preconv: Allow set default encodding

[Alexey Gladkov]

Greetings!

Following patch adds -D in the preconv to change the input encoding by 
default.

--
Rgrds, legion

--- groff-1.19.3/src/preproc/preconv/preconv.cpp.fix    2008-08-20 01:57:33 
+0400
+++ groff-1.19.3/src/preproc/preconv/preconv.cpp        2008-08-20 01:57:36 
+0400
@@ -22,8 +22,9 @@ Foundation, 51 Franklin St - Fifth Floor
 #include "lib.h"
 
 #include <assert.h>
 #include <stdlib.h>
+#include <string.h>
 #include <errno.h>
 #include "errarg.h"
 #include "error.h"
 #include "localcharset.h"
@@ -44,9 +45,9 @@ Foundation, 51 Franklin St - Fifth Floor
 #define MAX_VAR_LEN 100
 
 extern "C" const char *Version_string;
 
-const char *default_encoding;
+char *default_encoding = NULL;
 char user_encoding[MAX_VAR_LEN];
 char encoding_string[MAX_VAR_LEN];
 int debug_flag = 0;
 int raw_flag = 0;
@@ -1098,8 +1099,9 @@ usage(FILE *stream)
 {
   fprintf(stream, "usage: %s [ option ] [ files ]\n"
                  "\n"
                  "-d           show debugging messages\n"
+                 "-D encoding  specify default encoding\n"
                  "-e encoding  specify input encoding\n"
                  "-h           print this message\n"
                  "-r           don't add .lf requests\n"
                  "-v           print version number\n"
@@ -1119,23 +1121,18 @@ main(int argc, char **argv)
   // getopt() is called since the usage message shows the default
   // encoding.
   setlocale(LC_ALL, "");
   char *locale = getlocale(LC_CTYPE);
-  if (!locale || !strcmp(locale, "C") || !strcmp(locale, "POSIX"))
-    default_encoding = "latin1";
-  else
-    default_encoding = locale_charset();
-
   program_name = argv[0];
-  int opt;
+  int opt, const_default_locale = 1;
   static const struct option long_options[] = {
     { "help", no_argument, 0, 'h' },
     { "version", no_argument, 0, 'v' },
     { NULL, 0, 0, 0 }
   };
   // Parse the command line options.
   while ((opt = getopt_long(argc, argv,
-                           "de:hrv", long_options, NULL)) != EOF)
+                           "dD:e:hrv", long_options, NULL)) != EOF)
     switch (opt) {
     case 'v':
       printf("GNU preconv (groff) version %s %s iconv support\n",
             Version_string,
@@ -1157,8 +1154,19 @@ main(int argc, char **argv)
       }
       else
        user_encoding[0] = 0;
       break;
+    case 'D':
+      if (optarg) {
+       int len = strlen(optarg) + 1;
+       if ((default_encoding = (char *) malloc(len)) == NULL) {
+         fprintf(stderr, "malloc: %s\n", strerror(errno));
+         exit(1);
+       }
+       strncpy(default_encoding, optarg, len);
+       const_default_locale = 0;
+      }
+      break;
     case 'r':
       raw_flag = 1;
       break;
     case 'h':
@@ -1171,16 +1179,24 @@ main(int argc, char **argv)
       break;
     default:
       assert(0);
     }
+  if (!default_encoding) {
+    if (!locale || !strcmp(locale, "C") || !strcmp(locale, "POSIX"))
+       default_encoding = "latin1";
+    else
+       default_encoding = (char *) locale_charset();
+  }
   int nbad = 0;
   if (debug_flag)
     fprintf(stderr, "default encoding: `%s'\n", default_encoding);
   if (optind >= argc)
     nbad += !do_file("-");
   else
     for (int i = optind; i < argc; i++)
       nbad += !do_file(argv[i]);
+  if (!const_default_locale)
+    free(default_encoding);
   if (ferror(stdout) || fflush(stdout) < 0)
     fatal("output error");
   return nbad != 0;
 }
--- groff-1.19.3/src/preproc/preconv/preconv.man.fix    2008-08-20 01:57:36 
+0400
+++ groff-1.19.3/src/preproc/preconv/preconv.man        2008-08-20 01:57:36 
+0400
@@ -73,8 +73,12 @@ options of
 .B \-d
 Emit debugging messages to standard error (mainly the used encoding).
 .
 .TP
+.BI \-D encoding
+Specify default encoding if everything fails (See below).
+.
+.TP
 .BI \-e encoding
 Specify input encoding explicitly, overriding all other methods.
 This corresponds to
 .BR groff 's
--- groff-1.19.3/src/roff/groff/groff.cpp.fix   2008-08-20 01:57:33 +0400
+++ groff-1.19.3/src/roff/groff/groff.cpp       2008-08-20 02:08:27 +0400
@@ -134,9 +134,9 @@ int main(int argc, char **argv)
     { NULL, 0, 0, 0 }
   };
   while ((opt = getopt_long(
                  argc, argv,
-                 "abcCd:eEf:F:gGhiI:lkK:L:m:M:n:No:pP:r:RsStT:UvVw:W:XzZ",
+                 "abcCd:D:eEf:F:gGhiI:lkK:L:m:M:n:No:pP:r:RsStT:UvVw:W:XzZ",
                  long_options, NULL))
         != EOF) {
     char buf[3];
     buf[0] = '-';
@@ -155,8 +155,12 @@ int main(int argc, char **argv)
       Pargs += buf;
       Pargs += optarg;
       Pargs += '\0';
       break;
+    case 'D':
+      commands[PRECONV_INDEX].set_name(command_prefix, "preconv");
+      commands[PRECONV_INDEX].append_arg("-D", optarg);
+      break;
     case 'K':
       commands[PRECONV_INDEX].append_arg("-e", optarg);
       Kflag = 1;
       // fall through
@@ -742,9 +746,9 @@ void synopsis(FILE *stream)
 {
   fprintf(stream,
 "usage: %s [-abceghiklpstvzCENRSUVXZ] [-Fdir] [-mname] [-Tdev] [-ffam]\n"
 "       [-wname] [-Wname] [-Mdir] [-dcs] [-rcn] [-nnum] [-olist] [-Parg]\n"
-"       [-Karg] [-Larg] [-Idir] [files...]\n",
+"       [-Darg] [-Karg] [-Larg] [-Idir] [files...]\n",
          program_name);
 }
 
 void help()
@@ -789,8 +793,9 @@ void help()
 "-S\tenable safer mode (the default)\n"
 "-U\tenable unsafe mode\n"
 "-Idir\tsearch dir for soelim, troff, and grops.  Implies -s\n"
 "-Karg\tuse arg as input encoding.  Implies -k\n"
+"-Darg\tuse arg as default input encoding. Implies -k\n"
 "\n",
        stdout);
   exit(0);
 }
--- groff-1.19.3/src/roff/groff/groff.man.fix   2008-08-20 01:57:33 +0400
+++ groff-1.19.3/src/roff/groff/groff.man       2008-08-20 01:57:36 +0400
@@ -49,8 +49,9 @@ groff \- front-end for the groff documen
 .OP \-f fam
 .OP \-F dir
 .OP \-I dir
 .OP \-K arg
+.OP \-D arg
 .OP \-L arg
 .OP \-m name
 .OP \-M dir
 .OP \-n num
@@ -230,8 +231,19 @@ Implies
 .BR \-k .
 .
 .
 .TP
+.BI \-D\  arg
+Set default input encoding used by
+.B preconv
+to
+.IR arg .
+.
+Implies
+.BR \-k .
+.
+.
+.TP
 .B \-l
 Send the output to a spooler program for printing.
 .
 The command that should be used for this is specified by the