From: Pavel Charvat <pchar@camellia.ms.mff.cuni.cz>
Date: Sun, 5 Mar 2006 22:26:54 +0000 (+0100)
Subject: Charset conversion with allocation on the stack do not
X-Git-Tag: holmes-import~664
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=47fec72b685482a58561be68759e9a8f9059e6b5;p=libucw.git

Charset conversion with allocation on the stack do not
depend on string_table's lengths now.
---

diff --git a/charset/misc/gen-charconv b/charset/misc/gen-charconv
index a731d9ab..ab22b156 100755
--- a/charset/misc/gen-charconv
+++ b/charset/misc/gen-charconv
@@ -182,9 +182,6 @@ print "static unsigned char string_table[] = {\n";
 $i = 256;
 while ($i < $pstr) {
 	$w = $strval{$i};
-	if (length($w) > 3) {
-		print STDERR "Warning: Entries should have at most 3 characters, see charset/stk-charconv.c\n";
-	}
 	print length $w, ",";
 	foreach $x (unpack("C256", $w)) {
 		print " $x,";
diff --git a/charset/stk-charconv.c b/charset/stk-charconv.c
index 3fbc07bb..2d5c2066 100644
--- a/charset/stk-charconv.c
+++ b/charset/stk-charconv.c
@@ -8,25 +8,90 @@
 #include "charset/stk-charconv.h"
 #include <string.h>
 
-uns
-stk_conv_internal(struct conv_context *c, byte *s, uns in_cs, uns out_cs)
+#define INITIAL_MIN_SIZE	16
+#define INITIAL_SCALE		2
+
+void
+stk_conv_init(struct stk_conv_context *c, byte *s, uns in_cs, uns out_cs)
 {
-  /* We do not allocate anything for identical charsets. */
+  uns l = strlen(s);
+  c->count = 0;
+
+  /* For in_cs == out_cs, we emulate stk_strdup */
   if (in_cs == out_cs)
+  {
+    c->size[0] = c->request = l + 1;
+    c->buf[0] = s;
+    c->c.source = NULL;
+    return;
+  }
+ 
+  /* Initialization */
+  conv_init(&c->c);
+  conv_set_charset(&c->c, in_cs, out_cs);
+  c->c.source = s;
+  c->c.source_end = s + l + 1;
+  c->sum = 0;
+
+  /* Size of the first buffer */
+  if (l < (INITIAL_MIN_SIZE - 1) / INITIAL_SCALE)
+    c->request = INITIAL_MIN_SIZE;
+  else
+    c->request = l * INITIAL_SCALE + 1;
+}
+
+int
+stk_conv_step(struct stk_conv_context *c, byte *buf)
+{
+  /* Merge all buffers to the new one and exit */
+  if (!c->c.source)
+  {
+    c->c.dest_start = buf;
+    for (uns i = 0; i <= c->count; i++)
     {
-      c->dest_start = s;
-      return 0;
+      memcpy(buf, c->buf[i], c->size[i]);
+      buf += c->size[i];
     }  
+    return 0;
+  }
   
-  uns l = strlen(s);
+  /* Run conv_run using the new buffer */
+  c->buf[c->count] = c->c.dest_start = c->c.dest = buf;
+  c->c.dest_end = buf + c->request;
+  if (!(conv_run(&c->c) & CONV_SOURCE_END))
+  {
+
+    /* Buffer is too small, continue with a new one */
+    c->size[c->count++] = c->request;
+    c->sum += c->request;
+    c->request <<= 1; /* This can be freely changed */
+    return 1;
+  }
+
+  /* We have used only one buffer for the conversion, no merges are needed */
+  if (!c->count)
+    return 0;
   
-  conv_init(c);
-  conv_set_charset(c, in_cs, out_cs);
-  c->source = s;
-  c->source_end = s + l + 1;
-
-  /* Resulting string can be longer after the conversion.
-   * The following constant must be at least 3 for conversion to UTF-8
-   * and at least the maximum length of the strings in string_table for other charsets. */
-  return 3 * l + 1;
+  /* We can merge everything to the current buffer ... */
+  uns s = c->c.dest - c->c.dest_start;
+  if (c->sum + s <= c->request)
+  {
+    memmove(buf + c->sum, buf, s);
+    for (uns i = 0; i < c->count; i++)
+    {
+      memcpy(buf, c->buf[i], c->size[i]);
+      buf += c->size[i];
+    }
+    return 0;
+  }
+  
+  /* ... or we allocate a new one */
+  else
+  {
+    c->request = c->sum + s;
+    c->size[c->count] = s;
+    c->c.source = NULL;
+    return 1;
+  }
 }
+
diff --git a/charset/stk-charconv.h b/charset/stk-charconv.h
index d63ba746..d8ae585c 100644
--- a/charset/stk-charconv.h
+++ b/charset/stk-charconv.h
@@ -10,18 +10,27 @@
 #include "charset/charconv.h"
 #include <alloca.h>
 
-/* The following macros convert strings between given charsets (CONV_CHARSET_x).
- * The resulting string is allocated on the stack with the exception of cs_in == cs_out, 
- * when the pointer to the input string is returned. */
+/* The following macros convert strings between given charsets (CONV_CHARSET_x). */
 
 #define stk_conv(s, cs_in, cs_out) \
-    ({ struct conv_context _c; uns _l=stk_conv_internal(&_c, (s), (cs_in), (cs_out)); \
-    if (_l) { _c.dest=_c.dest_start=alloca(_l); _c.dest_end=_c.dest+_l; conv_run(&_c); } \
-    _c.dest_start; })
+    ({ struct stk_conv_context _c; stk_conv_init(&_c, (s), (cs_in), (cs_out)); \
+       while (stk_conv_step(&_c, alloca(_c.request))); _c.c.dest_start; })
 
 #define stk_conv_to_utf8(s, cs_in) stk_conv(s, cs_in, CONV_CHARSET_UTF8)
 #define stk_conv_from_utf8(s, cs_out) stk_conv(s, CONV_CHARSET_UTF8, cs_out)
-    
-uns stk_conv_internal(struct conv_context *c, byte *s, uns cs_in, uns cs_out);
+   
+/* Internal structure and routines. */
+
+struct stk_conv_context {
+  struct conv_context c;
+  uns count;
+  uns sum;
+  uns request;
+  byte *buf[16]; 
+  uns size[16];
+};
+
+void stk_conv_init(struct stk_conv_context *c, byte *s, uns cs_in, uns cs_out);
+int stk_conv_step(struct stk_conv_context *c, byte *buf);
 
 #endif