void put_local_vars_in_opts(Parse_Options opts) {
parse_options_set_verbosity(opts, local.verbosity);
parse_options_set_max_parse_time(opts, local.timeout);
parse_options_set_max_memory(opts, local.memory);
parse_options_set_linkage_limit(opts, local.linkage_limit);
parse_options_set_null_block(opts, local.null_block);
parse_options_set_islands_ok(opts, local.islands_ok);
parse_options_set_short_length(opts, local.short_length);
parse_options_set_echo_on(opts, local.echo_on);
parse_options_set_batch_mode(opts, local.batch_mode);
parse_options_set_panic_mode(opts, local.panic_mode);
parse_options_set_screen_width(opts, local.screen_width);
parse_options_set_allow_null(opts, local.allow_null);
parse_options_set_screen_width(opts, local.screen_width);
parse_options_set_display_on(opts, local.display_on);
parse_options_set_display_postscript(opts, local.display_postscript);
parse_options_set_display_constituents(opts, local.display_constituents);
parse_options_set_max_sentence_length(opts, local.max_sentence_length);
parse_options_set_display_bad(opts, local.display_bad);
parse_options_set_display_links(opts, local.display_links);
parse_options_set_display_walls(opts, local.display_walls);
parse_options_set_display_union(opts, local.display_union);
}
static void put_local_vars_in_opts(Parse_Options opts)
{
parse_options_set_verbosity(opts, local.verbosity);
parse_options_set_max_parse_time(opts, local.timeout);
parse_options_set_max_memory(opts, local.memory);
parse_options_set_linkage_limit(opts, local.linkage_limit);
parse_options_set_null_block(opts, local.null_block);
parse_options_set_islands_ok(opts, local.islands_ok);
parse_options_set_spell_guess(opts, local.spell_guess);
parse_options_set_short_length(opts, local.short_length);
parse_options_set_echo_on(opts, local.echo_on);
parse_options_set_cost_model_type(opts, local.cost_model);
parse_options_set_disjunct_costf(opts, local.max_cost);
parse_options_set_batch_mode(opts, local.batch_mode);
parse_options_set_panic_mode(opts, local.panic_mode);
parse_options_set_screen_width(opts, local.screen_width);
parse_options_set_allow_null(opts, local.allow_null);
parse_options_set_use_cluster_disjuncts(opts, local.use_cluster_disjuncts);
#ifdef USE_FAT_LINKAGES
parse_options_set_use_fat_links(opts, local.use_fat_links);
parse_options_set_display_union(opts, local.display_union);
#endif /* USE_FAT_LINKAGES */
#ifdef USE_SAT_SOLVER
parse_options_set_use_sat_parser(opts, local.use_sat_solver);
#endif
parse_options_set_use_viterbi(opts, local.use_viterbi);
parse_options_set_screen_width(opts, local.screen_width);
parse_options_set_display_on(opts, local.display_on);
parse_options_set_display_postscript(opts, local.display_postscript);
parse_options_set_display_constituents(opts, local.display_constituents);
parse_options_set_max_sentence_length(opts, local.max_sentence_length);
parse_options_set_display_bad(opts, local.display_bad);
parse_options_set_display_disjuncts(opts, local.display_disjuncts);
parse_options_set_display_links(opts, local.display_links);
parse_options_set_display_senses(opts, local.display_senses);
parse_options_set_display_walls(opts, local.display_walls);
}
int main(int argc, char * argv[])
{
FILE *input_fh = stdin;
Dictionary dict;
Sentence sent;
const char *language="en"; /* default to english, and not locale */
int pp_on=TRUE;
int af_on=TRUE;
int cons_on=TRUE;
int num_linkages, i;
char *input_string;
Label label = NO_LABEL;
const char *codeset;
/*############################################################################################*/
int num=0;
FILE *fp2,*fp_word_info,*fp_num,*fp_lname,*fp_rel,*fp_lcount,*fp_word_cat,*fp_word;
/*##############################################################################################*/
i = 1;
/*############################################################################################*/
if (argc > 4){
num =atoi(argv[4]);
num=num-1;
/// language = argv[1];
/*####################################################################################*/
//
// if ((argc > 1) && (argv[1][0] != '-')) {
/* the dictionary is the first argument if it doesn't begin with "-" */
// language = argv[1];
i++;
}
/*################################*/
language = argv[1];
/*################################*/
/* Get the locale from the environment...
* perhaps we should someday get it from the dictionary ??
*/
setlocale(LC_ALL, "");
/* Check to make sure the current locale is UTF8; if its not,
* then force-set this to the english utf8 locale
*/
codeset = nl_langinfo(CODESET);
if (!strstr(codeset, "UTF") && !strstr(codeset, "utf"))
{
setlocale(LC_CTYPE, "en_US.UTF-8");
}
/*########################################*/
for (; i<argc-3; i++) {
/*########################################*/
if (argv[i][0] == '-') {
if (strcmp("--version", argv[i])==0) {
printf("Version: %s\n", linkgrammar_get_version());
exit(0);
} else if (strcmp("-ppoff", argv[i])==0) {
pp_on = FALSE;
} else if (strcmp("-coff", argv[i])==0) {
cons_on = FALSE;
} else if (strcmp("-aoff", argv[i])==0) {
af_on = FALSE;
} else if (strcmp("-batch", argv[i])==0) {
} else if (strncmp("-!", argv[i],2)==0) {
} else {
print_usage(argv[0]);
}
} else {
print_usage(argv[0]);
}
}
opts = parse_options_create();
if (opts == NULL) {
fprintf(stderr, "%s: Fatal error: unable to create parse options\n", argv[0]);
exit(-1);
}
panic_parse_opts = parse_options_create();
if (panic_parse_opts == NULL) {
fprintf(stderr, "%s: Fatal error: unable to create panic parse options\n", argv[0]);
exit(-1);
}
setup_panic_parse_options(panic_parse_opts);
parse_options_set_max_sentence_length(opts, 170);
parse_options_set_panic_mode(opts, TRUE);
parse_options_set_max_parse_time(opts, 30);
parse_options_set_linkage_limit(opts, 1000);
parse_options_set_short_length(opts, 10);
/*##########################################*/
// parse_options_set_display_on(opts, TRUE);
/*##########################################*/
if(language && *language)
dict = dictionary_create_lang(language);
else
dict = dictionary_create_default_lang();
if (dict == NULL) {
/*###########################################################################*/
fprintf(stderr, "%s: Fatal error: Unable to open dictionary.\n", argv[1]);
/*##########################################################################*/
exit(-1);
}
/* process the command line like commands */
for (i=1; i<argc; i++) {
if ((strcmp("-pp", argv[i])==0) ||
(strcmp("-c", argv[i])==0) ||
(strcmp("-a", argv[i])==0))
{
i++;
}
else if ((argv[i][0] == '-') && (strcmp("-ppoff", argv[i])!=0) &&
(argv[i][0] == '-') && (strcmp("-coff", argv[i])!=0) &&
(argv[i][0] == '-') && (strcmp("-aoff", argv[i])!=0))
{
if (argv[i][1] == '!')
issue_special_command(argv[i]+2, opts, dict);
else
issue_special_command(argv[i]+1, opts, dict);
}
}
verbosity = parse_options_get_verbosity(opts);
/* Main input loop */
while (1)
{
input_string = fget_input_string(input_fh, stdout, opts);
if (NULL == input_string)
{
if (input_fh == stdin) break;
fclose (input_fh);
input_fh = stdin;
continue;
}
if ((strcmp(input_string, "quit\n")==0) ||
(strcmp(input_string, "exit\n")==0)) break;
/* We have to handle the !file command inline; its too hairy
* otherwise ... */
if (strncmp(input_string, "!file", 5) == 0)
{
char * filename = &input_string[6];
input_fh = fopen(filename, "r");
if (NULL == input_fh)
{
int perr = errno;
fprintf(stderr, "Error: %s (%d) %s\n",
filename, perr, strerror(perr));
input_fh = stdin;
continue;
}
continue;
}
if (special_command(input_string, dict)) continue;
if (parse_options_get_echo_on(opts)) {
printf("%s", input_string);
}
if (parse_options_get_batch_mode(opts)) {
label = strip_off_label(input_string);
}
sent = sentence_create(input_string, dict);
if (sent == NULL) continue;
if (sentence_length(sent) > parse_options_get_max_sentence_length(opts)) {
if (verbosity > 0) {
fprintf(stdout,
"Sentence length (%d words) exceeds maximum allowable (%d words)\n",
sentence_length(sent), parse_options_get_max_sentence_length(opts));
}
sentence_delete(sent);
continue;
}
/* First parse with cost 0 or 1 and no null links */
parse_options_set_disjunct_cost(opts, 2);
parse_options_set_min_null_count(opts, 0);
parse_options_set_max_null_count(opts, 0);
parse_options_reset_resources(opts);
num_linkages = sentence_parse(sent, opts);
//############################################################################################
if(num+1>num_linkages && num_linkages != 0)
{
sprintf(link_info_filename,"%s/%s_tmp/linkid_cat.txt",argv[2],argv[3]);
fp2 = fopen(link_info_filename,"a");
if(fp2==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp2,";~~~~~~~~~~\n");
fclose(fp2);
sprintf(link_info_filename,"%s/%s_tmp/linkid_word.txt",argv[2],argv[3]);
fp_word= fopen(link_info_filename, "a");
if(fp_word==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp_word,";~~~~~~~~~~\n");
fclose(fp_word);
sprintf(link_info_filename,"%s/%s_tmp/link_numeric_word.txt",argv[2],argv[3]);
fp_word_info= fopen(link_info_filename, "a");
if(fp_word_info==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp_word_info,";~~~~~~~~~~\n");
fclose(fp_word_info);
sprintf(link_info_filename,"%s/%s_tmp/link_name_expand.txt",argv[2],argv[3]);
fp_lname =fopen(link_info_filename,"a");
if(fp_lname==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp_lname,";~~~~~~~~~~\n");
fclose(fp_lname);
sprintf(link_info_filename,"%s/%s_tmp/link_relation_info.txt",argv[2],argv[3]);
fp_rel = fopen(link_info_filename,"a");
if(fp_rel==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp_rel,";~~~~~~~~~~\n");
fclose(fp_rel);
/* sprintf(link_info_filename,"%s/%s_tmp/constituents.txt",argv[2],argv[3]);
fp = fopen(link_info_filename,"a");
if(fp==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp,"\n");
fprintf(fp,";~~~~~~~~~~\n");
fclose(fp);*/
}
sprintf(link_info_filename,"%s/%s_tmp/linkage_count.txt",argv[2],argv[3]);
fp_lcount = fopen(link_info_filename,"a");
if(fp_lcount==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp_lcount, "(Found maximum of %d linkages )\n",num_linkages);
fprintf(fp_lcount,";~~~~~~~~~~\n");
fclose(fp_lcount);
//############################################################################################
if (num_linkages < 0) continue;
/* Now parse with null links */
if ((num_linkages == 0) && (!parse_options_get_batch_mode(opts))) {
if (verbosity > 0) //fprintf(stdout, "No complete linkages found.\n");
//############################################################################################
/* This part of the code written by Maha Laxmi and Shirisha Manju
* if no complete linkage is found then redirect the output to standard output in clips format*/
sprintf(link_info_filename,"%s/%s_tmp/number.txt",argv[2],argv[3]);
fp_num =fopen(link_info_filename,"a");
if(fp_num==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
/* sprintf(link_info_filename,"%s/%s_tmp/constituents.txt",argv[2],argv[3]);
fp_cons= fopen(link_info_filename,"a");
if(fp_cons==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}*/
sprintf(link_info_filename,"%s/%s_tmp/linkid_word.txt",argv[2],argv[3]);
fp_word= fopen(link_info_filename, "a");
if(fp_word==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
sprintf(link_info_filename,"%s/%s_tmp/link_numeric_word.txt",argv[2],argv[3]);
fp_word_info= fopen(link_info_filename, "a");
if(fp_word_info==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
sprintf(link_info_filename,"%s/%s_tmp/linkid_cat.txt",argv[2],argv[3]);
fp_word_cat= fopen(link_info_filename, "a");
if(fp_word_cat==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
sprintf(link_info_filename,"%s/%s_tmp/link_name_expand.txt",argv[2],argv[3]);
fp_lname =fopen(link_info_filename,"a");
if(fp_lname==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
sprintf(link_info_filename,"%s/%s_tmp/link_relation_info.txt",argv[2],argv[3]);
fp_rel = fopen(link_info_filename,"a");
if(fp_rel==NULL) {printf("Could not open %s for writing\n",link_info_filename);exit(1);}
fprintf(fp_word, "\n(No complete linkages found)\n");
fprintf(fp_word_cat, "\n(No complete linkages found)\n");
fprintf(fp_word_info, "\n(No complete linkages found)\n");
fprintf(fp_lname, "\n(No complete linkages found)\n");
fprintf(fp_rel,"\n(No complete linkages found)\n");
fprintf(fp_word_cat,";~~~~~~~~~~\n"); // fprintf(fp_cons,";~~~~~~~~~~\n");
fprintf(fp_lname,";~~~~~~~~~~\n"); fprintf(fp_rel,";~~~~~~~~~~\n");
fprintf(fp_word_info,";~~~~~~~~~~\n"); fprintf(fp_word,";~~~~~~~~~~\n");
/* fclose(fp_cons); */ fclose(fp_word_cat); fclose(fp_word_info);
fclose(fp_lname); fclose(fp_rel); fclose(fp_word);
continue;
/*###############################################################################################################*/
if (parse_options_get_allow_null(opts)) {
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 (parse_options_timer_expired(opts)) {
if (verbosity > 0) fprintf(stdout, "Timer is expired!\n");
}
if (parse_options_memory_exhausted(opts)) {
if (verbosity > 0) fprintf(stdout, "Memory is exhausted!\n");
}
if ((num_linkages == 0) &&
parse_options_resources_exhausted(opts) &&
parse_options_get_panic_mode(opts)) {
/* print_total_time(opts); */
if (verbosity > 0) fprintf(stdout, "Entering \"panic\" mode...\n");
parse_options_reset_resources(panic_parse_opts);
parse_options_set_verbosity(panic_parse_opts, verbosity);
num_linkages = sentence_parse(sent, panic_parse_opts);
if (parse_options_timer_expired(panic_parse_opts)) {
if (verbosity > 0) fprintf(stdout, "Timer is expired!\n");
}
}
/* print_total_time(opts); */
if (parse_options_get_batch_mode(opts)) {
//#######################################################################
batch_process_some_linkages(label, sent, opts,argv[2],argv[3]);
//#########################################################################
// batch_process_some_linkages(label, sent, opts);
}
/*###############################################################################*/
// send num as an argument to process_some_linkages
else {
int c = process_some_linkages(sent, opts,num,argv[2],argv[3]);
if (c == EOF) break;
}
/*################################################################################*/
/* else {
int c = process_some_linkages(sent, opts);
if (c == EOF) break;
}*/
sentence_delete(sent);
}
if (parse_options_get_batch_mode(opts)) {
/* print_time(opts, "Total"); */
/* fprintf(stderr,
"%d error%s.\n", batch_errors, (batch_errors==1) ? "" : "s");*/
}
parse_options_delete(panic_parse_opts);
parse_options_delete(opts);
dictionary_delete(dict);
// printf ("Bye.\n");
return 0;
}
int main(int argc, char * argv[]) {
Dictionary dict;
Sentence sent;
char *dictionary_file=NULL;
char *post_process_knowledge_file=NULL;
char *constituent_knowledge_file=NULL;
char *affix_file=NULL;
int pp_on=TRUE;
int af_on=TRUE;
int cons_on=TRUE;
int num_linkages, i;
char input_string[MAXINPUT];
Label label = NO_LABEL;
int parsing_space_leaked, reported_leak, dictionary_and_option_space;
i = 1;
if ((argc > 1) && (argv[1][0] != '-')) {
/* the dictionary is the first argument if it doesn't begin with "-" */
dictionary_file = argv[1];
i++;
}
for (; i<argc; i++) {
if (argv[i][0] == '-') {
if (strcmp("-pp", argv[i])==0) {
if ((post_process_knowledge_file != NULL) || (i+1 == argc))
print_usage(argv[0]);
post_process_knowledge_file = argv[i+1];
i++;
} else
if (strcmp("-c", argv[i])==0) {
if ((constituent_knowledge_file != NULL) || (i+1 == argc))
print_usage(argv[0]);
constituent_knowledge_file = argv[i+1];
i++;
} else
if (strcmp("-a", argv[i])==0) {
if ((affix_file != NULL) || (i+1 == argc)) print_usage(argv[0]);
affix_file = argv[i+1];
i++;
} else if (strcmp("-ppoff", argv[i])==0) {
pp_on = FALSE;
} else if (strcmp("-coff", argv[i])==0) {
cons_on = FALSE;
} else if (strcmp("-aoff", argv[i])==0) {
af_on = FALSE;
} else if (strcmp("-batch", argv[i])==0) {
} else if (strncmp("-!", argv[i],2)==0) {
} else {
print_usage(argv[0]);
}
} else {
print_usage(argv[0]);
}
}
if (!pp_on && post_process_knowledge_file != NULL) print_usage(argv[0]);
if (dictionary_file == NULL) {
dictionary_file = "4.0.dict";
fprintf(stderr, "No dictionary file specified. Using %s.\n",
dictionary_file);
}
if (af_on && affix_file == NULL) {
affix_file = "4.0.affix";
fprintf(stderr, "No affix file specified. Using %s.\n", affix_file);
}
if (pp_on && post_process_knowledge_file == NULL) {
post_process_knowledge_file = "4.0.knowledge";
fprintf(stderr, "No post process knowledge file specified. Using %s.\n",
post_process_knowledge_file);
}
if (cons_on && constituent_knowledge_file == NULL) {
constituent_knowledge_file = "4.0.constituent-knowledge";
fprintf(stderr, "No constituent knowledge file specified. Using %s.\n",
constituent_knowledge_file);
}
opts = parse_options_create();
if (opts == NULL) {
fprintf(stderr, "%s\n", lperrmsg);
exit(-1);
}
panic_parse_opts = parse_options_create();
if (panic_parse_opts == NULL) {
fprintf(stderr, "%s\n", lperrmsg);
exit(-1);
}
setup_panic_parse_options(panic_parse_opts);
parse_options_set_max_sentence_length(opts, 70);
parse_options_set_panic_mode(opts, TRUE);
parse_options_set_max_parse_time(opts, 30);
parse_options_set_linkage_limit(opts, 1000);
parse_options_set_short_length(opts, 10);
dict = dictionary_create(dictionary_file,
post_process_knowledge_file,
constituent_knowledge_file,
affix_file);
if (dict == NULL) {
fprintf(stderr, "%s\n", lperrmsg);
exit(-1);
}
/* process the command line like commands */
for (i=1; i<argc; i++) {
if ((strcmp("-pp", argv[i])==0) ||
(strcmp("-c", argv[i])==0) ||
(strcmp("-a", argv[i])==0)) {
i++;
} else if ((argv[i][0] == '-') && (strcmp("-ppoff", argv[i])!=0) &&
(argv[i][0] == '-') && (strcmp("-coff", argv[i])!=0) &&
(argv[i][0] == '-') && (strcmp("-aoff", argv[i])!=0)) {
issue_special_command(argv[i]+1, opts, dict);
}
}
dictionary_and_option_space = space_in_use;
reported_leak = external_space_in_use = 0;
verbosity = parse_options_get_verbosity(opts);
while (fget_input_string(input_string, stdin, stdout, opts)) {
if (space_in_use != dictionary_and_option_space + reported_leak) {
fprintf(stderr, "Warning: %d bytes of space leaked.\n",
space_in_use-dictionary_and_option_space-reported_leak);
reported_leak = space_in_use - dictionary_and_option_space;
}
if ((strcmp(input_string, "quit\n")==0) ||
(strcmp(input_string, "exit\n")==0)) break;
if (special_command(input_string, dict)) continue;
if (parse_options_get_echo_on(opts)) {
printf("%s", input_string);
}
if (parse_options_get_batch_mode(opts)) {
label = strip_off_label(input_string);
}
sent = sentence_create(input_string, dict);
if (sent == NULL) {
if (verbosity > 0) fprintf(stderr, "%s\n", lperrmsg);
if (lperrno != NOTINDICT) exit(-1);
else continue;
}
if (sentence_length(sent) > parse_options_get_max_sentence_length(opts)) {
sentence_delete(sent);
if (verbosity > 0) {
fprintf(stdout,
"Sentence length (%d words) exceeds maximum allowable (%d words)\n",
sentence_length(sent), parse_options_get_max_sentence_length(opts));
}
continue;
}
/* First parse with cost 0 or 1 and no null links */
parse_options_set_disjunct_cost(opts, 2);
parse_options_set_min_null_count(opts, 0);
parse_options_set_max_null_count(opts, 0);
parse_options_reset_resources(opts);
num_linkages = sentence_parse(sent, opts);
/* Now parse with null links */
if ((num_linkages == 0) && (!parse_options_get_batch_mode(opts))) {
if (verbosity > 0) fprintf(stdout, "No complete linkages found.\n");
if (parse_options_get_allow_null(opts)) {
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 (parse_options_timer_expired(opts)) {
if (verbosity > 0) fprintf(stdout, "Timer is expired!\n");
}
if (parse_options_memory_exhausted(opts)) {
if (verbosity > 0) fprintf(stdout, "Memory is exhausted!\n");
}
if ((num_linkages == 0) &&
parse_options_resources_exhausted(opts) &&
parse_options_get_panic_mode(opts)) {
print_total_time(opts);
if (verbosity > 0) fprintf(stdout, "Entering \"panic\" mode...\n");
parse_options_reset_resources(panic_parse_opts);
parse_options_set_verbosity(panic_parse_opts, verbosity);
num_linkages = sentence_parse(sent, panic_parse_opts);
if (parse_options_timer_expired(panic_parse_opts)) {
if (verbosity > 0) fprintf(stdout, "Timer is expired!\n");
}
}
print_total_time(opts);
if (parse_options_get_batch_mode(opts)) {
batch_process_some_linkages(label, sent, opts);
}
else {
process_some_linkages(sent, opts);
}
sentence_delete(sent);
if (external_space_in_use != 0) {
fprintf(stderr, "Warning: %d bytes of external space leaked.\n",
external_space_in_use);
}
}
if (parse_options_get_batch_mode(opts)) {
print_time(opts, "Total");
fprintf(stderr,
"%d error%s.\n", batch_errors, (batch_errors==1) ? "" : "s");
}
parsing_space_leaked = space_in_use - dictionary_and_option_space;
if (parsing_space_leaked != 0) {
fprintf(stderr, "Warning: %d bytes of space leaked during parsing.\n",
parsing_space_leaked);
}
parse_options_delete(panic_parse_opts);
parse_options_delete(opts);
dictionary_delete(dict);
if (space_in_use != parsing_space_leaked) {
fprintf(stderr,
"Warning: %d bytes of dictionary and option space leaked.\n",
space_in_use - parsing_space_leaked);
}
else if (parsing_space_leaked == 0) {
fprintf(stderr, "Good news: no space leaked.\n");
}
if (external_space_in_use != 0) {
fprintf(stderr, "Warning: %d bytes of external space leaked.\n",
external_space_in_use);
}
return 0;
}