docs

.gitignore

@@ -3,3 +3,4 @@ regtest
 *.so
 *.data
 .gdb_history
+gdb.txt

Makefile

@@ -12,3 +12,6 @@ run:
 	${OUT}
 
 test: run
+
+clean:
+	rm ${OUT}

debug/regex.pretty_print.py

@@ -7,15 +7,16 @@ class RegexPrinter:
 		#   Regular shit
 		s += "accepting_state = " + str(self.val['accepting_state']) + ", str = " + str(self.val['str']) + ",\n"
 		#   Delta
-		delta_t_ptr_ptr = gdb.lookup_type("delta_t").pointer().pointer()
+		delta_t_ptr_ptr_t = gdb.lookup_type("delta_t").pointer().pointer()
 		dt = self.val['delta_table']
 		s += "delta_table = {\n"
+		d0 = 0
 		for i in range(0, dt['element_count']):
 			s += "\t"
 			s += (
 					str(
 						(
-							dt['data'].cast(delta_t_ptr_ptr)
+							dt['data'].cast(delta_t_ptr_ptr_t)
 								+
 							i
 						).dereference().dereference()

documentation/README.md

@@ -29,6 +29,23 @@ State transition table look up
 		+ fail    --> return 
 	+ fail    --> return
 ---
+# Legend
+
+|      | Start | End |
+| :--: | :---: | :-: |
+| Line |   SOS | EOS |
+| Word |   WOS | EOW |
+
+#### SOS
+Start Of String
+
+#### EOS
+Start Of String
+
+#### WOS
+Start Of String
+
+---
 ##### HALT\_AND\_CATCH\_FIRE
 H&C is a special state signalling that we have hit a dead end.
 The reason why need it and we cant just instanly quick is backtracking.
@@ -89,7 +106,7 @@ and finally returns failiour.
 
 
 #### \<
-This is the SOW (start of word) operator.
+This is the SOW operator.
 SOW must match:
 ```
 ^myword

documentation/TODO.md

@@ -3,3 +3,5 @@
 [ ] arbitrary memory support (this probably covers UTF-8 support)
 [ ] documentation thats not shit
 [ ] HOOK\_ALL / OFFSHOOT width parameter inconsistency
+[ ] nesting is going real wild
+[ ] states could be optimized by chopping of the init padding if not flagged required

source/jeger.c

@@ -490,6 +490,11 @@ regex_t * regex_compile(const char * const pattern) {
 	char whitelist[64];
 	char blacklist[64];
 
+	static const int REGEX_PREVERSABLE_FLAGS = IS_AT_THE_BEGINNING
+	                                         | FORCE_START_OF_STRING
+	                                         | DO_FORBID_START_OF_STRING
+	                                         ;
+
 	compiler_state cs = {
 		.flags     = IS_AT_THE_BEGINNING,
 		.state     = JEGER_INIT_STATE,
@@ -502,7 +507,7 @@ regex_t * regex_compile(const char * const pattern) {
 		// Reset the compiler
 		whitelist[0]   = '\0';
 		blacklist[0]   = '\0';
-		cs.flags    &= (IS_AT_THE_BEGINNING | FORCE_START_OF_STRING);
+		cs.flags      &= REGEX_PREVERSABLE_FLAGS;
 		cs.width       = 1;
 		cs.match_width = 1;
 
@@ -657,6 +662,7 @@ regex_t * regex_compile(const char * const pattern) {
 			++cs.state;
 		}
 
+		// Purge SOS flag
 		cs.flags &= (~IS_AT_THE_BEGINNING);
 	}