make pdf/eps/svg formats utf8 aware too -- Peter Speck

git-svn-id: svn://svn.oetiker.ch/rrdtool/branches/1.2/program@612 a5681a0c-68f1-0310-ab6d-d61299d08faa

diff --git a/src/rrd_afm.c b/src/rrd_afm.c
index 2a2ec6cb415cc2f184f6b3d1d29358d57b84ca2f..f9acc161a782109d127ae01d9008236425b23ce9 100644 (file)
--- a/src/rrd_afm.c
+++ b/src/rrd_afm.c
@@ -13,7 +13,6 @@
 #include "rrd_afm.h"
 #include "rrd_afm_data.h"
 
-#include <stdlib.h>
 #include <stdio.h>
 
 #ifdef HAVE_STRING_H
@@ -187,24 +186,49 @@ static long afm_find_kern(const afm_fontinfo *fontinfo,
 }
 
 /* measure width of a text string */
-double afm_get_text_width ( double UNUSED(start), const char* font, double size,
-          double UNUSED(tabwidth), const char* text)
+double afm_get_text_width( double start, const char* font, double size,
+          double tabwidth, const char* text)
+{
+#ifdef HAVE_MBSTOWCS     
+    size_t clen = strlen(text) + 1;
+    wchar_t *cstr = malloc(sizeof(wchar_t) * clen); /* yes we are allocating probably too much here, I know */
+    int text_count = mbstowcs(cstr, text, clen);
+    double w;
+    if (text_count == -1)
+           text_count = mbstowcs(cstr, "Enc-Err", 6);
+#ifdef __APPLE__
+       while (text_count > 0) {
+               text_count--;
+               cstr[text_count] = afm_fix_osx_charset(cstr[text_count]); /* unsafe macro */
+       }
+#endif
+    w = afm_get_text_width_wide(start, font, size, tabwidth, cstr);
+    free(cstr);
+    return w;
+#else
+    return afm_get_text_width_wide(start, font, size, tabwidth, text);
+#endif
+}
+
+double afm_get_text_width_wide( double UNUSED(start), const char* font, double size,
+          double UNUSED(tabwidth), const afm_char* text)
 {
   const afm_fontinfo *fontinfo = afm_findfont(font);
   long width = 0;
   double widthf;
-  const unsigned char *up = (const unsigned char*)text;
+  const afm_char *up = text;
   DLOG((stderr, "================= %s\n", text));
-  if (fontinfo == NULL)
-    return size * strlen(text);
+  if (fontinfo == NULL) {
+      while (*up)
+         up++;
+    return size * (up - text);
+  }
   while (1) {
     afm_unicode ch1, ch2;
     int idx1, kern_idx;
     if ((ch1 = *up) == 0)
-      break;
-    ch1 = afm_host2unicode(ch1); /* unsafe macro */
+        break;
     ch2 = *++up;
-    ch2 = afm_host2unicode(ch2); /* unsafe macro */
     DLOG((stderr, "------------- Loop: %d + %d (%c%c)   at %d\n",
           ch1, ch2, ch1, ch2 ? ch2 : ' ',
          (up - (const unsigned char*)text) - 1));
@@ -217,7 +241,7 @@ double afm_get_text_width ( double UNUSED(start), const char* font, double size,
       if (ch1_new) {
         ch1 = ch1_new;
         idx1 = afm_find_char_index(fontinfo, ch1);
-        ch2 = afm_host2unicode(*++up);
+        ch2 = *++up;
         DLOG((stderr, "  -> idx1 = %d, ch2 = %d (%c)\n", 
             idx1, ch2, ch2 ? ch2 : ' '));
       }

diff --git a/src/rrd_afm.h b/src/rrd_afm.h
index cedc9949d30a42ae76e54aafee405e1c87651f1b..9d12e494cf49187314278f9a8532d86a825ac09f 100644 (file)
--- a/src/rrd_afm.h
+++ b/src/rrd_afm.h
@@ -7,6 +7,13 @@
 #ifndef  RRD_AFM_H
 #define RRD_AFM_H
 
+#include <stdlib.h>
+ 
+#ifdef HAVE_MBSTOWCS
+#define afm_char wchar_t
+#else
+#define afm_char unsigned char
+#endif
 /*
    If the font specified by the name parameter in the routes below
    is not found
@@ -21,8 +28,10 @@
 
 /* measure width of a text string */
 /* fontname can be full name or postscript name */
-double afm_get_text_width ( double start, const char* font, double size,
+double afm_get_text_width( double start, const char* font, double size,
                            double tabwidth, const char* text);
+double afm_get_text_width_wide( double start, const char* font, double size,
+                           double tabwidth, const afm_char* text);
 
 double afm_get_ascender(const char* font, double size);
 double afm_get_descender(const char* font, double size);
@@ -35,11 +44,11 @@ const char *afm_get_font_name(const char* font);
 #ifdef __APPLE__
 /* need charset conversion from macintosh to unicode. */
 extern const unsigned char afm_mac2iso[128];
-#define afm_host2unicode(c) \
-       ( (c) >= 128 ? afm_mac2iso[(c) - 128] : (c))
+#define afm_fix_osx_charset(c) \
+       ( (c) >= 128 && (c) <= 255 ? afm_mac2iso[(c) - 128] : (c))
 #else
 /* UNSAFE macro */
-#define afm_host2unicode(a_unsigned_char) ((unsigned int)(a_unsigned_char))
+#define afm_fix_osx_charset(x) (x)
 #endif
 
 #endif

diff --git a/src/rrd_gfx.c b/src/rrd_gfx.c
index d01b90487a5bbb03edc074e0181b503fe460498c..a241fc28bc9088dbd0d1e44e8cc5ac08405715df 100644 (file)
--- a/src/rrd_gfx.c
+++ b/src/rrd_gfx.c
@@ -415,6 +415,7 @@ gfx_string gfx_string_create(gfx_canvas_t *canvas, FT_Face face,const char *text
        have a witespace glyph inserted, but set its width such that the distance
     of the new right edge is x times tabwidth from 0,0 where x is an integer. */    
     unsigned int letter = cstr[n];
+       letter = afm_fix_osx_charset(letter); /* unsafe macro */
           
     gottab = 0;
     if (letter == '\\' && n+1 < string->count && cstr[n+1] == 't'){
@@ -1038,37 +1039,46 @@ static void svg_close_tag_empty_node(FILE *fp)
  
 static void svg_write_text(FILE *fp, const char *text)
 {
-   const unsigned char *p, *start, *last;
-   unsigned int ch;
-   p = (const unsigned char*)text;
-   if (!p)
-     return;
-   /* trim leading spaces */
-   while (*p == ' ')
-     p++;
-   start = p;
-   /* trim trailing spaces */
-   last = p - 1;
-   while ((ch = *p) != 0) {
-     if (ch != ' ')
-       last = p;
-     p++;
-  }
-  /* encode trimmed text */
-  p = start;
-  while (p <= last) {
+#ifdef HAVE_MBSTOWCS
+    size_t clen;
+    wchar_t *p, *cstr, ch;
+    int text_count;
+    if (!text)
+       return;
+    clen = strlen(text) + 1;
+    cstr = malloc(sizeof(wchar_t) * clen);
+    text_count = mbstowcs(cstr, text, clen);
+    if (text_count == -1)
+       text_count = mbstowcs(cstr, "Enc-Err", 6);
+    p = cstr;
+#else
+    const unsigned char *p = text, ch;
+    if (!p)
+       return;
+#endif
+  while (1) {
     ch = *p++;
-    ch = afm_host2unicode(ch); /* unsafe macro */
+    ch = afm_fix_osx_charset(ch); /* unsafe macro */
     switch (ch) {
+    case 0:
+#ifdef HAVE_MBSTOWCS     
+    free(cstr);
+#endif
+    return;
     case '&': fputs("&amp;", fp); break;
     case '<': fputs("&lt;", fp); break;
     case '>': fputs("&gt;", fp); break;
     case '"': fputs("&quot;", fp); break;
     default:
-      if (ch >= 127)
-       fprintf(fp, "&#%d;", ch);
+        if (ch == 32) {
+            if (p <= cstr + 1 || !*p || *p == 32)
+                fputs("&#160;", fp); /* non-breaking space in unicode */
+            else
+                fputc(32, fp);
+        } else if (ch < 32 || ch >= 127)
+       fprintf(fp, "&#%d;", (int)ch);
       else
-       putc(ch, fp);
+       putc((char)ch, fp);
      }
    }
 }
@@ -1707,10 +1717,26 @@ static void eps_write_linearea(eps_state *state, gfx_node_t *node)
 static void eps_write_text(eps_state *state, gfx_node_t *node)
 {
   FILE *fp = state->fp;
-  const char *p;
   const char *ps_font = afm_get_font_postscript_name(node->filename);
   int lineLen = 0;
   pdf_coords g;
+#ifdef HAVE_MBSTOWCS
+    size_t clen;
+    wchar_t *p, *cstr, ch;
+    int text_count;
+    if (!node->text)
+       return;
+    clen = strlen(node->text) + 1;
+    cstr = malloc(sizeof(wchar_t) * clen);
+    text_count = mbstowcs(cstr, node->text, clen);
+    if (text_count == -1)
+       text_count = mbstowcs(cstr, "Enc-Err", 6);
+    p = cstr;
+#else
+    const unsigned char *p = node->text, ch;
+    if (!p)
+       return;
+#endif
   pdf_calc(state->page_height, node, &g);
   eps_set_color(state, node->color);
   if (strcmp(ps_font, state->font) || node->size != state->font_size) {
@@ -1723,11 +1749,11 @@ static void eps_write_text(eps_state *state, gfx_node_t *node)
          fputs("T1 ", fp);
   fputs("(", fp);
   lineLen = 20;
-  p = node->text;
   while (1) {
-    unsigned char ch = *(unsigned char*)p;
+    ch = *p;
     if (!ch)
       break;
+       ch = afm_fix_osx_charset(ch); /* unsafe macro */
     if (++lineLen > 70) {
       fputs("\\\n", fp); /* backslash and \n */
       lineLen = 0;
@@ -1750,7 +1776,9 @@ static void eps_write_text(eps_state *state, gfx_node_t *node)
         fputs("\\t", fp);
         break;
       default:
-        if (ch >= 126 || ch < 32) {
+        if (ch > 255) {
+            fputc('?', fp);
+        } else if (ch >= 126 || ch < 32) {
           fprintf(fp, "\\%03o", ch);
           lineLen += 3;
         } else {
@@ -1759,6 +1787,9 @@ static void eps_write_text(eps_state *state, gfx_node_t *node)
       }
       p++;
   }
+#ifdef HAVE_MBSTOWCS
+  free(cstr);
+#endif
   if (node->angle) {
         /* can't use svg_write_number as 2 decimals is far from enough to avoid
                skewed text */
@@ -1918,6 +1949,17 @@ static void pdf_put(pdf_buffer *buf, const char *text, int len)
   buf->current_size += len;
 }
 
+static void pdf_put_char(pdf_buffer *buf, char c)
+{
+    if (buf->alloc_size >= buf->current_size + 1) {
+       buf->data[buf->current_size++] = c;
+    } else {
+       char tmp[1];
+       tmp[0] = (char)c;
+       pdf_put(buf, tmp, 1);
+    }
+}
+
 static void pdf_puts(pdf_buffer *buf, const char *text)
 {
   pdf_put(buf, text, strlen(text));
@@ -1948,18 +1990,23 @@ static void pdf_putnumber(pdf_buffer *buf, double d)
   pdf_puts(buf, tmp);
 }
 
-static void pdf_put_string_contents(pdf_buffer *buf, const char *text)
+static void pdf_put_string_contents_wide(pdf_buffer *buf, const afm_char *text)
 {
-    const char *p = text;
+    const afm_char *p = text;
     while (1) {
-       unsigned char ch = *(unsigned char*)p;
+       afm_char ch = *p;
+       ch = afm_fix_osx_charset(ch); /* unsafe macro */
        switch (ch) {
-           case 0: return;
+           case 0:
+               return;
            case '(':
+               pdf_puts(buf, "\\(");
+               break;
            case ')':
+               pdf_puts(buf, "\\)");
+               break;
            case '\\':
-               pdf_puts(buf, "\\");
-               pdf_put(buf, p, 1);
+               pdf_puts(buf, "\\\\");
                break;
            case '\n':
                pdf_puts(buf, "\\n");
@@ -1971,18 +2018,48 @@ static void pdf_put_string_contents(pdf_buffer *buf, const char *text)
                pdf_puts(buf, "\\t");
                break;
            default:
-               if (ch >= 126 || ch < 32) {
+               if (ch > 255) {
+                   pdf_put_char(buf, '?');
+               } else if (ch >= 126 || ch < 32) {
+                   pdf_put_char(buf, ch);
+               } else if (ch >= 0 && ch <= 255) {
                    char tmp[10];
-                   snprintf(tmp, sizeof(tmp), "\\%03o", ch);
+                   snprintf(tmp, sizeof(tmp), "\\%03o", (int)ch);
                    pdf_puts(buf, tmp);
-               } else {
-                   pdf_put(buf, p, 1);
                }
        }
        p++;
     }
 }
 
+static void pdf_put_string_contents(pdf_buffer *buf, const char *text)
+{
+#ifdef HAVE_MBSTOWCS
+    size_t clen = strlen(text) + 1;
+    wchar_t *cstr = malloc(sizeof(wchar_t) * clen);
+    int text_count = mbstowcs(cstr, text, clen);
+    if (text_count == -1)
+       text_count = mbstowcs(cstr, "Enc-Err", 6);
+    pdf_put_string_contents_wide(buf, cstr);
+#if 0
+    if (*text == 'W') {
+       fprintf(stderr, "Decoding utf8 for '%s'\n", text);
+       wchar_t *p = cstr;
+       char *pp = text;
+       fprintf(stderr, "sz wc = %d\n", sizeof(wchar_t));
+       while (*p) {
+           fprintf(stderr, "  %d = %c  versus %d = %c\n", *p, (char)*p, 255 & (int)*pp, *pp);
+           p++;
+           pp++;
+       }
+    }
+#endif
+    free(cstr);
+#else
+    pdf_put_string_contents_wide(buf, text);
+#endif
+}
+
 static void pdf_init_object(pdf_state *state, pdf_buffer *buf)
 {
   pdf_init_buffer(state, buf);