void print_args(args_t* args) {
int i, j;
printf("\nSTATES: %d/%d\n", args->num_fstates, args->num_states);
for (i = 0; i < args->num_states; i++) {
if (is_final_state(args, i)) {
printf("(%s), ", args->states[i]);
}
else {
printf("%s , ", args->states[i]);
}
}
printf("\b\b \n");
printf("\nSYMBOLS: %d\n", args->num_symbols);
for (i = 0; i < args->num_symbols; i++) {
printf("%c , ", args->symbols[i]);
}
printf("\b\b \n");
printf("\nTRANSITIONS:\n");
for (i = 0; i < args->num_states; i++) {
for (j = 0; j < args->num_symbols; j++) {
if (args->transitions[args->num_states * i + j] >= 0) {
printf("%03d ", args->transitions[args->num_states * i + j]);
}
else {
printf("--- ");
}
}
printf("\n");
}
printf("\n");
}
/**
* We create a copy of the given graph using the following rules if full_simplification is not null:
* - <E> transitions and graph calls are kept
* - all right contexts are ignored, replaced by an epsilon transition
* - all tags that don't match anything in the text (like $* $< and $>) are kept,
* because they can be involved into a E loop. We also add a real E transition.
* - all other transitions that matches something from the text are removed
*
* As a consequence, the resulting graph is only made of real E transitions,
* pseudo-E transitions, and graph calls and we can use it as follows:
* - if no final state is accessible, it means that the graph cannot match E
* - if the initial state is final, it means that the graph match E
* - otherwise, we don't know yet
*
*
* If full_simplification is null, we have to create a condition graph suitable for
* E loop and left recursion detection. For that purpose, we keep the graph as is,
* with only one modification: adding an E transition to skip right contexts. But still,
* we keep the context, because we also have to look at it for E loops and left recursions.
*
*/
static SingleGraph create_condition_graph(Fst2* fst2,int graph,int full_simplification) {
SingleGraph g=new_SingleGraph(INT_TAGS);
int initial_state=fst2->initial_states[graph];
int n_states=fst2->number_of_states_per_graphs[graph];
for (int i=initial_state;i<initial_state+n_states;i++) {
SingleGraphState dst=add_state(g);
Fst2State src=fst2->states[i];
if (is_initial_state(src)) {
set_initial_state(dst);
}
if (is_final_state(src)) {
set_final_state(dst);
}
Transition* t=src->transitions;
while (t!=NULL) {
if (full_simplification) {
deal_with_transition_v1(fst2,t,dst,initial_state);
} else {
deal_with_transition_v2(fst2,t,dst,initial_state);
}
t=t->next;
}
}
clean_condition_graph(g);
return g;
}
/**
* Replaces the given automaton by its complement one.
*/
void elag_complementation(language_t* language,SingleGraph A) {
int sink_state_index=A->number_of_states;
SingleGraphState sink_state=add_state(A);
/* The sink state is not final (because finalities will be reversed
* below), and its default transition loops back on itself */
sink_state->default_state=sink_state_index;
for (int q=0;q<A->number_of_states;q++) {
/* We reverse the finality of each state */
if (is_final_state(A->states[q])) {
unset_final_state(A->states[q]);
} else {
set_final_state(A->states[q]);
}
if (A->states[q]->default_state==-1) {
/* If there is no default transition, we create one that is
* tagged by anything but the non default ones */
symbol_t* s=LEXIC_minus_transitions(language,A->states[q]->outgoing_transitions);
if (s!=NULL) {
add_all_outgoing_transitions(A->states[q],s,sink_state_index);
/* We have added a single transition tagged by a symbol list. Now
* we replace it by a list of transitions, each one of them
* tagged with a single symbol */
flatten_transition(A->states[q]->outgoing_transitions);
/* Important to use free_symbols and not free_symbol, because
* s represents a symbol list */
free_symbols(s);
}
}
}
}
/**
* A debug printing of a SingleGraph.
*/
void print_graph(SingleGraph g) {
u_printf("--------------------------------\n");
if (g==NULL) {
u_printf("NULL graph\n");
u_printf("--------------------------------\n");
return;
}
for (int i=0;i<g->number_of_states;i++) {
SingleGraphState s=g->states[i];
if (is_initial_state(s)) {
u_printf("-> ");
} else {u_printf(" ");}
if (is_final_state(s)) {
u_printf("%d t ",i);
} else {
u_printf("%d : ",i);
}
u_printf("(def %d) \n\t",s->default_state);
Transition* t=s->outgoing_transitions;
while (t!=NULL) {
symbols_dump((symbol_t*)t->label);
u_printf(",%d ",t->state_number);
t=t->next;
}
u_printf("\n\n");
}
u_printf("--------------------------------\n");
}
// simulate NTM until it halts or nr_steps steps are reached
void NTM::simulate(uint max_nr_steps, bool verbose)
{
uint c_symbol;
uint c_state;
while(nr_steps <= max_nr_steps) {
list<TM_State> new_tm_states;
if ( verbose ) {
cout << "Nr steps: " << nr_steps << endl;
cout << "There are " << tm_states.size() << " states" << endl;
}
for (list<TM_State>::const_iterator it = tm_states.begin();
it != tm_states.end(); ++it) {
c_symbol = it->tape.read();
c_state = it->state;
if ( !is_final_state(c_state) ) {
for(list<Action>::const_iterator act_it =
trans_actions[c_state][c_symbol].begin();
act_it != trans_actions[c_state][c_symbol].end();
++act_it) {
TM_State tms(*it);
tms.state = act_it->state;
tms.tape.write(act_it->symbol);
tms.tape.move(act_it->move);
new_tm_states.push_front(tms);
// if(verbose) {
// print(tms);
// }
}
}
}
nr_steps++;
swap(tm_states, new_tm_states);
}
cout << "Done simulating!" << endl;
}
// Ex. 14
automaton* read_automaton(FILE* fptr) {
int state_c;
int symbol_c;
int initial_state;
int final_state_c;
int* final_states;
s_table_element* symbol_table;
state* state_table;
char line[MAX_LEN+1];
if(fseek(fptr, 0, SEEK_SET)) {
puts("Couldn't seek the start of the file!");
return NULL;
}
// Reading the number of states, number of symbols, and initial state
fgets(line, (int)MAX_LEN, fptr);
sscanf(line, "%d %d %d", &state_c, &symbol_c, &initial_state);
// Constructing the table of final states
final_states = read_final_states(fptr, &final_state_c);
// Constructing the symbol table
symbol_table = read_symbols(symbol_c, fptr);
// Constructing the state table
state_table = malloc(sizeof(state)*state_c);
if(state_table == NULL) {
puts("Couldn't allocate the state table");
return NULL;
}
int type;
for(int i = 0; i < state_c; ++i) {
if((i+1) == initial_state)
type = 1;
else if(is_final_state(final_states, final_state_c, i+1))
type = -1;
else
type = 0;
state_table[i] = new_state(i+1, type, NULL);
}
while(read_transition(state_table, fptr));
automaton* ret = malloc(sizeof(automaton));
if(ret == NULL)
return NULL;
ret->initial_state = initial_state;
ret->states = state_c;
ret->symbol_c = symbol_c;
ret->state_table = state_table;
ret->symbol_table = symbol_table;
ret->final_states = final_states;
ret->final_state_c = final_state_c;
return ret;
}
/**
* This function concatenates B at the end of A. A is modified.
*/
void elag_concat(language_t* language,SingleGraph A,SingleGraph B) {
int oldnb=A->number_of_states;
int* renumber=(int*)malloc(B->number_of_states*sizeof(int));
if (renumber==NULL) {
fatal_alloc_error("elag_concat");
}
int q;
/* We copy the states of B into A */
for (q=0;q<B->number_of_states;q++) {
renumber[q]=A->number_of_states;
add_state(A);
}
for (q=0;q<B->number_of_states;q++) {
A->states[renumber[q]]->outgoing_transitions=clone_transition_list(B->states[q]->outgoing_transitions,renumber,dup_symbol);
A->states[renumber[q]]->default_state=(B->states[q]->default_state!=-1)?renumber[B->states[q]->default_state]:-1;
if (is_final_state(B->states[q])) {
set_final_state(A->states[renumber[q]]);
}
}
/* Then, we concatenate A and B.
* 1) We replace default transitions that outgo from B's initial states
* by explicit transitions */
struct list_int* initials=get_initial_states(B);
for (struct list_int* tmp=initials;tmp!=NULL;tmp=tmp->next) {
explicit_default_transition(language,A,renumber[tmp->n]);
}
for (q=0;q<oldnb;q++) {
if (is_final_state(A->states[q])) {
/* Each final state of A becomes non final. Moreover, we have
* to explicit its default transition, because if not, the concatenation
* algorithm will modify the recognized language. */
unset_final_state(A->states[q]);
explicit_default_transition(language,A,q);
for (struct list_int* tmp=initials;tmp!=NULL;tmp=tmp->next) {
concat(&(A->states[q]->outgoing_transitions),clone_transition_list(A->states[renumber[tmp->n]]->outgoing_transitions,NULL,dup_symbol));
if (is_final_state(A->states[renumber[tmp->n]])) {
set_final_state(A->states[q]);
}
}
}
}
free(renumber);
free_list_int(initials);
}
//
// this function explores a sub-graph, considering tokens as strings
//
void explorer_sub_automate_normalization_string(Fst2* automate,int n,
struct normalization_tree* noeud_normalization,
unichar* output,struct norm_info** TEMP_LIST) {
Fst2State etat;
etat=automate->states[n];
if (is_final_state(etat)) {
// if we are in a final state
(*TEMP_LIST)=insert_in_norm_info_list(output,noeud_normalization,(*TEMP_LIST));
}
Transition* trans;
trans=etat->transitions;
unichar tmp[1000];
while (trans!=NULL) {
if (trans->tag_number<0) {
// case of a sub-graph
struct norm_info* TMP=NULL;
explorer_sub_automate_normalization_string(automate,automate->initial_states[-(trans->tag_number)],noeud_normalization,
output,&TMP);
while (TMP!=NULL) {
// we continue to explore the current automaton
explorer_sub_automate_normalization_string(automate,trans->state_number,TMP->node,
TMP->output,TEMP_LIST);
struct norm_info* z=TMP;
TMP=TMP->next;
free_norm_info(z);
}
}
else {
// normal transition
Fst2Tag etiq;
etiq=automate->tags[trans->tag_number];
u_strcpy(tmp,output);
u_strcat(tmp," ");
if (etiq->output!=NULL && u_strcmp(etiq->output,"")
&& u_strcmp(etiq->output,"<E>") && !only_spaces(etiq->output)) {
// we append the output if it exists and is not epsilon
u_strcat(tmp,etiq->output);
}
struct normalization_tree_transition* trans_norm;
trans_norm=get_trans_arbre_normalization_string(etiq->input,noeud_normalization->trans);
if (trans_norm==NULL) {
// if the transition does not exist in the tree, we create it
trans_norm=new_trans_arbre_normalization_string(etiq->input);
// we also create the destination node
trans_norm->node=new_normalization_tree();
trans_norm->next=noeud_normalization->trans;
noeud_normalization->trans=trans_norm;
}
explorer_sub_automate_normalization_string(automate,trans->state_number,trans_norm->node,
tmp,TEMP_LIST);
}
trans=trans->next;
}
}
/**
* Allocates, initializes and returns an array that associates
* a color (0 or 1) to each state of 'A', making sure that the
* state #0 will be colored with 0. '*nbColors' will be set to
* the number of colors that have been used (1 if all states
* have the same finality; 2 otherwise).
*/
int* init_colors(SingleGraph A,int *nbColors) {
int* color=(int*)calloc(A->number_of_states,sizeof(int));
if (color==NULL) {
fatal_alloc_error("init_colors");
}
/* bicolor will indicate if all states are of the same color (finality) or
* not */
bool bicolor=false;
if (is_final_state(A->states[0])) {
/* We distinguish two cases (initial state final or not), just
* to ensure that the color of the initial state #0 will be 0 */
for (int e=0;e<A->number_of_states;e++) {
color[e]=is_final_state(A->states[e])?0:(bicolor=true,1);
}
} else {
for (int e=0;e<A->number_of_states;e++) {
color[e]=is_final_state(A->states[e])?(bicolor=true,1):0;
}
}
(*nbColors)=(bicolor?2:1);
return color;
}
/**
* An Elag constraints is of the form: <=> .... <=> .... <=>
* This function looks for states that are pointed to by middle '<=>' transitions.
* It places their numbers into 'constraints' and it returns the size of this
* array, i.e. the number of constraints expressed by this Elag rule.
*/
int count_constraints(Fst2Automaton* aut,int* constraints) {
int source=0;
int e;
Transition* t;
int c;
int nbConstraints=0;
symbol_t* symbol;
SingleGraph automaton=aut->automaton;
for (t=automaton->states[0]->outgoing_transitions;t!=NULL && source==0;t=t->next) {
symbol=t->label;
if (symbol->type==S_EQUAL) {
if (t->state_number==0) {
fatal_error("Illegal cycle in grammar\n");
}
source=t->tag_number;
}
}
if (source==0) {
/* If there are no contraints */
return 0;
}
/* We look for '<=>' transitions, but only from the state 1, because
* from state 0, we would take into account all the '<=>' that begin rules. */
for (e=1;e<automaton->number_of_states;e++) {
for (t=automaton->states[e]->outgoing_transitions;t!=NULL;t=t->next) {
symbol=t->label;
if (t->state_number!=source && symbol->type==S_EQUAL && !is_final_state(automaton->states[t->state_number])) {
/* We don't take into account '<=>' transitions that go to final states because
* they are not middle '<=>' transitions. */
for (c=0;c<nbConstraints;c++) {
if (constraints[c]==t->state_number) {
/* We stop if the constraint is already in the list */
break;
}
}
if (c==nbConstraints) {
if (++nbConstraints>=ELAG_MAX_CONSTRAINTS) {
fatal_error("Too many constraints with same condition\n");
}
constraints[c]=t->state_number;
}
}
}
}
if (nbConstraints==0) {
fatal_error("Middle delimitor '<=>' not found\n");
}
return nbConstraints;
}
static E_MATCHING_STATUS get_status(SingleGraph g) {
if (g->number_of_states==0) {
return CHK_DOES_NOT_MATCH_E;
}
int i=get_initial_state(g);
if (i<0) {
fatal_error("Internal error in get_status: invalid negative initial state %d\n",i);
}
SingleGraphState s=g->states[i];
if (is_final_state(s)) {
/* If the initial state is final, it means that we can reach it without matching
* anything in the text */
return CHK_MATCHES_E;
}
return CHK_DONT_KNOW;
}
/**
* Creates a SingleGraph copy of the given .fst2 subgraph, using
* the same tag numeration.
*/
SingleGraph create_copy_of_fst2_subgraph(Fst2* fst2,int n) {
int n_states=fst2->number_of_states_per_graphs[n];
SingleGraph g=new_SingleGraph(n_states,INT_TAGS);
int shift=fst2->initial_states[n];
for (int i=0;i<n_states;i++) {
SingleGraphState dest=add_state(g);
Fst2State src=fst2->states[i+shift];
if (is_initial_state(src)) {
set_initial_state(dest);
}
if (is_final_state(src)) {
set_final_state(dest);
}
Transition* t=src->transitions;
while (t!=NULL) {
add_outgoing_transition(dest,t->tag_number,t->state_number);
t=t->next;
}
}
return g;
}
/**
* Returns 1 if we can match <E> from the current state, with or without
* conditions; 0 otherwise.
*/
int graph_matches_E(int initial_state,int current_state,const Fst2State* states,Fst2Tag* tags,
int current_graph,unichar** graph_names,
ConditionList conditions_for_states[],
ConditionList *graph_conditions) {
Transition* l;
Fst2State s;
int ret_value=0;
int ret;
*graph_conditions=NULL;
s=states[current_state];
if (is_final_state(s)) {
/* If we are arrived in a final state, then the graph matches <E> */
set_bit_mask(&(s->control),UNCONDITIONAL_E_MATCH);
return 1;
}
if (is_bit_mask_set(s->control,TMP_LOOP_MARK)) {
/* If we have a loop, we do nothing, because they will be
* dealt with later. */
return 0;
}
if (is_bit_mask_set(s->control,VISITED_MARK)) {
/* If we are in state that has already been visited */
if (is_bit_mask_set(s->control,UNCONDITIONAL_E_MATCH)) {
/* If this state can match <E> without conditions, then we have finished */
return 1;
}
if (is_bit_mask_set(s->control,CONDITIONAL_E_MATCH)) {
/* If this state can match <E> with conditions, then we have finished, but
* we copy the necessary conditions in 'graph_conditions'. */
*graph_conditions=clone_ConditionList(conditions_for_states[current_state-initial_state]);
return 1;
}
/* If the state has been visited and if it does not match <E>, then we return OK */
return 0;
}
set_bit_mask(&(s->control),VISITED_MARK);
set_bit_mask(&(s->control),TMP_LOOP_MARK);
l=s->transitions;
/* We look all the outgoing transitions */
while (l!=NULL) {
if (l->tag_number<0) {
/* If we have a subgraph, we test if it matches <E> */
*graph_conditions=NULL;
ret=graph_matches_E(initial_state,l->state_number,states,tags,current_graph,graph_names,conditions_for_states,graph_conditions);
if (ret==1) {
/* If the subgraph matches <E>, we say that the current state matches
* <E>, modulo the conditions to be verified */
set_bit_mask(&(s->control),CONDITIONAL_E_MATCH);
/* We insert the new condition in first position... */
insert_graph_in_conditions(-(l->tag_number),graph_conditions);
/* ...and we merge the new conditions with the existing ones for this state */
merge_condition_lists(&conditions_for_states[current_state-initial_state],*graph_conditions);
*graph_conditions=NULL;
}
ret_value=ret_value|ret;
}
else if (tags[l->tag_number]->control&1) {
/* If we have an <E> transition, we explore the rest of the graph from it */
*graph_conditions=NULL;
ret=graph_matches_E(initial_state,l->state_number,states,tags,current_graph,graph_names,conditions_for_states,graph_conditions);
if (ret==1) {
/* If we can match <E> from the <E>-transition's destination state, then
* we can match it from the current state. */
if (*graph_conditions==NULL) {
/* If there is no condition */
set_bit_mask(&(s->control),UNCONDITIONAL_E_MATCH);
}
else {
/* Otherwise, we add the condition to the existing ones */
set_bit_mask(&(s->control),CONDITIONAL_E_MATCH);
merge_condition_lists(&conditions_for_states[current_state-initial_state],*graph_conditions);
*graph_conditions=NULL;
}
}
ret_value=ret_value|ret;
}
l=l->next;
}
unset_bit_mask(&(s->control),TMP_LOOP_MARK);
*graph_conditions=clone_ConditionList(conditions_for_states[current_state-initial_state]);
return ret_value;
}
/**
* This function analyzes the given Elag rule automaton to find
* where the rule and constraint parts are. As a side effect, it builds
* a fst2 grammar ("foo.fst2" => "foo-conc.fst2") that can be used by
* the Locate program to match the <!> .... <!> .... <!> part of the rule.
*/
void split_elag_rule(elRule* rule, const VersatileEncodingConfig* vec,language_t* language) {
int c;
/* This array contains the numbers of the states that are pointed to by
* middle '<=>' of the constraints */
int constraints[ELAG_MAX_CONSTRAINTS];
int nbConstraints=count_constraints(rule->automaton,constraints);
/* +1 because we have to count the <!> .... <!> .... <!> part of the rule */
rule->nbContexts=nbConstraints+1;
rule->contexts=(elContext*)malloc(rule->nbContexts*sizeof(elContext));
if (rule->contexts==NULL) {
fatal_alloc_error("split_elag_rule");
}
for (c=0;c<rule->nbContexts;c++) {
rule->contexts[c].left=NULL;
rule->contexts[c].right=NULL;
}
int endR1=ELAG_UNDEFINED;
int endR2=ELAG_UNDEFINED;
int endC2=ELAG_UNDEFINED;
for (Transition* t=rule->automaton->automaton->states[0]->outgoing_transitions;t!=NULL;t=t->next) {
symbol_t* symbol=t->label;
switch (symbol->type) {
/* We split the unique <!> .... <!> .... <!> part */
case S_EXCLAM:
if (rule->contexts[0].left!=NULL) {
fatal_error("Too much '<!>' tags\n",rule->name);
}
rule->contexts[0].left=new_SingleGraph(PTR_TAGS);
/* We look for the end of the first part of the rule */
endR1=get_sub_automaton(rule->automaton->automaton,rule->contexts[0].left,t->state_number,0,S_EXCLAM);
rule->contexts[0].right=new_SingleGraph(PTR_TAGS);
endR2=get_sub_automaton(rule->automaton->automaton,rule->contexts[0].right,endR1,0,S_EXCLAM);
if (endR1==ELAG_UNDEFINED || endR2==ELAG_UNDEFINED
|| !is_final_state(rule->automaton->automaton->states[endR2])) {
fatal_error("split_elag_rule: %s: parse error in <!> part\n",rule->name);
}
break;
/* We split the nbConstraints <=> .... <=> .... <=> parts */
case S_EQUAL:
if (rule->contexts[1].left!=NULL) {
fatal_error("Non deterministic .fst2 file\n");
}
for (c=0;c<nbConstraints;c++) {
rule->contexts[c+1].left=new_SingleGraph(PTR_TAGS);
get_sub_automaton(rule->automaton->automaton,rule->contexts[c+1].left,t->state_number,1,constraints[c]);
rule->contexts[c+1].right=new_SingleGraph(PTR_TAGS);
endC2=get_sub_automaton(rule->automaton->automaton,rule->contexts[c+1].right,constraints[c],0,S_EQUAL);
if (endC2==ELAG_UNDEFINED || !is_final_state(rule->automaton->automaton->states[endC2])) {
fatal_error("split_elag_rule: %s: parse error in <=> part\n",rule->name);
}
}
break;
default: fatal_error("Left delimitor '<!>' or '<=>' missing\n");
}
}
if (rule->contexts[0].left==NULL) {
fatal_error("In grammar '%s': symbol '<!>' not found.\n",rule->name);
}
char buf[FILENAME_MAX];
remove_extension(rule->name,buf);
strcat(buf,"-conc.fst2");
/* We create the.fst2 to be used by Locate */
Fst2Automaton* locate=make_locate_automaton(rule,language);
save_automaton(locate,buf,vec,FST_LOCATE);
free_Fst2Automaton(locate,free_symbol);
}
void make_with_func(args_t* args, char* outfile){
int i, j, count;
if (get_args("test.auto", args)) {
}
else {
printf("Falha ao obter args\n");
}
//cria arquivo
FILE * arq = fopen(outfile, "w");
//insere includes
fprintf(arq, "#include <stdio.h>\n");
fprintf(arq, "#include <stdlib.h>\n");
fprintf(arq, "#define TAM 10\n");
fprintf(arq, "\n");
//variaveis globais
fprintf(arq, "char v[TAM];\n");
fprintf(arq, "int p = 0;\n");
fprintf(arq, "\n");
//protitipos das funcões
for(i = 0; i < args->num_states; i++){
fprintf(arq, "void e%d();\n", i);
}
fprintf(arq, "void sucesso();\n");
fprintf(arq, "void erro();\n");
fprintf(arq, "\n");
//main
fprintf(arq, "int main(void) {\n");
fprintf(arq, " p = 0;\n");
fprintf(arq, " printf(\"Sequencia:\"); \n");
fprintf(arq, " fgets(v, TAM, stdin);\n");
fprintf(arq, " e0();\n");
fprintf(arq, " return(0);\n");
fprintf(arq, "}\n");
fprintf(arq, "\n");
//funcoes dos estados
for (i = 0; i < args->num_states; i++) {
fprintf(arq, "\nvoid %s(){\n", args->states[i]);
count = 0;
for (j = 0; j < args->num_symbols; j++) {
if (args.transitions[i + j * args->num_states] != -1) {
if (count > 0){
fprintf(arq, " else \n");
fprintf(arq, " if(v[p] == '%c'){ \n", args->symbols[j]);
}
else{
fprintf(arq, " if(v[p] == '%c'){ \n", args->symbols[j]);
}
fprintf(arq, " p++;\n");
fprintf(arq, " e%d();\n", args->transitions[i + j * args->num_states]);
fprintf(arq, " }");
count++;
}
}
if (count > 0) {
fprintf(arq, " else {\n");
}
if (is_final_state(&args, i)){
fprintf(arq, " sucesso();\n");
fprintf(arq, " }\n");
fprintf(arq, "}\n");
}
else {
fprintf(arq, " erro();\n");
fprintf(arq, " }\n");
fprintf(arq, "}\n");
}
}
fprintf(arq, "\n");
//cria funcao sucesso
fprintf(arq, "void sucesso(){\n");
fprintf(arq, " printf(\" Sucesso!\");\n");
fprintf(arq, "}\n");
fprintf(arq, "\n");
//cria funcao erro
fprintf(arq, "void erro(){\n");
fprintf(arq, " printf(\" Erro!\");\n");
fprintf(arq, "}\n");
fprintf(arq, "\n");
//fecha arquivo
fclose(arq);
}
/**
* Saves the given automaton into the given .fst2 file.
*/
void fst_file_write(Elag_fst_file_out* fstf,const Fst2Automaton* A) {
Ustring* tag=new_Ustring();
void (*symbol_to_tag)(const symbol_t*,Ustring*)=NULL;
switch (fstf->type) {
case FST_TEXT:
symbol_to_tag=symbol_to_text_label;
break;
case FST_GRAMMAR:
symbol_to_tag=symbol_to_grammar_label;
break;
case FST_LOCATE:
symbol_to_tag=symbol_to_locate_label;
break;
default:
fatal_error("fst_file_write: invalid fstf->type: %d\n",fstf->type);
}
/* We save the graph number and name */
u_fprintf(fstf->f,"-%d %S\n",fstf->nb_automata+1,A->name);
int index;
unichar deflabel[]={'<','d','e','f','>',0};
for (int q=0;q<A->automaton->number_of_states;q++) {
SingleGraphState state=A->automaton->states[q];
u_fprintf(fstf->f,"%C ",is_final_state(state)?'t':':');
for (Transition* t=state->outgoing_transitions;t!=NULL;t=t->next) {
if (t->tag_number==-1) {
/* If we are in the case of an "EMPTY" transition created because
* the automaton was emptied as trim time */
u_strcpy(tag,"EMPTY");
} else {
symbol_t* symbol=t->label;
symbol_to_tag(symbol,tag);
}
if (fstf->type==FST_LOCATE) {
/* If we are saving a Locate .fst2, we have to perform
* some special things */
if (u_strcmp(tag->str, "<PNC>") == 0) {
PNC_trans_write(fstf, t->state_number);
} else if (u_strcmp(tag->str, "<CHFA>") == 0 || u_strcmp(tag->str, "<NB>") == 0) {
CHFA_trans_write(fstf, t->state_number);
} else if (u_strcmp(tag->str, "<.>") == 0) {
LEXIC_trans_write(fstf, t->state_number);
} else {
goto normal_output;
}
} else {
/* If we have a normal transition to print */
normal_output:
index=get_value_index(tag->str,fstf->labels);
u_fprintf(fstf->f,"%d %d ",index,t->state_number);
}
}
if (state->default_state!=-1) {
if (fstf->type!=FST_GRAMMAR) {
error("Unexpected <def> label in text/locate automaton\n");
}
index=get_value_index(deflabel,fstf->labels);
u_fprintf(fstf->f,"%d %d ",index,state->default_state);
}
u_fputc('\n',fstf->f);
}
u_fprintf(fstf->f,"f \n");
free_Ustring(tag);
fstf->nb_automata++;
}
/**
* This function explore the normalization grammar to construct
* the normalization tree. If the 'list' parameter is NULL, then we
* are in the main call to the main graph; otherwise, we are within
* a subgraph.
*/
void explore_normalization_fst2(Fst2* fst2,int current_state,
struct normalization_tree* node,
struct string_hash* tokens,const unichar* output,
const Alphabet* alph,struct norm_info** list) {
Fst2State state=fst2->states[current_state];
if (is_final_state(state)) {
/* If we are in a final state, we behave differently if we are in a subgraph
* or in the main call to the main graph. */
if (list!=NULL) {
(*list)=insert_in_norm_info_list(output,node,(*list));
}
else {
node->outputs=sorted_insert(output,node->outputs);
}
}
Transition* trans=state->transitions;
unichar tmp[1024];
while (trans!=NULL) {
if (trans->tag_number<0) {
/* Case of a subgraph call */
struct norm_info* tmp_list=NULL;
explore_normalization_fst2(fst2,fst2->initial_states[-(trans->tag_number)],node,
tokens,output,alph,&tmp_list);
while (tmp_list!=NULL) {
/* We continue to explore the current graph */
explore_normalization_fst2(fst2,trans->state_number,tmp_list->node,
tokens,tmp_list->output,alph,list);
struct norm_info* z=tmp_list;
tmp_list=tmp_list->next;
free_norm_info(z);
}
}
else {
/* If we have a normal transition */
Fst2Tag tag=fst2->tags[trans->tag_number];
u_strcpy(tmp,output);
u_strcat(tmp," ");
if (tag->output!=NULL && tag->output[0]!='\0' && u_strcmp(tag->output,"<E>") && !only_spaces(tag->output)) {
/* We append the output if it exists and is not epsilon */
u_strcat(tmp,tag->output);
}
if (!u_strcmp(tag->input,"<E>")) {
/* If we have an epsilon transition, we go on in the fst2, but
* we don't move in the normalization tree */
explore_normalization_fst2(fst2,trans->state_number,node,tokens,tmp,alph,list);
} else {
/* If we have a normal transition, we explore all the tokens that match it */
struct list_int* l=get_token_list_for_sequence(tag->input,alph,tokens);
while (l!=NULL) {
/* Then, we add a branch in the normalization tree for
* each token. Note that it may introduce combinatory explosions
* if the the fst2 matches large sequences */
struct normalization_tree_transition* trans_norm;
trans_norm=get_transition(l->n,node->trans);
if (trans_norm==NULL) {
/* If the transition does not exist in the tree, we create it */
trans_norm=new_normalization_tree_transition(l->n,new_normalization_tree(),node->trans);
node->trans=trans_norm;
}
explore_normalization_fst2(fst2,trans->state_number,trans_norm->node,
tokens,tmp,alph,list);
struct list_int* L=l;
l=l->next;
free(L);
}
}
}
trans=trans->next;
}
}
void scan_graph(int n_graph, // number of current graph
int e, // number of current state
int pos, //
int depth,
struct parsing_info** liste_arrivee,
unichar* mot_token_buffer,
struct fst2txt_parameters* p,Abstract_allocator prv_alloc_recycle) {
Fst2State etat_courant=p->fst2->states[e];
if (depth > MAX_DEPTH) {
error( "\n"
"Maximal stack size reached in graph %i!\n"
"Recognized more than %i tokens starting from:\n"
" ",
n_graph, MAX_DEPTH);
for (int i=0; i<60; i++) {
error("%S",p->buffer[p->current_origin+i]);
}
error("\nSkipping match at this position, trying from next token!\n");
p->output[0] = '\0'; // clear output
p->input_length = 0; // reset taille_entree
empty(p->stack); // clear output stack
if (liste_arrivee != NULL) {
while (*liste_arrivee != NULL) { // free list of subgraph matches
struct parsing_info* la_tmp=*liste_arrivee;
*liste_arrivee=(*liste_arrivee)->next;
la_tmp->next=NULL; // to don't free the next item
free_parsing_info(la_tmp, prv_alloc_recycle);
}
}
return;
// exit(1); // don't exit, try at next position
}
depth++;
if (is_final_state(etat_courant)) {
// if we are in a final state
p->stack->stack[p->stack->stack_pointer+1]='\0';
if (n_graph == 0) { // in main graph
if (pos>=p->input_length/*sommet>u_strlen(output)*/) {
// and if the recognized input is longer than the current one, it replaces it
u_strcpy(p->output,p->stack->stack);
p->input_length=(pos);
}
} else { // in a subgraph
(*liste_arrivee)=insert_if_absent(pos,-1,-1,(*liste_arrivee),p->stack->stack_pointer+1,
p->stack->stack,p->variables,NULL,NULL,-1,-1,NULL,-1, prv_alloc_recycle);
}
}
if (pos+p->current_origin==p->text_buffer->size) {
// if we are at the end of the text, we return
return;
}
int SOMMET=p->stack->stack_pointer+1;
int pos2;
/* If there are some letter sequence transitions like %hello, we process them */
if (p->token_tree[e]->transition_array!=NULL) {
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
/* we don't keep this line because of problems occur in sentence tokenizing
* if the return sequence is defautly considered as a separator like space
else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
*/
else pos2=pos;
int position=0;
unichar *token=mot_token_buffer;
if (p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION
|| (is_letter(p->buffer[pos2+p->current_origin],p->alphabet) && (pos2+p->current_origin==0 || !is_letter(p->buffer[pos2+p->current_origin-1],p->alphabet)))) {
/* If we are in character by character mode */
while (pos2+p->current_origin<p->text_buffer->size && is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
token[position++]=p->buffer[(pos2++)+p->current_origin];
if (p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION) {
break;
}
}
token[position]='\0';
if (position!=0 &&
(p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION || !(is_letter(token[position-1],p->alphabet) && is_letter(p->buffer[pos2+p->current_origin],p->alphabet)))) {
// we proceed only if we have exactly read the contenu sequence
// in both modes MERGE and REPLACE, we process the transduction if any
int SOMMET2=p->stack->stack_pointer;
Transition* RES=get_matching_tags(token,p->token_tree[e],p->alphabet);
Transition* TMP;
unichar* mot_token_new_recurse_buffer=NULL;
if (RES!=NULL) {
// we allocate a new mot_token_buffer for the scan_graph recursin because we need preserve current
// token=mot_token_buffer
mot_token_new_recurse_buffer=(unichar*)malloc(MOT_BUFFER_TOKEN_SIZE*sizeof(unichar));
if (mot_token_new_recurse_buffer==NULL) {
fatal_alloc_error("scan_graph");
}
}
while (RES!=NULL) {
p->stack->stack_pointer=SOMMET2;
Fst2Tag etiq=p->fst2->tags[RES->tag_number];
traiter_transduction(p,etiq->output);
int longueur=u_strlen(etiq->input);
unichar C=token[longueur];
token[longueur]='\0';
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_input_string(p->stack,token,0);
}
token[longueur]=C;
scan_graph(n_graph,RES->state_number,pos2-(position-longueur),depth,liste_arrivee,mot_token_new_recurse_buffer,p);
TMP=RES;
RES=RES->next;
free(TMP);
}
if (mot_token_new_recurse_buffer!=NULL) {
free(mot_token_new_recurse_buffer);
}
}
}
}
Transition* t=etat_courant->transitions;
while (t!=NULL) {
p->stack->stack_pointer=SOMMET-1;
// we process the transition of the current state
int n_etiq=t->tag_number;
if (n_etiq<0) {
// case of a sub-graph
struct parsing_info* liste=NULL;
unichar* pile_old;
p->stack->stack[p->stack->stack_pointer+1]='\0';
pile_old = u_strdup(p->stack->stack);
scan_graph((((unsigned)n_etiq)-1),p->fst2->initial_states[-n_etiq],pos,depth,&liste,mot_token_buffer,p);
while (liste!=NULL) {
p->stack->stack_pointer=liste->stack_pointer-1;
u_strcpy(p->stack->stack,liste->stack);
scan_graph(n_graph,t->state_number,liste->position,depth,liste_arrivee,mot_token_buffer,p);
struct parsing_info* l_tmp=liste;
liste=liste->next;
l_tmp->next=NULL; // to don't free the next item
free_parsing_info(l_tmp, prv_alloc_recycle);
}
u_strcpy(p->stack->stack,pile_old);
free(pile_old);
p->stack->stack_pointer=SOMMET-1;
}
else {
// case of a normal tag
Fst2Tag etiq=p->fst2->tags[n_etiq];
unichar* contenu=etiq->input;
int contenu_len_possible_match=u_len_possible_match(contenu);
if (etiq->type==BEGIN_OUTPUT_VAR_TAG) {
fatal_error("Unsupported $|XXX( tags in Fst2Txt\n");
}
if (etiq->type==END_OUTPUT_VAR_TAG) {
fatal_error("Unsupported $|XXX) tags in Fst2Txt\n");
}
if (etiq->type==BEGIN_VAR_TAG) {
// case of a $a( variable tag
//int old;
struct transduction_variable* L=get_transduction_variable(p->variables,etiq->variable);
if (L==NULL) {
fatal_error("Unknown variable: %S\n",etiq->variable);
}
//old=L->start;
if (p->buffer[pos+p->current_origin]==' ' && pos+p->current_origin+1<p->text_buffer->size) {
pos2=pos+1;
if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');
}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
L->start_in_tokens=pos2;
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
//L->start=old;
}
else if (etiq->type==END_VAR_TAG) {
// case of a $a) variable tag
//int old;
struct transduction_variable* L=get_transduction_variable(p->variables,etiq->variable);
if (L==NULL) {
fatal_error("Unknown variable: %S\n",etiq->variable);
}
//old=L->end;
if (pos>0)
L->end_in_tokens=pos-1;
else L->end_in_tokens=pos;
// BUG: qd changement de buffer, penser au cas start dans ancien buffer et end dans nouveau
scan_graph(n_graph,t->state_number,pos,depth,liste_arrivee,mot_token_buffer,p);
//L->end=old;
}
else if ((contenu_len_possible_match==5) && (!u_trymatch_superfast5(contenu,ETIQ_MOT_LN5))) {
// case of transition by any sequence of letters
if (p->buffer[pos+p->current_origin]==' ' && pos+p->current_origin+1<p->text_buffer->size) {
pos2=pos+1;
if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');
}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
unichar* mot=mot_token_buffer;
int position=0;
if (p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION ||
((pos2+p->current_origin)==0 || !is_letter(p->buffer[pos2+p->current_origin-1],p->alphabet))) {
while (pos2+p->current_origin<p->text_buffer->size && is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
mot[position++]=p->buffer[(pos2++)+p->current_origin];
}
mot[position]='\0';
if (position!=0) {
// we proceed only if we have read a letter sequence
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_output_string(p->stack,mot);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
}
else if ((contenu_len_possible_match==4) && (!u_trymatch_superfast4(contenu,ETIQ_NB_LN4))) {
// case of transition by any sequence of digits
if (p->buffer[pos+p->current_origin]==' ') {
pos2=pos+1;
if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');
}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
unichar* mot=mot_token_buffer;
int position=0;
while (pos2+p->current_origin<p->text_buffer->size && (p->buffer[pos2+p->current_origin]>='0')
&& (p->buffer[pos2+p->current_origin]<='9')) {
mot[position++]=p->buffer[(pos2++)+p->current_origin];
}
mot[position]='\0';
if (position!=0) {
// we proceed only if we have read a letter sequence
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_output_string(p->stack,mot);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
else if ((contenu_len_possible_match==5) && (!u_trymatch_superfast5(contenu,ETIQ_MAJ_LN5))) {
// case of upper case letter sequence
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
unichar* mot=mot_token_buffer;
int position=0;
if (p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION ||
((pos2+p->current_origin)==0 || !is_letter(p->buffer[pos2+p->current_origin-1],p->alphabet))) {
while (pos2+p->current_origin<p->text_buffer->size && is_upper(p->buffer[pos2+p->current_origin],p->alphabet)) {
mot[position++]=p->buffer[(pos2++)+p->current_origin];
}
mot[position]='\0';
if (position!=0 && !is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
// we proceed only if we have read an upper case letter sequence
// which is not followed by a lower case letter
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_input_string(p->stack,mot,0);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
}
else if ((contenu_len_possible_match==5) && (!u_trymatch_superfast5(contenu,ETIQ_MIN_LN5))) {
// case of lower case letter sequence
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
unichar* mot=mot_token_buffer;
int position=0;
if (p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION ||
(pos2+p->current_origin==0 || !is_letter(p->buffer[pos2+p->current_origin-1],p->alphabet))) {
while (pos2+p->current_origin<p->text_buffer->size && is_lower(p->buffer[pos2+p->current_origin],p->alphabet)) {
mot[position++]=p->buffer[(pos2++)+p->current_origin];
}
mot[position]='\0';
if (position!=0 && !is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
// we proceed only if we have read a lower case letter sequence
// which is not followed by an upper case letter
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_input_string(p->stack,mot,0);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
}
else if ((contenu_len_possible_match==5) && (!u_trymatch_superfast5(contenu,ETIQ_PRE_LN5))) {
// case of a sequence beginning by an upper case letter
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
unichar* mot=mot_token_buffer;
int position=0;
if (p->tokenization_policy==CHAR_BY_CHAR_TOKENIZATION ||
(is_upper(p->buffer[pos2+p->current_origin],p->alphabet) && (pos2+p->current_origin==0 || !is_letter(p->buffer[pos2+p->current_origin-1],p->alphabet)))) {
while (pos2+p->current_origin<p->text_buffer->size && is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
mot[position++]=p->buffer[(pos2++)+p->current_origin];
}
mot[position]='\0';
if (position!=0 && !is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
// we proceed only if we have read a letter sequence
// which is not followed by a letter
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_input_string(p->stack,mot,0);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
}
else if ((contenu_len_possible_match==5) && (!u_trymatch_superfast5(contenu,ETIQ_PNC_LN5))) {
// case of a punctuation sequence
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
unichar C=p->buffer[pos2+p->current_origin];
if (C==';' || C=='!' || C=='?' ||
C==':' || C==0xbf ||
C==0xa1 || C==0x0e4f || C==0x0e5a ||
C==0x0e5b || C==0x3001 || C==0x3002 ||
C==0x30fb) {
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push(p->stack,C);
}
scan_graph(n_graph,t->state_number,pos2+1,depth,liste_arrivee,mot_token_buffer,p);
}
else {
// we consider the case of ...
// BUG: if ... appears at the end of the buffer
if (C=='.') {
if ((pos2+p->current_origin+2)<p->text_buffer->size && p->buffer[pos2+p->current_origin+1]=='.' && p->buffer[pos2+p->current_origin+2]=='.') {
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the ... we have read
push(p->stack,C);push(p->stack,C);push(p->stack,C);
}
scan_graph(n_graph,t->state_number,pos2+3,depth,liste_arrivee,mot_token_buffer,p);
} else {
// we consider the . as a normal punctuation sign
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push(p->stack,C);
}
scan_graph(n_graph,t->state_number,pos2+1,depth,liste_arrivee,mot_token_buffer,p);
}
}
}
}
else if ((contenu_len_possible_match==3) && (!u_trymatch_superfast3(contenu,ETIQ_E_LN3))) {
// case of an empty sequence
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
scan_graph(n_graph,t->state_number,pos,depth,liste_arrivee,mot_token_buffer,p);
}
else if ((contenu_len_possible_match==3) && (!u_trymatch_superfast3(contenu,ETIQ_CIRC_LN3))) {
// case of a new line sequence
if (p->buffer[pos+p->current_origin]=='\n') {
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push(p->stack,'\n');
}
scan_graph(n_graph,t->state_number,pos+1,depth,liste_arrivee,mot_token_buffer,p);
}
}
else if ((contenu_len_possible_match==1) && (!u_trymatch_superfast1(contenu,'#')) && (!(etiq->control&RESPECT_CASE_TAG_BIT_MASK))) {
// case of a no space condition
if (p->buffer[pos+p->current_origin]!=' ') {
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
scan_graph(n_graph,t->state_number,pos,depth,liste_arrivee,mot_token_buffer,p);
}
}
else if ((contenu_len_possible_match==1) && (!u_trymatch_superfast1(contenu,' '))) {
// case of an obligatory space
if (p->buffer[pos+p->current_origin]==' ') {
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push(p->stack,' ');
}
scan_graph(n_graph,t->state_number,pos+1,depth,liste_arrivee,mot_token_buffer,p);
}
}
else if ((contenu_len_possible_match==3) && (!u_trymatch_superfast5(contenu,ETIQ_L_LN3))) {
// case of a single letter
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
if (is_letter(p->buffer[pos2+p->current_origin],p->alphabet)) {
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push(p->stack,p->buffer[pos2+p->current_origin]);
}
scan_graph(n_graph,t->state_number,pos2+1,depth,liste_arrivee,mot_token_buffer,p);
}
}
else {
// case of a normal letter sequence
if (p->buffer[pos+p->current_origin]==' ') {pos2=pos+1;if (p->output_policy==MERGE_OUTPUTS) push(p->stack,' ');}
//else if (buffer[pos+origine_courante]==0x0d) {pos2=pos+2;if (MODE==MERGE) empiler(0x0a);}
else pos2=pos;
if (etiq->control&RESPECT_CASE_TAG_BIT_MASK) {
// case of exact case match
int position=0;
while (pos2+p->current_origin<p->text_buffer->size && p->buffer[pos2+p->current_origin]==contenu[position]) {
pos2++; position++;
}
if (contenu[position]=='\0' && position!=0 &&
!(is_letter(contenu[position-1],p->alphabet) && is_letter(p->buffer[pos2+p->current_origin],p->alphabet))) {
// we proceed only if we have exactly read the contenu sequence
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_input_string(p->stack,contenu,0);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
else {
// case of variable case match
// the letter sequences may have been caught by the arbre_etiquette structure
int position=0;
unichar* mot=mot_token_buffer;
while (pos2+p->current_origin<p->text_buffer->size && is_equal_or_uppercase(contenu[position],p->buffer[pos2+p->current_origin],p->alphabet)) {
mot[position++]=p->buffer[(pos2++)+p->current_origin];
}
mot[position]='\0';
if (contenu[position]=='\0' && position!=0 &&
!(is_letter(contenu[position-1],p->alphabet) && is_letter(p->buffer[pos2+p->current_origin],p->alphabet))) {
// we proceed only if we have exactly read the contenu sequence
// in both modes MERGE and REPLACE, we process the transduction if any
traiter_transduction(p,etiq->output);
if (p->output_policy==MERGE_OUTPUTS /*|| etiq->transduction==NULL || etiq->transduction[0]=='\0'*/) {
// if we are in MERGE mode, we add to ouput the char we have read
push_input_string(p->stack,mot,0);
}
scan_graph(n_graph,t->state_number,pos2,depth,liste_arrivee,mot_token_buffer,p);
}
}
}
}
t=t->next;
}
}
