New option --tag; Some minor fixes.

git-svn-id: file:///svnroot/swis/trunk@11 05ba3e8d-823b-0410-8fb2-de0ee4edb5ba

2 files changed, 41 insertions, 7 deletions

diff --git a/src/html-strip.l b/src/html-strip.l
index b42f618..f337c5d 100644
--- a/src/html-strip.l
+++ b/src/html-strip.l
@@ -25,7 +25,7 @@ size_t idx;
 iconv_t cd = INVALID_ICONV_CD;
 
 static void
-parse_content_type()
+parse_content_type ()
 {
   char *start, *p;
   size_t len;
@@ -53,7 +53,7 @@ parse_content_type()
 	      if (*p == ';' || isspace (*p))
 		{
 		  *p = 0;
-	      break;
+		  break;
 		}
 	      p++;
 	    }
@@ -166,7 +166,7 @@ WS [ \t]+
       "ß"  output ("ß");
       "&Eth;"    output (yytext[1] == 'E' ? "Ð" : "ð");
       "ÿ"   output ("ÿ");
-      "&#"[0-9]{1,3}";" { /* FIXME */; }
+      "&#"[0-9]{1,5}";" { /* FIXME */; }
       " "   output (" ");
       "&"[^;]*";" output (" ");
       . {
@@ -185,12 +185,13 @@ WS [ \t]+
 <ELEMENT>{
       \"[^\"]*\" ; 
       '[^']*'    ;
-      ">"        { BEGIN (INITIAL); }
+      ">"        { BEGIN (INITIAL); output (" "); }
       .          ;
 }
 %%
 
 char **input_file;
+int tag_option;
 
 int
 open_input ()
@@ -206,7 +207,11 @@ open_input ()
           if (!yyin)
             error (1, errno, "cannot open input file %s", name);
         }
-      return 0;
+      if (tag_option)
+        {
+           fprintf (yyout ? yyout : stdout, "\n> %s\n", name);
+        }
+       return 0;
     }
   return 1;
 }
@@ -235,6 +240,7 @@ struct option options[] = {
   { "help", no_argument, NULL, 'h' },
   { "version", no_argument, NULL, 'v' },
   { "output", required_argument, NULL, 'o' },
+  { "tag", no_argument, NULL, 't' },
   { NULL }
 };
 
@@ -248,6 +254,7 @@ usage ()
   printf ("\nOptions are:\n");
   printf ("  -d, --debug                    output debugging info\n");
   printf ("  -o, --output=FILE              direct output to FILE instead of stdout\n");
+  printf ("  -t, --tag                      tag each output block with the source file name");
   printf ("\n");
   printf ("  -h, --help                     print this help list\n");
   printf ("  -v, --version                  print program version and exit\n");
@@ -263,7 +270,7 @@ main (int argc, char **argv)
   program_name = argv[0];
   yy_flex_debug = 0;
 
-  while ((c = getopt_long (argc, argv, "dho:v", options, NULL)) != EOF)
+  while ((c = getopt_long (argc, argv, "dho:tv", options, NULL)) != EOF)
     {
       switch (c)
 	{
@@ -284,6 +291,10 @@ main (int argc, char **argv)
 	  if (!yyout)
 	    error (1, errno, "cannot open output file %s", optarg);
 	  break;
+
+	case 't':
+	  tag_option = 1;
+	  break;
 	  
 	case 'v':
 	  version_etc (stdout, "html-strip", PACKAGE_NAME, VERSION, 
diff --git a/src/word-split.c b/src/word-split.c
index 59a05b7..59b11d6 100644
--- a/src/word-split.c
+++ b/src/word-split.c
@@ -39,6 +39,7 @@ usage ()
   printf ("\nOptions are:\n");
   /*  printf ("  -d, --debug                    output debugging info\n"); */
   printf ("  -o, --output=FILE              direct output to FILE instead of stdout\n");
+  printf ("  -t, --tag                      preserve file name tags\n");
   printf ("\n");
   printf ("  -h, --help                     print this help list\n");
   printf ("  -v, --version                  print program version and exit\n");
@@ -49,6 +50,7 @@ usage ()
 char **input_file;
 FILE *input;
 FILE *output;
+int tag_option;
 
 int
 open_input ()
@@ -75,10 +77,27 @@ int
 word_split ()
 {
   unsigned wc;
+  int after_newline;
+  
   while ((wc = fgetc (input)) != EOF)
     {
       if (wc < 0x80)
 	{
+	  if (tag_option)
+	    {
+	      if (after_newline && wc == '>')
+		{
+		  do
+		    fputc (wc, output);
+		  while ((wc = fgetc (input)) != EOF && wc != '\n');
+		  fputc ('\n', output);
+		  after_delim_output = 1;
+		  after_newline = 0;
+		  continue;
+		}
+	      after_newline = wc == '\n';
+	    }
+	  
 	  if (isalnum (wc))
 	    {
 	      fputc (wc, output);
@@ -134,7 +153,7 @@ main (int argc, char **argv)
   
   program_name = argv[0];
 
-  while ((c = getopt_long (argc, argv, "dho:v", options, NULL)) != EOF)
+  while ((c = getopt_long (argc, argv, "dho:tv", options, NULL)) != EOF)
     {
       switch (c)
 	{
@@ -156,6 +175,10 @@ main (int argc, char **argv)
 	  if (!output)
 	    error (1, errno, "cannot open output file %s", optarg);
 	  break;
+
+	case 't':
+	  tag_option = 1;
+	  break;
 	  
 	case 'v':
 	  version_etc (stdout, "word-split", PACKAGE_NAME, VERSION,