Chapter 1
Flow of Program

<..TeX4ht copyright..>
 %
 % This work may be distributed and/or modified under the
 % conditions of the LaTeX Project Public License, either
 % version 1.3c of this license or (at your option) any
 % later version. The latest version of this license is in
 %   http://www.latex-project.org/lppl.txt
 % and version 1.3c or later is part of all distributions
 % of LaTeX version 2005/12/01 or later.
 %
 % This work has the LPPL maintenance status "maintained".
 %
 % The Current Maintainer of this work
 % is the TeX4ht Project <http://tug.org/tex4ht>.
 %
 % If you modify this program, changing the
 % version identification would be appreciated.-_-_-

<..TeX4ht copywrite..>
 |<TeX4ht copyright|>
 \immediate\write-1{version |version}
 -_-_-

1.1 Preamble

A command-line program running in the console environment of the operating system (MS-DOS, MS-DOS window of Windows 3.1, Unix shell, etc). The input is a heavily specialized DVI file. The main output is the HTML/XML and ‘.css’, also ‘.lg’, ‘.idv’, and ‘.tmp’.

<..tex4ht.c..>
 /* tex4ht.c (2014-07-21-21:39), generated from tex4ht-c.tex
    Copyright (C) 2009-2012 TeX Users Group
    Copyright (C) 1996-2009 Eitan M. Gurari
 <.TeX4ht copyright.> */
 <.tex4ht.h.>
 -_-_-

<..tex4ht.h..>
 /* Compiler options (uncommented | command line), as comments:
 
        Classic C (CC)             default
 #define ANSI                      ansi-c, c++
 #define DOS_C
 #define HAVE_STRING_H             <string.h>
 #define HAVE_DIRENT_H             <dirent.h>
 #define HAVE_SYS_NDIR_H           <sys/ndir.h>
 #define HAVE_SYS_DIR_H            <sys/dir.h>
 #define HAVE_NDIR_H               <ndir.h>
 #define HAVE_IO_H                 <io.h>
 #define HAVE_UNISTD_H             <unistd.h>
 #define WIN32
 #define KPATHSEA
 #define CDECL                     .......
 #define BCC32                     bordland c++
 
 *************************************************
     Tex4ht variables                            *
     (uncommented | command line)                *
 ----------------------------------------------- */
 <.h-defines.>
 /* ******************************************** */
 <.define MSVC-1-52.>
 <.bordland: bcc32 -DBCC32 tex4ht.>
 <.bordland: bcc -DBCC -mh tex4ht.>
 <.DJGPP.>
 #ifdef DOS_C
 #define DOS
 #endif
 #ifdef DOS
 #define DOS_WIN32
 #define HAVE_STRING_H   <string.h>
 #endif
 #ifdef WIN32
 #define DOS_WIN32
 #ifndef KPATHSEA
 #define HAVE_STRING_H   <string.h>
 #endif
 #endif
 <.h-defines DOS-WIN32.>
 <.h-defines KWIN32.>
 <.h-include.>
 <.h-defines 2.>
 -_-_-

<..bordland: bcc32 -DBCC32 tex4ht..>
 #ifdef BCC32
 #define DOS_WIN32
 #define ANSI
 #define HAVE_DIRENT_H
 #define PLATFORM "ms-win32"
 #endif
 -_-_-

The ‘-mh’ below calls for ‘Huge Memory Model’.

<..bordland: bcc -DBCC -mh tex4ht..>
 #ifdef BCC
 #define DOS_C
 #define ANSI
 #define HAVE_DIRENT_H
 #define PLATFORM "ms-dos"
 #ifndef PATH_MAX
 #define PATH_MAX 256
 #endif
 #endif
 -_-_-

[more]

microsoft visual c options: msdos application (.exe), large, ss=sd

<..h-include..>
 #ifdef KPATHSEA
 #include <kpathsea/config.h>
 #endif
 #include <stdio.h>    EOF, FILE 
 #include <stdlib.h>   EXIT_FAILURE 
 -_-_-

<..defines..>
 #ifndef EXIT_FAILURE
 #define EXIT_FAILURE 1
 #endif
 -_-_-

<..h-include..>+
 #ifdef HAVE_STRING_H
 #include <string.h>
 #endif
 -_-_-

Under ANSI C, all malloc stuff is declared in ‘<stdlib.h>’ (which you also include), hence this non-standard header is redundant.

MSVC-1-52-DOS

<..define MSVC-1-52NO..>
 #ifdef MSVC_1_52_DOS_LIB
 #define  MSVC_1_52_DOS
 #endif
 #ifdef MSVC_1_52_DOS
 #define DOS
 #define NOSUBDIR
 #endif
 #ifdef MSVC_1_52_DOS_LIB
 #undef NOSUBDIR
 #endif
 -_-_-

<..if not-def MSVC-1-52NO..>
 #ifndef MSVC_1_52_DOS
 -_-_-

<..endif not-def MSVC-1-52NO..>
 #endif
 -_-_-

KPATHSEA Windows

<..h-defines KWIN32..>
 #ifdef KPATHSEA
 #ifdef WIN32
 #define KWIN32
 #endif
 #endif
 -_-_-

Here are the promised patch. However I have not yet tested it very much. It is unlikely that I test it with non-kpathsea setting. At least the patch compiles in both kpathsea and non-kpathsea setting. The following patch is very similar to some code in kpathsea / web2c.

<..h-include..>+
 #ifdef WIN32
 #ifdef KPATHSEA
 #undef CDECL
 #define CDECL                     __cdecl
 #else
 #include <windows.h>
 #endif
 #else
 #ifdef KPATHSEA
 #define CDECL
 #endif
 #endif
 -_-_-

<..if not-WIN32..>
 #ifndef WIN32
 -_-_-

<..endif not-WIN32..>
 #endif
 -_-_-

<..header functions..>
 #ifdef WIN32
 static BOOL sigint_handler(ARG_I(DWORD));
 #endif
 -_-_-

<..functions..>
 #ifdef WIN32
 
 
 
 static BOOL  sigint_handler(dwCtrlType)  DWORD dwCtrlType
 ;{
   <.avoid warning of unused dwCtrlType.>
   err_i(32);
   return FALSE;       return value obligatory 
 }
 #endif
 -_-_-

<..avoid warning of unused dwCtrlType..>
 if( dwCtrlType ){ (IGNORED) printf(" "); }
 -_-_-

<..CDECL..>
 #ifdef CDECL
 CDECL
 #endif
 -_-_-

Forgetting ‘_pascal’ and ‘_cdecl’ modifiers: Each C function may be either ‘_pascal’ or ‘_cdecl’. This modifier defines how parameters are passed to it. Default for Smalltalk definition is ‘_cdecl’. Default for C functions depends on compiler settings, and you can use other types uncompatible with Smalltalk. In Windows API 16-bit functions are ‘_pascal’ and 32-bit ‘_cdecl’.

The following Shared across recursive calls, it acts like a stack.

<..vars..>
 #ifdef WIN32
 static U_CHAR dirname[PATH_MAX];
 #endif
 -_-_-

Recursive directory search relies on some Windows API similar to opendir/readdir

<..file = search WIN32 subdir (name,str)..>
 {
     WIN32_FIND_DATA find_file_data;
     HANDLE hnd;
     int proceed;
     (IGNORED) strcpy((char *) dirname, (char *) str);
     strct(dirname, "/*.*");        "*.*" or "*" -- seems equivalent. 
     hnd = FindFirstFile(dirname, &find_file_data);
     if (hnd != INVALID_HANDLE_VALUE) {
       <.valid WIN32 handle.>
     }
 }
 -_-_-

<..valid WIN32 handle..>
 proceed = 1;
 while (proceed) {
   if( !eq_str(find_file_data.cFileName, ".")  &&
       !eq_str(find_file_data.cFileName, "..") )
     {
       (IGNORED) strcpy((char *)  str+n, (char *) find_file_data.cFileName );
       str[n-1] = dir_path_slash(str);
       if (find_file_data.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
   if( (file = search_file_ext(name, str, flags)) != NULL ){
     FindClose(hnd);
     return file; }
       }
     }
   proceed = FindNextFile(hnd, &find_file_data);
 }
 FindClose(hnd);
 -_-_-

SIGFPE is handled by the C library, SIGSEGV too but not generated by the system, so it does nothing but is harmless

SetConsoleCtrlHandler is need to catch Ctrl+C and Ctrl+Break under Windows, SIGINT is not generated by the system.

<..KWIN32 signals..>
 SetConsoleCtrlHandler((PHANDLER_ROUTINE)sigint_handler, TRUE);
 -_-_-

DJGPP

<..DJGPP..>
 #ifdef __DJGPP__
 #define DOS_WIN
 #define ANSI
 #ifndef HAVE_STRING_H
 #define HAVE_STRING_H 1
 #endif
 #endif
 -_-_-

<..vars..>+
 static BOOL dos_file_names =
 #ifdef DOS_GIF_FILE
   TRUE
 #else
   FALSE
 #endif
 ;
 -_-_-

Set dos_file_names to TRUE for all DOS platforms, FALSE on Unix and WIN32, and compute at runtime for DJGPP.

<..DJGPP gif names for pictorial symbols..>
 #if !defined(DOS_GIF_FILE) && !defined(WIN32) && defined(__DJGPP__)
    dos_file_names = !_use_lfn(NULL);
 #endif
 -_-_-

<..extra options..>
 switch( *(p+2) ){
   case ’s’:{   dos_file_names = TRUE;  break; }
    default:{ bad_arg; }
 }
 -_-_-

<..DJGPP signals..>
 #ifdef __DJGPP__
   if (s != SIGINT && s != SIGQUIT)
     exit(EXIT_FAILURE);
 #endif
 -_-_-

Compatibity

<..defines..>+
 #if INT_MAX < 2147483647L  2^31
 #define LONG L
 #endif
 -_-_-

Lint does not like non-null pointers to be cast. ‘free(void *)’ instead of ‘free()’ will make lint to stop complaining, but is this really the problem. For this function we have a definition ‘void free(void *)’.

‘1<<20’ should be ‘1L<<20’ for short integer.

Integers on Dos they are just 16 bits!!! Also, ordering of bytes is inconsistence.

<..h-include..>+
 #include <limits.h>  INT_MAX 
 -_-_-

<..defines..>+
 #ifdef LONG
 #define INTEGER long
 #else
 #define INTEGER int
 #endif
 #define U_CHAR char
 -_-_-

<..resplit argv for windows..>
 #ifdef WIN32
   /* The idea here is to split options apart at spaces: a single argument
      "-foo -bar" becomes the two options "-foo" and "-bar".  We need to
      do this for Windows because mk4ht passes this sort of combined
      option in one string to scripts like htlatex.{unix,bat}.  In the
      Unix case, the shell resplits words when calling t4ht and tex4ht,
      so the program see two options.  But this does not happen with the
      .bat; %4, for instance, remains "-foo -bar".  So we fix it here.  */
   if (argc > 2) {
     int  i, nargc;
     char **nargv, **pnargv, **pargv;
 
     nargv = (char **) xmalloc (2 * argc * sizeof (char *));
     pnargv = nargv;
     pargv = argv;
     *pnargv++ = xstrdup (*pargv++);
     *pnargv++ = xstrdup (*pargv++);
     nargc = 2;
 
     for (i=2; i < argc; i++) {
       char *p, *q, *r;
       p = q = *pargv++;
       while (*p == ’ ’ || *p == ’\t’) {
         p++;
         q++;
       }
       while (*p != ’ ’ && *p != ’\t’ && *p) {
         p++;
         if (*p == ’\0’) {
           *pnargv++ = xstrdup(q);
           nargc++;
         } else if (*p == ’ ’ || *p == ’\t’) {
           r = p;
           while (*p == ’ ’ || *p == ’\t’)
             p++;
           if (*p == ’-’ || *p == ’\0’) {
             *r = ’\0’;
             *pnargv++ = xstrdup(q);
             nargc++;
             q = p;
           }
         }
       }
     }
 
     nargv[nargc] = NULL;
     argv = nargv;
     argc = nargc;
   }
 #endif
 -_-_-

Signals

<..h-include..>+
 #include <signal.h>
 -_-_-

<..header functions..>+
 static  void
 <.CDECL.>
 sig_err(ARG_I(int));
 -_-_-

<..functions..>+
 
 static void
 <.CDECL.>
 sig_err(s)  int s
 ;{
   (void) signal(s,SIG_IGN);  ignore the signal
   switch( s ){
 #ifdef SIGSEGV
     case SIGSEGV: err_i(30);
 #endif
     case SIGFPE : err_i(31);
 #if defined(SIGINT) && !defined(WIN32)
     case SIGINT : err_i(32);
 #endif
   }
   <.DJGPP signals.>
 }
 -_-_-

DJGPP:(sig_err): If got fatal signal other than SIGINT or SIGQUIT, don’t return to caller, since the program will be aborted otherwise.

<..set signals..>
 
 #ifdef SIGSEGV
   (void) signal(SIGSEGV,sig_err);
 #endif
   (void) signal(SIGFPE,sig_err);
 #ifdef WIN32
   <.KWIN32 signals.>
 #else
 #ifdef SIGINT
   (void) signal(SIGINT,sig_err);    Control-c, user interrupt
 #endif
 #endif
 -_-_-

SIGSEGV, SIGILL, SIGTERM not implemented in MS-DOS. They are supplied just for compatibility. Looks like that SIGINT is defined for windows but not for dos. [more]

Msvc recommends not using printf, but we ignoring this recommendation here with the assumption that the recommendation relates to I/O interrupts that are not considered here.

<..signals messages: 30-32..>
 "Illegal storage address\n", 30 segmentation
 "Floating-point\n",          31 
 "Interrupt with Cntr-C\n",   32 
 -_-_-

1.2 Kpathsea

<..KPATHSEA h-include..>
 #ifdef KPATHSEA
 #include <kpathsea/c-errno.h>
 #include <kpathsea/c-ctype.h>
 #include <kpathsea/c-fopen.h>
 #include <kpathsea/c-pathmx.h>
 #include <kpathsea/proginit.h>
 #include <kpathsea/tex-file.h>
 #include <kpathsea/tex-make.h>
 #include <kpathsea/variable.h>
 #include <signal.h>
 #if !defined(_AMIGA) && !defined(WIN32)
 #include <sys/time.h>
 #endif
 #include <fcntl.h>
 #include <setjmp.h>
 #endif /* KPATHSEA */
 -_-_-

• The ‘__cdecl’ is defined in KPATHSEA: kpathsea/c-protos.h is defined to nothing for compilers other than MS.
• Compiling for kpathsea:
  • Download http://www.xs4all.nl/˜infovore/tex4htk.tgz.
  • At texk, issue the commands ‘./configure’ and ‘make’.

• In the presence of kpathsea, there is no reason for the explicit t-records in tex4ht.env. All tex4ht needs is just the information available to latex about the tfm files, and kpathsea already embeds it.
• To my understanding, the value of @texmf@ is known to kpathsea within some variable (TEXMF?). The value of that variable can be accessed by tex4ht through ‘kpse_var_value( "TEXMF" )’. So i-records similar to the following ones may provide to tex4ht the information needed to locate the files.
    ikpatsea{TEXMF}\tex4ht\ht-fonts\ 
    ikpatsea{.....}\tex4ht\ht-fonts\ 
   
  

  Of course, I rather have texht request kpathsea to perform the search, but I don’t know how to request it through a command similar to the following one.

       extended_kpse_find_file(filename, var, path, ...) 
                                         ^^^ 
                                       "TEXMF" 
                                              ^^^^^ 
                                             "\tex4ht\ht-fonts\" 
  

  I don’t yet follow the details of ‘client_path’ but I suspect it might provide support for instructions similar to the above.

1.3 Outline

<..tex4ht.c..>+
 <.defines.>
 <.types.>
 <.vars.>
 <.header functions.>
 <.functions.>
 
 int <.CDECL.> main(argc, argv)
        int  argc;
        U_CHAR **argv
 ;{  <.main’s vars.>
    <.resplit argv for windows.>
    <.set signals.>
    <.program signature.>
    <.init traces.>
    <.main’s init.>
    <.main’s body.>
    return 0;
 }
 -_-_-

The operating system claims that ‘Compilation exited abnormally with code 255’ even on cases that come back through return. why? Was this corrected?

<..program signature..>
 (IGNORED) printf("----------------------------\n");
 #ifndef KPATHSEA
 #ifdef PLATFORM
    (IGNORED) printf("tex4ht.c (2014-07-21-21:39 %s)\n",PLATFORM);
 #else
    (IGNORED) printf("tex4ht.c (2014-07-21-21:39)\n");
 #endif
 #else
 #ifdef PLATFORM
    (IGNORED) printf("tex4ht.c (2014-07-21-21:39 %s kpathsea)\n",PLATFORM);
 #else
    (IGNORED) printf("tex4ht.c (2014-07-21-21:39 kpathsea)\n");
 #endif
 #endif
 for(i=0; i<argc; i++){
     (IGNORED) printf("%s%s ", (i>1)?"\n  " : "", argv[i]); }
 (IGNORED) printf("\n");
 -_-_-

Main’s Body

<..main’s body..>
 {      long   file_len;
   <.scan job args.>
   <.find tex4ht.env.>
   <.get from tex4ht.env file.>
   <.file-len = size of file.>
   <.open log file.>
   <.export TEX4HTFONTSET.>
   <.scan postamble.>
   (IGNORED) fclose(dot_file);
 }
 <.scan preamble.>
 <.scan pages.>
 put_char(’\n’);put_char(’\n’);
 <.close active output files.>
 {    <.idv vars.>
   <.extract dvi chunks.> }
 -_-_-

<..vars..>+
 static FILE*  dot_file;
 -_-_-

<..main’s vars..>
 register INTEGER  i;
 int  ch;
 -_-_-

Utilities

<..defines..>+
 #define m_alloc(typ,n) (typ *) malloc_chk((int) ((n) * sizeof(typ)))
 -_-_-

<..header functions..>+
 static void* malloc_chk(ARG_I(int));
 -_-_-

<..functions..>+
 
 static void* malloc_chk( n ) int n
 ;{      void* p;
    if((p = (void *) malloc( (size_t) n)) == NULL ) bad_mem;
    return p;
 }
 -_-_-

<..header functions..>+
 static void* r_alloc(ARG_II(void *, size_t));
 -_-_-

<..functions..>+
 
 static void* r_alloc( q, n )
       void   *q;
       size_t  n
 ;{    void*   p;
    if((p = (void *) realloc( q, (size_t) n)) == NULL) bad_mem;
    return p;
 }
 -_-_-

1.4 Installers and Makefiles

1.5 Command Line Switches

<..scan job args..>
 {      int i;
        U_CHAR *p;
        const U_CHAR *in_name = "", *out_name = "";
   <.kpathsea arg 0.>
   <.DJGPP gif names for pictorial symbols.>
   if( argc == 1 ){ bad_arg; }
   for(i=1; i<argc; i++) {
     if( *( p=argv[i] ) == ’-’ ){ <.scan flaged args.> }
     else in_name = argv[i];
   }
   <.post scan comm ln args.>
   <.open input file.>
   { <.pre open output file.> }
 }
 -_-_-

If ‘strlen() argv[i] ) == 2’ the switch is followed by a space to be deleted when more input is awaited.

<..scan flaged args..>
 if( (int) strlen((char *)  argv[i] ) == 2 ){
    if( ++i == argc ) bad_arg;
 }
 switch( *(p+1) ){
   case ’b’:{ <.get start char-gif from com ln.>     break; }
   case ’c’:{ <.env blocks selector.>  break;}
   case ’e’:{ <.get .env directory from com ln.>break; }
   case ’f’:{ <.get file name wo dir.>break; }
   case ’F’:{ <.get ignored chars.>             break; }
   case ’g’:{ <.get gif-type from com ln.>      break; }
   case ’h’:{ <.trace context.>  break; }
   case ’i’:{ <.get htf font dir from com ln.>  break; }
   case ’l’:{ <.get tex4ht.fls from com ln.>  break; }
   case ’P’:{ <.permission for system calls.>  break; }
   case ’S’:{ <.get gif script from com ln.>    break; }
   case ’s’:{ <.get cascade style sheet data.>    break; }
   case ’t’:{ <.get tfm font dir from com ln.>  break; }
   case ’u’:{ <.get unicode entity representations.>  break; }
   case ’v’:{ <.idv version replacement num.>  break; }
   case ’x’:{ <.extra options.>  break; }
   case ’.’:{ <.default extension.>  break; }
    default:{ bad_arg; }
 }
 -_-_-

The following is to allow for commands ‘htlatex -f/../source/foo.tex’ which draw files from other directories for latex, but use the current directory for tex4ht.c (and t4ht.c). The character immediately after ‘-f’ is a directory indicator character. The ‘-f/’ requests a search for the indicated file to take place in the current directory, not in a remote one.

<..get file name wo dir..>
 p = p + 2;
 in_name = p + (int) strlen((char *)  p );
 while( *in_name != *p ){ in_name--; }
 in_name++;
 -_-_-

Suboptions for ‘-h’.

<..trace context..>
 {
   char trace = *(p+2);
   if (trace == ’A’ || trace == ’e’) { <.context for warn and errs.> }
   if (trace == ’A’ || trace == ’f’) { <.dump htf files.> }
   if (trace == ’A’ || trace == ’F’) { <.trace htf search.> }
   if (trace == ’A’ || trace == ’s’) { <.trace specials in dvi.> }
   if (trace == ’A’ || trace == ’g’) { <.cmd line trace groups.> }
   if (trace == ’A’ || trace == ’v’) { <.dump env file.> }
   if (trace == ’A’ || trace == ’V’) { <.trace env search.> }
   else { bad_arg; }
 }
 -_-_-

The following is incomplete!!!

<..command line options..>
 "improper command line\ntex4ht [-f<path-separator-ch>]in-file[.dvi]\n"
 "   [-.<ext>]            replacement to default file extension name .dvi\n"
 "   [-c<tag name>]       choose named segment in env file\n"
 "   [-e<env-file>]\n"
 "   [-f<path-separator-ch>]        remove path from the file name\n"
 "   [-F<ch-code>]        replacement for missing font characters; 0--255; default 0\n"
 "   [-g<bitmap-file-ext>]\n"
 "   [-h(e|f|F|g|s|v|V)]  trace: e-errors/warnings, f-htf, F-htf search\n"
 "                             g-groups, s-specials, v-env, V-env search, A-all\n"
 "   [-i<htf-font-dir>]\n"
 "   [-l<bookkeeping-file>]\n"
 "   [-P(*|<filter>)]     permission for system calls: *-always, filter\n"
 "   [-S<image-script>]\n"
 "   [-s<css-file-ext>]   default: -s4cs; multiple entries allowed\n"
 "   [-t<tfm-font-dir>]\n"
 "   [-u10]               base 10 for unicode characters\n"
 "   [-utf8]              utf-8 encoding for unicode characters\n"
 "   [-v<idv version>]    replacement for the given dvi version\n"
 "   [-xs]           ms-dos file names for automatically generated gifs\n"
 -_-_-

Line breaks within command lines should be physically given.

1.6 I/O Files

<..defines..>+
 #ifndef PATH_MAX
 #define PATH_MAX 512
 #endif
 -_-_-

Open Input

<..vars..>+
 static FILE* dvi_file;
 -_-_-

<..main’s init..>
 dvi_file = stdin;
 -_-_-

Below: If input is from stdin which is not a console device, switch it to binary mode.

Warning ‘statement with no effect’ under ‘gcc -Wall’ on non-Windows. This is not a problem; SET˙BINARY is a no-op.

<..open input file..>
 if( *in_name != ’\0’ ){ <.file = non-piped input file.> }
 #ifdef KWIN32
    else if (!isatty(fileno(stdin))) SET_BINARY(fileno(stdin));
 #endif
 -_-_-

<..file = non-piped input file..>
       BOOL tag;
 job_name_n = (int) strlen( in_name );
 job_name = m_alloc(char, job_name_n+6);
 (IGNORED) strcpy((char *) job_name, in_name);
 tag = job_name_n < 3;
 if( !tag ){
    tag = !eq_str( job_name+job_name_n-<.dvi ext name length.>,<.dvi ext name.>);
 }
 if( tag ){
    job_name_n+=<.dvi ext name length.>; (IGNORED) strct(job_name, <.dvi ext name.>);
 }
 if( (dvi_file = fopen(job_name, READ_BIN_FLAGS)) == NULL )
    { <.try jobname with extension.> }
 <.open idv file.>
 -_-_-

<..dvi ext name length..>
 (
   (ext==NULL)? 4 : (int) strlen((char *) ext)
 )
 -_-_-

<..dvi ext name..>
 (
   (ext==NULL)? ".dvi" : ext
 )
 -_-_-

<..vars..>+
 static U_CHAR *ext = NULL;
 -_-_-

<..default extension..>
 ext = p+1;
 -_-_-

The following tries to remove an extension from the filename, before adding the ‘.dvi’ extension. This can help with ht scripts of the form

in cases that the users insert for ‘$2’ file names with extension.

<..try jobname with extension..>
 {                             int i;
    for(i=job_name_n-5; i; i--){
      if( job_name[i] == ’.’ ){
        job_name[i] = ’\0’;
        job_name_n = i + <.dvi ext name length.>;
        (IGNORED) strct(job_name, <.dvi ext name.>);
        break;
    } }
    if( (dvi_file = fopen(job_name, READ_BIN_FLAGS)) == NULL ){
       warn_i_str(1, job_name); bad_in_file(job_name);
 }  }
 -_-_-

<..vars..>+
 static char* job_name;
 static int   job_name_n;
 -_-_-

strcat should be in string.h, but c++ doesn’t find it there. We use it just for concatenating an extension of file name. Should have the interface ‘char *strcat( (char *, const U_CHAR *) );’.

<..header functions..>+
 static void strct( ARG_II(char *, const U_CHAR *) );
 -_-_-

<..functions..>+
 
 static void strct( str1, str2 )
      U_CHAR * str1;
      const U_CHAR * str2
 
 ;{   U_CHAR * ch;
    ch = str1 + (int) strlen((char *) str1);
    (IGNORED) strcpy((char *)  ch, str2 );
 }
 -_-_-

Open Root Html File

<..pre open output file..>
    U_CHAR *name=0;
 if( *out_name == ’\0’ )
   { if( *in_name == ’\0’ ){ <.name = "tex4htput.html".> }
     else                  { <.name = derived from input file name.> }
   }
 else{ <.name = output file from arg list.> }
 no_root_file = name;
 -_-_-

We want to wait with actually opening the file output file until the last minute, to allow a special to change the extension name from the tex file. That last minute is taken to be when a cll to openning ort closing a file occurs, or when output is to be written into the file.

<..open output file..>
 if( no_root_file ){  open_o_file(); }
 -_-_-

<..header functions..>+
 static void open_o_file( ARG_I(void) );
 -_-_-

<..functions..>+
 static void open_o_file(MYVOID)
 {
    <.open root tex4ht file.>
    cur_o_file = root_file = open_html_file(no_root_file);
    no_root_file = (char *) 0;
 }
 -_-_-

<..open root tex4ht file..>
     struct files_rec* p;
 p = m_alloc(struct files_rec, 1);
 if( opened_files != (struct files_rec*) 0 ) opened_files->prev = p;
 p->prev = (struct files_rec *) 0;
 p->next = opened_files;     opened_files = p;
 p->name = no_root_file;
 p->file =
 -_-_-

<..cond insert new line..>
 if( !no_root_file ){
   <.end text accent.>
   if( !<.radical-line-off.> ){  put_char(’\n’); }
 }
 -_-_-

<..change ext of root file..>
 if( no_root_file ){
    U_CHAR *name;
    name = m_alloc(char, 256);
    (IGNORED) strcpy((char *)  name, (char *) no_root_file );
    free((void *)  no_root_file);
    no_root_file = name;
    name += (size_t) strlen((char *) name);  while( *(--name) != ’.’ ); name++;
    while( special_n-- ){
        if( (no_root_file+253) == name ) name--;
        *name++ = get_char();
   }
    *name = ’\0’;
 } else {
           U_CHAR str[256], *p;
    p = str;  while( special_n-- ){ *p++ = get_char(); }  *p = ’\0’;
    warn_i_str(43,str);
 }
 -_-_-

<..vars..>+
 static U_CHAR *no_root_file;
 -_-_-

<..name = "tex4htput.html"..>
 bad_arg;
 -_-_-

<..name = derived from input file name..>
 int n = (int) strlen((char *)  job_name );
 name = m_alloc(char, 6 + n);
 (IGNORED) strcpy((char *) name, (char *) job_name);
 n -= 4; *(name + n) = ’\0’;
 (IGNORED) strct(name,".html");
 #ifdef HTM
 name[n+4] =’\0’;
 #endif
 -_-_-

<..name = output file from arg list..>
    int tag = 1;
    int n = (int) strlen( out_name );
 name = m_alloc(char, 6 + n);
 (IGNORED) strcpy((char *) name, out_name);
 while( n-- )   tag = tag && (*(name+n) != ’.’) ;
 if( tag ) (IGNORED) strct(name,".html");
 #ifdef HTM
 name[n+4] = ’\0’;
 #endif
 -_-_-

<..header functions..>+
 static FILE* open_html_file( ARG_I(char*) );
 -_-_-

<..functions..>+
 
 static FILE* open_html_file(name)
      char* name
 ;{   FILE* file;
      char* str;
   str = m_alloc(char, (int) strlen((char *) name) +  1);
   (IGNORED) strcpy((char *) str, (char *) name);
   (IGNORED) printf(" file %s\n", str);
   (IGNORED) fprintf(log_file, "File: %s\n", str);
   if( (file = fopen(str, WRITE_TEXT_FLAGS)) == NULL )  bad_in_file(str);
   free((void *)  str);
   return file;
 }
 -_-_-