#!/usr/bin/env perl
require 5.003;
use warnings;

#
# Copyright by The HDF Group.
# All rights reserved.
#
# This file is part of HDF4.  The full HDF4 copyright notice, including
# terms governing use, modification, and redistribution, is contained in
# the COPYING file, which can be found at the root of the source code
# distribution tree, or in https://www.hdfgroup.org/licenses.
# If you do not have access to either file, you may request a copy from
# help@hdfgroup.org.
#
# Purpose: Given an input file containing the output from a build of the
#          library, gather the file names and line numbers, alias
#          identical types of warnings together into a single bin and
#          report the number of warnings for each type of warning, each file
#          and the total count of warnings

# Perl modules/settings
use strict;
use Getopt::Std;

# Global variables, for accumulating information
my $totalcount = 0;
my $notecount = 0;
my $dupcount = 0;
my %warn_count = ();
my $warncount;
my %warn_file = ();
my %warn_file_line = ();
my %warn_file_line_offset = ();
my %file_count = ();
my $filecount;
my $ignorecount = 0;
my @ignorenames;
my %ignored_files = ();
my %warn_file_indices = ();
my %file_warn_indices = ();
my @warn_match_strings;
my @file_match_strings;
my %file_warn = ();
my %file_warn_line = ();
my $current_warning = 0;
my $current_file = 0;
my $warn_index;
my $genericize = 1;

# Info about last name / line / offset for file
my $last_c_name;
my $last_fort_name;
my $last_fort_line;
my $last_fort_offset;

# Display usage
sub do_help {
    print "Usage: 'warnhist [-h, --help] [-t <prefix>] [-w <n>] [-W] [-f <n>] [-F] [-s <warning string list>] [-S <file string list] [-l] [-u] [-i <name list>] [file]'\n";
    print "\t-h, --help\tDisplay this usage\n";
    print "\t-t <prefix>\tTrim pathname prefix from filenames, <prefix>\n";
    print "\t-w <n>\tDisplay files for a given warning index list, <n>\n";
    print "\t\t<n> can be a single value, a range, or a comma separated list\n";
    print "\t\tFor example: '0' or '0,4' or '8-10' or '0,2-4,8-10,13'\n";
    print "\t-W\tDisplay files for all warnings\n";
    print "\t-f <n>\tDisplay warnings for a given file index list, <n>\n";
    print "\t\t<n> can be a single value, a range, or a comma separated list\n";
    print "\t\tFor example: '0' or '0,4' or '8-10' or '0,2-4,8-10,13'\n";
    print "\t-F\tDisplay warnings for all files\n";
    print "\t-s <warning string list>\tDisplay files for warnings which contain a string, <warning string list>\n";
    print "\t\t<warning string list> is a comma separated list, with no spaces\n";
    print "\t\tFor example: 'Wunused-dummy-argument' or 'Wunused-dummy-argument,Wunused-variable'\n";
    print "\t-S <file string list>\tDisplay warnings for files which contain a string, <file string list>\n";
    print "\t\t<file string list> is a comma separated list, with no spaces\n";
    print "\t\tFor example: 'H5Fint' or 'H5Fint,H5Gnode'\n";
    print "\t-l\tDisplay line numbers for file/warning\n";
    print "\t-u\tLeave 'unique' types in warnings, instead of genericizing them\n";
    print "\t-i <name list>\tIgnore named files, <name list>\n";
    print "\t\t<name list> is a comma separated list, with no spaces\n";
    print "\t\tFor example: 'H5LTparse' or 'H5LTparse,H5LTanalyze'\n";
    print "\tfile\tFilename containing build output\n";
    print "\t\tIf no file is given, standard input is used.\n";
    exit;
}

sub main::HELP_MESSAGE {
    do_help();
}

# declare the Perl command line flags/options we want to allow
my %options=();
getopts("FWhut:w:f:s:S:i:l", \%options);

# Display usage, if requested
if($options{h}) {
    do_help();
}

# Parse list of file names to ignore
if(exists $options{i}) {
    @ignorenames = split /,/, $options{i};
#print STDERR @ignorenames;
}

# Parse list of warning indices to expand file names
if(exists $options{w}) {
    my @tmp_indices;

    @tmp_indices = split /,/, $options{w};
#print STDERR @tmp_indices;
    for my $x (@tmp_indices) {
#print STDERR "x = '$x'\n";
        if($x =~ /\-/) {
            my $start_index;
            my $end_index;

#print STDERR "matched = '$x'\n";
            ($start_index, $end_index) = split /\-/, $x;
#print STDERR "start_index = '$start_index', end_index = '$end_index'\n";
            for my $y ($start_index..$end_index) {
#print STDERR "y = '$y'\n";
                if(!exists $warn_file_indices{$y}) {
                    $warn_file_indices{$y} = $y;
                }
            }
        }
        else {
            if(!exists $warn_file_indices{$x}) {
                $warn_file_indices{$x} = $x;
            }
        }
    }
#foreach (sort keys %warn_file_indices) {
#    print STDERR "$_ : $warn_file_indices{$_}\n";
#}
}

# Parse list of warning strings to expand file names
if(exists $options{s}) {
    @warn_match_strings = split /,/, $options{s};
# print STDERR @warn_match_strings;
}

# Parse list of file indices to expand warnings
if(exists $options{f}) {
    my @tmp_indices;

    @tmp_indices = split /,/, $options{f};
#print STDERR @tmp_indices;
    for my $x (@tmp_indices) {
#print STDERR "x = '$x'\n";
        if($x =~ /\-/) {
            my $start_index;
            my $end_index;

#print STDERR "matched = '$x'\n";
            ($start_index, $end_index) = split /\-/, $x;
#print STDERR "start_index = '$start_index', end_index = '$end_index'\n";
            for my $y ($start_index..$end_index) {
#print STDERR "y = '$y'\n";
                if(!exists $file_warn_indices{$y}) {
                    $file_warn_indices{$y} = $y;
                }
            }
        }
        else {
            if(!exists $file_warn_indices{$x}) {
                $file_warn_indices{$x} = $x;
            }
        }
    }
#foreach (sort keys %warn_file_indices) {
#    print STDERR "$_ : $warn_file_indices{$_}\n";
#}
}

# Parse list of warning strings for files to expand warnings
if(exists $options{S}) {
    @file_match_strings = split /,/, $options{S};
# print STDERR @file_match_strings;
}

# Check if warnings should stay unique and not be "genericized"
if($options{u}) {
    $genericize = 0;
}

PARSE_LINES:
while (<>) {
    my $name;
    my $line;
    my $prev_line;
    my $toss;
    my $offset;
    my $warning;
    my $extra;
    my $extra2;

    # Retain last FORTRAN compile line, which comes a few lines before warning
    if($_ =~ /.*\.[fF]90:.*/) {
        ($last_fort_name, $last_fort_line, $last_fort_offset) = split /\:/, $_;
        ($last_fort_line, $toss) = split /\./, $last_fort_line;
    }

    # Retain last C/C++ compile line, which possibly comes a few lines before warning
    if($_ =~ /.*[A-Za-z0-9_]\.[cC]:.*/) {
        ($last_c_name, $toss) = split /\:/, $_;
    }

    # Retain C/C++ compile line, which comes with the line of warning
    if($_ =~ /.*[A-Za-z0-9_]\.[chC]\(.*[0-9]\):.*#.*/) {
        $last_c_name = $_;
    }

    # Skip lines that don't have the word "warning"
    next if $_ !~ /[Ww]arning/;

    # Skip warnings from linker
    next if $_ =~ /ld: warning:/;

    # Skip warnings from build_py and install_lib
    next if $_ =~ /warning: (build_py|install_lib)/;

    # Skip variables with the word 'warning' in them
    next if $_ =~ /_warning_/;

    # Skip AMD Optimizing Compiler (aocc) lines "<#> warning(s) generated."
    next if $_ =~ / warnings? generated\./;

    # "Hide" the C++ '::' symbol until we've parsed out the parts of the line
    while($_ =~ /\:\:/) {
	$_ =~ s/\:\:/@@@@/g;
    }

    # Check for weird formatting of warning message
    $line = "??";
    $offset = "??";
    if($_ =~ /^cc1: warning:.*/) {
        $name = $last_c_name;
        ($toss, $toss, $warning, $extra, $extra2) = split /\:/, $_;
    # Check for CMAKE build with warning on first line and no filename
    } elsif($_ =~ /^\s*[Ww]arning:.*/) {
        $name = $last_c_name;
        ($toss, $warning, $extra, $extra2) = split /\:/, $_;
    # Check for file-scope gcc Fortran warning output
    } elsif($_ =~ /f\d\d\d: Warning:/) {
        # These are interspersed with the "compiling a file" output
        # when compiling with `make -j` and thus difficult to tie to
        # any particular file. They are due to things like inappropriate
        # build options and don't have a line number.
        #
        # They start with f, as in f951
        $name = "(generic)";
        $line = int(rand(1000000)); # Hack to avoid counting as duplictates

        ($warning) = $_ =~ /\[(.*)\]/x;
    # Check for FORTRAN warning output
    } elsif($_ =~ /^Warning:.*/) {
        $name = $last_fort_name;
        $line = $last_fort_line;
        $offset = $last_fort_offset;
        ($toss, $warning, $extra, $extra2) = split /\:/, $_;

        # Check for improperly parsed filename or line
        if($name =~ /^$/) {
            print "Filename is a null string! Input line #$. is: '$_'";
            next
        }
        if($line =~ /^$/) {
            print "Line is a null string! Input line #$. is: '$_'";
            next
        }
    # Check for non-GCC warning (Solaris/Oracle?)
    } elsif($_ =~ /^\".*, line [0-9]+: *[Ww]arning:.*/) {
        ($name, $toss, $warning, $extra, $extra2) = split /\:/, $_;
        ($name, $line) = split /\,/, $name;
	    $name =~ s/^\"//g;
	    $name =~ s/\"$//g;
	    $line =~ s/^\s*line\s*//g;
    # Check for Intel icc warning
    } elsif($_ =~ /.*[A-Za-z0-9_]\.[chC]\(.*[0-9]\):.*#.*/) {
        ($last_c_name, $toss, $warning) = split /\:/, $last_c_name;
        ($name, $line) = split /\(/, $last_c_name;
        $line =~ s/\)//g;
    } else {
        # Check for 'character offset' field appended to file & line #
	# (This is probably specific to GCC)
        if($_ =~ /^.*[0-9]+\:[0-9]+\:/) {
            ($name, $line, $offset, $toss, $warning, $extra, $extra2) = split /\:/, $_;
        } else {
            ($name, $line, $toss, $warning, $extra, $extra2) = split /\:/, $_;
        }
    }

    # Check for extra ':' followed by more text in original warning string,
    #  and append the ':' and text back onto the parsed warning
    #  (Use 'length $extra' idiom to avoid warning when $extra is undefined)
    if(length $extra ) {
        $warning = join ':', $warning, $extra;
    }
    if(length $extra2 ) {
        $warning = join ':', $warning, $extra2;
    }

    # Restore the C++ '::' symbol now that we've parsed out the parts of the line
    while($warning =~ /@@@@/) {
        $warning =~ s/@@@@/\:\:/g;
    }

    # Trim leading '..' paths from filename
    while($name =~ /^\.\.\//) {
        $name =~ s/^\.\.\///g;
    }

    # Check for trimming prefix
    if((exists $options{t}) && ($name =~ /$options{t}/)) {
        $name =~ s/^$options{t}\///g;
    }

    # Check for ignored file
    if(exists $options{i}) {
        for my $x (@ignorenames) {
            if($name =~ /$x/) {
                $ignorecount++;
                if(!(exists $ignored_files{$name})) {
                    $ignored_files{$name} = $name;
                }
                next PARSE_LINES;
            }
        }
    }

    # Check for improperly parsed warning (usually an undefined warning string)
    if(!defined $warning) {
        print "Warning Undefined! Input line is: '$_'";
        next
    }

    # Get rid of leading & trailing whitespace
    $warning =~ s/^\s//g;
    $warning =~ s/\s$//g;

    # Check for improperly parsed warning
    if($warning =~ /^$/) {
        print "Warning is a null string! Input line is: '$_'";
        next
    }

    # Convert all quotes to '
    $warning =~ s/‘/'/g;
    $warning =~ s/’/'/g;
    $warning =~ s/"/'/g;

#
# These skipped messages & "genericizations" may be specific to GCC

    # Skip supplemental warning message
    if($warning =~ /near initialization for/) {
	$notecount++;
        next
    }

    # Skip C++ supplemental warning message
    if($warning =~ /in call to/) {
	$notecount++;
        next
    }

    # Skip GCC warning that should be a note
    if($_ =~ /\(this will be reported only once per input file\)/) {
	$notecount++;
        next
    }

    if($genericize) {
	# Eliminate C/C++ "{aka <some type>}" and "{aka '<some type>'}" info
	if($warning =~ /\s(\{|\()aka '?[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#]+[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#\ ]*'?(\}|\))/) {
	    $warning =~ s/\s(\{|\()aka '?[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#]+[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#\ ]*'?(\}|\))//g;
	}

	# Genericize C/C++ '<some type>', printf format '%<some format>', and
	# "unknown warning group" into '-'
	if($warning =~ /'[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#\-\=]+[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#\-\=\ ]*'/) {
	    $warning =~ s/'[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#\-\=]+[A-Za-z_0-9\(\)\*\,\[\]\.\<\>\&\:\+\#\-\=\ ]*'/'-'/g;
	}
	if($warning =~ /'%[\#0\-\ \+]*[,;\:_]?[0-9\*]*\.?[0-9\*]*[hjltzL]*[aAcdeEfFgGinosuxX]'/) {
	    $warning =~ s/'%[\#0\-\ \+]*[,;\:_]?[0-9\*]*\.?[0-9\*]*[hjltzL]*[aAcdeEfFgGinosuxX]'/'-'/g;
	}

	# Genericize C/C++ "<macro>" warnings into "-"
	if($warning =~ /"[A-Za-z_0-9]*"/) {
	    $warning =~ s/"[A-Za-z_0-9]*"/"-"/g;
	}

	# Genericize [GCC?] C/C++ warning text about suggested attribute
	if($warning =~ /attribute=[A-Za-z_0-9]*\]/) {
	    $warning =~ s/=[A-Za-z_0-9]*\]/=-\]/g;
	}

	# Genericize FORTRAN "at (<n>)" into "at (-)", "REAL(<n>)" into "REAL(-)",
	# 	and "INTEGER(<n>)" into "INTEGER(-)"
	if($warning =~ /.*at\s\([0-9]+\).*/) {
	    $warning =~ s/at\s\([0-9]+\)/at \(-\)/g;
	}
	if($warning =~ /.*REAL\([0-9]+\).*/) {
	    $warning =~ s/REAL\([0-9]+\)/REAL\(-\)/g;
	}
	if($warning =~ /.*INTEGER\([0-9]+\).*/) {
	    $warning =~ s/INTEGER\([0-9]+\)/INTEGER\(-\)/g;
	}

	# Genericize standalone numbers in warnings
	if($warning =~ /(\s|')-?[0-9]+(\s|')/) {
	    $warning =~ s/-?[0-9]+/-/g;
	}

	# Genericize unusual GCC/G++/GFORTRAN warnings that aren't handled above
	if($warning =~ /\[deprecation\] [A-Za-z_0-9]*\([A-Za-z_,0-9]*\) in [A-Za-z_0-9]* has been deprecated.*/) {
	    $warning =~ s/[A-Za-z_0-9]*\([A-Za-z_,0-9]*\) in [A-Za-z_0-9]*/-\(-\) in -/g;
	}
    }
# <end possible GCC-specific code>

    # Check if we've already seen this warning on this line in this file
    # (Can happen for warnings from inside header files)
    if( !exists $warn_file_line_offset{$warning}{$name}{$line}{$offset} ) {
	# Increment count for [generic] warning
	$warn_count{$warning}++;
	$warn_file{$warning}{$name}++;
	$warn_file_line{$warning}{$name}{$line}++;
	$warn_file_line_offset{$warning}{$name}{$line}{$offset}++;

	# Increment count for filename
	$file_count{$name}++;
	$file_warn{$name}{$warning}++;
	$file_warn_line{$name}{$warning}{$line}++;

	# Increment total count of warnings
	$totalcount++;
    }
    else {
	# Increment count of duplicate warnings
	$dupcount++;
    }

#    print STDERR "name = $name\n";
#    print STDERR "line = $line\n";
#    print STDERR "offset = $offset\n";
#    print STDERR "warning = \"$warning\"\n";
}

print "Total unique [non-ignored] warnings: $totalcount\n";
print "Ignored notes / supplemental warning lines [not counted in unique warnings]: $notecount\n";
print "Duplicated warning lines [not counted in unique warnings]: $dupcount\n";
print "Total ignored warnings: $ignorecount\n";
$warncount = keys %warn_count;
print "Total unique kinds of warnings: $warncount\n";
$filecount = keys %file_count;
print "Total files with warnings: $filecount\n\n";

# Print warnings in decreasing frequency
print "# of Warnings by frequency (file count)\n";
print "=======================================\n";
for my $x (sort {$warn_count{$b} <=> $warn_count{$a}} keys(%warn_count)) {
    printf ("[%2d] %4d (%2d) - %s\n", $current_warning++, $warn_count{$x}, scalar(keys %{$warn_file{$x}}), $x);
    if((exists $options{W}) || (exists $options{w}) || (exists $options{s})) {
        my $curr_index = $current_warning - 1;
	my $match = 0;

	# Check for string from list in current warning
	if(exists $options{s}) {
	    for my $y (@warn_match_strings) {
# print STDERR "y = '$y'\n";
		if($x =~ /$y/) {
# print STDERR "matched warning = '$x'\n";
		    $match = 1;
		    last;
		}
	    }
	}

	# Check if current warning index matches
        if((exists $warn_file_indices{$curr_index}) && $curr_index == $warn_file_indices{$curr_index}) {
	    $match = 1;
	}

        if($match) {
            for my $y (sort {$warn_file{$x}{$b} <=> $warn_file{$x}{$a}} keys(%{$warn_file{$x}})) {
                printf ("\t%4d - %s\n", $warn_file{$x}{$y}, $y);
                if(exists $options{l}) {
		    my $lines = join ", ", sort {$a <=> $b} keys %{$warn_file_line{$x}{$y}};
                    printf("\t\tLines: $lines \n");
                }
            }
        }
    }
}

# Print warnings in decreasing frequency, by filename
print "\n# of Warnings by filename (warning type)\n";
print   "========================================\n";
for my $x (sort {$file_count{$b} <=> $file_count{$a}} keys(%file_count)) {
    printf ("[%3d] %4d (%2d) - %s\n", $current_file++, $file_count{$x}, scalar(keys %{$file_warn{$x}}), $x);
    if((exists $options{F}) || (exists $options{f}) || (exists $options{S})) {
        my $curr_index = $current_file - 1;
	my $match = 0;

	# Check for string from list in current file
	if(exists $options{S}) {
	    for my $y (@file_match_strings) {
# print STDERR "y = '$y'\n";
		if($x =~ /$y/) {
# print STDERR "matched warning = '$x'\n";
		    $match = 1;
		    last;
		}
	    }
	}

	# Check if current file index matches
        if((exists $file_warn_indices{$curr_index}) && $curr_index == $file_warn_indices{$curr_index}) {
	    $match = 1;
	}

        if($match) {
            for my $y (sort {$file_warn{$x}{$b} <=> $file_warn{$x}{$a}} keys(%{$file_warn{$x}})) {
                printf ("\t%4d - %s\n", $file_warn{$x}{$y}, $y);
                if(exists $options{l}) {
		    my $lines = join ", ", sort {$a <=> $b} keys %{$file_warn_line{$x}{$y}};
                    printf("\t\tLines: $lines \n");
                }
            }
        }
    }
}

# Print names of files that were ignored
# Check for ignored file
if(exists $options{i}) {
    print "\nIgnored filenames\n";
    print   "=================\n";
    for my $x (sort keys(%ignored_files)) {
	print "$x\n";
    }
}