#!/usr/bin/perl
#
# mv_suffix [-nf] files|directory...
#
# Using the file command, try to determine and set the correct filename suffix
# of the given files or recursive directories, and fix them.
#
# OPTIONS:
#    -f   Fix the suffix problems found (default)
#    -n   Dry Run, list command to fix, don't actually run them
#
# This is typically used after downloading files from the WWW as often the
# suffix is not correct, or is a HTML error file rather than the expected
# data. This command lets me find and re-download such problems.
#
####
#
# Anthony Thyssen,  19 March 1995
# 19 Feb 2018  Script renamed from "fix_suffix" to "mv_suffix"
#              and re-factored to make simplier

select((select(STDOUT), $| = 1)[$[]);

my $mv='merge';
my $FIX = 1;
   $FIX = shift   if $ARGV[0] eq '-f';    # force fix
   $FIX = 0,shift if $ARGV[0] eq '-n';    # dry run

if ( $FIX ) {
  print STDERR "Running the following commands...\n";
} else {
  print STDERR "DRY RUN... Commands to fix suffixes...\n";
}


map { s/'/'\\''/; $_ = "\'$_\'"; }  @ARGV;
open(FILE, "find @ARGV \\( -name .xvpics -prune \\) -o \\( -type f -print0 \\) |
               xargs -0 file |") or die("open \"file|\" failed: $!\n");


while( <FILE> ) {
  chomp;
  my ($file, $type) = split(/:\s*/, $_, 2);

  # Non file types
  next if $type =~ /\bdirectory\b/;
  next if $type =~ /\bspecial\b/;

  if ( $type =~ /\btext\b/ ) {
    fix_it($file, 'html') if $type =~ /\bHTML\b/;
    # Text files could many different suffixes:
    # For example:  txt, info, data, keys, csv, json
    # All of which "file" can not properly identify.
    # It is basically imposible to properly correct text file suffixes.
    #
    #next if $file =~ /\bREADME[^\/]*$/i; # skip README files
    next;
  }

  if ( $type =~ /\bimage data\b/ ) {
    fix_it($file, 'jpg'),next if $type =~ /\bJPEG\b/;
    fix_it($file, 'png'),next if $type =~ /\bPNG\b/;
    fix_it($file, 'gif'),next if $type =~ /\bGIF\b/;
    next;
  }

  if ( $type =~ /\bAudio\b/ ) {
    fix_it($file, 'mp3'  ),next if $type =~ /\bMPEG ADTS, layer III\b/;
    fix_it($file, 'm4a'  ),next if $type =~ /\bM4A\b/;
    fix_it($file, 'flac' ),next if $type =~ /\bFLAC\b/;
  }

  # no major catagory for these in 'file'
  fix_it($file, 'pdf'  ),next if $type =~ /\bPDF\b/;
  fix_it($file, 'avi'  ),next if $type =~ /\bAVI\b/;
  fix_it($file, 'mp4'  ),next if $type =~ /\bMP4\b/;
  fix_it($file, 'mpg'  ),next if $type =~ /\bMPEG\b/;
  fix_it($file, 'mov'  ),next if $type =~ /\bMOV\b/;
  fix_it($file, 'mkv'  ),next if $type =~ /\bMatroska\b/;
  fix_it($file, 'empty'),next if $type =~ /\bempty\b/;
}

# -------------------------
sub fix_it {
  my ( $file, $new_suffix) = @_;

  my ($name, $suffix);
  if ( $file =~ /\.(\w+)\.(\d+)$/ ) {  # suffix with a 'merge' number
    $name    = "$`_$2";
    $suffix .= "$1.$2";  # force suffix replacement if file type is known
  }
  elsif ( $file =~ /\.(\w+)$/ ) {     # just the suffix
    $name   = $`;
    $suffix = $1;
  }
  else {
    $name=$file; $suffix='';
  }

  #print "$file => \"$name\".\"$suffix\"\n===> \"$new_suffix\"\n";

  if ( $suffix ne $new_suffix ) {
    print "$mv \"$file\" \"$name.$new_suffix\"\n";
    return unless $FIX;
    $mv eq 'mv' ? rename( $file, "$name.$new_suffix")
                : system( $mv, $file, "$name.$new_suffix" )
                ;
  }
}