#!/usr/bin/env perl

# Print names of .class files with no corresponding .java file or file
# containing the definition of the class.
# All the calls to external grep are very slow; perhaps figure out a way of
# optimizing this.

use strict;
use English;
$WARNING = 1;

my $debug = 0;

# Directories known to contain .class files without source; don't report these.
my @no_source = (
		 'com/oroinc/',
		 'gnu/regexp/',
		 'Metamata1_1dx.class',
		 'Metamata1_1ax.class',
		 'dec/trek/',
		 'jtrek/',
		 'jasmine/',
		 'mocha/'
		 );

my $class_files = `find . -name '*.class' -print`;
foreach my $class_file (split(/[ \t\n\r]+/, $class_files)) {
  my $has_no_source = 0;
  foreach my $nosource (@no_source) {
    if ($class_file =~ /$nosource/) {
      $has_no_source = 1;
      last;
    }
  }
  if ($has_no_source) {
    next;
  }

  my $class = $class_file;
  if ($debug) { print "$class_file\n"; }
  $class =~ s/.class$//;
  $class =~ s/\$[0-9]+$//;
  $class =~ s/.*\///;
  my $dir = $class_file;
  if (!($dir =~ s/\/[^\/]*$//)) {
    die "no slash??";
    $dir = ".";
  }
  my $java_file = "$class.java";
  if (-e $java_file) {
    if ($debug) { print "$java_file exists\n"; }
    next;
  }

  # Inner classes have names like "Outer$Inner", except that inner classes
  # defined within a method have names like "Outer$[N]Inner", where N is
  # composed of decimal digits.  An "Inner" identifier cannot start with
  # decimal digits.
  if ($class =~ /\$[0-9]*/) {
    my $inner = $POSTMATCH;
    my $outer = $PREMATCH;
    # Consider only the most deeply nested inner class.
    $inner =~ s/^.*\$[0-9]*([^\$]+)$/$1/;
    if ($debug) { print "Trying outer/inner: $outer $inner; file will be $dir/$outer.java\n"; }
    if ((-e "$dir/$outer.java")
	&& (system("grep -q -s '\\(class\\|interface\\) $inner\\b' $dir/$outer.java") == 0)) {
      if ($debug) { print "$dir/$outer.java contains '(class|interface) $inner'\n"; }
      next;
    }
    $class = $inner;
  }

  # If we didn't find the definition in a specific file, then try all files
  # in the directory.  This is not done from the start because it is more
  # expensive than checking a single file.
  if (system("grep -q -s '\\(class\\|interface\\) $class\\b' $dir/*.java") == 0) {
    if ($debug) { print "$dir/*.java contains '(class|interface) $class'\n"; }
    next;
  }

  if ($debug) { print "FAILURE: $class_file\n"; }
  print "$class_file\n";
}