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
  36 usage() {
  37         printf("Clear text from lib.ru for Pocket PC, Version 0.2.\n");
  38         printf("Copyright (c) 2004 Teodor Sigaev <teodor@sigaev.ru>\n");
  39         printf("       All rights reserved.\n");
  40         printf("Usage:\n   clrlibru [-i INPUTFILE] [-o OUTPUT] [-l NUMSPACE]\n");
  41         exit(0);
  42 }
  43
  44 char* RemoveTag[]={
  45         "select",
  46         "head",
  47         "div",
  48         "a",
  49         "form",
  50         NULL
  51 };
  52
  53 static int
  54 is_rtag(char *tag, int len) {
  55         char **ptr=RemoveTag;
  56
  57         tag[len]='\0';
  58         while( *ptr ) {
  59                 if ( strcmp(tag, *ptr) == 0 )
  60                         return 1;
  61                 ptr++;
  62         }
  63         return 0;
  64 }
  65
  66
  67 static char *optarg = NULL;
  68 static int current=1;
  69
  70 int
  71 mgetopt(int argn, char* argv[], char *option) {
  72         char key;
  73
  74         if ( current >= argn ) return -1;
  75
  76         key = *(argv[current]+1);
  77         if ( *(argv[current]) == '-' || *(argv[current]) == '/' ) {
  78                 char *ptr = strchr( option, key );
  79                 if ( ptr == NULL ) {
  80                         printf( "Unknown option: %s\n", argv[current]);
  81                         exit(1);
  82                 }
  83                 current++;
  84                 if ( *(ptr+1) == ':' ) {
  85                         if (current < argn) {
  86                                 optarg=argv[current];
  87                                 current++;
  88                         } else {
  89                                 printf("No value for -%c\n" ,key);
  90                                 exit(1);
  91                         }
  92                 } else
  93                         optarg=NULL;
  94         } else {
  95                 printf("Unknown option: %s\n", argv[current]);
  96                 exit(1);
  97         }
  98         return (int)key;
  99 }
 100
 101 #define INTXT   0
 102 #define INTAG   1
 103 #define FINDEND 2
 104 #define INDROPTAG       3
 105 #define INDROPINTAG     4
 106 #define INHEADTAG       5
 107 #define INDROPINCLSTAG  6
 108
 109
 110 int
 111 main(int argn, char *argv[]) {
 112         int ch;
 113         FILE    *in=stdin, *out=stdout;
 114         int state=INTXT;
 115         char    buf[8192];
 116         int lenbuf=0,closelen=0;
 117         int spacelen=4;
 118
 119         while((ch = mgetopt(argn, argv, "l:i:o:h?"))!=-1) {
 120                 switch (ch) {
 121                         case 'i':
 122                                 if ( (in=fopen(optarg, "r"))==NULL) {
 123                                         printf("Can't open file %s\n", optarg);
 124                                         exit(1);
 125                                 }
 126                                 break;
 127                         case 'o':
 128                                 if ( (out=fopen(optarg, "w"))==NULL) {
 129                                         printf("Can't open file %s\n",optarg);
 130                                         exit(1);
 131                                 }
 132                                 break;
 133                         case 'l':
 134                                 spacelen = atoi(optarg);
 135                                 if ( spacelen < 0 ) {
 136                                         printf("-l should be >= 0\n");
 137                                         exit(1);
 138                                 }
 139                                 break;
 140                         case 'h':
 141                         case '?':
 142                         default:
 143                                 usage();
 144                 }
 145         }
 146
 147         while( (ch=getc(in)) != EOF ) {
 148                 if ( state==INTXT ) {
 149                         if ( ch == '<' ) {
 150                                 state=INHEADTAG;
 151                                 lenbuf=0;
 152                         } else if ( ch == '\n' ) {
 153                                 state=FINDEND;
 154                                 lenbuf=1;
 155                                 *buf = ch;
 156                         } else if ( ch != '\r' )
 157                                 fputc(ch,out);
 158                 } else if ( state==INHEADTAG ) {
 159                         if ( isalpha(ch) ) {
 160                                 buf[ lenbuf ] = tolower(ch);
 161                                 lenbuf++;
 162                         } else if ( ch == '>' ) {
 163                                 if ( is_rtag(buf,lenbuf) ) {
 164                                         state = INDROPTAG;
 165                                         closelen=0;
 166                                 } else {
 167                                         state=INTXT;
 168                                         fputc(' ',out);
 169                                 }
 170                         } else if ( lenbuf == 0 && ch != '/' ) {
 171                                 fputc('<',out); fputc(ch,out);
 172                                 state=INTXT;
 173                         } else {
 174                                 if ( is_rtag(buf,lenbuf) ) {
 175                                         state = INDROPTAG;
 176                                         closelen=0;
 177                                 } else {
 178                                         state=INTAG;
 179                                         fputc(' ',out);
 180                                 }
 181                         }
 182                 } else if ( state==INTAG ) {
 183                         if ( ch == '>' ) {
 184                                 state=INTXT;
 185                                 fputc(' ',out);
 186                         }
 187                 } else if ( state == INDROPTAG ) {
 188                         if ( ch == '<' ) {
 189                                 state=INDROPINTAG;
 190                                 closelen=0;
 191                         }
 192                 } else if ( state == INDROPINTAG ) {
 193                         if ( ch == '/' )
 194                                 state=INDROPINCLSTAG;
 195                         else
 196                                 state=INDROPTAG;
 197                 } else if ( state == INDROPINCLSTAG ) {
 198                         if ( isalpha(ch) ) {
 199                                 if ( closelen < lenbuf && tolower(ch) == buf[closelen] ) {
 200                                         closelen++;
 201                                         if ( closelen==lenbuf )
 202                                                 state=INTAG;
 203                                 } else
 204                                         state=INDROPTAG;
 205                         } else
 206                                 state=INDROPTAG;
 207                 } else if ( state==FINDEND ) {
 208                         if ( ch == ' ' ) {
 209                                 buf[ lenbuf ] = ch;
 210                                 lenbuf++;
 211                                 if ( lenbuf > spacelen ) {
 212                                         fwrite(buf, sizeof(char), lenbuf, out);
 213                                         state=INTXT;
 214                                 }
 215                         } else if ( ch=='\n' ) {
 216                                 buf[ lenbuf ] = ch;
 217                                 lenbuf++;
 218                                 fwrite(buf, sizeof(char), lenbuf, out);
 219                                 state=INTXT;
 220                         } else if ( ch !='\r' ) {
 221                                 state=INTXT;
 222                                 fputc(' ',out);
 223                                 ungetc(ch,in);
 224                         }
 225                 } else {
 226                         printf("Unknown state: %d\n", state);
 227                         exit(1);
 228                 }
 229         }
 230
 231         if ( in!=stdin )
 232                 fclose(in);
 233         if ( out!=stdout )
 234                 fclose(out);
 235
 236         return 0;
 237 }
 238