clrlibru.c

   1 /*
   2  * Copyright (c) 2004 Teodor Sigaev <teodor@sigaev.ru>
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *        notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *        notice, this list of conditions and the following disclaimer in the
  12  *        documentation and/or other materials provided with the distribution.
  13  * 3. Neither the name of the author nor the names of any co-contributors
  14  *        may be used to endorse or promote products derived from this software
  15  *        without specific prior written permission.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY CONTRIBUTORS ``AS IS'' AND ANY EXPRESS
  18  * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  19  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED. IN NO EVENT SHALL CONTRIBUTORS BE LIABLE FOR ANY
  21  * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
  23  * GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  24  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
  25  * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
  26  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN
  27  * IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  28  */
  29
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <string.h>
  33 #include <ctype.h>
  34
  35 static void pushoutstr(FILE *out, char *buf, int len);
  36
  37 static void
  38 usage() {
  39         printf("Clear text from lib.ru for Pocket PC, Version 0.2.\n");
  40         printf("Copyright (c) 2004 Teodor Sigaev <teodor@sigaev.ru>\n");
  41         printf("       All rights reserved.\n");
  42         printf("Usage:\n   clrlibru [-i INPUTFILE] [-o OUTPUT] [-l NUMSPACE]\n");
  43         exit(0);
  44 }
  45
  46 static char* RemoveTag[]={
  47         "select",
  48         "head",
  49         "div",
  50         "a",
  51         "form",
  52         "script",
  53         "style",
  54         "title",
  55         NULL
  56 };
  57
  58 static int
  59 is_rtag(char *tag, int len) {
  60         char **ptr=RemoveTag;
  61
  62         tag[len]='\0';
  63         while( *ptr ) {
  64                 if ( strcmp(tag, *ptr) == 0 )
  65                         return 1;
  66                 ptr++;
  67         }
  68         return 0;
  69 }
  70
  71 typedef struct {
  72         char    *quote;
  73         char    *str;
  74         int     len;
  75 } Quote;
  76
  77 static Quote quoteChange[] = {
  78         {"quot", "\"", -1},
  79         {"laquo", "\"", -1},
  80         {"raquo", "\"", -1},
  81         {"quot", "\"", -1},
  82         {"lt", "<", -1},
  83         {"gt", ">", -1},
  84         {"nbsp", " ", -1},
  85         {"mdash", "-", -1},
  86         {"amp", "&", -1},
  87         {"shy", "", -1},
  88         {NULL, NULL, -1}
  89 };
  90
  91 static int
  92 pushoutquot(FILE *out, char *buf, int buflen) {
  93         Quote   *ptr = quoteChange;
  94
  95         buf[buflen]='\0';
  96         while( ptr->quote ) {
  97                 if ( strcmp( ptr->quote, buf ) == 0 ) {
  98                         if ( ptr->len < 0 )
  99                                 ptr->len = strlen( ptr->str );
 100                         pushoutstr( out, ptr->str, ptr->len );
 101                         return 1;
 102                 }
 103                 ptr++;
 104         }
 105         return 0;
 106 }
 107
 108
 109 static char *optarg = NULL;
 110 static int current=1;
 111
 112 int
 113 mgetopt(int argn, char* argv[], char *option) {
 114         char key;
 115
 116         if ( current >= argn ) return -1;
 117
 118         key = *(argv[current]+1);
 119         if ( *(argv[current]) == '-' || *(argv[current]) == '/' ) {
 120                 char *ptr = strchr( option, key );
 121                 if ( ptr == NULL ) {
 122                         printf( "Unknown option: %s\n", argv[current]);
 123                         exit(1);
 124                 }
 125                 current++;
 126                 if ( *(ptr+1) == ':' ) {
 127                         if (current < argn) {
 128                                 optarg=argv[current];
 129                                 current++;
 130                         } else {
 131                                 printf("No value for -%c\n" ,key);
 132                                 exit(1);
 133                         }
 134                 } else
 135                         optarg=NULL;
 136         } else {
 137                 printf("Unknown option: %s\n", argv[current]);
 138                 exit(1);
 139         }
 140         return (int)key;
 141 }
 142
 143 #define INTXT           0
 144 #define INTAG           1
 145 #define FINDEND         2
 146 #define INDROPTAG       3
 147 #define INDROPINTAG     4
 148 #define INHEADTAG       5
 149 #define INDROPINCLSTAG  6
 150 #define WAITAFTERRED    7
 151 #define COMMENTBEGIN1   8
 152 #define COMMENTBEGIN2   9
 153 #define COMMENTIN       10
 154 #define COMMENTEND1     11
 155 #define COMMENTEND2     12
 156 #define INQUOTE         13
 157
 158 #define BUFFERLENGTH    8192
 159 #define REDSTRING       "  "
 160
 161 typedef enum TypeOut {
 162         Char,
 163         NewLine,
 164         Tag,
 165         Paragraph,
 166         None
 167 } TypeOut;
 168
 169
 170 static void
 171 pushout( FILE *out, TypeOut type, int value ) {
 172         static TypeOut PrevType=None;
 173         static int  prevvalue=0;
 174         static int newlinecount=0;
 175
 176         if ( type == Char ) {
 177                 if ( !(( PrevType==Tag || PrevType==None || PrevType==NewLine ) && ( value == ' ' || value == '\t' )) ) {
 178                         newlinecount=0;
 179                         fputc(value ,out);
 180                 }
 181                 prevvalue = value;
 182         } else if ( type == NewLine ) {
 183                 if ( newlinecount < 2 )
 184                         fputc('\n', out);
 185                 newlinecount++;
 186         } else if ( type != PrevType ) {
 187                 switch(type) {
 188                         case Tag:
 189                                 if ( !(PrevType==None || PrevType==NewLine || ( PrevType==Char && ( prevvalue == ' ' || prevvalue == '\t' || prevvalue == '\n'  ) )) ) {
 190                                         newlinecount=0;
 191                                         fputc(' ', out);
 192                                 }
 193                                 break;
 194                         case Paragraph:
 195                                 pushout(out, NewLine, 0);
 196                                 fwrite(REDSTRING, sizeof(char), strlen(REDSTRING), out);
 197                                 newlinecount=0;
 198                                 break;
 199                         default:
 200                                 printf("Unknown type: %d", type);
 201                                 exit(1);
 202                 }
 203         }
 204         PrevType = type;
 205 }
 206
 207 static void
 208 pushoutstr(FILE *out, char *buf, int len) {
 209         char *ptr=buf;
 210         while( ptr-buf<len ) {
 211                 pushout(out, Char, (int)(*ptr));
 212                 ptr++;
 213         }
 214 }
 215
 216 int
 217 main(int argn, char *argv[]) {
 218         int ch;
 219         FILE    *in=stdin, *out=stdout;
 220         int state=INTXT;
 221         char    buf[BUFFERLENGTH];
 222         int lenbuf=0,closelen=0;
 223         int spacelen=4;
 224
 225         while((ch = mgetopt(argn, argv, "l:i:o:h?"))!=-1) {
 226                 switch (ch) {
 227                         case 'i':
 228                                 if ( (in=fopen(optarg, "r"))==NULL) {
 229                                         printf("Can't open file %s\n", optarg);
 230                                         exit(1);
 231                                 }
 232                                 break;
 233                         case 'o':
 234                                 if ( (out=fopen(optarg, "w"))==NULL) {
 235                                         printf("Can't open file %s\n",optarg);
 236                                         exit(1);
 237                                 }
 238                                 break;
 239                         case 'l':
 240                                 spacelen = atoi(optarg);
 241                                 if ( spacelen < 0 || spacelen >= BUFFERLENGTH ) {
 242                                         printf("-l should be >= 0 and < %d\n", BUFFERLENGTH);
 243                                         exit(1);
 244                                 }
 245                                 break;
 246                         case 'h':
 247                         case '?':
 248                         default:
 249                                 usage();
 250                 }
 251         }
 252
 253         while( (ch=getc(in)) != EOF ) {
 254                 if ( ch == '\r' )
 255                         continue;
 256
 257                 if ( state==INTXT ) {
 258                         if ( ch == '<' ) {
 259                                 state=INHEADTAG;
 260                                 lenbuf=0;
 261                         } else if ( ch == '\n' ) {
 262                                 state=FINDEND;
 263                                 lenbuf=1;
 264                         } else if ( ch=='&' ) {
 265                                 *buf='&';
 266                                 lenbuf=1;
 267                                 state=INQUOTE;
 268                         } else {
 269                                 pushout(out, Char, ch);
 270                         }
 271                 } else if ( state==INHEADTAG ) {
 272                         if ( isalpha(ch) ) {
 273                                 if ( lenbuf < BUFFERLENGTH-1 ) {
 274                                         buf[ lenbuf ] = tolower(ch);
 275                                         lenbuf++;
 276                                 }
 277                         } else if ( ch == '!' ) {
 278                                 state = COMMENTBEGIN1;
 279                         } else if ( ch == '>' ) {
 280                                 if ( is_rtag(buf,lenbuf) ) {
 281                                         state = INDROPTAG;
 282                                         closelen=0;
 283                                 } else {
 284                                         state=INTXT;
 285                                         if ( lenbuf==0 )
 286                                                 pushoutstr(out, "<>", 2);
 287                                         else
 288                                                 pushout(out, Tag, 0);
 289                                 }
 290                         } else if ( lenbuf == 0 && ch != '/' ) {
 291                                 pushout(out, Char, '<');
 292                                 pushout(out, Char, ch);
 293                                 state=INTXT;
 294                         } else {
 295                                 if ( is_rtag(buf,lenbuf) ) {
 296                                         state = INDROPTAG;
 297                                         closelen=0;
 298                                 } else {
 299                                         state=INTAG;
 300                                 }
 301                         }
 302                 } else if ( state==INTAG ) {
 303                         if ( ch == '>' ) {
 304                                 state=INTXT;
 305                                 pushout(out, Tag, 0);
 306                         }
 307                 } else if ( state == INDROPTAG ) {
 308                         if ( ch == '<' ) {
 309                                 state=INDROPINTAG;
 310                                 closelen=0;
 311                         }
 312                 } else if ( state == INDROPINTAG ) {
 313                         if ( ch == '/' )
 314                                 state=INDROPINCLSTAG;
 315                         else
 316                                 state=INDROPTAG;
 317                 } else if ( state == INDROPINCLSTAG ) {
 318                         if ( isalpha(ch) ) {
 319                                 if ( closelen < lenbuf && tolower(ch) == buf[closelen] ) {
 320                                         closelen++;
 321                                         if ( closelen==lenbuf )
 322                                                 state=INTAG;
 323                                 } else
 324                                         state=INDROPTAG;
 325                         } else
 326                                 state=INDROPTAG;
 327                 } else if ( state==FINDEND ) {
 328                         if ( ch == ' ' || ch == '\t' ) {
 329                                 lenbuf++;
 330                                 if ( lenbuf > spacelen ) {
 331                                         pushout( out, Paragraph, 0 );
 332                                         state=WAITAFTERRED;
 333                                 }
 334                         } else if ( ch=='\n' ) {
 335                                 pushout( out, NewLine, 0 );
 336                                 pushout( out, NewLine, 0 );
 337                                 lenbuf++;
 338                         } else {
 339                                 state=INTXT;
 340                                 pushout(out, Char, ' ');
 341                                 ungetc(ch,in);
 342                         }
 343                 } else if ( state==WAITAFTERRED ) {
 344                         if ( !isspace(ch) ) {
 345                                 ungetc(ch,in);
 346                                 state=INTXT;
 347                         }
 348                 } else if ( state==COMMENTBEGIN1 ) {
 349                         if ( ch == '-' ) {
 350                                 state = COMMENTBEGIN2;
 351                         } else {
 352                                 pushoutstr(out, "<!", 2);
 353                                 ungetc(ch,in);
 354                                 state=INTXT;
 355                         }
 356                 } else if ( state==COMMENTBEGIN2 ) {
 357                         if ( ch == '-' ) {
 358                                 state = COMMENTIN;
 359                         } else {
 360                                 pushoutstr(out, "<!-", 2);
 361                                 ungetc(ch,in);
 362                                 state=INTXT;
 363                         }
 364                 } else if ( state==COMMENTIN ) {
 365                         if ( ch == '-' )
 366                                 state = COMMENTEND1;
 367                 } else if ( state==COMMENTEND1 ) {
 368                         state = ( ch == '-' ) ? COMMENTEND2 : COMMENTIN;
 369                 } else if ( state==COMMENTEND2 ) {
 370                         if ( ch == '>' )
 371                                 state = INTXT;
 372                         else if ( ch != '-' )
 373                                 state = COMMENTIN;
 374                 } else if ( state==INQUOTE ) {
 375                         if ( isalpha( ch ) && lenbuf < BUFFERLENGTH-2 ) {
 376                                 buf[ lenbuf ] = ch;
 377                                 lenbuf++;
 378                         } else if ( ch == ';' && lenbuf>1 && pushoutquot( out, buf+1, lenbuf-1 ) ) {
 379                                 state = INTXT;
 380                         } else {
 381                                 pushoutstr(out, buf, lenbuf);
 382                                 state = INTXT;
 383                                 ungetc(ch,in);
 384                         }
 385                 } else {
 386                         printf("Unknown state: %d\n", state);
 387                         exit(1);
 388                 }
 389         }
 390
 391         if ( in!=stdin )
 392                 fclose(in);
 393         if ( out!=stdout )
 394                 fclose(out);
 395
 396         return 0;
 397 }
 398