Fixed grohtml handling of any named glyph for glyph indexes < 0x80.

Cosmetic changes to `.html-begin', `.html-end', `.html-image' which
are now `.begin', `.end', `.image'.

* src/devices/grohtml/post-html.cc: Adding UNICODE_DESC_START.
(html_printer::add_to_sbuf): Changing type of `code' parameter.
Use add_char_to_sbuf().
(to_unicode): New function.
(char_translate_to_html): Changing type of `ch' parameter.
Use `to_unicode()'.
(html_printer::~html_printer): Comment out doctype string.
* src/preproc/html/pre-html.cc (write_end_image): Use `.end'
instead of `.html-end'.
(write_start_image): Use `.begin' and `.image' instead of
`.html-begin' and `.html-image'.
* src/roff/troff/input.cc: Rename `html_level' to `begin_level'.
(html_begin): Renamed to ...
(begin): This.
(html_end): Renamed to ...
(end): This.
(html_image): Renamed to ...
(image): This.
(init_html_requests): Renamed to ...
(init_markup_requests): This.
* tmac/www.tmac: Updated.

6 files changed, 95 insertions, 49 deletions

diff --git a/ChangeLog b/ChangeLog
index 6edcf6b1..575e1e3a 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,31 @@
+2001-03-04  Gaius Mulley  <gaius@glam.ac.uk>
+
+	Fixed grohtml handling of any named glyph for glyph indexes < 0x80. 
+	Cosmetic changes to `.html-begin', `.html-end', `.html-image' which
+	are now `.begin', `.end', `.image'.
+
+	* src/devices/grohtml/post-html.cc: Adding UNICODE_DESC_START.
+	(html_printer::add_to_sbuf): Changing type of `code' parameter.
+	Use add_char_to_sbuf().
+	(to_unicode): New function.
+	(char_translate_to_html): Changing type of `ch' parameter.
+	Use `to_unicode()'.
+	(html_printer::~html_printer): Comment out doctype string.
+	* src/preproc/html/pre-html.cc (write_end_image): Use `.end'
+	instead of `.html-end'.
+	(write_start_image): Use `.begin' and `.image' instead of
+	`.html-begin' and `.html-image'.
+	* src/roff/troff/input.cc: Rename `html_level' to `begin_level'.
+	(html_begin): Renamed to ...
+	(begin): This.
+	(html_end): Renamed to ...
+	(end): This.
+	(html_image): Renamed to ...
+	(image): This.
+	(init_html_requests): Renamed to ...
+	(init_markup_requests): This.
+	* tmac/www.tmac: Updated.
+
 2001-02-28  Bram  <bram@avontuur.org>
 
 	* src/libs/libgroff/font.cc (font_widths_cache): Fixing syntax of
diff --git a/src/devices/grohtml/post-html.cc b/src/devices/grohtml/post-html.cc
index 7f6490ca..e025cb48 100644
--- a/src/devices/grohtml/post-html.cc
+++ b/src/devices/grohtml/post-html.cc
@@ -52,6 +52,9 @@ Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
 #define BASE_POINT_SIZE                10            /* 10 points is the base size ie html size 3 */
 #define CENTER_TOLERANCE                2            /* how many pixels off center will we still  */
 #define ANCHOR_TEMPLATE       "heading%d"            /* if simple anchor is set we use this       */
+#define UNICODE_DESC_START           0x80            /* all character entities above this are     */
+                                                     /* either encoded by their glyph names or if */
+                                                     /* there is no name then we use &#nnn;       */
 
 typedef enum {CENTERED, LEFT, RIGHT, INLINE} TAG_ALIGNMENT;
 
@@ -985,7 +988,7 @@ class html_printer : public printer {
   void  terminate_current_font        (void);
   void  flush_font                    (void);
   void  add_char_to_sbuf              (unsigned char code);
-  void  add_to_sbuf                   (char code, const char *name);
+  void  add_to_sbuf                   (unsigned char code, const char *name);
   void  write_title                   (int in_head);
   void  determine_diacritical_mark    (const char *name, const environment *env);
   int   sbuf_continuation             (unsigned char code, const char *name, const environment *env, int w);
@@ -2127,15 +2130,19 @@ void html_printer::add_char_to_sbuf (unsigned char code)
  *  add_to_sbuf - adds character code or name to the sbuf.
  */
 
-void html_printer::add_to_sbuf (char code, const char *name)
+void html_printer::add_to_sbuf (unsigned char code, const char *name)
 {
+  if (code == 255) stop();
+
   if (name == 0) {
     add_char_to_sbuf(code);
   } else {
     if (sbuf_style.f != NULL) {
       char *html_glyph = get_html_translation(sbuf_style.f, name);
 
-      if (html_glyph != NULL) {
+      if (html_glyph == NULL) {
+	add_char_to_sbuf(code);
+      } else {
 	int   l          = strlen(html_glyph);
 	int   i;
 
@@ -2300,11 +2307,24 @@ char *get_html_translation (font *f, const char *name)
 }
 
 /*
+ *  to_unicode - returns a unicode translation of char, ch.
+ */
+
+static char *to_unicode (unsigned char ch)
+{
+  static char buf[20];
+
+  stop();
+  sprintf(buf, "&#%u;", (unsigned int)ch);
+  return( buf );
+}
+
+/*
  *  char_translate_to_html - convert a single non escaped character
  *                           into the appropriate html character.
  */
 
-int char_translate_to_html (font *f, char *buf, int buflen, char ch, int b, int and_single)
+int char_translate_to_html (font *f, char *buf, int buflen, unsigned char ch, int b, int and_single)
 {
   if (and_single) {
     int    t, l;
@@ -2314,6 +2334,9 @@ int char_translate_to_html (font *f, char *buf, int buflen, char ch, int b, int
     name[0] = ch;
     name[1] = (char)0;
     translation = get_html_translation(f, name);
+    if ((translation == NULL) && (ch >= UNICODE_DESC_START)) {
+      translation = to_unicode(ch);
+    }
     if (translation) {
       l = strlen(translation);
       t = max(0, min(l, buflen-b));
@@ -2461,6 +2484,9 @@ void html_printer::write_title (int in_head)
       html.put_string(title.text);
       html.put_string("</h1>\n\n");
     }
+  } else if (in_head) {
+    // place empty title tags to help conform to `tidy'
+    html.put_string("<title></title>\n");
   }
 }
 
@@ -2507,6 +2533,7 @@ html_printer::~html_printer()
 {
   current_paragraph->done_para();
   html.set_file(stdout);
+  // fputs("<!doctype html public \"-//IETF//DTD HTML 4.0//EN\">\n", stdout);
   fputs("<html>\n", stdout);
   fputs("<head>\n", stdout);
   fputs("<meta name=\"generator\" content=\"groff -Thtml, see www.gnu.org\">\n", stdout);
diff --git a/src/preproc/html/pre-html.cc b/src/preproc/html/pre-html.cc
index efd98e4e..2f00f786 100644
--- a/src/preproc/html/pre-html.cc
+++ b/src/preproc/html/pre-html.cc
@@ -258,7 +258,7 @@ void makeFileName ()
 
 static void write_end_image (int is_html)
 {
-  writeString(".html-end \\{\\\n");
+  writeString(".end \\{\\\n");
   if (is_html) {
     /*
      *  emit image name and enable output
@@ -283,21 +283,21 @@ static void write_end_image (int is_html)
 
 static void write_start_image (IMAGE_ALIGNMENT pos, int is_html)
 {
-  writeString(".html-begin \\{\\\n");
+  writeString(".begin \\{\\\n");
   switch (pos) {
 
   case LEFT:
-    writeString(".    html-image l ");
+    writeString(".    image l ");
     break;
   case RIGHT:
-    writeString(".    html-image r ");
+    writeString(".    image r ");
     break;
   case INLINE:
-    writeString(".    html-image i ");
+    writeString(".    image i ");
     break;
   case CENTERED:
   default:
-    writeString(".    html-image c ");
+    writeString(".    image c ");
   }
   writeString(image_template); writeString(".png\n");
   if (is_html) {
diff --git a/src/roff/troff/input.cc b/src/roff/troff/input.cc
index 85bf4735..ae864fe0 100644
--- a/src/roff/troff/input.cc
+++ b/src/roff/troff/input.cc
@@ -111,7 +111,7 @@ int compatible_flag = 0;
 int ascii_output_flag = 0;
 int suppress_output_flag = 0;
 int is_html = 0;
-int html_level = 0;		// number of nested .html-begin requests
+int begin_level = 0;		// number of nested .begin requests
 
 int tcommand_flag = 0;
 int safer_flag = 1;		// safer by default
@@ -4477,46 +4477,46 @@ void else_request()
 }
 
 /*
- *  html_begin - if this is the outermost html_begin request then execute the
+ *  begin - if this is the outermost html_begin request then execute the
  *               rest of the line, else skip line
  */
 
-void html_begin()
+void begin()
 {
-  html_level++;
-  if (html_level == 1)
+  begin_level++;
+  if (begin_level == 1)
     begin_alternative();
   else
     skip_alternative();
 }
 
 /*
- *  html_end - if this is the outermost html_end request then execute the
- *  rest of the line, else skip line
+ *  end - if this is the outermost html_end request then execute the
+ *        rest of the line, else skip line
  */
 
-void html_end()
+void end()
 {
-  html_level--;
-  if (html_level == 0)
+  begin_level--;
+  if (begin_level == 0)
     begin_alternative();
   else
     skip_alternative();
-  if (html_level < 0)
-    html_level = 0;
+  if (begin_level < 0)
+    begin_level = 0;
 }
 
 /*
- *  html_image - implements the directive `.html_image {l|r|c|i} filename'
- *               which places the filename into a node which is later
- *               written out
+ *  image - implements the directive `.image {l|r|c|i} filename'
+ *          which places the filename into a node which is later
+ *          written out
  *
- *               . either as a special in the form of an image tag for -Thtml
- *               . or as an image region definition for all other devices
+ *          . either as a special in the form of an image tag for -Thtml
+ *          . or as an image region definition for all other devices
  *
  */
 
-void html_image()
+void image()
 {
   if (has_arg()) {
     char position = tok.ch();
@@ -6154,7 +6154,7 @@ int main(int argc, char **argv)
   init_column_requests();
 #endif /* COLUMN */
   init_node_requests();
-  init_html_requests();
+  init_markup_requests();
   number_reg_dictionary.define(".T", new constant_reg(tflag ? "1" : "0"));
   init_registers();
   init_reg_requests();
@@ -6268,11 +6268,11 @@ void get_output_registers(int *minx, int *miny, int *maxx, int *maxy)
   *maxy = output_reg_maxy_contents;
 }
 
-void init_html_requests()
+void init_markup_requests()
 {
-  init_request("html-begin", html_begin);
-  init_request("html-end", html_end);
-  init_request("html-image", html_image);
+  init_request("begin", begin);
+  init_request("end", end);
+  init_request("image", image);
 }
 
 void init_input_requests()
diff --git a/src/roff/troff/node.cc b/src/roff/troff/node.cc
index 5422e91f..458ed484 100644
--- a/src/roff/troff/node.cc
+++ b/src/roff/troff/node.cc
@@ -3432,16 +3432,7 @@ void suppress_node::put(troff_output_file *out, const char *s)
 }
 
 /*
- *  We prefer to remember the last position, rather than have a .html-start
- *  followed by html-end-center, html-end-left as it is more natural to
- *  express an image by:
- *
- *    .html-image-left
- *
- *    .html-image-end
- *
- *  similar to other troff commands, although this method is slightly more
- *  messy to implement.
+ *  We need to remember the start of the image and its name.
  */
 
 static char last_position = 0;
diff --git a/tmac/www.tmac b/tmac/www.tmac
index 0297a265..fb6e146a 100644
--- a/tmac/www.tmac
+++ b/tmac/www.tmac
@@ -154,14 +154,14 @@
 .\" HTML-DO-IMAGE - tells troff to issue an image marker which can be read back by pre-html
 .\"
 .de HTML-DO-IMAGE
-.  if r ps4html .html-begin \{\
-.         html-image \\$2 \\$1.png
+.  if r ps4html .begin \{\
+.         image \\$2 \\$1.png
 .         bp
 .         tl '''
 \O0\O1
 .  \}
-.  if '\*(.T'html' .html-begin \{
-.         html-image \\$2 \\$1.png
+.  if '\*(.T'html' .begin \{
+.         image \\$2 \\$1.png
 \O0
 .  \}
 ..
@@ -169,10 +169,10 @@
 .\" HTML-IMAGE-END - terminates an image for html
 .\"
 .de HTML-IMAGE-END
-.  if r ps4html .html-end \{\
+.  if r ps4html .end \{\
 \O2\O1
 .  \}
-.  if '\*(.T'html' .html-end \{
+.  if '\*(.T'html' .end \{
 \O2\O1
 .  \}
 ..