X-Git-Url: http://git.shiar.nl/barcat.git/blobdiff_plain/b43c5deb53ccc3cfea0ac6ac7aa2f03ce1c7a186..e8e00f42da0ae898ded80b3f79b2d9f2ebef7a44:/graph diff --git a/graph b/graph index 16af69a..ea17d45 100755 --- a/graph +++ b/graph @@ -1,43 +1,135 @@ #!/usr/bin/env perl -use 5.014; +use 5.018; use warnings; -use List::Util qw( max sum ); +use utf8; +use List::Util qw( min max sum ); +use open qw( :std :utf8 ); +use experimental qw( lexical_subs ); -our $VERSION = '1.00'; +our $VERSION = '1.02'; -use Getopt::Long '2.33'; -GetOptions(\my %opt, +use Getopt::Long '2.33', qw( :config gnu_getopt ); +sub podexit { + require Pod::Usage; + Pod::Usage::pod2usage(-exitval => 0, -perldocopt => '-oman', @_); +} +my %opt; +GetOptions(\%opt, + 'color|c!', + 'C' => sub { $opt{color} = 0 }, + 'field|f=s', + 'interval|t:i', + 'trim|length|l=s' => sub { + my ($optname, $optval) = @_; + $optval =~ s/%$// and $opt{trimpct}++; + $optval =~ m/^-?[0-9]+$/ or die( + "Value \"$optval\" invalid for option $optname", + " (number or percentage expected)\n" + ); + $opt{trim} = $optval; + }, + 'value-length=i', + 'markers|m=s', + 'unmodified|u!', 'width|w=i', + 'usage|h' => sub { podexit() }, + 'help' => sub { podexit(-verbose => 2) }, ) or exit 64; # EX_USAGE + $opt{width} ||= $ENV{COLUMNS} || 80; +$opt{color} //= -t *STDOUT; # enable on tty +$opt{trim} *= $opt{width} / 100 if $opt{trimpct}; + +if (defined $opt{interval}) { + $opt{interval} ||= 1; + $SIG{ALRM} = sub { + show_lines(); + alarm $opt{interval}; + }; + alarm $opt{interval}; +} + +$SIG{INT} = 'IGNORE'; # continue after assumed eof + +my (@lines, @values); +my $anchor = !defined $opt{field} ? qr/\A/ : + $opt{field} =~ /^[0-9]+$/ ? qr/(?:\S*\h+){$opt{field}}\K/ : + $opt{field}; +while (readline) { + s/\r?\n\z//; + s/^\h*// unless $opt{unmodified}; + push @values, s/$anchor ( \h* -? [0-9]* \.? [0-9]+ |)/\n/x && $1; + if (defined $opt{trim}) { + my $trimpos = abs $opt{trim}; + if ($trimpos <= 1) { + $_ = substr $_, 0, 1; + } + elsif (length > $trimpos) { + substr($_, $trimpos - 1) = '…'; + } + } + push @lines, $_; +} + +$SIG{INT} = 'DEFAULT'; + +sub show_lines { -my @lines = readline or exit; -chomp for @lines; -my @values = map { s/^\h*([0-9]*)// and $1 } @lines; -my @order = sort { $b <=> $a } @values; -my $lenval = 1 + int log($order[0]) / log 10; # max string length -my $len = 1 + max map { length } @lines; # left padding -my $size = ($opt{width} - $lenval - $len) / $order[0]; # bar multiplication +state $nr = 0; +@lines and @lines > $nr or return; + +my @order = sort { $b <=> $a } grep { length } @values; +my $maxval = $order[0]; +my $minval = min $order[-1], 0; +my $lenval = $opt{'value-length'} // max map { length } @order; +my $len = defined $opt{trim} && $opt{trim} <= 0 ? -$opt{trim} + 1 : + max map { length $values[$_] && length $lines[$_] } 0 .. $#lines; # left padding +my $size = ($maxval - $minval) && + ($opt{width} - $lenval - $len) / ($maxval - $minval); # bar multiplication -sub orderpos { ($order[$_[0]] + $order[$_[0] + .5]) / 2 * $size } my @barmark; -$barmark[ sum(@values) / @values * $size ] = '='; # average -$barmark[ orderpos($#order / 2) ] = '+'; # mean -defined and $_ = "\e[36m$_\e[0m" for @barmark; +if ($opt{markers} // 1 and $size > 0) { + my sub orderpos { (($order[$_[0]] + $order[$_[0] + .5]) / 2 - $minval) * $size } + $barmark[ (sum(@order) / @order - $minval) * $size ] = '='; # average + $barmark[ orderpos($#order * .31731) ] = '>'; + $barmark[ orderpos($#order * .68269) ] = '<'; + $barmark[ orderpos($#order / 2) ] = '+'; # mean + $barmark[ -$minval * $size ] = '|' if $minval < 0; # zero + defined and $opt{color} and $_ = "\e[36m$_\e[0m" for @barmark; + + state $lastmax = $maxval; + if ($maxval > $lastmax) { + print ' ' x ($lenval + $len); + printf "\e[90m" if $opt{color}; + printf '%-*s', + ($lastmax - $minval) * $size + .5, + '-' x (($values[$nr - 1] - $minval) * $size); + print "\e[92m" if $opt{color}; + say '+' x (($maxval - $lastmax - $minval) * $size + .5); + print "\e[0m" if $opt{color}; + $lastmax = $maxval; + } +} -for my $nr (0 .. $#lines) { +while ($nr <= $#lines) { my $val = $values[$nr]; - my $color = - $val == $order[0] ? 32 : # max - $val == $order[-1] ? 31 : # min - 90; - printf "\e[%sm", $color if $color; - printf "%*s", $lenval, $val; - print "\e[0m" if $color; - printf '%-*s', $len, $lines[$nr]; - print $barmark[$_] // '-' for 1 .. $val * $size; + if (length $val) { + my $color = !$opt{color} ? 0 : + $val == $order[0] ? 32 : # max + $val == $order[-1] ? 31 : # min + 90; + $val = sprintf "%*s", $lenval, $val; + $val = "\e[${color}m$val\e[0m" if $color; + } + my $line = $lines[$nr] =~ s/\n/$val/r; + printf '%-*s', $len + length($val), $line; + print $barmark[$_] // '-' for 1 .. $size && (($values[$nr] || 0) - $minval) * $size; say ''; + $nr++; +} + } +show_lines(); __END__ @@ -49,16 +141,90 @@ graph - append bar chart to input numbers B [] [] -cat ... | uniq -c | graph - =head1 DESCRIPTION -Each line starting with a number is given a bar to visualise relative sizes. +Visualizes relative sizes of values read from input (file(s) or STDIN). +Contents are concatenated similar to I, +but numbers are reformatted and a bar graph is appended to each line. =head1 OPTIONS =over +=item -c, --[no-]color + +Force colored output of values and bar markers. +Defaults on if output is a tty, +disabled otherwise such as when piped or redirected. + +=item -f, --field=(|) + +Compare values after a given number of whitespace separators, +or matching a regular expression. + +Unspecified or I<-f0> means values are at the start of each line. +With I<-f1> the second word is taken instead. +A string can indicate the starting position of a value +(such as I<-f:> if preceded by colons), +or capture the numbers itself, +for example I<-f'(\d+)'> for the first digits anywhere. + +=item -t, --interval[=] + +Interval time to output partial progress. + +=item -l, --length=[-][%] + +Trim line contents (between number and bars) +to a maximum number of characters. +The exceeding part is replaced by an abbreviation sign, +unless C<--length=0>. + +Prepend a dash (i.e. make negative) to enforce padding +regardless of encountered contents. + +=item -m, --markers= + +Statistical positions to indicate on bars. +Cannot be customized yet, +only disabled by providing an empty argument. + +Any value enables all marker characters: + +=over 2 + +=item B<=> + +Average: +the sum of all values divided by the number of counted lines. + +=item B<+> + +Mean, median: +the middle value or average between middle values. + +=item B<<> + +Standard deviation left of the mean. +Only 16% of all values are lower. + +=item B<< > >> + +Standard deviation right of the mean. +The part between B<< <--> >> encompass all I results, +or 68% of all entries. + +=back + +=item -u, --unmodified + +Do not strip leading whitespace. +Keep original value alignment, which may be significant in some programs. + +=item --value-length= + +Reserved space for numbers. + =item -w, --width= Override the maximum number of columns to use. @@ -66,6 +232,69 @@ Appended graphics will extend to fill up the entire screen. =back +=head1 EXAMPLES + +Commonly used after counting, such as users on the current server: + + users | sed 's/ /\n/g' | sort | uniq -c | graph + +Letter frequencies in text files: + + cat /usr/share/games/fortunes/*.u8 | + perl -CO -nE 'say for grep length, split /\PL*/, uc' | + sort | uniq -c | graph + +Memory usage of user processes: + + ps xo %mem,pid,cmd | graph -l40 + +Sizes (in megabytes) of all root files and directories: + + du -d0 -m * | graph + +Number of HTTP requests per day: + + cat log/access.log | cut -d\ -f4 | cut -d: -f1 | uniq -c | graph + +Any kind of database query with leading counts: + + echo 'SELECT count(*),schemaname FROM pg_tables GROUP BY 2' | + psql -t | graph -u + +Exchange rate USD/EUR history from CSV download provided by ECB: + + curl https://sdw.ecb.europa.eu/export.do \ + -Gd 'node=SEARCHRESULTS&q=EXR.D.USD.EUR.SP00.A&exportType=csv' | + grep '^[12]' | graph -f',\K' --value-length=7 + +Total population history from the World Bank dataset (XML): + + curl http://api.worldbank.org/v2/country/1W/indicator/SP.POP.TOTL | + xmllint --xpath '//*[local-name()="date" or local-name()="value"]' - | + sed -r 's,,\n,g; s,(<[^>]+>)+, ,g' | graph -f1 + +Movies per year from prepared JSON data: + + curl https://github.com/prust/wikipedia-movie-data/raw/master/movies.json | + jq '.[].year' | uniq -c | graph + +Pokémon height comparison: + + curl https://github.com/Biuni/PokemonGO-Pokedex/raw/master/pokedex.json | + jq -r '.pokemon[] | [.height,.num,.name] | join(" ")' | graph + +Git statistics, such commit count by year: + + git log --pretty=%ci | cut -b-4 | uniq -c | graph + +Or the most frequent authors: + + git shortlog -sn | graph | head -3 + +Latency history: + + ping google.com | graph -f'time=\K' -t + =head1 AUTHOR Mischa POSLAWSKY