00001 #include "mrilib.h"
00002 
00003 /*----------------------------------------------------------------------
00004    history:
00005 
00006    ...         [rwcox]
00007      - Initial Version(s)
00008 
00009    24 Mar 2005 [rickr]
00010      - added options: -help, -new_char, -new_string, -unescape
00011   ----------------------------------------------------------------------
00012 */
00013 
00014    
00015 void help_n_exit( void );
00016 int  suck_file( char *fname , char **fbuf ) ;
00017 
00018 int main( int argc , char * argv[] )
00019 {
00020    int nbuf , ntarg , ii,jj, nfind , ff ;
00021    char *fbuf , *targ ;
00022    char *jstr=strdup("AFNI-rules-") ; int njstr=strlen(jstr) ;
00023 
00024    int   ac, unesc = 0; /* for -unescape option    24 Mar 2005 [rickr] */
00025    int   use_newstr = 0;
00026    char  newchar = 'x';
00027 
00028    int   nfname=0 ;    /* for globbing */
00029    char **fname=NULL ;
00030 
00031    /* help? */
00032 
00033    if( argc < 3 ) help_n_exit();
00034 
00035    /* Check for arguments.  If we don't get an exact match, continue, */
00036    /* allowing the user to start a target string with '-'.            */
00037    for( ac = 1; ac < argc && argv[ac][0] == '-'; ac++ ){
00038       if( strcmp(argv[ac], "-help") == 0 )
00039          help_n_exit();
00040 
00041       if( strcmp(argv[ac], "-new_char") == 0 ){
00042          ac++;
00043          if( ac >= argc ){
00044             fprintf(stderr,"** -new_char option requires an argument\n");
00045             exit(1);
00046          }
00047          
00048          newchar = argv[ac][0];
00049       }
00050 
00051       if( strcmp(argv[ac], "-new_string") == 0 ){
00052          ac++;
00053          if( ac >= argc ){
00054             fprintf(stderr,"** -new_string option requires an argument\n");
00055             exit(1);
00056          }
00057          
00058          jstr = strdup(argv[ac]);
00059          njstr = strlen(jstr);
00060          use_newstr = 1;
00061       }
00062 
00063       if( strcmp(argv[ac], "-unescape") == 0 )
00064          unesc = 1;
00065    }
00066 
00067    if( ac > argc-2 ){
00068       fprintf(stderr,"** missing target string or input files\n");
00069       fprintf(stderr,"   (please see 'strblast -help')\n");
00070       exit(1);
00071    }
00072 
00073    machdep() ;
00074 
00075    /* load the target */
00076 
00077    targ = argv[ac] ; ntarg = strlen(targ) ; ac++ ;
00078    if( ntarg < 1 ){
00079       fprintf(stderr,"** Can't enter an empty target string!\n") ;
00080       exit(1) ;
00081    }
00082    if( unesc ){ /* let's leave argv alone, so dup the string */
00083       char * tnew = strdup(targ);
00084       if( !tnew ){ fprintf(stderr,"** cannot dup targetstring?!\n"); exit(1); }
00085       for(ii=0, jj = 0; ii<ntarg; ii++, jj++){
00086               if(tnew[ii] == '\\' && tnew[ii+1] == 't'){tnew[jj] = '\t'; ii++;}
00087          else if(tnew[ii] == '\\' && tnew[ii+1] == 'n'){tnew[jj] = '\n'; ii++;}
00088          else if(tnew[ii] == '\\' && tnew[ii+1] == 'r'){tnew[jj] = '\r'; ii++;}
00089          else if(ii > jj) tnew[jj] = tnew[ii];
00090       }
00091       tnew[jj] = '\0';  /* and terminate */
00092 
00093       /* now for the ol' switcheroo... */
00094       targ = tnew;  ntarg = jj;
00095    }
00096 
00097    /* if the replacement string is too long, truncate it */
00098    if( ntarg < njstr ){
00099       if( use_newstr ){ jstr[ntarg] = '\0' ; njstr = ntarg; }
00100       else            { jstr[0] = newchar  ; njstr = 1;     }
00101    }
00102 
00103    /* get input filenames */
00104 
00105    MCW_warn_expand(1) ;
00106    MCW_file_expand( argc-ac , argv+ac , &nfname , &fname ) ;
00107    MCW_warn_expand(0) ;
00108    if( nfname == 0 ){
00109       fprintf(stderr,"** No files found from command line!\n") ;
00110       exit(1) ;
00111    }
00112 
00113    /* loop over files */
00114 
00115    for( ff=0 ; ff < nfname ; ff++ ){
00116 
00117       /* read it all into memory */
00118 
00119       fbuf = NULL ;
00120       nbuf = suck_file( fname[ff] , &fbuf ) ;
00121       if( nbuf < ntarg || fbuf == NULL ){
00122          fprintf(stderr,"** Can't read input file %s\n",fname[ff]) ;
00123          if( fbuf != NULL ) free(fbuf) ;
00124          continue ;
00125       }
00126 
00127       /* scan for start character */
00128 
00129       for( nfind=ii=0 ; ii < nbuf-ntarg ; ii++ ){
00130 
00131          if( fbuf[ii] == targ[0] ){  /* if find it, check rest of string */
00132 
00133             for( jj=1; jj < ntarg && fbuf[ii+jj]==targ[jj] ; jj++ ) ; /* nada */
00134 
00135             if( jj == ntarg ){  /* found it */
00136                nfind++ ;
00137                for( jj=0 ; jj < njstr ; jj++ ) fbuf[ii+jj] = jstr[jj] ;
00138                for(      ; jj < ntarg ; jj++ ) fbuf[ii+jj] = newchar ;
00139             }
00140             ii += ntarg-1 ;
00141          }
00142       }
00143 
00144       if( nfind > 0 ){
00145          FILE *fp ;
00146          fprintf(stderr,"++ Found %d copies of target %s in file %s\n",
00147                  nfind,targ,fname[ff] ) ;
00148          fp = fopen( fname[ff] , "wb" ) ;
00149          if( fp == NULL ){
00150             fprintf(stderr,"** Can't open file %s for output!\n",fname[ff]) ;
00151             exit(1) ;
00152          }
00153          fwrite( fbuf , 1 , nbuf , fp ) ; fclose(fp) ;
00154       } else {
00155          fprintf(stderr,"++ Found no copies of target %s in file %s\n",
00156                  targ , fname[ff] ) ;
00157       }
00158 
00159       free(fbuf) ;
00160 
00161    } /* end of loop over files */
00162 
00163    exit(0) ;
00164 }
00165 
00166 /*------------------------------------------------------------------*/
00167 
00168 int suck_file( char *fname , char **fbuf )
00169 {
00170    int len , fd , ii ;
00171    char * buf ;
00172 
00173    if( fname == NULL || fname[0] == '\0' || fbuf == NULL ) return 0 ;
00174 
00175    len = THD_filesize( fname ) ;
00176    if( len <= 0 ) return 0 ;
00177 
00178    buf = (char *) malloc( sizeof(char) * (len+4) ) ;
00179    if( buf == NULL ) return 0 ;
00180 
00181    fd = open( fname , O_RDONLY ) ;
00182    if( fd < 0 ) return 0 ;
00183 
00184    ii = read( fd , buf , len ) ;
00185    close( fd ) ;
00186    if( ii <= 0 ){ free(buf) ; return 0; }
00187    *fbuf = buf ; return ii ;
00188 }
00189 
00190 void help_n_exit( void )
00191 {
00192    printf("Usage: strblast [options] TARGETSTRING filename ...\n"
00193           "Finds exact copies of the target string in each of\n"
00194           "the input files, and replaces all characters with\n"
00195           "some junk string.\n"
00196           "\n"
00197           "options:\n"
00198           "\n"
00199           "  -help              : show this help\n"
00200           "\n"
00201           "  -new_char CHAR     : replace TARGETSTRING with CHAR (repeated)\n"
00202           "\n"
00203           "      This option is used to specify what TARGETSTRING is\n"
00204           "      replaced with.  In this case, replace it with repeated\n"
00205           "      copies of the character CHAR.\n"
00206           "\n"
00207           "  -new_string STRING : replace TARGETSTRING with STRING\n"
00208           "\n"
00209           "      This option is used to specify what TARGETSTRING is\n"
00210           "      replaced with.  In this case, replace it with the string\n"
00211           "      STRING.  If STRING is not long enough, then CHAR from the\n"
00212           "      -new_char option will be used to complete the overwrite\n"
00213           "      (or the character 'x', by default).\n"
00214           "\n"
00215           "  -unescape          : parse TARGETSTRING for escaped characters\n"
00216           "                       (includes '\\t', '\\n', '\\r')\n"
00217           "\n"
00218           "      If this option is given, strblast will parse TARGETSTRING\n"
00219           "      replacing any escaped characters with their encoded ASCII\n"
00220           "      values.\n"
00221           "\n"
00222           "Examples:\n"
00223           "  strings I.001 | more # see if Subject Name is present\n"
00224           "  strblast 'Subject Name' I.*\n"
00225           "\n"
00226           "  strblast -unescape \"END OF LINE\\n\"       infile.txt\n"
00227           "  strblast -new_char \" \" \"BAD STRING\"      infile.txt\n"
00228           "  strblast -new_string \"GOOD\" \"BAD STRING\" infile.txt\n"
00229           "\n"
00230           "Notes and Warnings:\n"
00231           "  * strblast will modify the input files irreversibly!\n"
00232           "      You might want to test if they are still usable.\n"
00233           "  * strblast reads files into memory to operate on them.\n"
00234           "      If the file is too big to fit in memory, strblast\n"
00235           "      will fail.\n"
00236           "  * strblast  will do internal wildcard expansion, so\n"
00237           "      if there are too many input files for your shell to\n"
00238           "      handle, you can do something like\n"
00239           "         strblast 'Subject Name' 'I.*'\n"
00240           "      and strblast will expand the 'I.*' wildcard for you.\n"
00241          ) ;
00242    exit(0) ;
00243 }