Regexps in 32-mode

Author: dboulytchev

runtime/runtime.c

@@ -1,12 +1,9 @@
 /* Runtime library */
 
-#define _GNU_SOURCE 1
+# define _GNU_SOURCE 1
 
-#include "runtime.h"
-#include <regex.h>
-
-#include "gc.h"
-#include "runtime_common.h"
+# include "runtime.h"
+# include "gc.h"
 
 extern size_t __gc_stack_top, __gc_stack_bottom;
 
@@ -486,7 +483,7 @@ extern regex_t *Lregexp (char *regexp) {
 
   int res = regcomp(regexp_compiled, regexp, REG_EXTENDED);
 
-  // printf("Lregexp: got compiled regexp %p, for string %s\n", regexp_compiled, regexp);
+  //printf("Lregexp: got compiled regexp %p, for string %s\n", regexp_compiled, regexp);
 
   if (res != 0) {
       char buf[100];
@@ -506,7 +503,7 @@ extern aint LregexpMatch (regex_t *b, char *s, aint pos) {
 
   int res = regexec(b, s + UNBOX(pos), (size_t) 1, &match, 0);
 
-  // printf("regexpMatch %p: %s, res=%d so=%d eo=%d\n", b, s + UNBOX(pos), res, match.rm_so, match.rm_eo);
+  //printf("regexpMatch %p: %s, res=%d so=%d eo=%d\n", b, s + UNBOX(pos), res, match.rm_so, match.rm_eo);
 
   if (res == 0 && match.rm_so == 0) {
       return BOX(match.rm_eo);
@@ -826,7 +823,7 @@ extern void *Bclosure (aint* args, aint bn) {
 extern void *Barray (aint* args, aint bn) {
   data   *r;
   aint     n = UNBOX(bn);
-
+  
   PRE_GC();
 
   for (aint i = 0; i < n; i++) {

runtime32/runtime.c

@@ -630,38 +630,40 @@ extern void* Lsubstring (void *subj, int p, int l) {
             subject length=%d)", pp, ll, LEN(d->tag));
 }
 
-extern struct re_pattern_buffer *Lregexp (char *regexp) {
-  regex_t *b = (regex_t*) malloc (sizeof (regex_t));
+extern regex_t *Lregexp (char *regexp) {
+  regex_t *regexp_compiled = (regex_t *) malloc(sizeof(regex_t));
 
-  /* printf ("regexp: %s,\t%x\n", regexp, b); */
+  memset(regexp_compiled, 0, sizeof(regex_t));
 
-  memset (b, 0, sizeof (regex_t));
+  int res = regcomp(regexp_compiled, regexp, REG_EXTENDED);
 
-  int n = (int) re_compile_pattern (regexp, strlen (regexp), b);
+  //printf("Lregexp: got compiled regexp %p, for string %s\n", regexp_compiled, regexp);
 
-  if (n != 0) {
-    failure ("%", strerror (n));
-  };
+  if (res != 0) {
+      char buf[100];
+      regerror(res, regexp_compiled, buf, 100);
+      failure("%s", buf);
+  }
 
-  return b;
+  return regexp_compiled;
 }
 
-extern int LregexpMatch (struct re_pattern_buffer *b, char *s, int pos) {
-  int res;
+extern void* LregexpMatch (regex_t *b, char *s, int pos) {
+  regmatch_t match;
 
   ASSERT_BOXED("regexpMatch:1", b);
   ASSERT_STRING("regexpMatch:2", s);
   ASSERT_UNBOXED("regexpMatch:3", pos);
 
-  res = re_match (b, s, LEN(TO_DATA(s)->tag), UNBOX(pos), 0);
+  int res = regexec(b, s + UNBOX(pos), (size_t) 1, &match, 0);
 
-  /* printf ("regexpMatch %x: %s, res=%d\n", b, s+UNBOX(pos), res); */
+  //printf ("regexpMatch %p: %s, res=%d so=%d eo=%d\n", b, s + UNBOX(pos), res, match.rm_so, match.rm_eo);
 
-  if (res) {
-    return BOX (res);
+  if (res == 0 && match.rm_so == 0) {
+      return BOX(match.rm_eo);
+  } else {
+      return BOX(-1);
   }
-
-  return BOX (res);
 }
 
 extern void* Bstring (void*);

src/X86_32.ml

@@ -706,16 +706,37 @@ class env prg =
         let n   = String.length x     in
         let buf = Buffer.create (n*2) in
         let rec iterate i =
-          if i < n
-          then (
-            (match x.[i] with
-            | '"'  -> Buffer.add_string buf "\\\""
-            | '\n' -> Buffer.add_string buf "\n"
-            | '\t' -> Buffer.add_string buf "\t"
-            | c    -> Buffer.add_char buf c
-            );
-            iterate (i+1)
-          )
+          if i < n then
+            match x.[i] with
+            | '"' ->
+                Buffer.add_char buf '\\';
+                Buffer.add_char buf '"';
+                iterate (i + 1)
+            | '\\' -> (
+                if i + 1 >= n then (
+                  Buffer.add_char buf '\\';
+                  Buffer.add_char buf '\\')
+                else
+                  match x.[i + 1] with
+                  | 'n' ->
+                      Buffer.add_char buf '\\';
+                      Buffer.add_char buf 'n';
+                      iterate (i + 2)
+                  | 't' ->
+                      Buffer.add_char buf '\\';
+                      Buffer.add_char buf 't';
+                      iterate (i + 2)
+                  | 'r' ->
+                      Buffer.add_char buf '\\';
+                      Buffer.add_char buf 'r';
+                      iterate (i + 2)
+                  | _ ->
+                      Buffer.add_char buf '\\';
+                      Buffer.add_char buf '\\';
+                      iterate (i + 1))
+            | c ->
+              Buffer.add_char buf c;
+              iterate (i + 1)
         in
         iterate 0;
         Buffer.contents buf