#!/usr/bin/env perl

# If you've mapped a network drive, use this line:
chdir "/Volumes/Corpora/en/brown/";
# If you've logged in to jones, comment out the previous line and
# uncomment this line:
#chdir "/Volumes/Data/Corpora/en/brown/";

@all_files = <*>;

# Q: what do we need to do to keep track of the number of occurrences
# of a particular search?

$counter = 0;

# loop over all files in the directory
foreach (@all_files) {
  # get appropriate files from directory
  if (/\d/) {
    # open each file, one at a time
    $file_name = $_;
    open FILE, '<', $file_name;

    # read each line in the file
    while (<FILE>) {
      # remove trailing newline
      chomp;

      # case #1: help/verb + following verb
      while (m{\b(
		 help\w*?/v\w*?   # help
		 \s+              # (space)
		 \w+/v\w*?        # following verb
	       )\b}gix) {         # 'x' option allows us to be RE on
                                  # separate lines
	print "$file_name: $1\n";
	$counter++;
      }

      # case #2: help/verb + to + following verb
#       while (m{\b(
# 		 help\w*?/v\w*?   # help
# 		 \s+              # (space)
# 		 to/to            # to
# 		 \s+              # (space)
# 		 \w+/v\w*?        # following verb
# 	       )\b}gix) {
# 	print "$file_name: $1\n";
#       }

      # case #3: help/verb + NP + following verb
#       while (m{\b(
# 		 help\w*?/v\w*?   # help
# 		 \s+              # (space)
# 		 \w+/[np]\w*?     # tag starts with 'n' or 'p'
# 		 \s+              # (space)
# 		 \w+/v\w*?        # following verb
# 	       )\b}gix) {
# 	print "$file_name: $1\n";
#       }

      # case #4: help/verb + NP + to + following verb
#       while (m{\b(
# 		 help\w*?/v\w*?   # help
# 		 \s+              # (space)
# 		 \w+/[np]\w*?     # tag starts with 'n' or 'p'
# 		 \s+              # (space)
# 		 to/to            # to
# 		 \s+              # (space)
# 		 \w+/v\w*?        # following verb
# 	       )\b}gix) {
# 	print "$file_name: $1\n";
#       }

    }
  }
}

print "Number of occurrences = $counter\n";