clrlibru.c

   1 #include <stdio.h>
   2 #include <stdlib.h>
   3 #include <string.h>
   4 #include <ctype.h>
   5
   6 static void
   7 usage() {
   8         printf("Clear text from lib.ru for Pocket PC, Version 0.2.\n");
   9         printf("Author: Teodor Sigaev <teodor@sigaev.ru>\n");
  10         printf("Usage:\n   clrlibru [-i INPUTFILE] [-o OUTPUT] [-l NUMSPACE]\n");
  11         exit(0);
  12 }
  13
  14 char* RemoveTag[]={
  15         "select",
  16         "head",
  17         "div",
  18         "a",
  19         "form",
  20         NULL
  21 };
  22
  23 static int
  24 is_rtag(char *tag, int len) {
  25         char **ptr=RemoveTag;
  26
  27         tag[len]='\0';
  28         while( *ptr ) {
  29                 if ( strcmp(tag, *ptr) == 0 )
  30                         return 1;
  31                 ptr++;
  32         }
  33         return 0;
  34 }
  35
  36
  37 static char *optarg = NULL;
  38 static int current=1;
  39
  40 int
  41 mgetopt(int argn, char* argv[], char *option) {
  42         char key;
  43
  44         if ( current >= argn ) return -1;
  45
  46         key = *(argv[current]+1);
  47         if ( *(argv[current]) == '-' || *(argv[current]) == '/' ) {
  48                 char *ptr = strchr( option, key );
  49                 if ( ptr == NULL ) {
  50                         printf( "Unknown option: %s\n", argv[current]);
  51                         exit(1);
  52                 }
  53                 current++;
  54                 if ( *(ptr+1) == ':' ) {
  55                         if (current < argn) {
  56                                 optarg=argv[current];
  57                                 current++;
  58                         } else {
  59                                 printf("No value for -%c\n" ,key);
  60                                 exit(1);
  61                         }
  62                 } else
  63                         optarg=NULL;
  64         } else {
  65                 printf("Unknown option: %s\n", argv[current]);
  66                 exit(1);
  67         }
  68         return (int)key;
  69 }
  70
  71 #define INTXT   0
  72 #define INTAG   1
  73 #define FINDEND 2
  74 #define INDROPTAG       3
  75 #define INDROPINTAG     4
  76 #define INHEADTAG       5
  77 #define INDROPINCLSTAG  6
  78
  79
  80 int
  81 main(int argn, char *argv[]) {
  82         int ch;
  83         FILE    *in=stdin, *out=stdout;
  84         int state=INTXT;
  85         char    buf[8192];
  86         int lenbuf=0,closelen=0;
  87         int spacelen=4;
  88
  89         while((ch = mgetopt(argn, argv, "l:i:o:h?"))!=-1) {
  90                 switch (ch) {
  91                         case 'i':
  92                                 if ( (in=fopen(optarg, "r"))==NULL) {
  93                                         printf("Can't open file %s\n", optarg);
  94                                         exit(1);
  95                                 }
  96                                 break;
  97                         case 'o':
  98                                 if ( (out=fopen(optarg, "w"))==NULL) {
  99                                         printf("Can't open file %s\n",optarg);
 100                                         exit(1);
 101                                 }
 102                                 break;
 103                         case 'l':
 104                                 spacelen = atoi(optarg);
 105                                 if ( spacelen < 0 ) {
 106                                         printf("-l should be >= 0\n");
 107                                         exit(1);
 108                                 }
 109                                 break;
 110                         case 'h':
 111                         case '?':
 112                         default:
 113                                 usage();
 114                 }
 115         }
 116
 117         while( (ch=getc(in)) != EOF ) {
 118                 if ( state==INTXT ) {
 119                         if ( ch == '<' ) {
 120                                 state=INHEADTAG;
 121                                 lenbuf=0;
 122                         } else if ( ch == '\n' ) {
 123                                 state=FINDEND;
 124                                 lenbuf=1;
 125                                 *buf = ch;
 126                         } else if ( ch != '\r' )
 127                                 fputc(ch,out);
 128                 } else if ( state==INHEADTAG ) {
 129                         if ( isalpha(ch) ) {
 130                                 buf[ lenbuf ] = tolower(ch);
 131                                 lenbuf++;
 132                         } else if ( ch == '>' ) {
 133                                 if ( is_rtag(buf,lenbuf) ) {
 134                                         state = INDROPTAG;
 135                                         closelen=0;
 136                                 } else {
 137                                         state=INTXT;
 138                                         fputc(' ',out);
 139                                 }
 140                         } else if ( lenbuf == 0 && ch != '/' ) {
 141                                 fputc('<',out); fputc(ch,out);
 142                                 state=INTXT;
 143                         } else {
 144                                 if ( is_rtag(buf,lenbuf) ) {
 145                                         state = INDROPTAG;
 146                                         closelen=0;
 147                                 } else {
 148                                         state=INTAG;
 149                                         fputc(' ',out);
 150                                 }
 151                         }
 152                 } else if ( state==INTAG ) {
 153                         if ( ch == '>' ) {
 154                                 state=INTXT;
 155                                 fputc(' ',out);
 156                         }
 157                 } else if ( state == INDROPTAG ) {
 158                         if ( ch == '<' ) {
 159                                 state=INDROPINTAG;
 160                                 closelen=0;
 161                         }
 162                 } else if ( state == INDROPINTAG ) {
 163                         if ( ch == '/' )
 164                                 state=INDROPINCLSTAG;
 165                         else
 166                                 state=INDROPTAG;
 167                 } else if ( state == INDROPINCLSTAG ) {
 168                         if ( isalpha(ch) ) {
 169                                 if ( closelen < lenbuf && tolower(ch) == buf[closelen] ) {
 170                                         closelen++;
 171                                         if ( closelen==lenbuf )
 172                                                 state=INTAG;
 173                                 } else
 174                                         state=INDROPTAG;
 175                         } else
 176                                 state=INDROPTAG;
 177                 } else if ( state==FINDEND ) {
 178                         if ( ch == ' ' ) {
 179                                 buf[ lenbuf ] = ch;
 180                                 lenbuf++;
 181                                 if ( lenbuf > spacelen ) {
 182                                         fwrite(buf, sizeof(char), lenbuf, out);
 183                                         state=INTXT;
 184                                 }
 185                         } else if ( ch=='\n' ) {
 186                                 buf[ lenbuf ] = ch;
 187                                 lenbuf++;
 188                                 fwrite(buf, sizeof(char), lenbuf, out);
 189                                 state=INTXT;
 190                         } else if ( ch !='\r' ) {
 191                                 state=INTXT;
 192                                 fputc(' ',out);
 193                                 ungetc(ch,in);
 194                         }
 195                 } else {
 196                         printf("Unknown state: %d\n", state);
 197                         exit(1);
 198                 }
 199         }
 200
 201         if ( in!=stdin )
 202                 fclose(in);
 203         if ( out!=stdout )
 204                 fclose(out);
 205
 206         return 0;
 207 }
 208