Made conversion interface more general.

[gedcom-parse.git] / utf8 / utf8-locale.c

diff --git a/utf8/utf8-locale.c b/utf8/utf8-locale.c
index 28eeb4b24c8c1b7806c2b2a258f748fc7c9c6aad..115c66ee7312e66bec78e797abaa4f2cc95bc9e2 100644 (file)
--- a/utf8/utf8-locale.c
+++ b/utf8/utf8-locale.c
@@ -9,156 +9,49 @@
 /* $Id$ */
 /* $Name$ */
 
+#include "utf8.h"
 #include <stdlib.h>
-#include <iconv.h>
 #include <assert.h>
-#include <errno.h>
-#include <string.h>
-#include "config.h"
 #include "libcharset.h"
-#include "utf8.h"
-
-#define INITIAL_OUTSIZE 256
 
-static iconv_t utf8_to_locale = (iconv_t) -1;
-static iconv_t locale_to_utf8 = (iconv_t) -1;
-static char*   outbuffer = NULL;
-static size_t  outbufsize = 0;
-static const char* the_unknown = "?";
+static convert_t locale_conv = NULL;
 
 void convert_set_unknown(const char* unknown)
 {
-  the_unknown = unknown;
+  conversion_set_unknown(locale_conv, unknown);
 }
 
 void close_conversion_contexts()
 {
-  iconv_close(utf8_to_locale);
-  iconv_close(locale_to_utf8);
-  utf8_to_locale = (iconv_t) -1;
-  locale_to_utf8 = (iconv_t) -1;
-  free(outbuffer);
+  cleanup_utf8_conversion(locale_conv);
 }
 
 int open_conversion_contexts()
 {
-  assert(utf8_to_locale == (iconv_t) -1);
-  assert(locale_to_utf8 == (iconv_t) -1);
-  utf8_to_locale = iconv_open(locale_charset(), "UTF-8");
-  if (utf8_to_locale == (iconv_t) -1)
-    return -1;
+  assert (locale_conv == NULL);
+  locale_conv = initialize_utf8_conversion(locale_charset());
+
+  if (locale_conv) {
+    atexit(close_conversion_contexts);
+    return 0;
+  }
   else {
-    locale_to_utf8 = iconv_open("UTF-8", locale_charset());
-    if (locale_to_utf8 == (iconv_t) -1) {
-      close_conversion_contexts();
-      return -1;
-    }
-    else {
-      outbufsize = INITIAL_OUTSIZE;
-      outbuffer = (char*)malloc(outbufsize);
-      atexit(close_conversion_contexts);
-      return 0;
-    }
+    return -1;
   }
 }
 
 char* convert_utf8_to_locale(const char* input, int *conv_fails)
 {
-  size_t insize  = strlen(input);
-  size_t outsize;
-  ICONV_CONST char *inptr  = (ICONV_CONST char*) input;
-  char   *outptr;
-  size_t nconv;
+  if (!locale_conv)
+    open_conversion_contexts();
 
-  if (utf8_to_locale == (iconv_t) -1 && (open_conversion_contexts() == -1)) {
-    if (conv_fails != NULL) *conv_fails = insize;
-    return NULL;
-  }
-  assert(utf8_to_locale != (iconv_t) -1);
-  /* make sure we start from an empty state */
-  iconv(utf8_to_locale, NULL, NULL, NULL, NULL);
-  if (conv_fails != NULL) *conv_fails = 0;
-  /* set up output buffer (empty it) */
-  outptr  = outbuffer;
-  outsize = outbufsize;
-  memset(outbuffer, 0, outbufsize);
-  nconv = iconv(utf8_to_locale, &inptr, &insize, &outptr, &outsize);
-  while (nconv == (size_t)-1) {
-    if (errno == E2BIG) {
-      /* grow the output buffer */
-      size_t outlen;
-      outlen     = outptr - outbuffer;
-      outbufsize *= 2;
-      outbuffer  = realloc(outbuffer, outbufsize);
-      outptr     = outbuffer + outlen;
-      outsize    = outbufsize - outlen;
-      memset(outptr, 0, outsize);
-    }
-    else if (errno == EILSEQ) {
-      /* skip over character */
-      const char* unkn_ptr = the_unknown;
-      if (conv_fails != NULL) (*conv_fails)++;
-      if ((*inptr & 0x80) == 0) {
-       /* an ASCII character, just skip one (this case is very improbable) */
-       inptr++; insize--;
-      }
-      else {
-       /* a general UTF-8 character, skip all 0x10xxxxxx bytes */
-       inptr++; insize--;
-       while ((*inptr & 0xC0) == 0x80) {
-         inptr++; insize--;
-       }
-      }
-      /* append the "unknown" string to the output */
-      while (*unkn_ptr) { *outptr++ = *unkn_ptr++; outsize--; }
-    }
-    else {
-      /* EINVAL should not happen, since we convert entire strings */
-      /* EBADF is an error which should be captured by the assert above */
-      if (conv_fails != NULL) *conv_fails += insize;
-      return NULL;
-    }
-    nconv = iconv(utf8_to_locale, &inptr, &insize, &outptr, &outsize);
-  }
-  return outbuffer;
+  return convert_from_utf8(locale_conv, input, conv_fails);
 }
 
 char* convert_locale_to_utf8(const char* input)
 {
-  size_t insize  = strlen(input);
-  size_t outsize;
-  ICONV_CONST char *inptr  = (ICONV_CONST char*) input;
-  char   *outptr;
-  size_t nconv;
+  if (!locale_conv)
+    open_conversion_contexts();
 
-  if (locale_to_utf8 == (iconv_t) -1 && (open_conversion_contexts() == -1))
-    return NULL;
-  assert(locale_to_utf8 != (iconv_t) -1);
-  /* make sure we start from an empty state */
-  iconv(locale_to_utf8, NULL, NULL, NULL, NULL);
-  /* set up output buffer (empty it) */
-  outptr  = outbuffer;
-  outsize = outbufsize;
-  memset(outbuffer, 0, outbufsize);
-  nconv = iconv(locale_to_utf8, &inptr, &insize, &outptr, &outsize);
-  while (nconv == (size_t)-1) {
-    if (errno == E2BIG) {
-      /* grow the output buffer */
-      size_t outlen;
-      outlen     = outptr - outbuffer;
-      outbufsize *= 2;
-      outbuffer  = realloc(outbuffer, outbufsize);
-      outptr     = outbuffer + outlen;
-      outsize    = outbufsize - outlen;
-      memset(outptr, 0, outsize);
-    }
-    else {
-      /* EILSEQ should not happen, because UTF-8 can represent anything */
-      /* EINVAL should not happen, since we convert entire strings */
-      /* EBADF is an error which should be captured by the assert above */
-      return NULL;
-    }
-    nconv = iconv(locale_to_utf8, &inptr, &insize, &outptr, &outsize);
-  }
-  return outbuffer;
+  return convert_to_utf8(locale_conv, input);
 }