print_diagram(linkage, true, 800);
std::cout << diagram << std::endl;
linkage_free_diagram( diagram );
// end print diagram
#endif
std::vector<std::string> labels;
// 1. find the S_link
// S* except there is an SJ* because then S* except Spx
// two cases: there is SJ* and there is not SJ*
// TODO: VJlp VJrp same as SJ but to predications
// TODO: SFut SFst what the f**k? ###FIXED###
// TODO: His form was shining like the light not working ###FIXED###
// TODO: Car is mine not working ###FIXED###
// TODO: The little brown bear has eaten all of the honey not working ###FIXED###
// REGEXES
std::regex SJ_( "SJ.*" );
std::regex VJ_( "VJ.*");
std::regex subject( "(Ss.*)|(SFut)|(Sp\*.*)" );
std::regex Spx( "Spx.*" );
// TODO:fix theese initializer list not allowed ###FIXED###
std::regex predicate( "(Pv.*)|(Pg.*)|(PP.*)|(I.*)|(TO)|(MVi.*)" );
// TODO: make one from theese // (Sp.*)|(Ss.*) ###FIXED###
std::regex noun_adject_object ( "(O.*)|(Os.*)|(Op.*)|(MVpn.*)|(Pa.*)|(MVa.*)" );
std::regex preposition ( "(MVp.*)|(Pp.*)|(OF)|(TO)" );
std::regex prep_object ( "(J.*)|(TI)|(I.*)|(ON)" );
// TODO: problems with matching!! Pg*!! ###FIXED###
// TODO: problems with matching!! Mvp.*!! ###FIXED###
bool s_found = false;
bool p_found = false;
bool o_found = false;
bool SJ = false;
// search for SJ.s labels
for( auto label: labels )
{
if( std::regex_match( label, SJ_ ) )
{
SJ = true;
break;
}
}
// multiple subject in the sentence
if( SJ )
{
// SPls left -> first subject
// SPrs right -> second subject
// Spx right -> predicate
// SJ-s are multiple subjects
std::string temp;
// go through every linkage
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// get their label
std::string l = linkage_get_link_label( linkage, i );
// if there is an SJl* label
if( std::regex_match( l, std::regex( "SJl.*" ) ) )
{
// SJls left side
triplet.s = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
triplet.cut( triplet.s );
temp = triplet.s + " ";
// and word
triplet.s = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
triplet.cut( triplet.s );
temp += triplet.s + " ";
// find SJr*
for( int j = 0; j < linkage_get_num_links( linkage ); ++j )
{
std::string m = linkage_get_link_label( linkage, j );
if( std::regex_match( m, std::regex( "SJr.*" ) ) )
{
triplet.s = linkage_get_word( linkage, linkage_get_link_rword( linkage, j ) );
triplet.cut();
temp += triplet.s;
triplet.s = temp;
s_found = true;
#ifdef DEBUG
std::cout << "Subject found: " << triplet.s << std::endl;
#endif
break;
} // if
} // for
break;
} // if
} // for
// now we have the subject
// find Spx and its right side will be the starter predicate
std::string current_word;
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
std::string l = linkage_get_link_label( linkage, i );
if( std::regex_match( l, std::regex( "Spx.*" ) ) )
{
triplet.p = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
current_word = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
}
}
// from now all the same as on the else branch !!!!
bool predicate_match = false;
// search for the linkage that has triplet.s as left!
do
{
predicate_match = false;
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// every linkage's left word
std::string word_i = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
// every linkage's label
std::string l = linkage_get_link_label( linkage, i );
if( std::regex_match( l, predicate ) && word_i == current_word )
{
// found predicate
triplet.p = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
current_word = triplet.p;
predicate_match = true;
break;
}
}
}
while( predicate_match );
// we now have the predicate too
// TODO: multiple predicates!
p_found = true;
#ifdef DEBUG
std::cout << "Predicate found: " << triplet.p << std::endl;
#endif
// ###COPY BEGIN###
// search for noun object or adjective object
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// get every linkage label
std::string l = linkage_get_link_label( linkage, i );
// get the left word of every linkage
std::string l_word = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
// if thete is a label that match AND its left word is the predicate
if( std::regex_match( l, noun_adject_object ) && triplet.p == l_word )
{
// then the object is that linkage's right word
triplet.o = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
triplet.cut( triplet.o );
o_found = true;
#ifdef DEBUG
std::cout << "Adjective or noun object found: " << triplet.o << std::endl;
#endif
} // if
} // for
// still not found object, then search for preposition
if( !o_found )
{
// go through every linkage
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// get the linkage's label
std::string l = linkage_get_link_label( linkage, i );
// and left word
std::string word_i = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
// if there is a linkage which is a preposition and its left word is the predicate
if( std::regex_match( l, preposition ) && triplet.p == word_i )
{
// found preposition
// search for prep_object
// then the temp will contain the preposition label's right word
std::string temp = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
#ifdef DEBUG
std::cout << "Preposition found! and its rigth word is: " << temp << std::endl;
#endif
for( int j = 0; j < linkage_get_num_links( linkage ); ++j )
{
// every linkages
std::string m = linkage_get_link_label( linkage, j );
// every left word
std::string word_j = linkage_get_word( linkage, linkage_get_link_lword( linkage, j ) );
// if there is a label with match and its left is exactly the preposition's right
if( std::regex_match( m, prep_object ) && temp == word_j )
{
triplet.o = linkage_get_word( linkage, linkage_get_link_lword( linkage, j ) );
triplet.cut(triplet.o);
triplet.o += " ";
// save o
std::string temp = triplet.o;
triplet.o = linkage_get_word( linkage, linkage_get_link_rword( linkage, j ) );
triplet.cut(triplet.o);
temp += triplet.o;
triplet.o = temp;
o_found = true;
#ifdef DEBUG
std::cout << "Object found: " << triplet.o << std::endl;
#endif
} // if( std::regex_match( m, prep_object ) && temp == word_j ) END
} // for J END
} // if( std::regex_match( l, preposition ) && triplet.p == word_i ) END
} // for I END
} // if( !o_found ) END
if( s_found && p_found && o_found )
{
// TODO: cut the words itself not the whole triplet
// have to cut every word itself
// triplet.cut();
triplet.cut(triplet.s);
triplet.cut(triplet.p);
triplets.push_back( triplet );
s_found = false;
p_found = false;
o_found = false;
}
// ###COPY END###
}
else // only one subject
{
// except Spx!!!
// S left -> subject
// S right -> predicate at first
// if the word next to S right, is an element of Pv*, Pg* PP*, I*, TO, MVi*
// then the new predicate will be that word
std::string current_word;
// search for subject (S_link)
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// get the linkage's label
std::string l = linkage_get_link_label( linkage, i );
if( std::regex_match( l, subject ) )
{
// subject found
triplet.s = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
s_found = true;
current_word = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
triplet.p = current_word;
#ifdef DEBUG
std::cout << "Subject found: " << triplet.s << std::endl;
#endif
break;
}
}
if( s_found )
{
bool predicate_match = false;
// search for the linkage that has triplet.s as left!
do
{
predicate_match = false;
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// every linkage's left word
std::string l_word = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
// every linkage's label
std::string l = linkage_get_link_label( linkage, i );
if( std::regex_match( l, predicate ) && l_word == current_word )
{
// found predicate
triplet.p = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
current_word = triplet.p;
predicate_match = true;
break;
}
} // for END
} while( predicate_match );
p_found = true;
#ifdef DEBUG
std::cout << "Predicate found: " << triplet.p << std::endl;
#endif
} // if( s_found ) END
// subject and predicate found
// search for object
// from k to linkage_get_num_links( linkage )
// if there is any of the noun, adjective od preposition object then that
// label's right will give the object.
// !!! search only between labels that has triplet.p as left word !!!!!
// search for noun object or adjective objects
// go through all links
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// get every linkage label
std::string l = linkage_get_link_label( linkage, i );
// get the left word of every linkage
std::string word_i = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
// if thete is a label that match AND its left word is the predicate
if( std::regex_match( l, noun_adject_object ) && triplet.p == word_i )
{
// then the object is that linkage's right word
triplet.o = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
o_found = true;
triplet.cut(triplet.o);
#ifdef DEBUG
std::cout << "Adjective or noun object found: " << triplet.o << std::endl;
#endif
} // if END
} // for END
// still not found object, then search for preposition
if( !o_found )
{
// go through every linkage
for( int i = 0; i < linkage_get_num_links( linkage ); ++i )
{
// get the linkage's label
std::string l = linkage_get_link_label( linkage, i );
// and left word
std::string word_i = linkage_get_word( linkage, linkage_get_link_lword( linkage, i ) );
// if there is a linkage which is a preposition and its left word is the predicate
if( std::regex_match( l, preposition ) && triplet.p == word_i )
{
// found preposition
// search for prep_object
// then the temp will contain the preposition label's right word
std::string temp = linkage_get_word( linkage, linkage_get_link_rword( linkage, i ) );
#ifdef DEBUG
std::cout << "Preposition found! and its rigth word is: " << temp << std::endl;
#endif
// start search from there
for( int j = 0; j < linkage_get_num_links( linkage ); ++j )
{
// every linkages
std::string m = linkage_get_link_label( linkage, j );
// every left word
std::string word_j = linkage_get_word( linkage, linkage_get_link_lword( linkage, j ) );
#ifdef DEBUG
if( std::regex_match( m, prep_object ) )
std::cout << m << " DOES match to (J.*)|(TI)|(I.*)|(ON)" << std::endl;
#endif
// if there is a label with match and its left is exactly the preposition's right
if( std::regex_match( m, prep_object ) && temp == word_j )
{
triplet.o = linkage_get_word( linkage, linkage_get_link_lword( linkage, j ) );
triplet.cut(triplet.o);
triplet.o += " ";
// save o
std::string temp = triplet.o;
triplet.o = linkage_get_word( linkage, linkage_get_link_rword( linkage, j ) );
triplet.cut(triplet.o);
temp += triplet.o;
triplet.o = temp;
#ifdef DEBUG
std::cout << "Object found: " << triplet.o << std::endl;
#endif
o_found = true;
}
} // for
} // if
} // for
} // if( o_found ) END
if( s_found && p_found && o_found )
{
// TODO: cut the words itself not the whole triplet ###FIXED###
// have to cut every word itself
// triplet.cut();
triplet.cut(triplet.s);
triplet.cut(triplet.p);
triplets.push_back( triplet );
s_found = false;
p_found = false;
o_found = false;
}
} // end else
linkage_delete ( linkage );
} // if( num_linkages > 0 ) END
// **************************************************************************
// HTTPヘッダを受信して解析する。
//
// 処理するのはGETのみ。GET以外のメソッドが来たらエラー
// 今のところ、URIとuser_agent、Range、Hostを解析。
// URIは、URIデコードもやる。
//
// return: 0 正常終了
// return: 0以外 エラー
// **************************************************************************
static int http_header_receive(int accept_socket, HTTP_RECV_INFO *http_recv_info_p)
{
int result = 0;
int recv_len;
char line_buf[1024]; // 大きめに。
char work_buf[1024];
char work_buf2[1024];
char split1[1024];
char split2[1024];
int ret;
int i;
// ================================
// １行づつ HTTPヘッダを受信
// ================================
for (i=0;;i++){
// 1行受信 実行。
recv_len = line_receive(accept_socket, line_buf, sizeof(line_buf));
// 受信した内容をチェック。
if ( recv_len == 0 ){ // 空行検知。ヘッダ受信終了。
break;
}else if ( recv_len < 0 ){ // 受信失敗
return ( -1 );
}
// debug. 受信したヘッダ表示
debug_log_output("'%s'(%d byte)\n", line_buf, recv_len );
// --------------------------
// GETメッセージチェック
// --------------------------
if ( i == 0 ){ // １行目のみチェック
debug_log_output("%d:URI Check start.'%s'\n", accept_socket,line_buf);
// GETある？
if ( strstr(line_buf, "GET") != NULL ){
http_recv_info_p->isGet = 1;
}else if ( strstr(line_buf, "HEAD") != NULL ){
http_recv_info_p->isGet = 2;
}else if ( strstr(line_buf, "POST") != NULL ){
http_recv_info_p->isGet = 3;
}else{
debug_log_output("'GET' not found. error.%d",accept_socket);
return ( -1 );
}
// 最初のスペースまでを削除。
cut_before_character(line_buf, ' ');
// 次にスペースが出てきたところの後ろまでを削除。
cut_after_character(line_buf, ' ');
// ===========================
// GETオプション部解析
// ===========================
// REQUEST_URI用・Proxy用に値を保存
strncpy(http_recv_info_p->request_uri, line_buf, sizeof(http_recv_info_p->request_uri));
// '?'が存在するかチェック。
if ( strchr(line_buf, '?') != NULL ){
strncpy(work_buf, line_buf, sizeof(work_buf));
// '?'より前をカット
cut_before_character(work_buf, '?' );
while ( 1 ){
memset(split1, 0, sizeof(split1));
memset(split2, 0, sizeof(split2));
// 最初に登場する'&'で分割
ret = sentence_split(work_buf, '&', split1, split2 );
if ( ret == 0 ){ // 分割成功
strncpy(work_buf, split2, sizeof(work_buf));
}else if (strlen(work_buf) > 0){ // まだwork_bufに中身ある？
strncpy( split1, work_buf, sizeof(split1));
strncpy( work_buf, "", sizeof(work_buf));
}else{ // 処理終了
break;
}
// -------------------------------------
// GETした内容 解析開始
// 超安直。いいのかこんな比較で。
// -------------------------------------
// URIデコード
uri_decode(work_buf2, sizeof(work_buf2), split1, sizeof(split1) );
// "action="あるか調査。
if (strncasecmp( work_buf2, "action=", strlen("action=") ) == 0 ){
// = より前を削除
cut_before_character(work_buf2, '=');
// 構造体に値を保存。
strncpy(http_recv_info_p->action, work_buf2, sizeof(http_recv_info_p->action));
continue;
}
}
}
debug_log_output("http_recv_info_p->action = '%s'", http_recv_info_p->action);
// URIデコード
cut_after_character(line_buf, '?');
uri_decode(work_buf, sizeof(work_buf), line_buf, sizeof(line_buf) );
strncpy(line_buf, work_buf, sizeof(line_buf));
debug_log_output("URI(decoded):'%s'\n", line_buf);
convert_language_code(line_buf, work_buf, sizeof(work_buf), CODE_AUTO, CODE_EUC);
debug_log_output("URI(decoded,euc,FYI):'%s'\n", work_buf);
// 構造体に保存
strncpy(http_recv_info_p->recv_uri, line_buf, sizeof(http_recv_info_p->recv_uri));
//httpから始まってる場合には、http://以降の最初の'/'の前でカット
if( strncmp(http_recv_info_p->recv_uri,"http://",7)==0){
char* ptr = strstr(http_recv_info_p->recv_uri+7,"/");
if( ptr ){
strcpy(http_recv_info_p->recv_uri,ptr);
}
}
continue;
}
// User-agent切り出し
if ( strncasecmp(line_buf, HTTP_USER_AGENT, strlen(HTTP_USER_AGENT) ) == 0 ){
debug_log_output("User-agent: Detect.\n");
// ':'より前を切る
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
// 構造体に保存
strncpy( http_recv_info_p->user_agent, line_buf, sizeof(http_recv_info_p->user_agent));
continue;
}
// Rangeあるかチェック
if ( strncasecmp(line_buf, HTTP_RANGE, strlen(HTTP_RANGE) ) == 0 ){
debug_log_output("%s Detect.\n", HTTP_RANGE);
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
// recv_range にRangeの中身保存
strncpy(http_recv_info_p->recv_range, line_buf, sizeof(http_recv_info_p->recv_range));
// '=' より前を切る
cut_before_character(line_buf, '=');
// '-'で前後に分割。
sentence_split(line_buf, '-', work_buf, work_buf2);
debug_log_output("wrok_buf='%s'\n", work_buf);
debug_log_output("wrok_buf2='%s'\n", work_buf2);
// 値を文字列→数値変換
http_recv_info_p->range_start_pos = strtoull(work_buf, NULL, 10);
if ( strlen(work_buf2) > 0 ){
http_recv_info_p->range_end_pos = strtoull(work_buf2, NULL, 10);
}
debug_log_output("range_start_pos=%d\n", http_recv_info_p->range_start_pos);
debug_log_output("range_end_pos=%d\n", http_recv_info_p->range_end_pos);
continue;
}
// Hostあるかチェック
if ( strncasecmp(line_buf, HTTP_HOST, strlen(HTTP_HOST) ) == 0 ){
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
strncpy(http_recv_info_p->recv_host, line_buf, sizeof(http_recv_info_p->recv_host));
debug_log_output("%s Detect. %s '%s'", HTTP_HOST, HTTP_HOST, http_recv_info_p->recv_host);
continue;
}
// Content-Lengthあるかチェック
if ( strncasecmp(line_buf, HTTP_CONTENT_LENGTH1, strlen(HTTP_CONTENT_LENGTH1) ) == 0 ){
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
strncpy(http_recv_info_p->content_length, line_buf, sizeof(http_recv_info_p->content_length));
debug_log_output("%s Detect. %s '%s'", HTTP_CONTENT_LENGTH1, HTTP_CONTENT_LENGTH1, http_recv_info_p->content_length);
continue;
}
// Content-TYPEあるかチェック
if ( strncasecmp(line_buf, HTTP_CONTENT_TYPE1, strlen(HTTP_CONTENT_TYPE1) ) == 0 ){
// ':' より前を切る。
cut_before_character(line_buf, ':');
cut_first_character(line_buf, ' ');
strncpy(http_recv_info_p->content_type, line_buf, sizeof(http_recv_info_p->content_type));
debug_log_output("%s Detect. %s '%s'", HTTP_CONTENT_TYPE1, HTTP_CONTENT_TYPE1, http_recv_info_p->content_type);
continue;
}
}
return result;
}
