beef up lexer a little

git-svn-id: file:///home/lennart/svn/public/sse/trunk@18 5fbabb74-0606-0410-a5e4-b5cc6a42724e

3 files changed, 63 insertions, 35 deletions

diff --git a/feed/Makefile b/feed/Makefile
index aeb8883..94c2540 100644
--- a/feed/Makefile
+++ b/feed/Makefile
@@ -1,4 +1,5 @@
-CLAGS=-Wextra -g -O2 -pipe
+CFLAGS=-Wextra -O3 -pipe
+#CFLAGS=-Wextra -g -O0 -pipe
 LIBS=-lfl
 
 all: sse_lex_c
diff --git a/feed/sse_lex_c.l b/feed/sse_lex_c.l
index 52f46a7..4674b73 100644
--- a/feed/sse_lex_c.l
+++ b/feed/sse_lex_c.l
@@ -4,6 +4,48 @@
 
 #include <stdio.h>
 
+static void print_with_subwords(const char *w) {
+    const char *e;
+
+    printf("%s\n", w);
+
+    if ((e = strchr(w, '_'))) {
+
+        /* Split at dashes.*/
+
+        do {
+            for (; *e == '_'; e++);
+
+            if (strlen(e) < 3)
+                break;
+            
+            printf("S:%s\n", e);
+
+            e = strchr(e, '_');
+        } while (e);
+
+    } else {
+
+        /* Split a place where an uppercase letter follows a lowercase
+         * letter */
+
+        e = w;
+        for (;;) {
+            for (; *e && !(islower(*e) && isupper(*(e+1))); e++);
+
+            if (!*e)
+                break;
+            
+            e++;
+            
+            if (strlen(e) < 3)
+                break;
+
+            printf("S:%s\n", e);
+        }
+    }
+        
+}
     
 %}
 
@@ -46,40 +88,7 @@ NIDCHAR [^_a-zA-Z0-9]
 <CHAR>. ;
 <CHARESC>. { BEGIN CHAR; }
 
-<DEF>auto{NIDCHAR}  |
-<DEF>break{NIDCHAR}  |
-<DEF>case{NIDCHAR}  |
-<DEF>char{NIDCHAR}  |
-<DEF>const{NIDCHAR}  |
-<DEF>continue{NIDCHAR}  |
-<DEF>default{NIDCHAR}  |
-<DEF>do{NIDCHAR}  |
-<DEF>double{NIDCHAR}  |
-<DEF>else{NIDCHAR}  |
-<DEF>enum{NIDCHAR}  |
-<DEF>extern{NIDCHAR}  |
-<DEF>float{NIDCHAR}  |
-<DEF>for{NIDCHAR}  |
-<DEF>goto{NIDCHAR}  |
-<DEF>if{NIDCHAR}  |
-<DEF>int{NIDCHAR}  |
-<DEF>long{NIDCHAR}  |
-<DEF>register{NIDCHAR}  |
-<DEF>return{NIDCHAR}  |
-<DEF>short{NIDCHAR}  |
-<DEF>signed{NIDCHAR}  |
-<DEF>sizeof{NIDCHAR}  |
-<DEF>static{NIDCHAR}  |
-<DEF>struct{NIDCHAR}  |
-<DEF>switch{NIDCHAR}  |
-<DEF>typedef{NIDCHAR}  |
-<DEF>union{NIDCHAR}  |
-<DEF>unsigned{NIDCHAR}  |
-<DEF>void{NIDCHAR}  |
-<DEF>volatile{NIDCHAR}  |
-<DEF>while{NIDCHAR} { yyless(yyleng-1);  }
-
-<DEF>[a-zA-Z_][a-zA-Z_0-9]{3,} { printf("%s\n", yytext); }
+<DEF>[a-zA-Z_][a-zA-Z_0-9]{3,} { print_with_subwords(yytext); }
 
 <DEF>[0-9]+ ;
 <DEF>0x[0-9a-fA-F]+ ;
@@ -89,6 +98,7 @@ NIDCHAR [^_a-zA-Z0-9]
 
 %%
 
+
 int main(int argc, char *argv[]) {
 
     if (argc <= 1)
diff --git a/feed/test-lex-c.c b/feed/test-lex-c.c
index d43f10f..9b984bb 100644
--- a/feed/test-lex-c.c
+++ b/feed/test-lex-c.c
@@ -118,3 +118,20 @@ int main(int argc, char argv[]) {
 
     FOO_BAR(blafasel);
 }
+
+
+
+  dies_ist_ein_test
+___DIES___IST___EIN__TEST_______
+
+  alpha beta gamma
+  _________
+
+
+  DiesIstEinTest
+
+  Dies_Ist_Ein_Test___
+
+  ArschTritt
+
+  deppVomLande