lynx-dev patch for multilingual bookmarks...

[Leonid Pauzner]

This is a long avaiting patch, it was discussed with Klaus about a year ago...
It was proposed to store non-ascii characters in bookmark titles with
a more consistent and independent way, by keeping them as &#UUUU;

The patch 'almost' working: I cannot manage to cast
'long' integer with a buffer string (see near the last line).

Leonid.




diff -u old/lybookma.c ./lybookma.c
--- old/lybookma.c      Fri Aug  7 15:47:58 1998
+++ ./lybookma.c        Wed Aug 12 01:59:46 1998
@@ -7,7 +7,8 @@
 #include <LYSignal.h>
 #include <LYSystem.h>
 #include <LYKeymap.h>
-#include <LYCharUtils.h>
+#include <LYCharUtils.h> /* need for META charset */
+#include <LYCharSets.h>  /* need for LYHaveCJKCharacterSet */
 #include <LYCurses.h>
 #include <GridText.h>

@@ -178,6 +179,9 @@
     return(newfile);
 }

+PRIVATE  BOOLEAN have8bit PARAMS((char *Title));
+PRIVATE  char* title_convert8bit PARAMS((char *Title));
+
 /*
  *  Adds a link to a bookmark file, creating the file
  *  if it doesn't already exist, and making sure that
@@ -279,9 +283,15 @@
      * Create the Title with any left-angle-brackets
      * converted to &lt; entities and any ampersands
      * converted to &amp; entities.  - FM
+     *
+     *  Covnert 8-bit letters to &#xUUUU to avoid dependencies
+     *  from display character set.
+     *  Do NOT convert any 8-bit chars if we have CJK display.
      */
     StrAllocCopy(Title, string_buffer);
     LYEntify(&Title, TRUE);
+    if (have8bit(Title) && (!LYHaveCJKCharacterSet))
+       StrAllocCopy(Title, title_convert8bit(Title));

     /*
      * Create the bookmark file, if it doesn't exist already,
@@ -880,7 +890,7 @@
  *  _statusline() so that any multibyte/CJK characters in the
  *  string will be handled properly. - FM
  */
- PUBLIC void LYMBM_statusline  ARGS1(
+PUBLIC void LYMBM_statusline  ARGS1(
        char *,         text)
 {
     if (LYMultiBookmarks == TRUE && user_mode == NOVICE_MODE) {
@@ -890,4 +900,82 @@
     } else {
        _statusline(text);
     }
+}
+
+/*
+ * Check whether string have 8 bit chars.
+ */
+PRIVATE  BOOLEAN have8bit ARGS1(char *, Title)
+{
+    CONST char *p = Title;
+
+    for ( ; *p; p++) {
+       if ((unsigned char)*p > 127)
+       return(TRUE);
+    }
+return(FALSE); /* if we came here */
+}
+
+/*
+ *  Ok, title have 8-bit characters and they are in display charset.
+ *  Bookmarks is a permanent file. To avoid dependencies from display
+ *  character set which may be changed with time
+ *  we store 8-bit characters as numeric character reference (NCR),
+ *  so where the character encoded as unicode number in form of &#xUUUU;
+ *
+ *  Starting with version 2.7.2 Lynx support NCR's translation
+ *  as part of I18N and HTML4.0
+ *  We do not want use META charset tag in bookmarks file:
+ *  is will never be changed later :-(
+ *
+ *  To make bookmarks more readable for human (&#xXXXX certainly not)
+ *  we add a comment with '7-bit approximation' from the converted string.
+ *  This is a valid HTML and bookmarks code.
+ *
+ */
+PRIVATE  char* title_convert8bit ARGS1(char *, Title)
+{
+    CONST char *p = Title;
+    char *q;
+    char *comment = NULL;
+    char *ncr     = NULL;
+    char *buf = NULL;
+
+    for ( ; *p; p++) {
+       LYstrncpy(q, p, 1);
+       if ((unsigned char)*q <= 127) {
+           StrAllocCat(comment, q);
+           StrAllocCat(ncr, q);
+       } else {
+       int charset_in, charset_out, uck;
+       long unicode;
+       char replace_buf [10];
+
+       charset_in  = current_char_set;
+       charset_out = UCGetLYhndl_byMIME("us-ascii");
+       uck = UCTransCharStr(replace_buf, sizeof(replace_buf), *q,
+                             charset_in, charset_out, YES);
+       if (uck >0)
+       StrAllocCat(comment, replace_buf);  /* 7-bit approx */
+
+       unicode = UCTransToUni( *q, charset_in);
+
+       StrAllocCat(ncr, "&#");
+       /* StrAllocCat(ncr, ???unicode);  casting from long to text string ? */
+
+       StrAllocCat(ncr, ";");
+
+       }
+    }
+
+    /*
+     * valid bookmark should be a single line (no linebreaks!).
+     */
+    StrAllocCat(buf, "<!-- ");
+    StrAllocCat(buf, comment);
+    StrAllocCat(buf, " -->");
+    StrAllocCat(buf, ncr);
+
+return(buf);
+
 }