anon/jeger
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

bak

Browse Source
This commit is contained in:
anon 2023-09-23 12:41:15 +02:00
parent 8080b44613
commit 283e1e48e7
4 changed files with 56 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
documentation/README.md
View File

@ -98,5 +98,13 @@ SOW must match:
Not only that, this combination is key,
either it has to be the start of the string
or there has to be at least something which is not a symbol char.
With out the last condition "eexample" would match "\\\<exaplme\\\>"
With out the last condition "eexample" would match "\\\<exaple\\\>"
as the iteration of `regex_match()` reaches "example".
From a more practical perspective:
``` C
\<myword\>
// Must match
"myword"
" myword"
```

1
documentation/TODO.md
View File

@ -2,3 +2,4 @@
[ ] UTF-8 support
[ ] arbitrary memory support (this probably covers UTF-8 support)
[ ] documentation thats not shit
[ ] HOOK\_ALL / OFFSHOOT width parameter inconsistency

42
source/jeger.c
View File

@ -9,6 +9,8 @@
#include <limits.h>
#include <stdlib.h>
#define JEGER_SOS_STATE 0
#define JEGER_NSOS_STATE 1
#define JEGER_INIT_STATE 2
// ------------------
@ -481,6 +483,8 @@ regex_t * regex_compile(const char * const pattern) {
.blacklist = blacklist,
};
bool fucku = true;
for (const char * s = pattern; *s != '\00';) {
assert(!is_quantifier(*s) && "Pattern starts with quantifier.");
// Reset the compiler
@ -539,21 +543,37 @@ regex_t * regex_compile(const char * const pattern) {
s += 1;
} break;
case '<': {
unsigned true_inc = 1;
if (cs.flags & IS_AT_THE_BEGINNING
&& !(cs.flags & DO_CATCH)
&& !(cs.flags & IS_NEGATIVE)
&& whitelist[0] == '\0') {
// ---
cs.flags |= INCREMENT_STATE;
fucku = false;
strcat(whitelist, JEGER_CHAR_symbol_chars);
// ---
ABSOLUTE_OFFSHOOT( JEGER_SOS_STATE, JEGER_INIT_STATE+1, 0, 0, regex);
ABSOLUTE_OFFSHOOT(JEGER_INIT_STATE, JEGER_INIT_STATE+2, 1, 0, regex);
HOOK_ALL(0, whitelist, HALT_AND_CATCH_FIRE, &cs, regex);
// ---
++cs.state;
cs.width = 0;
HOOK_ALL(0, whitelist, +1, &cs, regex);
cs.width = 1;
OFFSHOOT(0, +1, 1, 0, &cs, regex);
// ---
} else {
HOOK_ALL(0, whitelist, +1, &cs, regex);
if ((cs.flags & DO_CATCH)
|| (cs.flags & IS_NEGATIVE)) {
OFFSHOOT(0, +1, 1, 1, &cs, regex);
OFFSHOOT(+1, +2, 1, 1, &cs, regex);
++true_inc;
} else {
cs.flags |= INCREMENT_STATE;
}
cs.flags |= IS_NEGATIVE;
if (cs.flags & IS_AT_THE_BEGINNING) {
ABSOLUTE_OFFSHOOT(0, JEGER_INIT_STATE + true_inc, 0, 0, regex);
OFFSHOOT(0, +1, 1, 0, &cs, regex);
}
cs.flags |= IS_NEGATIVE;
strcat(blacklist, JEGER_CHAR_symbol_chars);
//OFFSHOOT(0 + (true_inc-1), +true_inc, 1, 0, &cs, regex);
s += 1;
} break;
case '>': {
@ -622,11 +642,13 @@ regex_t * regex_compile(const char * const pattern) {
}
// Init state hookups
ABSOLUTE_OFFSHOOT(0, JEGER_INIT_STATE, 0, 0, regex);
if (fucku) {
ABSOLUTE_OFFSHOOT(JEGER_SOS_STATE, JEGER_INIT_STATE, 0, 0, regex);
}
if (cs.flags & FORCE_START_OF_STRING) {
ABSOLUTE_OFFSHOOT(1, HALT_AND_CATCH_FIRE, 0, 0, regex);
ABSOLUTE_OFFSHOOT(JEGER_NSOS_STATE, HALT_AND_CATCH_FIRE, 0, 0, regex);
} else {
ABSOLUTE_OFFSHOOT(1, JEGER_INIT_STATE, 0, 0, regex);
ABSOLUTE_OFFSHOOT(JEGER_NSOS_STATE, JEGER_INIT_STATE, 0, 0, regex);
}
regex->accepting_state = cs.state;

8
source/main.cpp
View File

@ -107,6 +107,14 @@ signed main() {
TEST(R"del(.\<print\>.)del", "printf", false);
TEST(R"del(.\<print\>.)del", "fprintf", false);
puts("");
TEST(R"del(\<while\>)del", "while", true);
TEST(R"del(\<while\>)del", " while ", true);
TEST(R"del(\<while\>)del", "9while ", true);
TEST(R"del(\<while\>)del", "for while {", true);
TEST(R"del(\<while\>)del", "for while{", true);
if (test_counter == passed_tests) {
fputs("\033[32m", stdout);
} else {