Bugfix.

[chaz/tar] / src / transform.c

diff --git a/src/transform.c b/src/transform.c
index acdf552cf3b88d9237152e5004a305faf3b20122..fc3eb48779c599f345bb495fb50ddef94c8926fb 100644 (file)
--- a/src/transform.c
+++ b/src/transform.c
@@ -1,9 +1,9 @@
 /* This file is part of GNU tar. 
 /* This file is part of GNU tar. 

-   Copyright (C) 2006 Free Software Foundation, Inc.
+   Copyright (C) 2006, 2007, 2008 Free Software Foundation, Inc.

 
    This program is free software; you can redistribute it and/or modify it
    under the terms of the GNU General Public License as published by the
 
    This program is free software; you can redistribute it and/or modify it
    under the terms of the GNU General Public License as published by the

-   Free Software Foundation; either version 2, or (at your option) any later
+   Free Software Foundation; either version 3, or (at your option) any later

    version.
 
    This program is distributed in the hope that it will be useful, but
    version.
 
    This program is distributed in the hope that it will be useful, but


@@ -21,18 +21,24 @@
 
 enum transform_type
   {
 
 enum transform_type
   {

-    transform_none,

     transform_first,
     transform_global
     transform_first,
     transform_global

-  }
-transform_type = transform_none;
-static regex_t regex;
-static struct obstack stk;
+  };

 
 enum replace_segm_type
   {
     segm_literal,   /* Literal segment */
     segm_backref,   /* Back-reference segment */
 
 enum replace_segm_type
   {
     segm_literal,   /* Literal segment */
     segm_backref,   /* Back-reference segment */

+    segm_case_ctl   /* Case control segment (GNU extension) */
+  };
+
+enum case_ctl_type
+  {
+    ctl_stop,       /* Stop case conversion */ 
+    ctl_upcase_next,/* Turn the next character to uppercase */ 
+    ctl_locase_next,/* Turn the next character to lowercase */
+    ctl_upcase,     /* Turn the replacement to uppercase until ctl_stop */
+    ctl_locase      /* Turn the replacement to lowercase until ctl_stop */

   };
 
 struct replace_segm
   };
 
 struct replace_segm


@@ -45,35 +51,59 @@ struct replace_segm
     {
       char *ptr;
       size_t size;
     {
       char *ptr;
       size_t size;

-    } literal;
-    size_t ref;
+    } literal;                /* type == segm_literal */   
+    size_t ref;               /* type == segm_backref */
+    enum case_ctl_type ctl;   /* type == segm_case_ctl */ 

   } v;
 };
 
   } v;
 };
 

-static struct replace_segm *repl_head, *repl_tail;
-static segm_count;
+struct transform
+{
+  struct transform *next;
+  enum transform_type transform_type;
+  unsigned match_number;
+  regex_t regex;
+  /* Compiled replacement expression */
+  struct replace_segm *repl_head, *repl_tail;
+  size_t segm_count; /* Number of elements in the above list */
+};
+
+\f
+static struct transform *transform_head, *transform_tail;
+
+static struct transform *
+new_transform ()
+{
+  struct transform *p = xzalloc (sizeof *p);
+  if (transform_tail)
+    transform_tail->next = p;
+  else
+    transform_head = p;
+  transform_tail = p;
+  return p;
+}

 
 static struct replace_segm *
 
 static struct replace_segm *

-add_segment (void)
+add_segment (struct transform *tf)

 {
   struct replace_segm *segm = xmalloc (sizeof *segm);
   segm->next = NULL;
 {
   struct replace_segm *segm = xmalloc (sizeof *segm);
   segm->next = NULL;

-  if (repl_tail)
-    repl_tail->next = segm;
+  if (tf->repl_tail)
+    tf->repl_tail->next = segm;

   else
   else

-    repl_head = segm;
-  repl_tail = segm;
-  segm_count++;
+    tf->repl_head = segm;
+  tf->repl_tail = segm;
+  tf->segm_count++;

   return segm;
 }
 
 static void
   return segm;
 }
 
 static void

-add_literal_segment (char *str, char *end)
+add_literal_segment (struct transform *tf, char *str, char *end)

 {
   size_t len = end - str;
   if (len)
     {
 {
   size_t len = end - str;
   if (len)
     {

-      struct replace_segm *segm = add_segment ();
+      struct replace_segm *segm = add_segment (tf);

       segm->type = segm_literal;
       segm->v.literal.ptr = xmalloc (len + 1);
       memcpy (segm->v.literal.ptr, str, len);
       segm->type = segm_literal;
       segm->v.literal.ptr = xmalloc (len + 1);
       memcpy (segm->v.literal.ptr, str, len);


@@ -83,40 +113,41 @@ add_literal_segment (char *str, char *end)
 }
 
 static void
 }
 
 static void

-add_char_segment (int chr)
+add_char_segment (struct transform *tf, int chr)

 {
 {

-  struct replace_segm *segm = add_segment ();
+  struct replace_segm *segm = add_segment (tf);

   segm->type = segm_literal;
   segm->v.literal.ptr = xmalloc (2);
   segm->v.literal.ptr[0] = chr;
   segm->v.literal.ptr[1] = 0;
   segm->type = segm_literal;
   segm->v.literal.ptr = xmalloc (2);
   segm->v.literal.ptr[0] = chr;
   segm->v.literal.ptr[1] = 0;

-  segm->v.literal.size = 2;
+  segm->v.literal.size = 1;

 }
 
 static void
 }
 
 static void

-add_backref_segment (size_t ref)
+add_backref_segment (struct transform *tf, size_t ref)

 {
 {

-  struct replace_segm *segm = add_segment ();
+  struct replace_segm *segm = add_segment (tf);

   segm->type = segm_backref;
   segm->v.ref = ref;
 }
 
   segm->type = segm_backref;
   segm->v.ref = ref;
 }
 

-void
-set_transform_expr (const char *expr)
+static void
+add_case_ctl_segment (struct transform *tf, enum case_ctl_type ctl)
+{
+  struct replace_segm *segm = add_segment (tf);
+  segm->type = segm_case_ctl;
+  segm->v.ctl = ctl;
+}
+
+static const char *
+parse_transform_expr (const char *expr)

 {
   int delim;
   int i, j, rc;
   char *str, *beg, *cur;
   const char *p;
   int cflags = 0;
 {
   int delim;
   int i, j, rc;
   char *str, *beg, *cur;
   const char *p;
   int cflags = 0;

-
-  if (transform_type == transform_none)
-    obstack_init (&stk);
-  else
-    {
-      /* Redefinition of the transform expression */
-      regfree (&regex);
-    }
+  struct transform *tf = new_transform ();

 
   if (expr[0] != 's')
     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 
   if (expr[0] != 's')
     USAGE_ERROR ((0, 0, _("Invalid transform expression")));


@@ -140,12 +171,12 @@ set_transform_expr (const char *expr)
     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 
   /* Check flags */
     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 
   /* Check flags */

-  transform_type = transform_first;
-  for (p = expr + j + 1; *p; p++)
+  tf->transform_type = transform_first;
+  for (p = expr + j + 1; *p && *p != ';'; p++)

     switch (*p)
       {
       case 'g':
     switch (*p)
       {
       case 'g':

-       transform_type = transform_global;
+       tf->transform_type = transform_global;

        break;
 
       case 'i':
        break;
 
       case 'i':


@@ -156,26 +187,36 @@ set_transform_expr (const char *expr)
        cflags |= REG_EXTENDED;
        break;
        
        cflags |= REG_EXTENDED;
        break;
        

+      case '0': case '1': case '2': case '3': case '4':
+      case '5': case '6': case '7': case '8': case '9':
+       tf->match_number = strtoul (p, (char**) &p, 0);
+       p--;
+       break;
+

       default:
       default:

-       USAGE_ERROR ((0, 0, _("Unknown flag in transform expression")));
+       USAGE_ERROR ((0, 0, _("Unknown flag in transform expression: %c"),
+                     *p));

       }
 
       }
 

+  if (*p == ';')
+    p++;
+  

   /* Extract and compile regex */
   str = xmalloc (i - 1);
   memcpy (str, expr + 2, i - 2);
   str[i - 2] = 0;
 
   /* Extract and compile regex */
   str = xmalloc (i - 1);
   memcpy (str, expr + 2, i - 2);
   str[i - 2] = 0;
 

-  rc = regcomp (&regex, str, cflags);
+  rc = regcomp (&tf->regex, str, cflags);

   
   if (rc)
     {
       char errbuf[512];
   
   if (rc)
     {
       char errbuf[512];

-      regerror (rc, &regex, errbuf, sizeof (errbuf));
+      regerror (rc, &tf->regex, errbuf, sizeof (errbuf));

       USAGE_ERROR ((0, 0, _("Invalid transform expression: %s"), errbuf));
     }
 
   if (str[0] == '^' || str[strlen (str) - 1] == '$')
       USAGE_ERROR ((0, 0, _("Invalid transform expression: %s"), errbuf));
     }
 
   if (str[0] == '^' || str[strlen (str) - 1] == '$')

-    transform_type = transform_first;
+    tf->transform_type = transform_first;

   
   free (str);
 
   
   free (str);
 


@@ -191,69 +232,101 @@ set_transform_expr (const char *expr)
        {
          size_t n;
          
        {
          size_t n;
          

-         add_literal_segment (beg, cur);
+         add_literal_segment (tf, beg, cur);

          switch (*++cur)
            {
            case '0': case '1': case '2': case '3': case '4':
            case '5': case '6': case '7': case '8': case '9':
              n = strtoul (cur, &cur, 10);
          switch (*++cur)
            {
            case '0': case '1': case '2': case '3': case '4':
            case '5': case '6': case '7': case '8': case '9':
              n = strtoul (cur, &cur, 10);

-             if (n > regex.re_nsub)
+             if (n > tf->regex.re_nsub)

                USAGE_ERROR ((0, 0, _("Invalid transform replacement: back reference out of range")));
                USAGE_ERROR ((0, 0, _("Invalid transform replacement: back reference out of range")));

-             add_backref_segment (n);
+             add_backref_segment (tf, n);

              break;
 
            case '\\':
              break;
 
            case '\\':

-             add_char_segment ('\\');
+             add_char_segment (tf, '\\');

              cur++;
              break;
 
            case 'a':
              cur++;
              break;
 
            case 'a':

-             add_char_segment ('\a');
+             add_char_segment (tf, '\a');

              cur++;
              break;
              
            case 'b':
              cur++;
              break;
              
            case 'b':

-             add_char_segment ('\b');
+             add_char_segment (tf, '\b');

              cur++;
              break;
              
            case 'f':
              cur++;
              break;
              
            case 'f':

-             add_char_segment ('\f');
+             add_char_segment (tf, '\f');

              cur++;
              break;
              
            case 'n':
              cur++;
              break;
              
            case 'n':

-             add_char_segment ('\n');
+             add_char_segment (tf, '\n');

              cur++;
              break;
              
            case 'r':
              cur++;
              break;
              
            case 'r':

-             add_char_segment ('\r');
+             add_char_segment (tf, '\r');

              cur++;
              break;
              
            case 't':
              cur++;
              break;
              
            case 't':

-             add_char_segment ('\t');
+             add_char_segment (tf, '\t');

              cur++;
              break;
              
            case 'v':
              cur++;
              break;
              
            case 'v':

-             add_char_segment ('\v');
+             add_char_segment (tf, '\v');

              cur++;
              break;
 
            case '&':
              cur++;
              break;
 
            case '&':

-             add_char_segment ('&');
+             add_char_segment (tf, '&');

              cur++;
              break;
              cur++;
              break;

-
+             
+           case 'L':
+             /* Turn the replacement to lowercase until a `\U' or `\E'
+                is found, */
+             add_case_ctl_segment (tf, ctl_locase);
+             cur++;
+             break;
+ 
+           case 'l':
+             /* Turn the next character to lowercase, */
+             add_case_ctl_segment (tf, ctl_locase_next);
+             cur++;
+             break;
+             
+           case 'U':
+             /* Turn the replacement to uppercase until a `\L' or `\E'
+                is found, */
+             add_case_ctl_segment (tf, ctl_upcase);
+             cur++;
+             break;
+             
+           case 'u':
+             /* Turn the next character to uppercase, */
+             add_case_ctl_segment (tf, ctl_upcase_next);
+             cur++;
+             break;
+             
+           case 'E':
+             /* Stop case conversion started by `\L' or `\U'. */
+             add_case_ctl_segment (tf, ctl_stop);
+             cur++;
+             break;
+  

            default:
              /* Try to be nice */
              {
                char buf[2];
                buf[0] = '\\';
                buf[1] = *cur;
            default:
              /* Try to be nice */
              {
                char buf[2];
                buf[0] = '\\';
                buf[1] = *cur;

-               add_literal_segment (buf, buf + 2);
+               add_literal_segment (tf, buf, buf + 2);

              }
              cur++;
              break;
              }
              cur++;
              break;


@@ -262,60 +335,168 @@ set_transform_expr (const char *expr)
        }
       else if (*cur == '&')
        {
        }
       else if (*cur == '&')
        {

-         add_literal_segment (beg, cur);
-         add_backref_segment (0);
+         add_literal_segment (tf, beg, cur);
+         add_backref_segment (tf, 0);

          beg = ++cur;
        }
       else
        cur++;
     }
          beg = ++cur;
        }
       else
        cur++;
     }

-  add_literal_segment (beg, cur);
+  add_literal_segment (tf, beg, cur);
+
+  return p;
+}
+
+void
+set_transform_expr (const char *expr)
+{
+  while (*expr)
+    expr = parse_transform_expr (expr);
+}
+
+/* Run case conversion specified by CASE_CTL on array PTR of SIZE
+   characters. Returns pointer to statically allocated storage. */
+static char *
+run_case_conv (enum case_ctl_type case_ctl, char *ptr, size_t size)
+{
+  static char *case_ctl_buffer;
+  static size_t case_ctl_bufsize;
+  char *p;

   
   

+  if (case_ctl_bufsize < size)
+    {
+      case_ctl_bufsize = size;
+      case_ctl_buffer = xrealloc (case_ctl_buffer, case_ctl_bufsize);
+    }
+  memcpy (case_ctl_buffer, ptr, size);
+  switch (case_ctl)
+    {
+    case ctl_upcase_next:
+      case_ctl_buffer[0] = toupper (case_ctl_buffer[0]);
+      break;
+      
+    case ctl_locase_next:
+      case_ctl_buffer[0] = tolower (case_ctl_buffer[0]);
+      break;
+      
+    case ctl_upcase:
+      for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
+       *p = toupper (*p);
+      break;
+      
+    case ctl_locase:
+      for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
+       *p = tolower (*p);
+      break;
+
+    case ctl_stop:
+      break;
+    }
+  return case_ctl_buffer;

 }
 
 }
 

-bool
-_transform_name_to_obstack (char *input)
+\f
+static struct obstack stk;
+static bool stk_init;
+
+void
+_single_transform_name_to_obstack (struct transform *tf, char *input)

 {
   regmatch_t *rmp;
 {
   regmatch_t *rmp;

-  char *p;

   int rc;
   int rc;

+  size_t nmatches = 0;
+  enum case_ctl_type case_ctl = ctl_stop,  /* Current case conversion op */
+                     save_ctl = ctl_stop;  /* Saved case_ctl for \u and \l */

   
   

-  if (transform_type == transform_none)
-    return false;
-
-  rmp = xmalloc ((regex.re_nsub + 1) * sizeof (*rmp));
+  /* Reset case conversion after a single-char operation */
+#define CASE_CTL_RESET()  if (case_ctl == ctl_upcase_next     \
+                             || case_ctl == ctl_locase_next) \
+                            {                                 \
+                              case_ctl = save_ctl;            \
+                              save_ctl = ctl_stop;            \
+                           }
+  
+  rmp = xmalloc ((tf->regex.re_nsub + 1) * sizeof (*rmp));

 
   while (*input)
     {
       size_t disp;
 
   while (*input)
     {
       size_t disp;

+      char *ptr;

       
       

-      rc = regexec (&regex, input, regex.re_nsub + 1, rmp, 0);
+      rc = regexec (&tf->regex, input, tf->regex.re_nsub + 1, rmp, 0);

       
       if (rc == 0)
        {
          struct replace_segm *segm;
       
       if (rc == 0)
        {
          struct replace_segm *segm;

-
+         

          disp = rmp[0].rm_eo;
 
          if (rmp[0].rm_so)
            obstack_grow (&stk, input, rmp[0].rm_so);
          disp = rmp[0].rm_eo;
 
          if (rmp[0].rm_so)
            obstack_grow (&stk, input, rmp[0].rm_so);

-         
-         for (segm = repl_head; segm; segm = segm->next)
+
+         nmatches++;
+         if (tf->match_number && nmatches < tf->match_number)
+           {
+             obstack_grow (&stk, input, disp);
+             input += disp;
+             continue;
+           }
+
+         for (segm = tf->repl_head; segm; segm = segm->next)

            {
              switch (segm->type)
                {
                case segm_literal:    /* Literal segment */
            {
              switch (segm->type)
                {
                case segm_literal:    /* Literal segment */

-                 obstack_grow (&stk, segm->v.literal.ptr,
-                               segm->v.literal.size);
+                 if (case_ctl == ctl_stop)
+                   ptr = segm->v.literal.ptr;
+                 else
+                   {
+                     ptr = run_case_conv (case_ctl,
+                                          segm->v.literal.ptr,
+                                          segm->v.literal.size);
+                     CASE_CTL_RESET();
+                   }
+                 obstack_grow (&stk, ptr, segm->v.literal.size);

                  break;
              
                case segm_backref:    /* Back-reference segment */
                  if (rmp[segm->v.ref].rm_so != -1
                      && rmp[segm->v.ref].rm_eo != -1)
                  break;
              
                case segm_backref:    /* Back-reference segment */
                  if (rmp[segm->v.ref].rm_so != -1
                      && rmp[segm->v.ref].rm_eo != -1)

-                   obstack_grow (&stk,
-                                 input + rmp[segm->v.ref].rm_so,
-                             rmp[segm->v.ref].rm_eo - rmp[segm->v.ref].rm_so);
+                   {
+                     size_t size = rmp[segm->v.ref].rm_eo
+                                     - rmp[segm->v.ref].rm_so;
+                     ptr = input + rmp[segm->v.ref].rm_so;
+                     if (case_ctl != ctl_stop)
+                       {
+                         ptr = run_case_conv (case_ctl, ptr, size);
+                         CASE_CTL_RESET();
+                       }
+                     
+                     obstack_grow (&stk, ptr, size);
+                   }

                  break;
                  break;

+
+               case segm_case_ctl:
+                 switch (segm->v.ctl)
+                   {
+                   case ctl_upcase_next:
+                   case ctl_locase_next:
+                     switch (save_ctl)
+                       {
+                       case ctl_stop:
+                       case ctl_upcase:
+                       case ctl_locase:
+                         save_ctl = case_ctl;
+                       default:
+                         break;
+                       }
+                     /*FALL THROUGH*/
+                     
+                   case ctl_upcase:
+                   case ctl_locase:
+                   case ctl_stop:
+                     case_ctl = segm->v.ctl;
+                   }

                }
            }
        }
                }
            }
        }


@@ -327,7 +508,7 @@ _transform_name_to_obstack (char *input)
 
       input += disp;
 
 
       input += disp;
 

-      if (transform_type == transform_first)
+      if (tf->transform_type == transform_first)

        {
          obstack_grow (&stk, input, strlen (input));
          break;
        {
          obstack_grow (&stk, input, strlen (input));
          break;


@@ -336,25 +517,42 @@ _transform_name_to_obstack (char *input)
 
   obstack_1grow (&stk, 0);
   free (rmp);
 
   obstack_1grow (&stk, 0);
   free (rmp);

-  return true;
+}
+
+bool
+_transform_name_to_obstack (char *input, char **output)
+{
+  struct transform *tf;
+
+  if (!stk_init)
+    {
+      obstack_init (&stk);
+      stk_init = true;
+    }
+  
+  for (tf = transform_head; tf; tf = tf->next)
+    {
+      _single_transform_name_to_obstack (tf, input);
+      input = obstack_finish (&stk);
+    }
+  *output = input;
+  return transform_head != NULL;

 }
   
 bool
 }
   
 bool

-transform_name_fp (char **pinput, char *(*fun)(char *))
+transform_name_fp (char **pinput, char *(*fun)(char *, void *), void *dat)

 {
 {

-    char *str, *p;
-    bool ret = _transform_name_to_obstack (*pinput);
+    char *str;
+    bool ret = _transform_name_to_obstack (*pinput, &str);

     if (ret)
       {
     if (ret)
       {

-       str = obstack_finish (&stk);
-       assign_string (pinput, fun ? fun (str) : str);
+       assign_string (pinput, fun ? fun (str, dat) : str);

        obstack_free (&stk, str);
       }
     else if (fun)
       {
        obstack_free (&stk, str);
       }
     else if (fun)
       {

-       str = *pinput;

        *pinput = NULL;
        *pinput = NULL;

-       assign_string (pinput, fun (str));
+       assign_string (pinput, fun (str, dat));

        free (str);
        ret = true;
       }
        free (str);
        ret = true;
       }


@@ -364,23 +562,6 @@ transform_name_fp (char **pinput, char *(*fun)(char *))
 bool
 transform_name (char **pinput)
 {
 bool
 transform_name (char **pinput)
 {

-  return transform_name_fp (pinput, NULL);
+  return transform_name_fp (pinput, NULL, NULL);

 }
 
 }
 

-#if 0
-void
-read_and_transform_loop ()
-{
-  char buf[512];
-  while (fgets (buf, sizeof buf, stdin))
-    {
-      char *p = buf + strlen (buf);
-      if (p[-1] == '\n')
-       p[-1] = 0;
-      if (transform_name (buf, &p))
-       printf ("=> %s\n", p);
-      else
-       printf ("=\n");
-    }
-}
-#endif