*** 1.7/vtt.c	Tue Sep  7 17:34:05 2004
--- vtt.c	Tue Sep  7 22:25:59 2004
***************
*** 1,10 ****
  #include "stdio.h"
  #include "string.h"
  #include "ctype.h"
- #define MAXLEN 512
  
! /*
!    Voynich Transcription Tool 
! */
  
  /* Global variables */
--- 1,25 ----
+ /*
+    Voynich Transcription Tool 
+    by Rene Zandbergen
+    
+    Last edited on 2004-09-07 22:25:59 by stolfi
+    Based on vtt 1.7 1998/05/09 RZ
+    Changes:
+    * Allowed extra "."s in locus name.
+    * Allowed "#" inside {}-comments.
+    * Added "-v" (print program version) option.
+    * Added return values to "main" procedure.
+    * Safer allocation of "folname" and "locname".
+    * Added check for line buffer overflow.
+    * Changed "if(x = y)" to "if((x = y) != 0)" to avoid warnings.
+    Should check for overflow in "folname" and "locname", too.
+ */
+ 
  #include "stdio.h"
  #include "string.h"
  #include "ctype.h"
  
! /* Maximum line length (including final '\0'): */
! #define MAXLEN 512
  
  /* Global variables */
***************
*** 27,35 ****
  int infarg= -1; /* Argument of input file name */
  int oufarg= -1; /* Argument of output file name */
  char auth=' ';  /* Name of transcriber */
! char invar[27]; /* List of 'include page' options */
! char exvar[27]; /* List of 'exclude page' options */ 
  int npgopt = 0; /* Number of include/exclude options */
- FILE *fin, *fout; /* File handles */
  
  /* Some file stats */
--- 42,52 ----
  int infarg= -1; /* Argument of input file name */
  int oufarg= -1; /* Argument of output file name */
+ FILE *fin, *fout; /* File handles */
+ 
+ /* Selection options */
  char auth=' ';  /* Name of transcriber */
! char invar[27]; /* List of 'include page' options. */
! char exvar[27]; /* List of 'exclude page' options. */ 
  int npgopt = 0; /* Number of include/exclude options */
  
  /* Some file stats */
***************
*** 48,55 ****
  int  newfoli=0      /* 1 if there are < > without a locus (i.e. newpage)*/;
  char cwarn=' '      /* Warning character */;
! char *folname= "      "  /* Folio name, e.g. f85r3 */;
! char *locname= "     "   /* Locus name, e.g. C3, or blank */;
  char lineauth= ' '       /* Name of transcriber or blank */;
! char pgvar[27]      /* List of page variable settings */;
  
  /* These track the text */
--- 65,72 ----
  int  newfoli=0      /* 1 if there are < > without a locus (i.e. newpage)*/;
  char cwarn=' '      /* Warning character */;
! char folname[8]     /* Folio name, e.g. "f85r3" */;
! char locname[8]     /* Locus name, e.g. "C3", "L3.8", or blank */;
  char lineauth= ' '  /* Name of transcriber or blank */;
! char pgvar[27]      /* Page variable settings; pgvar[0] not used. */;
  
  /* These track the text */
***************
*** 107,111 ****
    char cb;
  
!   while ((cb = buf[ii]) && ii < width) {
      /* Dedicated check for comments.
         Sets voycode if outside comments */
--- 124,128 ----
    char cb;
  
!   while ((cb = buf[ii]) != '\0' && ii < width) {
      /* Dedicated check for comments.
         Sets voycode if outside comments */
***************
*** 164,170 ****
  /* Decide whether to select page based on comparing page
     options in pgvar[] with include and exclude options
!    given by user. Page variable <at> abused to select 
     loci, making this a line selection option instead.
!    The value of pgvar[0] is set in GetLine. If iopt=0 check
     only pgvar[0]. Otherwise check all others.
  */
--- 181,187 ----
  /* Decide whether to select page based on comparing page
     options in pgvar[] with include and exclude options
!    given by user. Page variable "@" abused to select 
     loci, making this a line selection option instead.
!    The value of pgvar[0] is set in PrepLine. If iopt=0 check
     only pgvar[0]. Otherwise check all others.
  */
***************
*** 520,524 ****
    /* RTF: hash comment always in comment font */
    if (hash) font = 0;
!   while (cb = buf[index]) {
      /* RTF: if not hash comment: keep track of font */
      if (hash == 0) {
--- 537,541 ----
    /* RTF: hash comment always in comment font */
    if (hash) font = 0;
!   while ((cb = buf[index]) != '\0') {
      /* RTF: if not hash comment: keep track of font */
      if (hash == 0) {
***************
*** 552,555 ****
--- 569,575 ----
    char sign, optn, val;
  
+   /* Why not indeed: */
+   char *what = "@(#)vtt\t1.7e1\t2004/09/07\tJS\n";
+ 
    for (i=0; i<=26; i++) {
      invar[i]=' '; exvar[i]=' ';
***************
*** 657,660 ****
--- 677,683 ----
                 ast=3; brack=2; /* Remove line with ? * or [..] */}
               break;
+            case 'v':
+              fprintf(stderr, "%s\n", what);
+              break;
             case 'w':
               if (val == '0') wrap=0;       /* Maintain line wrapping */
***************
*** 922,926 ****
          return -1;       /* Correct EOF */
        } else {
!         buf[index] = 0;  /* Add a null character for safety */
          fprintf (stderr, "EOF at record pos. %3d\n", index);        
          fprintf (stderr, "Line read so far: %s\n", buf); 
--- 945,949 ----
          return -1;       /* Correct EOF */
        } else {
!         buf[index] = '\0';  /* Add a null character for safety */
          fprintf (stderr, "EOF at record pos. %3d\n", index);        
          fprintf (stderr, "Line read so far: %s\n", buf); 
***************
*** 938,951 ****
      /* Check for comment line */
  
!     if (cget == '#') {
!       /* A hash symbol. Not allowed when unwrapping */
!       if (concat > 0) {
!         buf[index] = 0;  /* Add a null character for safety */
!         fprintf (stderr, "Hash comment after continuation\n");        
!         fprintf (stderr, "Line read so far: %s\n", buf); 
!         return 1;
!       }
!       /* Check for valid hash comment line */
!       if (index == 0) hash = 1;
      }
      
--- 961,968 ----
      /* Check for comment line */
  
!     if (cget == '#' && index == 0) {
!       /* Hash symbols in invalid positions should be caught by PrepLine. */
!       /* That includes "#" in column 1 of continuation lines. */
!       hash = 1;
      }
      
***************
*** 980,983 ****
--- 997,1006 ----
        /* Add the character to the buffer */
        buf[index++] = cget;
+       if (index >= MAXLEN) {
+         fprintf (stderr, "Line too long - %d chars\n", index);
+         buf[index-1] = '\0';
+         fprintf (stderr, "Line read so far: %s\n", buf); 
+         return -2;
+       }
        if (blank == 0) hastext = 1;
       
***************
*** 985,989 ****
        if (cget == '\n') {
          cr = 1;
!         buf[index] = 0; /* Add null at end of line */
        }
      } /* End if ignore == 0 */
--- 1008,1012 ----
        if (cget == '\n') {
          cr = 1;
!         buf[index] = '\0'; /* Add null at end of line */
        }
      } /* End if ignore == 0 */
***************
*** 1014,1018 ****
    trackinit();
  
!   while (cget = buf1[ind1]) {
      
      /* For hash comment just copy: */
--- 1037,1041 ----
    trackinit();
  
!   while ((cget = buf1[ind1]) != '\0') {
      
      /* For hash comment just copy: */
***************
*** 1023,1028 ****
        /* No hash comment: do all the processing */
        
!       /* 1. Check for hash later in line and change to = */
!       if (cget == '#') {
          cwarn = cget;
          cget = '=';
--- 1046,1051 ----
        /* No hash comment: do all the processing */
        
!       /* 1. Check for unprotected hash later in line and change to = */
!       if (cget == '#' && incomm == 0) {
          cwarn = cget;
          cget = '=';
***************
*** 1134,1138 ****
              case '>': /* Terminate (optionally) locus ID */
                if (inlocu > 0) {
!                 locname[inlocu] = 0; inlocu = -2;
                }
                if (newfoli < 0) {
--- 1157,1161 ----
              case '>': /* Terminate (optionally) locus ID */
                if (inlocu > 0) {
!                 locname[inlocu] = '\0'; inlocu = -2;
                }
                if (newfoli < 0) {
***************
*** 1141,1146 ****
                }
                break;
!             case '.': /* Start of locus */
                inlocu = -1; newfoli = 0;
                break;
              case ';': 
--- 1164,1175 ----
                }
                break;
!             case '.':
!               if (inlocu <= 0) {
!                  /* Start of locus name */
                  inlocu = -1; newfoli = 0;
+               } else {
+                 /* Dot is part of locus name: */
+                 locname[inlocu-1] = cget;
+               }
                break;
              case ';': 
***************
*** 1150,1154 ****
                }
                /* Terminate locus ID */
!               locname[inlocu] = 0; inlocu = -2;
                /* Start for author ID */
                inauth = -1;
--- 1179,1183 ----
                }
                /* Terminate locus ID */
!               locname[inlocu] = '\0'; inlocu = -2;
                /* Start for author ID */
                inauth = -1;
***************
*** 1246,1250 ****
      
      /* Set pgvar[0] if a new locus was read */
!     if (hasfoli && (newfoli == 0)) {
        pgvar[0] = locname[0];
        if (pgvar[0] >= '0' && pgvar[0] <= '9') pgvar[0] = 'P';
--- 1275,1279 ----
      
      /* Set pgvar[0] if a new locus was read */
!     if (hasfoli != 0 && newfoli == 0) {
        pgvar[0] = locname[0];
        if (pgvar[0] >= '0' && pgvar[0] <= '9') pgvar[0] = 'P';
***************
*** 1282,1286 ****
    trackinit(); index = 0;
  
!   while (cb = buf[index]) {
  
      /* Track the text */
--- 1311,1315 ----
    trackinit(); index = 0;
  
!   while ((cb = buf[index]) != '\0') {
  
      /* Track the text */
***************
*** 1357,1361 ****
    trackinit(); index = 0;
    /* fprintf(stderr,"%s\n",buf); */
!   while (cb = buf[index]) {
  
      /* Track the text */
--- 1386,1390 ----
    trackinit(); index = 0;
    /* fprintf(stderr,"%s\n",buf); */
!   while ((cb = buf[index]) != '\0') {
  
      /* Track the text */
***************
*** 1493,1497 ****
  char *buf;
  {
!   int index=0, indout=0, eol=0, output, ii;
    char cb;
    char wrapbuf[MAXLEN];
--- 1522,1526 ----
  char *buf;
  {
!   int index=0, indout=0, output, ii;
    char cb;
    char wrapbuf[MAXLEN];
***************
*** 1522,1526 ****
    trackinit();
  
!   while (cb = buf[index]) {
      output = 1;
      TrackLight(cb, index);
--- 1551,1555 ----
    trackinit();
  
!   while ((cb = buf[index]) != '\0') {
      output = 1;
      TrackLight(cb, index);
***************
*** 1588,1606 ****
  /*-----------------------------------------------------------*/
  
! void main(argc, argv)
  int argc;
  char *argv[];
  {
    char orig[MAXLEN], buf1[MAXLEN], buf2[MAXLEN];
!   int ilin = 0, iprep = 0, iproc = 0, iout = 0;
    int selpage, selloc;
  
!   /* Why not indeed: */
!   char *what = "@(#)vtt\t1.7\t1998/05/09\tRZ\n";
  
    /* Parse command line options */
    if (ParseOpts(argc, argv)) {
      fprintf (stderr, "%s\n", "Error parsing command line");
!     return;
    }
   
--- 1617,1635 ----
  /*-----------------------------------------------------------*/
  
! int main(argc, argv)
  int argc;
  char *argv[];
  {
    char orig[MAXLEN], buf1[MAXLEN], buf2[MAXLEN];
!   int ilin = 0, iprep = 0, iproc = 0;
    int selpage, selloc;
    
!   /* Just in case: */
!   folname[0] = '\0'; locname[0] = '\0';
  
    /* Parse command line options */
    if (ParseOpts(argc, argv)) {
      fprintf (stderr, "%s\n", "Error parsing command line");
!     return 1;
    }
   
***************
*** 1608,1612 ****
    if (DumpOpts(argc, argv)) {
      fprintf (stderr, "%s\n", "Error opening file(s)");
!     return;
    }  
    fprintf (stderr, "\n%s\n", "Starting...");
--- 1637,1641 ----
    if (DumpOpts(argc, argv)) {
      fprintf (stderr, "%s\n", "Error opening file(s)");
!     return 1;
    }  
    fprintf (stderr, "\n%s\n", "Starting...");
***************
*** 1656,1664 ****
          fprintf (stderr, "%7d lines after wrapping\n", nlwrap);
        }
!       return;
      }
      else if (ilin > 0) {
        fprintf (stderr, "%s\n", "Error reading line from stdin");
!       return;
      }
      nlread += 1; nlpart += 1;
--- 1685,1693 ----
          fprintf (stderr, "%7d lines after wrapping\n", nlwrap);
        }
!       return 0;
      }
      else if (ilin > 0) {
        fprintf (stderr, "%s\n", "Error reading line from stdin");
!       return 1;
      }
      nlread += 1; nlpart += 1;
***************
*** 1673,1677 ****
        fprintf (stderr, "%s\n", "Error preprocessing line");
        fprintf (stderr, "Line: %s\n", orig);
!       return;
      }
      if (cwarn != ' ') {
--- 1702,1706 ----
        fprintf (stderr, "%s\n", "Error preprocessing line");
        fprintf (stderr, "Line: %s\n", orig);
!       return 1;
      }
      if (cwarn != ' ') {
***************
*** 1692,1696 ****
  
      /* Check locus type only if page in principle OK */
!     if (selloc = selpage) selloc = usepage(0);
    
      /* Only continue if page/locus as requested */
--- 1721,1725 ----
  
      /* Check locus type only if page in principle OK */
!     if ((selloc = selpage) != 0) selloc = usepage(0);
    
      /* Only continue if page/locus as requested */
***************
*** 1703,1707 ****
          fprintf (stderr, "%s", "Error processing spaces\n");
          fprintf (stderr, "Line: %s\n", orig);
!         return;
        }
  
--- 1732,1736 ----
          fprintf (stderr, "%s", "Error processing spaces\n");
          fprintf (stderr, "Line: %s\n", orig);
!         return 1;
        }
  
***************
*** 1712,1716 ****
          fprintf (stderr, "%s\n", "Error processing uncertain readings");
          fprintf (stderr, "Line: %s\n", orig);
!         return;
        } else {
  
--- 1741,1745 ----
          fprintf (stderr, "%s\n", "Error processing uncertain readings");
          fprintf (stderr, "Line: %s\n", orig);
!         return 1;
        } else {
  
***************
*** 1725,1729 ****
            if (PutLine(buf2) != 0) {
              fprintf (stderr, "Line: %s\n", orig);
!             return;
            }
          } else {
--- 1754,1758 ----
            if (PutLine(buf2) != 0) {
              fprintf (stderr, "Line: %s\n", orig);
!             return 0;
            }
          } else {
***************
*** 1733,1735 ****
--- 1762,1765 ----
      }  
    }
+   return 0;
  }