freeswan/doc/utils/html2four.c

   1 /*
   2         extract headers from HTML files
   3         in format suitable for turning into permuted index
   4 */
   5
   6 #include <ctype.h>
   7 #include <stdlib.h>
   8 #include <stdio.h>
   9 #include <string.h>
  10
  11 /*
  12         maximum sizes for input line and for name in <a> tag
  13 */
  14 #define MAX_LINE  512
  15 #define MAX_NAME   64
  16
  17 /*
  18         functions
  19         all return 0 for OK, 1 for errors
  20 */
  21 int do_file( char *, FILE * ) ;
  22 int parse_line( char * ) ;
  23 int print_line( char *, char *) ;
  24 int print_header_problem( char * ) ;
  25 int sanity() ;
  26
  27 void die( char * ) ;
  28
  29 char    *prog_name ;
  30 int     max_level ;
  31 char    *current_file ;
  32
  33 int main(int argc, char* argv[])
  34 {
  35         char *p ;
  36         int temp, done, status ;
  37         FILE *fp ;
  38
  39         prog_name = *argv ;
  40         argc--,argv++ ;
  41
  42         max_level = 9 ;
  43         if(argc && *argv )      {
  44                 p = *argv ;
  45                 if( p[0] == '-' )       {
  46                         if( isdigit(p[1]) && p[2] == '\0' )     {
  47                                 max_level = p[1] - 0 ;
  48                                 argc-- ;
  49                                 argv++ ;
  50                         }
  51                         else die("unknown option") ;
  52         }       }
  53
  54         status = done = 0 ;
  55         if( argc == 0)  {
  56                 if( (status = do_file("STDIN", stdin)) == 0 )
  57                         done++ ;
  58         }
  59         else    {
  60 /*
  61                 printf("ARGC = %d\n", argc ) ;
  62 */
  63                 while( argc-- ) {
  64                         p = *argv++ ;
  65 /*
  66                         printf("ARGV P %s %s\n", *argv, p) ;
  67 */
  68                         if( p == NULL ) {
  69                                 fprintf(stderr, "%s: null filename pointer\n", prog_name) ;
  70                                 status++ ;
  71                         }
  72                         else if( (fp = fopen(p,"r")) == NULL )  {
  73                                 fprintf(stderr, "%s: cannot open file %s\n", prog_name, p) ;
  74                                 status++ ;
  75                         }
  76                         else    {
  77                                 if( (temp = do_file(p, fp)) != 0 )
  78                                         status++ ;
  79                                 done++ ;
  80                                 fclose(fp) ;
  81                         }
  82                         fflush(stderr) ;
  83                         fflush(stdout) ;
  84                 }
  85         }
  86 /*
  87         printf("%s: %d files processed, %d with errors\n", prog_name, done, status) ;
  88 */
  89         return( status ? 1 : 0 ) ;
  90 }
  91
  92 void die( char *message )
  93 {
  94         fflush(stdout) ;
  95         fprintf(stderr, "%s: %s\n", prog_name, message) ;
  96         exit(1) ;
  97 }
  98
  99 int header_flags[10] ;
 100 int in_header ;
 101
 102 char buffer[MAX_LINE+1] ;
 103 char label[MAX_NAME+1] ;
 104
 105 int do_file( char *file, FILE *fp )
 106 {
 107         int i, status, x, y ;
 108         char *base, *p ;
 109
 110         status = 0 ;
 111         in_header = 0 ;
 112         label[0] = '\0' ;
 113         for( i = 0 ; i < 10 ; i++ )
 114                 header_flags[i] = 0 ;
 115         current_file = file ;
 116
 117         while( base = fgets(buffer, MAX_LINE, fp) )     {
 118                 // count < and > characters in line
 119                 for( x = y = 0, p = base ; *p ; p++ )
 120                         switch( *p )    {
 121                                 case '<':
 122                                         x++ ;
 123                                         break ;
 124                                 case '>':
 125                                         y++ ;
 126                                         break ;
 127                                 default:
 128                                         break ;
 129                         }
 130                 // skip line if no < or >
 131                 if( x == 0 && y == 0 )
 132                         continue ;
 133                 // report error for unequal count
 134                 else if( x != y )       {
 135                         if( strncmp( base, "<!--", 4) && strncmp(base, "-->", 3) )      {
 136                                 fflush(stdout) ;
 137                                 fprintf(stderr, "%s in file %s: unequal < > counts %d %d\n",
 138                                         prog_name, file, x, y ) ;
 139                                 fprintf(stderr, "%s: %s\n", prog_name, base) ;
 140                                 fflush(stderr) ;
 141                                 status = 1 ;
 142                         }
 143                         continue ;
 144                 }
 145                 // parse lines containing tags
 146                 else
 147                         if( parse_line(base) )
 148                                 status = 1 ;
 149                 // check that header labelling is sane
 150                 for( i = x = y = 0 ; i < 10 ; i++ )     {
 151                         // count non-zero entries
 152                         if( x = header_flags[i] )
 153                                 y++ ;
 154                         // should be in 0 or 1 headers at a time
 155                         if( x > 1 || x < 0 )
 156                                 status = 1 ;
 157                 }
 158                 if( y > 1 )
 159                         status = 1 ;
 160         }
 161         return status ;
 162 }
 163
 164 int parse_line( char *data )
 165 {
 166         char *p, *q, *end ;
 167         int x ;
 168
 169         // set end pointer
 170         for( end = data ; *end ; end++ )
 171                 ;
 172         // trim off trailing returns or newlines
 173         for( p = end - 1, q = end ; q > data ; p--,q-- )        {
 174                 switch( *p )    {
 175                         case '\012':
 176                         case '\015':
 177                                 *p = '\0' ;
 178                                 continue ;
 179                         default:
 180                                 break ; // out of switch()
 181                 }
 182                 break ; // out of for()
 183         }
 184         end = q ;
 185         p = data ;
 186         while( p < end )        {
 187                 // find tag delimiters
 188                 if( *p == '<')  {
 189                         for( q = p + 1 ; *q ; q++ )
 190                                 if( *q == '<' || *q == '>' )
 191                                         break ;
 192                         // if we find another '<'
 193                         // restart tag search from it
 194                         if( *q == '<' ) {
 195                                 p = q ;
 196                                 continue ;
 197                         }
 198                         // "<>" is not interesting
 199                         if( q == p + 1 )        {
 200                                 fflush(stdout) ;
 201                                 fprintf(stderr, "%s: null tag\n", prog_name) ;
 202                                 fprintf(stderr, "%s: line\n", prog_name, data) ;
 203                                 fflush(stderr) ;
 204                                 p = q + 1 ;
 205                                 continue ;
 206                         }
 207                         // ignore delimiters once found
 208                         *q = '\0' ;
 209                         p++ ;
 210                         // p points to tag contents, null terminated
 211                         switch( *p )    {
 212                         // save contents of <a name= > tags
 213                         case 'a' :
 214                         case 'A' :
 215                                 if(      p[1] == ' ' &&
 216                                         (p[2] == 'n' || p[2] == 'N') &&
 217                                         (p[3] == 'a' || p[3] == 'A') &&
 218                                         (p[4] == 'm' || p[4] == 'M') &&
 219                                         (p[5] == 'e' || p[5] == 'E') &&
 220                                          p[6] == '=' )
 221                                 strncpy(label, p + 7, MAX_NAME) ;
 222                                 break ;
 223                         case 'b' :
 224                         case 'B' :
 225                                 if(     in_header && strlen(p) == 2 &&
 226                                         (p[1] == 'r' || p[1] == 'R') )
 227                                         putchar(' ') ;
 228                                 break ;
 229                         // header tags
 230                         case 'h' :
 231                         case 'H' :
 232                                 if( strlen(p) == 2 && isdigit(p[1]) )   {
 233                                         if( in_header )
 234                                                 fprintf(stderr, "%s: bad header nesting in %s\n",
 235                                                         prog_name, current_file) ;
 236                                         x = p[1] - '0' ;
 237                                         in_header = 1 ;
 238                                         header_flags[x]++ ;
 239                                         printf("%s\t%s\tH%d\t", current_file, label, x) ;
 240                                 }
 241                                 break ;
 242                         // only care about end-of-header
 243                         case '/':
 244                                 p++ ;
 245                                 switch( *p )    {
 246                                 case 'h' :
 247                                 case 'H' :
 248                                         if( strlen(p) == 2 && isdigit(p[1]) )   {
 249                                                 if( ! in_header )
 250                                                         fprintf(stderr, "%s: bad header nesting in %s\n",
 251                                                                 prog_name, current_file) ;
 252                                                 x = p[1] - '0' ;
 253                                                 in_header = 0 ;
 254                                                 header_flags[x]-- ;
 255                                                 printf("\n") ;
 256                                         }
 257                                         break ;
 258                                 }
 259                                 break ;
 260                         // uninteresting tag, look for next
 261                         default :
 262                                 break ;
 263                         }
 264                 // tag done, point p beyond it
 265                 p = q + 1 ;
 266                 }
 267                 else if( in_header )    {
 268                         if( isprint(*p) && *p != '\n' )
 269                                 putchar(*p) ;
 270                         else
 271                                 putchar(' ');
 272                         p++ ;
 273                 }
 274                 else
 275                         p++ ;
 276         }
 277         return(0) ;
 278 }
 279
 280 int print_line( char *tag, char *text)
 281 {
 282         printf("%%s\ts\t%s\t%s\t\n", current_file, label, tag, text) ;
 283         return 0 ;
 284 }
 285
 286 int print_header_problem( char *file )
 287 {
 288         int i ;
 289         fflush(stdout) ;
 290         fprintf(stderr, "%s: HEADER TAG PROBLEM in file %s\n", prog_name, file) ;
 291         fprintf(stderr, "%s: counts", prog_name) ;
 292         for ( i = 0 ; i < 10 ; i++ )
 293                 fprintf(stderr, "\t%d", i) ;
 294         fprintf(stderr,"\n") ;
 295         fflush(stderr) ;
 296         return(0) ;
 297 }
 298