66 int direction,
bool word_end,
UNICHAR_ID unichar_id,
70 " direction %d word_end %d unichar_id %d, exploring node:\n",
71 node_ref,
next_node, direction, word_end, unichar_id);
72 if (node_ref != NO_EDGE) {
76 if (node_ref == NO_EDGE)
return false;
79 nodes_[node_ref]->forward_edges :
nodes_[node_ref]->backward_edges;
80 int vec_size = vec.
size();
86 while (start <= end) {
87 k = (start + end) >> 1;
91 *edge_ptr = &(vec[k]);
94 }
else if (compare == 1) {
101 for (
int i = 0; i < vec_size; ++i) {
107 *edge_ptr = &(edge_rec);
117 int direction,
bool word_end,
120 &(
nodes_[node1]->forward_edges) : &(
nodes_[node1]->backward_edges);
124 while (search_index < vec->size() &&
126 (*vec)[search_index]) == 1) {
130 search_index = vec->
size();
133 link_edge(&edge_rec, node2, marker_flag, direction, word_end, unichar_id);
137 (*vec)[edge_index] = edge_rec;
138 }
else if (search_index < vec->size()) {
139 vec->
insert(edge_rec, search_index);
159 unichar_id, &back_edge_ptr, &back_edge_index));
171 if (word.
length() <= 0)
return false;
174 for (
int i = 0; i < word.
length(); ++i) {
182 bool marker_flag =
false;
185 int32_t still_finding_chars =
true;
186 int32_t word_end =
false;
187 bool add_failed =
false;
193 for (i = 0; i < word.
length() - 1; ++i) {
195 marker_flag = (repetitions !=
nullptr) ? (*repetitions)[i] :
false;
197 if (still_finding_chars) {
199 unichar_id, &edge_ptr, &edge_index);
202 edge_index, last_node);
205 still_finding_chars =
false;
215 still_finding_chars =
false;
223 if (!still_finding_chars) {
227 if (the_next_node == 0) {
232 marker_flag, word_end, unichar_id)) {
237 last_node = the_next_node;
242 marker_flag = (repetitions !=
nullptr) ? (*repetitions)[i] :
false;
244 if (still_finding_chars &&
246 unichar_id, &edge_ptr, &edge_index)) {
250 marker_flag, unichar_id);
256 !
add_new_edge(last_node, the_next_node, marker_flag,
true, unichar_id))
260 tprintf(
"Re-initializing document dictionary...\n");
275 static int sort_strings_by_dec_length(
const void* v1,
const void* v2) {
276 const auto *s1 =
static_cast<const STRING *
>(v1);
277 const auto *s2 =
static_cast<const STRING *
>(v2);
278 return s2->
length() - s1->length();
286 word_list.
sort(sort_strings_by_dec_length);
296 word_file = fopen(filename,
"rb");
297 if (word_file ==
nullptr)
return false;
299 while (fgets(line_str,
sizeof(line_str), word_file) !=
nullptr) {
301 STRING word_str(line_str);
304 tprintf(
"Read %d words so far\n", word_count);
308 tprintf(
"Read %d words total.\n", word_count);
316 for (
int i = 0; i < words.
size(); ++i) {
328 tprintf(
"Error: word '%s' not in DAWG after adding it\n",
357 bool is_alpha = unicharset.
get_isalpha(unichar_id);
379 }
else if (ch ==
'd') {
381 }
else if (ch ==
'n') {
383 }
else if (ch ==
'p') {
385 }
else if (ch ==
'a') {
387 }
else if (ch ==
'A') {
390 return INVALID_UNICHAR_ID;
397 tprintf(
"please call initialize_patterns() before read_pattern_list()\n");
401 FILE *pattern_file = fopen(filename,
"rb");
402 if (pattern_file ==
nullptr) {
403 tprintf(
"Error opening pattern file %s\n", filename);
407 int pattern_count = 0;
415 const char *str_ptr = string;
416 int step = unicharset.
step(str_ptr);
419 UNICHAR_ID curr_unichar_id = INVALID_UNICHAR_ID;
420 if (step == 1 && *str_ptr ==
'\\') {
422 if (*str_ptr ==
'\\') {
426 tprintf(
"Please provide at least %d concrete characters at the"
437 if (curr_unichar_id == INVALID_UNICHAR_ID) {
444 step = unicharset.
step(str_ptr);
446 if (step == 1 && *str_ptr ==
'\\' && *(str_ptr+1) ==
'*') {
447 repetitions_vec[repetitions_vec.
size()-1] =
true;
449 step = unicharset.
step(str_ptr);
453 tprintf(
"Invalid user pattern %s\n",
string);
458 tprintf(
"Inserting expanded user pattern %s\n",
464 tprintf(
"Error: failed to insert pattern '%s'\n",
string);
470 tprintf(
"Read %d valid patterns from %s\n", pattern_count, filename);
472 fclose(pattern_file);
481 unichar_id, &edge_ptr, &edge_index));
489 }
else if (node1 == 0) {
517 for (
int i = 0; i <
nodes_.
size(); i++) reduced_nodes[i] =
false;
519 delete[] reduced_nodes;
530 node_ref_map[i+1] = node_ref_map[i] +
nodes_[i]->forward_edges.
size();
532 int num_forward_edges = node_ref_map[i];
536 auto edge_array =
new EDGE_RECORD[num_forward_edges];
541 for (j = 0; j < end; ++j) {
552 delete[] node_ref_map;
562 tprintf(
"\nCollapsing node %" PRIi64
":\n", node);
586 curr_word_end, curr_unichar_id);
589 curr_word_end, curr_unichar_id,
590 &edge_ptr, &edge_index));
597 next_node2_num_edges, next_node2);
613 bool did_something =
false;
614 for (
int i = edge_index; i < backward_edges->
size() - 1; ++i) {
616 UNICHAR_ID curr_unichar_id = INVALID_UNICHAR_ID;
617 while (i < backward_edges->size()) {
618 if (!
DeadEdge((*backward_edges)[i])) {
620 if (curr_unichar_id != unichar_id)
return did_something;
625 if (i == backward_edges->
size())
break;
626 const EDGE_RECORD &edge_rec = (*backward_edges)[i];
628 for (
int j = i + 1; j < backward_edges->
size(); ++j) {
629 const EDGE_RECORD &next_edge_rec = (*backward_edges)[j];
630 if (
DeadEdge(next_edge_rec))
continue;
632 if (next_id != unichar_id)
break;
638 did_something =
true;
643 return did_something;
647 int num_edges = edges->
size();
648 if (num_edges <= 1)
return;
651 for (
int i = 0; i < num_edges; ++i) {
656 for (
int i = 0; i < num_edges; ++i)
657 (*edges)[i] = sort_vec[i].data;
670 while (edge_index < backward_edges.
size()) {
671 if (
DeadEdge(backward_edges[edge_index]))
continue;
675 &backward_edges, reduced_nodes));
676 while (++edge_index < backward_edges.
size()) {
678 if (!
DeadEdge(backward_edges[edge_index]) &&
id != unichar_id)
break;
681 reduced_nodes[node] =
true;
688 for (
int i = 0; i < backward_edges.
size(); ++i) {
689 if (
DeadEdge(backward_edges[i]))
continue;
698 if (node == NO_EDGE)
return;
703 for (
int dir = 0; dir < 2; ++dir) {
712 for (i = 0; (dir == 0 ? i < num_fwd : i < num_bkw) &&
713 i < max_num_edges; ++i) {
718 if (dir == 0 ? i < num_fwd : i < num_bkw)
tprintf(
"...");