initial commit: support C/C++ comments

6 files changed, 338 insertions, 0 deletions

diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..949524c
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,39 @@
+
+CC     = gcc
+CFLAGS = -ggdb3 `pkg-config --cflags --libs eina`
+BIN = eo_tokenizer
+SRC = main.c eo_tokenizer.c
+
+.PHONY: clean
+.SUFFIXES: .rl .xml .dot .png
+
+.rl.c:
+	ragel -s -o $@ $<
+
+.rl.xml:
+	ragel -x -o $@ $<
+
+.rl.dot:
+	ragel -V -p -S eo_tokenizer -o $@ $<
+
+.dot.png:
+	dot -Tpng >$@ $<
+
+.c.o:
+	$(CC) -c -Wall -I. $(CFLAGS) -D$(STD) -o $@ $<
+
+all: $(BIN)
+
+eo_tokenizer.c: eo_tokenizer.rl
+eo_tokenizer.xml: eo_tokenizer.rl
+eo_tokenizer.dot: eo_tokenizer.rl
+eo_tokenizer.png: eo_tokenizer.dot
+
+eo_tokenizer: main.c eo_tokenizer.c eo_tokenizer.h
+
+test-comments: eo_tokenizer
+	EINA_LOG_LEVELS=eo_toknz:4 ./eo_tokenizer ./data/comments.c
+
+clean:
+	rm -f *~ *.o *.dot *.png *.xml $(BIN).c $(BIN)
+
diff --git a/data/comments.c b/data/comments.c
new file mode 100644
index 0000000..5fd0216
--- /dev/null
+++ b/data/comments.c
@@ -0,0 +1,20 @@
+/* comment:1-1 */

+

+/*

+   comment:3-4    */ 

+/*

+ *  comment:5-7

+ */

+   

+// comment:9-9

+

+ /*

+comment:11-15

+ *

+ *

+ * *//*comment 15-16   

+*/

+

+ Evas_object {

+    inherits { Eo_Base, Evas_Common_Interface }

+}

diff --git a/eo_tokenizer.h b/eo_tokenizer.h
new file mode 100644
index 0000000..536cdda
--- /dev/null
+++ b/eo_tokenizer.h
@@ -0,0 +1,63 @@
+#ifndef __EO_TOKENIZER_H__
+#define __EO_TOKENIZER_H__
+
+#include <Eina.h>
+
+extern int _eo_tokenizer_log_dom;
+#undef EINA_LOG_DOMAIN_DEFAULT
+#define EINA_LOG_DOMAIN_DEFAULT _eo_tokenizer_log_dom
+
+#ifdef CRITICAL
+#undef CRITICAL
+#endif
+#define CRITICAL(...) EINA_LOG_DOM_CRIT(_eo_tokenizer_log_dom, __VA_ARGS__)
+
+#ifdef ERR
+#undef ERR
+#endif
+#define ERR(...) EINA_LOG_DOM_ERR(_eo_tokenizer_log_dom, __VA_ARGS__)
+
+#ifdef WRN
+#undef WRN
+#endif
+#define WRN(...) EINA_LOG_DOM_WARN(_eo_tokenizer_log_dom, __VA_ARGS__)
+
+#ifdef INF
+#undef INF
+#endif
+#define INF(...) EINA_LOG_DOM_INFO(_eo_tokenizer_log_dom, __VA_ARGS__)
+
+#ifdef DBG
+#undef DBG
+#endif
+#define DBG(...) EINA_LOG_DOM_DBG(_eo_tokenizer_log_dom, __VA_ARGS__)
+
+#define BUFSIZE 256
+
+typedef struct _eo_tokenizer {
+     /* ragel vars */
+     int cs;        /* current machine state */
+     int act;       /* last pattern matched */
+     char *ts;      /* current token match start */
+     char *te;      /* current token match end */
+     char *p;       /* data start */
+     char *pe;      /* data end */
+     char *eof;     /* eof = (EOF ? pe : NULL) */
+     /* int stack[10]; /1* state stack used by fret fcall ... *1/ */
+     /* int top;       /1* stack pointer *1/ */
+
+     const char *source;
+     int current_line;
+     int current_nesting;
+     int max_nesting;
+     int token_start_line;
+     char buf[BUFSIZE];
+} Eo_Tokenizer;
+
+Eo_Tokenizer* eo_tokenizer_get();
+
+Eina_Bool eo_tokenizer_walk(Eo_Tokenizer *toknz, const char *source);
+
+void eo_tokenizer_free(Eo_Tokenizer *toknz);
+
+#endif /* __EO_TOKENIZER_H__ */
diff --git a/eo_tokenizer.rl b/eo_tokenizer.rl
new file mode 100644
index 0000000..b10218d
--- /dev/null
+++ b/eo_tokenizer.rl
@@ -0,0 +1,165 @@
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "eo_tokenizer.h"
+
+%%{
+   machine common;
+
+   action inc_line {
+      toknz->current_line += 1;
+   }
+
+   action save_start_line {
+      toknz->token_start_line = toknz->current_line;
+   }
+
+   cr                = '\n';
+   cr_neg            = "[^\n]";
+   ws                = "[ \t\r]";
+   newline           = cr @inc_line;
+   ignore            = (0x00..0x20 - cr)* newline?;
+
+   c_comment         = "/*" ( any | '\n' @inc_line )* :>> "*/";
+   cpp_comment       = "//" (any - cr)* newline;
+   comment           = ( c_comment | cpp_comment ) > save_start_line;
+
+}%%
+
+%%{
+   machine eo_tokenizer;
+   include common;
+
+   access toknz->;
+   variable p toknz->p;
+   variable pe toknz->pe;
+   variable eof toknz->eof;
+
+   action show_comment {
+      DBG("comment %03d:%03d", toknz->token_start_line, toknz->current_line);
+   }
+
+   action show_ignore {
+      DBG("ignore %d (%d)", toknz->current_line, (toknz->te - toknz->ts));
+   }
+
+   action error {
+      /* TODO find a more elegant way,
+       *  -> cant't use $err() or $!error on scanner
+       *  -> fgoto another machine to eat the line, ts is set to NULL
+       */
+      char buf[BUFSIZE];
+      char *s, *d;
+      for (s = fpc, d = buf; (s <= toknz->pe); s++)
+        {
+           if ( (int)*s == 13 || (int)*s == 10)
+             break;
+           *d++ = *s;
+        }
+      *d = '\0';
+      ERR("error line %d : %s...", toknz->current_line, buf);
+      toknz->cs = eo_tokenizer_error;
+      fbreak;
+   }
+
+   main := |*
+      ignore => show_ignore;
+      comment => show_comment;
+      any => error;
+   *|;
+
+}%%
+
+%% write data;
+
+Eina_Bool
+eo_tokenizer_walk(Eo_Tokenizer *toknz, const char *source)
+{
+   INF("tokenize %s...", source);
+   toknz->source = source;
+
+   FILE *stream;
+
+   int done = 0;
+   int have = 0;
+
+   stream = fopen(toknz->source, "r");
+   if (!stream)
+     {
+        ERR("unable to read in %s", toknz->source);
+        return EINA_FALSE;
+     }
+
+   %% write init;
+
+   while (!done)
+     {
+        int len;
+        int space;
+
+        toknz->p = toknz->buf + have;
+        space = BUFSIZE - have;
+
+        if (space == 0)
+          {
+             ERR("out of buffer space");
+             fclose(stream);
+             exit(EXIT_FAILURE);
+          }
+
+        len = fread(toknz->p, 1, space, stream);
+        if (len == 0) break;
+        toknz->pe = toknz->p + len;
+
+        if (len < space)
+          {
+             toknz->eof = toknz->pe;
+             done = 1;
+          }
+
+        %% write exec;
+
+        if ( toknz->cs == %%{ write error; }%% )
+            break;
+
+        if ( toknz->ts == 0 )
+          have = 0;
+        else
+          {
+             /* There is a prefix to preserve, shift it over. */
+             have = toknz->pe - toknz->ts;
+             memmove( toknz->buf, toknz->ts, have);
+             toknz->te = toknz->buf + (toknz->te - toknz->ts);
+             toknz->ts = toknz->buf;
+          }
+     }
+
+   fclose(stream);
+
+   return EINA_TRUE;
+}
+
+Eo_Tokenizer*
+eo_tokenizer_get()
+{
+   Eo_Tokenizer *toknz = calloc(1, sizeof(Eo_Tokenizer));
+   if (!toknz) return NULL;
+
+   toknz->ts = NULL;
+   toknz->te = NULL;
+   /* toknz->top = 0; */
+   toknz->source = NULL;
+   toknz->max_nesting = 10;
+   toknz->current_line = 1;
+   toknz->current_nesting = 0;
+   toknz->token_start_line = 0;
+
+   return toknz;
+}
+
+void
+eo_tokenizer_free(Eo_Tokenizer *toknz)
+{
+   free(toknz);
+}
+
diff --git a/main.c b/main.c
new file mode 100644
index 0000000..2b420f2
--- /dev/null
+++ b/main.c
@@ -0,0 +1,50 @@
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+
+#include "eo_tokenizer.h"
+
+int _eo_tokenizer_log_dom = -1;
+
+int main(int argc, char **argv)
+{
+   char *fpath;
+   Eo_Tokenizer *toknz;
+
+   if (argc < 2)
+     {
+        fprintf(stderr, "usage %s input_file\n", argv[0]);
+        exit(EXIT_FAILURE);
+     }
+
+   fpath = argv[1];
+
+   eina_init();
+
+   eina_log_color_disable_set(EINA_FALSE);
+   _eo_tokenizer_log_dom = eina_log_domain_register("eo_toknz", EINA_COLOR_CYAN);
+
+   if (access(fpath, F_OK) != 0)
+     {
+        ERR("error accessing file %s : %s", fpath, strerror(errno));
+        exit(EXIT_FAILURE);
+     }
+
+   toknz = eo_tokenizer_get(fpath);
+   if (!toknz)
+     {
+        ERR("can't create eo_tokenizer");
+        eina_shutdown();
+        exit(EXIT_FAILURE);
+     }
+
+   eo_tokenizer_walk(toknz, fpath);
+
+   eo_tokenizer_free(toknz);
+
+   eina_shutdown();
+
+   return EXIT_SUCCESS;
+}
+
diff --git a/tok.c b/tok.c
new file mode 120000
index 0000000..a66afa9
--- /dev/null
+++ b/tok.c
@@ -0,0 +1 @@
+eo_tokenizer.rl
\ No newline at end of file