|
Tesseract
3.02
|
#include <unicharmap.h>
Classes | |
| struct | UNICHARMAP_NODE |
Public Member Functions | |
| UNICHARMAP () | |
| ~UNICHARMAP () | |
| void | insert (const char *const unichar_repr, UNICHAR_ID id) |
| UNICHAR_ID | unichar_to_id (const char *const unichar_repr) const |
| UNICHAR_ID | unichar_to_id (const char *const unichar_repr, int length) const |
| bool | contains (const char *const unichar_repr) const |
| bool | contains (const char *const unichar_repr, int length) const |
| int | minmatch (const char *const unichar_repr) const |
| void | clear () |
Definition at line 27 of file unicharmap.h.
| UNICHARMAP::UNICHARMAP | ( | ) |
Definition at line 25 of file unicharmap.cpp.
:
nodes(0) {
}
| UNICHARMAP::~UNICHARMAP | ( | ) |
Definition at line 29 of file unicharmap.cpp.
{
if (nodes != 0)
delete[] nodes;
}
| void UNICHARMAP::clear | ( | ) |
Definition at line 154 of file unicharmap.cpp.
{
if (nodes != 0)
{
delete[] nodes;
nodes = 0;
}
}
| bool UNICHARMAP::contains | ( | const char *const | unichar_repr | ) | const |
Definition at line 101 of file unicharmap.cpp.
{
const char* current_char = unichar_repr;
UNICHARMAP_NODE* current_nodes = nodes;
assert(*unichar_repr != '\0');
while (current_nodes != 0 && *(current_char + 1) != '\0') {
current_nodes =
current_nodes[static_cast<unsigned char>(*current_char)].children;
++current_char;
}
return current_nodes != 0 && *(current_char + 1) == '\0' &&
current_nodes[static_cast<unsigned char>(*current_char)].id >= 0;
}
| bool UNICHARMAP::contains | ( | const char *const | unichar_repr, |
| int | length | ||
| ) | const |
Definition at line 120 of file unicharmap.cpp.
{
const char* current_char = unichar_repr;
UNICHARMAP_NODE* current_nodes = nodes;
assert(*unichar_repr != '\0');
assert(length > 0 && length <= UNICHAR_LEN);
while (current_nodes != 0 && (length > 1 && *(current_char + 1) != '\0')) {
current_nodes =
current_nodes[static_cast<unsigned char>(*current_char)].children;
--length;
++current_char;
}
return current_nodes != 0 && (length == 1 || *(current_char + 1) == '\0') &&
current_nodes[static_cast<unsigned char>(*current_char)].id >= 0;
}
| void UNICHARMAP::insert | ( | const char *const | unichar_repr, |
| UNICHAR_ID | id | ||
| ) |
Definition at line 76 of file unicharmap.cpp.
{
const char* current_char = unichar_repr;
UNICHARMAP_NODE** current_nodes_pointer = &nodes;
assert(*unichar_repr != '\0');
assert(id >= 0);
do {
if (*current_nodes_pointer == 0)
*current_nodes_pointer = new UNICHARMAP_NODE[256];
if (*(current_char + 1) == '\0') {
(*current_nodes_pointer)
[static_cast<unsigned char>(*current_char)].id = id;
return;
}
current_nodes_pointer =
&((*current_nodes_pointer)
[static_cast<unsigned char>(*current_char)].children);
++current_char;
} while (true);
}
| int UNICHARMAP::minmatch | ( | const char *const | unichar_repr | ) | const |
Definition at line 140 of file unicharmap.cpp.
{
const char* current_char = unichar_repr;
UNICHARMAP_NODE* current_nodes = nodes;
while (current_nodes != NULL && *current_char != '\0') {
if (current_nodes[static_cast<unsigned char>(*current_char)].id >= 0)
return current_char + 1 - unichar_repr;
current_nodes =
current_nodes[static_cast<unsigned char>(*current_char)].children;
++current_char;
}
return 0;
}
| UNICHAR_ID UNICHARMAP::unichar_to_id | ( | const char *const | unichar_repr | ) | const |
Definition at line 36 of file unicharmap.cpp.
{
const char* current_char = unichar_repr;
UNICHARMAP_NODE* current_nodes = nodes;
assert(*unichar_repr != '\0');
do {
if (*(current_char + 1) == '\0')
return current_nodes[static_cast<unsigned char>(*current_char)].id;
current_nodes =
current_nodes[static_cast<unsigned char>(*current_char)].children;
++current_char;
} while (true);
}
| UNICHAR_ID UNICHARMAP::unichar_to_id | ( | const char *const | unichar_repr, |
| int | length | ||
| ) | const |
Definition at line 54 of file unicharmap.cpp.
{
const char* current_char = unichar_repr;
UNICHARMAP_NODE* current_nodes = nodes;
assert(*unichar_repr != '\0');
assert(length > 0 && length <= UNICHAR_LEN);
do {
if (length == 1 || *(current_char + 1) == '\0')
return current_nodes[static_cast<unsigned char>(*current_char)].id;
current_nodes =
current_nodes[static_cast<unsigned char>(*current_char)].children;
++current_char;
--length;
} while (true);
}