#!/usr/bin/perl
#
# mkPromptFiles.pl    Version 1.1
#                     (C) Copyright 1999
#                     Chr. Draxler, University of Munich
#                     draxler@phonetik.uni-muenchen.de
#
# mkPromptFiles.pl makes prompt files for the SpeechDat-Car
# recordings. It reads the prompt contents from text files
# according to the CCD (corpus code, the two-letter id code
# associated with each recorded item), creates a prompt 
# definition file line which consists of the fields
#
#    CCD;PromptDefinitionFile;LineIndex
#
# and collects these lines in an array. Once the array is
# complete, the elements are extracted in random order
# and written to a prompt file in DOS format with no 
# empty lines.
#
# NOTES:
# 1) LineIndex > 0 in the prompt file, but for array access
#    in the script, it must be decremented by 1 because
#    array indexes start with 0
# 2) the script always creates at least two prompt files, one
#    with the end digit "0", one with "1".
#
# HISTORY:
#  1.1: - added link between L2 and O3 CCDs; this link requires
#       a file with spelled items (i.e. with a blank between
#       letters; these items must be in the same order as the 
#       non-spelled items in the original file.
#       - usage message is printed if number of arguments is less
#       than six (if command line arguments are used)
#  1.0: first release
#
# --- arguments needed for script to run --------------------
#
# on the Mac, they are asked for via dialogs. On other 
# systems, they are passed as command line arguments.
#
# Use the version that's compatible with your machine and
# comment out the other one.
#

$srcDir = MacPerl::Ask('Source directory',':ITEM:');
$targetDir = MacPerl::Ask('Target directory',':PROMPT:');
$txtExt = MacPerl::Ask('Text file extension','.TXT');
$pmtExt = '.TXT';
$lowSessNo = MacPerl::Ask('First session number','50');
$highSessNo = MacPerl::Ask('Last session number','60');

#if(int(@ARGV) < 6) {
#   die "Usage\n\nmkPromptFiles.pl SRCDIR TARGETDIR TXTEXT PMTEXT LOWSESSNO HIGHSESSNO\n";
#} else {
#   $srcDir = $ARGV[0];
#   $targetDir = $ARGV[1];
#   $txtExt = $ARGV[2];
#   $pmtExt = $ARGV[3];
#   $lowSessNo = $ARGV[4];
#   $highSessNo = $ARGV[5];
#}

# --- define an associative array with CCD codes and the ----
# source files from which to take the corresponding prompt
# text or data

%ccdSource = (
 'A1','VoiceAct',
 'A2','VoiceAct',
 'B1','TenDigits',
 'C1','SheetNumber',
 'C2','TelNumberSpon',
 'C3','CreditCard',
 'C4','PinCode',
 'C5','TelNumber',
 'C6','TelNumber',
 'C7','TelNumber',
 'D1','DateSpon',
 'D2','DateAbs',
 'D3','DateRel',
 'E1','KeywordSentences',
 'E2','KeywordSentences',
 'I1','IsolatedDigit',
 'I2','IsolatedDigit',
 'I3','IsolatedDigit',
 'I4','IsolatedDigit',
 'L1','NameSponSpell',
 'L2','CitySpell',
 'L3','WordSpell',
 'L4','WordSpell',
 'L5','WordSpell',
 'L6','WordSpell',
 'L7','WordArtSpell',
 'M1','Amounts',
 'N1','NaturalNumber',
 'O1','NameSpon',
 'O2','BirthCity',
 'O3','Cities',
 'O4','Cities',
 'O5','Companies',
 'O6','Companies',
 'O7','Names',
 'S1','Sentences',
 'S2','Sentences',
 'S3','Sentences',
 'S4','Sentences',
 'S5','Sentences',
 'S6','Sentences',
 'S7','Sentences',
 'S8','Sentences',
 'S9','Sentences',
 'T1','TimeSpon',
 'T2','TimeAna',
 'W1','Words',
 'W2','Words',
 'W3','Words',
 'W4','Words',
 'P1','KeywordsDep',
 'P2','KeywordsDep',
 'Z0','SponSentences',
 'Z1','SponSentences',
 'Z2','SponSentences',
 'Z3','SponSentences',
 'Z4','SponSentences',
 'Z5','SponSentences',
 'Z6','SponSentences',
 'Z7','SponSentences',
 'Z8','SponSentences',
 'Z9','SponSentences',
 '00','AppWords',
 '01','AppWords',
 '02','AppWords',
 '03','AppWords',
 '04','AppWords',
 '05','AppWords',
 '06','AppWords',
 '07','AppWords',
 '08','AppWords',
 '09','AppWords',
 '10','AppWords',
 '11','AppWords',
 '12','AppWords',
 '13','AppWords',
 '14','AppWords',
 '15','AppWords',
 '16','AppWords',
 '17','AppWords',
 '18','AppWords',
 '19','AppWords',
 '20','AppWords',
 '21','AppWords',
 '22','AppWords',
 '23','AppWords',
 '24','AppWords',
 '25','AppWords',
 '26','AppWords',
 '27','AppWords',
 '28','AppWords',
 '29','AppWords',
 '30','AppWords',
 '31','AppWords',
 '32','AppWords',
 '33','AppWords',
 '34','AppWords',
 '35','AppWords',
 '36','AppWords',
 '37','AppWords',
 '38','AppWords',
 '39','AppWords',
 '40','AppWords',
 '41','AppWords',
 '42','AppWords',
 '43','AppWords',
 '44','AppWords',
 '45','AppWords',
 '46','AppWords',
 '47','AppWords',
 '48','AppWords',
 '49','AppWords',
 '50','AppWords',
 '51','AppWords',
 '52','AppWords',
 '53','AppWords',
 '54','AppWords',
 '55','AppWords',
 '56','AppWords',
 '57','AppWords',
 '58','AppWords',
 '59','AppWords',
 '60','AppWords',
 '61','AppWords',
 '62','AppWords',
 '63','AppWords',
 '64','AppWords',
 '65','AppWords',
 '66','AppWords',
);

# --- now create prompt files ---------------------------------------

$sessNo = $lowSessNo;
$i=0;
while($sessNo <= $highSessNo) {

   if($sessNo < 10) {
      $sessCode = '00' . $sessNo . $i;
   } elsif($sessNo < 100) {
      $sessCode = '0' . $sessNo . $i;
   } else {
      $sessCode = $sessNo . $i;
   }

   if($i == 1) {
      $sessNo++;
      $i = 0;
   } else {
      print "session: $sessNo\n";
      $i = 1;
   }

   $promptFile = $targetDir . 'prmt' . $sessCode . $pmtExt;
   open(PROMPT,">$promptFile") ||
      die "Could not open $promptFile";

   # --- initialize counters ----------------------------------------
   $oldCcd = '';
   $inFile = '';
   $promptCnt = 0;
   @prompts = ();

   foreach $ccd (sort(keys %ccdSource)) {

      if($ccdSource{$ccd} ne $oldCcd) {
         # --- if $inFile was open, close now -----------------------
         if($inFile ne '') {
            close(IN);
         }

         # --- open $inFile for prompt material ---------------------
         $inFile = $srcDir . $ccdSource{$ccd} . $txtExt;
         open(IN,$inFile) ||
            die "Could not open $inFile";

         # --- initialize items array -------------------------------
         @items = ();
         $promptFileCnt = 0;

         while(<IN>) {
            chop;
            # --- remove any new line, carriage return, etc. --------
            s/[\n\r\f]//g;
            if($_ ne '') {
               $items[$promptFileCnt] = $_;
               $promptFileCnt++;
            }
         }

         # --- how many items from this file are used? --------------
         $itemFileCnt = 0;
         foreach $itemKey (keys(%ccdSource)) {
            if($ccdSource{$itemKey} eq $ccdSource{$ccd}) {
               $itemFileCnt++;
            }
         }

         # --- initialize $lineIndex --------------------------------
         $lineIndex = 0;

         # --- update $oldCcd to detect next change of $ccdSource ---
         $oldCcd = $ccdSource{$ccd};
      }

      if($promptFileCnt > $itemFileCnt) {
         $lineIndex = int(rand(@items)+1);
      } else {
         $lineIndex++;
      }

      # --- handle special cases, e.g. CCDs linked to each other ----
      # L2 and O2 are linked; L2 must come before O3 in the array
      #    @ccdSource; the file containing data for L2 and O3
      #    must have the same sequence of entries
      if ($ccd eq 'L2') {
         $lineIndexL2 = $lineIndex;
      } elsif ($ccd eq 'O3') {
         $lineIndex = $lineIndexL2;
      } 

      # --- get prompt text, format if, and print to log output -----
      $lineText = $items[$lineIndex-1];
      $lineText =~ s/[\r\n\f]//g;
      printf "%s;%s;%d;%s\n",$ccd,$ccdSource{$ccd},$lineIndex,$lineText;

      # --- store prompt file line in array -------------------------
      $prompts[$promptCnt]="$ccd;$ccdSource{$ccd};$lineIndex";
      $promptCnt++;
   }

   # --- permutate lines of prompt array and print to prompt file ---
   # remove any new lines, carriage returns, etc from $promptLine 
   # and print it

   $promptLine = splice(@prompts,int(rand(@prompts)),1);
   $promptLine =~ s/[\r\n\f]//g;
   printf PROMPT "%s",$promptLine;

   # --- then print rest of lines preceded by CR/LF -----------------
   while(@prompts) {
      $promptLine = splice(@prompts,int(rand(@prompts)),1);

      # --- remove any new lines, carriage returns, etc. ------------
      $promptLine =~ s/[\r\n\f]//g;

      printf PROMPT "\r\n%s",$promptLine;
      #print $promptLine;
   }

   close(PROMPT);
   close(IN);
   print "\n\n";
}