--- psiconv/trunk/program/psiconv/gen_html4.c	1999/12/02 17:18:55	34
+++ psiconv/trunk/program/psiconv/gen_html4.c	2002/05/10 15:55:55	147
@@ -18,13 +18,25 @@
  * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  */
 
+/*
+    2002/Apr.    Keita KAWABE
+    A: <META HTTP-EQUIV="Content-Type" CONTENT="text/html"> header
+       was added.
+
+    B: Support for narrow build Asian Psions added.
+       If the encoding_type is PSICONV_ENCODING_UTF8,
+       B-1:  use utf8_table for character conversion.
+       B-2:  add the "Charset=UTF-8" in the Content-Type header.
+       Otherwise proceed as normal.
+*/
+
 #include "config.h"
 #include <stdio.h>
 #include <string.h>
 #include <ctype.h>
 #include <stdlib.h>
-#include "data.h"
-#include "list.h"
+#include "psiconv/data.h"
+#include "psiconv/list.h"
 #include "gen.h"
 #include "psiconv.h"
 
@@ -32,6 +44,10 @@
 #define TRUE (0==0)
 #endif
 
+#ifdef DMALLOC
+#include "dmalloc.h"
+#endif
+
 /*
  * Various string tables for HTML4 settings
  */
@@ -88,6 +104,45 @@
              "&uuml;", "&yacute;", "&thorn;", "&yuml;"
 };
 
+static const char *utf_table[0x100] = {
+  /* 0x00 */ "", "", "", "", "", "", "<BR>\n", "<BR>\n",
+  /* 0x08 */ "\n<HR>\n", "\t", "", "", "", "", "", "",
+  /* 0x10 */ " ", "", "", "", "", "", "", "",
+  /* 0x18 */ "", "", "", "", "", "", "", "",
+  /* 0x20 */ " ", "!", "&quot;", "#", "$", "%", "&amp;", "'",
+  /* 0x28 */ "(", ")", "*", "+", ",", "-", ".", "/",
+  /* 0x30 */ "0", "1", "2", "3", "4", "5", "6", "7",
+  /* 0x38 */ "8", "9", ":", ";", "&lt;", "=", "&gt;", "?",
+  /* 0x40 */ "@", "A", "B", "C", "D", "E", "F", "G",
+  /* 0x48 */ "H", "I", "J", "K", "L", "M", "N", "O",
+  /* 0x50 */ "P", "Q", "R", "S", "T", "U", "V", "W",
+  /* 0x58 */ "X", "Y", "Z", "[", "\\", "]", "^", "_",
+  /* 0x60 */ "`", "a", "b", "c", "d", "e", "f", "g",
+  /* 0x68 */ "h", "i", "j", "k", "l", "m", "n", "o",
+  /* 0x70 */ "p", "q", "r", "s", "t", "u", "v", "w",
+  /* 0x78 */ "x", "y", "z", "{", "|", "}", "~", "",
+  /* 0x80 */ "\x80", "\x81", "\x82", "\x83", "\x84", "\x85", "\x86", "\x87",
+  /* 0x88 */ "\x88", "\x89", "\x8a", "\x8b", "\x8c", "\x8d", "\x8e", "\x8f",
+  /* 0x90 */ "\x90", "\x91", "\x92", "\x93", "\x94", "\x95", "\x96", "\x97",
+  /* 0x98 */ "\x98", "\x99", "\x9a", "\x9b", "\x9c", "\x9d", "\x9e", "\x9f",
+  /* 0xa0 */ "\xa0", "\xa1", "\xa2", "\xa3", "\xa4", "\xa5", "\xa6", "\xa7",
+  /* 0xa8 */ "\xa8", "\xa9", "\xaa", "\xab", "\xac", "\xad", "\xae", "\xaf",
+  /* 0xb0 */ "\xb0", "\xb1", "\xb2", "\xb3", "\xb4", "\xb5", "\xb6", "\xb7",
+  /* 0xb8 */ "\xb8", "\xb9", "\xba", "\xbb", "\xbc", "\xbd", "\xbe", "\xbf",
+  /* 0xc0 */ "\xc0", "\xc1", "\xc2", "\xc3", "\xc4", "\xc5", "\xc6", "\xc7",
+  /* 0xc8 */ "\xc8", "\xc9", "\xca", "\xcb", "\xcc", "\xcd", "\xce", "\xcf",
+  /* 0xd0 */ "\xd0", "\xd1", "\xd2", "\xd3", "\xd4", "\xd5", "\xd6", "\xd7",
+  /* 0xd8 */ "\xd8", "\xd9", "\xda", "\xdb", "\xdc", "\xdd", "\xde", "\xdf",
+  /* 0xe0 */ "\xe0", "\xe1", "\xe2", "\xe3", "\xe4", "\xe5", "\xe6", "\xe7",
+  /* 0xe8 */ "\xe8", "\xe9", "\xea", "\xeb", "\xec", "\xed", "\xee", "\xef",
+  /* 0xf0 */ "\xf0", "\xf1", "\xf2", "\xf3", "\xf4", "\xf5", "\xf6", "\xf7",
+  /* 0xf8 */ "\xf8", "\xf9", "\xfa", "\xfb", "\xfc", "\xfd", "\xfe", "\xff",
+};
+
+/* a flag to indicate the use of UTF8 */
+static psiconv_encoding encoding=PSICONV_ENCODING_CP1252;
+
+
 /* The order of these must match the enum psiconv_border_kind (data.h) */
 static const char *border_strings[] = {
     "none", "solid", "double", "dotted", "dashed", "dashed", "dotted"
@@ -119,24 +174,30 @@
 static void fput_text(FILE * of, const char *text, int length, int *pwasspace) {
     int j;
     int space;
+    char ** table;
+    if (encoding == PSICONV_ENCODING_UTF8){
+      table = (char**) utf_table;
+    }else{
+      table=(char**) char_table;
+    }
     
     for (j = 0; j < length; j++) {
 	space = (text[j] == ' ');
 	if (*pwasspace && space)
 	    fputs("&nbsp;", of);
 	else
-	    fputs(char_table[(unsigned char) (text[j])], of);
+	    fputs(table[(unsigned char) (text[j])], of);
 	*pwasspace = space;
     }
 }
 
 static const char *screenfont_name(psiconv_u8 screenfont)
 {
-  if (screenfont == 1)
+  if (screenfont == psiconv_font_sansserif)
     return "sans-serif";
-  else if (screenfont == 2)
+  else if (screenfont == psiconv_font_nonprop)
     return "monospace";
-  else if (screenfont == 3)
+  else if (screenfont == psiconv_font_serif)
     return "serif";
   else
     return "serif";
@@ -212,13 +273,13 @@
     const psiconv_paragraph_layout old,
     const psiconv_paragraph_layout new)
 {
-    return ((old->top_space != new->top_space) ||
+    return ((old->space_above != new->space_above) ||
     	    (old->indent_right != new->indent_right) ||
-    	    (old->bottom_space != new->bottom_space) ||
+    	    (old->space_below != new->space_below) ||
     	    (old->indent_left != new->indent_left) ||
     	    (old->indent_first != new->indent_first) ||
     	    (old->justify_hor != new->justify_hor) ||
-    	    (old->interline_exact != new->interline_exact) ||
+    	    (old->linespacing_exact != new->linespacing_exact) ||
 	    borders_different(old->left_border, new->left_border) ||
 	    borders_different(old->right_border, new->right_border) ||
 	    borders_different(old->top_border, new->top_border) ||
@@ -229,12 +290,12 @@
     const psiconv_paragraph_layout old,
     const psiconv_paragraph_layout new)
 {
-    if (!old || (old->top_space != new->top_space))
-	fprintf(of, "margin-top:%.1fpt;", new->top_space);
+    if (!old || (old->space_above != new->space_above))
+	fprintf(of, "margin-top:%.1fpt;", new->space_above);
     if (!old || (old->indent_right != new->indent_right))
 	fprintf(of, "margin-right:%.2fcm;", new->indent_right);
-    if (!old || (old->bottom_space != new->bottom_space))
-	fprintf(of, "margin-bottom:%.1fpt;", new->bottom_space);
+    if (!old || (old->space_below != new->space_below))
+	fprintf(of, "margin-bottom:%.1fpt;", new->space_below);
     if (!old || 
 	(old->bullet->on != new->bullet->on) ||
 	(old->bullet->indent != new->bullet->indent) ||
@@ -254,11 +315,11 @@
     }
     if (!old || (old->justify_hor != new->justify_hor))
 	fprintf(of, "text-align:%s;", justify_strings[new->justify_hor]);
-    if (!old || (old->interline_exact != new->interline_exact)) {
-	if (!old || old->interline_exact)
+    if (!old || (old->linespacing_exact != new->linespacing_exact)) {
+	if (!old || old->linespacing_exact)
 	    fputs("line-height:normal;", of);
 	else
-	    fprintf(of, "line-height:%.1fpt;", new->interline);
+	    fprintf(of, "line-height:%.1fpt;", new->linespacing);
     }
     fput_border(of, "left", old ? old->left_border : NULL, 
     		new->left_border);
@@ -288,7 +349,7 @@
 	    (old->bold != new->bold) ||
 	    (old->super_sub != new->super_sub) ||
 	    (old->underline != new->underline) ||
-	    (old->strike_out != new->strike_out));
+	    (old->strikethrough != new->strikethrough));
 }
 
 static void fput_character_layout(FILE * of,
@@ -317,13 +378,13 @@
     if (!old || (old->super_sub != new->super_sub))
     	fprintf(of, "vertical-align:%s;", vertical_strings[new->super_sub]);
     if (!old || (old->underline != new->underline) ||
-        (old->strike_out != new->strike_out)) {
+        (old->strikethrough != new->strikethrough)) {
 	fputs("text-decoration:", of);
-	if (new->strike_out)
+	if (new->strikethrough)
 	    fputs("line-through ", of);
 	if (new->underline)
 	    fputs("underline", of);
-	else if (!new->strike_out)
+	else if (!new->strikethrough)
 	    fputs("none", of);
 	fputs(";", of);
     }
@@ -344,6 +405,13 @@
     int spans = 0;
     int wasspace = 0;
     int i, j, loc;
+    char ** table;
+    
+    if (encoding == PSICONV_ENCODING_UTF8){
+      table= (char**) utf_table;
+    }else{
+      table= (char**) char_table;
+    }
     
     layout_stack = calloc(psiconv_list_length(para->in_lines),
     			  sizeof(*layout_stack));
@@ -386,7 +454,7 @@
 		    base_para->bullet->color->green, 
 		    base_para->bullet->color->blue);
 	fprintf(of, "\">%s </SPAN>", 
-		char_table[base_para->bullet->character]);
+		table[base_para->bullet->character]);
 	wasspace = TRUE;
     }
     
@@ -447,31 +515,44 @@
     fputs("}\n", of);
 }
 
-static int psiconv_gen_html4(FILE * of, const psiconv_file file, 
-                             const char *dest)
+static int psiconv_gen_html4(const char *filename, const psiconv_file file, 
+			     const char *dest, const psiconv_encoding encoding_type)
 {
+    FILE * of;
     int i;
     psiconv_paragraph para;
     psiconv_word_style normal, sty;
     psiconv_word_f wf;
+    
+    encoding=encoding_type;
 
     if (file->type == psiconv_word_file)
       wf = file->file;
     else {
       /* Fall back on the normal HTML generator */
-      return psiconv_gen_html(of,file,dest);
+      return psiconv_gen_html(filename,file,dest,encoding_type);
     }
 
+    if (! (of = fopen(filename,"w"))) 
+        return -1;
+
     fputs("<!doctype html public \"-//W3C//DTD HTML 4.0 STRICT//EN\">", of);
     fputs("\n<HTML>\n<HEAD>\n  <META NAME=\"GENERATOR\"", of);
     fputs(" CONTENT=\"psiconv-" VERSION "\">\n", of);
-    
-    if (psiconv_list_length(wf->page_sec->header->text->paragraphs) > 0) {
-	fputs("  <TITLE>", of);
-	para = psiconv_list_get(wf->page_sec->header->text->paragraphs, 0);
-	i = 0;
-	fput_text(of, para->text, strlen(para->text), &i);
-	fputs("</TITLE>\n", of);
+
+    fputs("  <META HTTP-EQUIV=\"Content-Type\" CONTENT=\"text/html", of);
+    if (encoding == PSICONV_ENCODING_UTF8)
+      fputs("; CHARSET=UTF-8", of);
+    fputs("\">\n", of);
+    
+    if (wf->page_sec->header->text) {
+        if (psiconv_list_length(wf->page_sec->header->text->paragraphs) > 0) {
+    	    fputs("  <TITLE>", of);
+	    para = psiconv_list_get(wf->page_sec->header->text->paragraphs, 0);
+	    i = 0;
+	    fput_text(of, para->text, strlen(para->text), &i);
+	    fputs("</TITLE>\n", of);
+        }
     }
     
     normal = wf->styles_sec->normal;
@@ -486,15 +567,18 @@
 	fput_style(of, normal, sty, wf);
     }
     fputs("  --></STYLE>\n</HEAD>\n<BODY>\n", of);
-    for (i = 0; i < psiconv_list_length(wf->paragraphs); i++) {
-	para = psiconv_list_get(wf->paragraphs, i);
-	fput_para(of, para, wf);
+    if (wf->page_sec->header->text) {
+        for (i = 0; i < psiconv_list_length(wf->paragraphs); i++) {
+	    para = psiconv_list_get(wf->paragraphs, i);
+	    fput_para(of, para, wf);
+        }
     }
     fputs("</BODY>\n</HTML>\n", of);
-    return 0;
+
+    return fclose(of);
 }
 
-static struct psiconv_fileformat ff =
+static struct psiconv_fileformat_s ff =
   {
     "HTML4",
     "HTML 4.0 with cascading style sheets",