TALK

I just gave a talk about this at SCaLE 17x. Here are the video of the talk and the "slides".

NAME

feedgnuplot - General purpose pipe-oriented plotting tool

SYNOPSIS

Simple plotting of piped data:

$ seq 5 | awk '{print 2*$1, $1*$1}'
2 1
4 4
6 9
8 16
10 25

$ seq 5 | awk '{print 2*$1, $1*$1}' |
  feedgnuplot               \
    --lines                 \
    --points                \
    --title "Test plot"     \
    --y2 1                  \
    --unset key             \
    --unset grid

Simple real-time plotting example: plot how much data is received on the wlan0 network interface in bytes/second. This plot updates at 1Hz, and shows the last 10sec of history. The plot shown here is the final state of a sample run

$ while true; do
    sleep 1;
    cat /proc/net/dev;
  done \
  | gawk '/wlan0/ {if(b) {print $2-b; N++; fflush()} b=$2} N==15 {exit}' \
  | feedgnuplot            \
      --lines              \
      --title "wlan0 throughput" \
      --stream             \
      --xlen 10            \
      --ylabel 'Bytes/sec' \
      --xlabel seconds     \
      --unset key          \
      --unset grid

DESCRIPTION

This is a flexible, command-line-oriented frontend to Gnuplot. It creates plots from data coming in on STDIN or given in a filename passed on the commandline. Various data representations are supported, as is hardcopy output and streaming display of live data. For a tutorial and a gallery please see the guide at https://github.com/dkogan/feedgnuplot/blob/master/guide/guide.org

A simple example:

$ seq 5 | awk '{print 2*$1, $1*$1}' | feedgnuplot

You should see a plot with two curves. The awk command generates some data to plot and the feedgnuplot reads it in from STDIN and generates the plot. The awk invocation is just an example; more interesting things would be plotted in normal usage. No commandline-options are required for the most basic plotting. Input parsing is flexible; every line need not have the same number of points. New curves will be created as needed.

The most commonly used functionality of gnuplot is supported directly by the script. Anything not directly supported can still be done with options such as --set, --cmds --style, etc. Arbitrary gnuplot commands can be passed in with --cmds. For example, to turn off the grid, you can pass in --cmds 'unset grid'. Commands --set and --unset exists to provide nicer syntax, so this is equivalent to passing --unset grid. As many of these options as needed can be passed in. To add arbitrary curve styles, use --style curveID extrastyle. Pass these more than once to affect more than one curve.

To apply an extra style to all the curves that lack an explicit --style, pass in --styleall extrastyle. In the most common case, the extra style is with something. To support this more simply, you can pass in --with something instead of --styleall 'with something'. --styleall and --with are mutually exclusive. Furthermore any curve-specific --style overrides the global --styleall or --with setting.

Data formats

By default, each value present in the incoming data represents a distinct data point, as demonstrated in the original example above (we had 10 numbers in the input and 10 points in the plot). If requested, the script supports more sophisticated interpretation of input data

Domain selection

If --domain is passed in, the first value on each line of input is interpreted as the X-value for the rest of the data on that line. Without --domain the X-value is the line number, and the first value on a line is a plain data point like the others. Default is --nodomain. Thus the original example above produces 2 curves, with 1,2,3,4,5 as the X-values. If we run the same command with --domain:

$ seq 5 | awk '{print 2*$1, $1*$1}' | feedgnuplot --domain

we get only 1 curve, with 2,4,6,8,10 as the X-values. As many points as desired can appear on a single line, but all points on a line are associated with the X-value at the start of that line.

Curve indexing

We index the curves in one of 3 ways: sequentially, explicitly with a --dataid or by --vnlog headers.

By default, each column represents a separate curve. The first column (after any domain) is curve 0. The next one is curve 1 and so on. This is fine unless sparse data is to be plotted. With the --dataid option, each point is represented by 2 values: a string identifying the curve, and the value itself. If we add --dataid to the original example:

$ seq 5 | awk '{print 2*$1, $1*$1}' | feedgnuplot --dataid --autolegend

we get 5 different curves with one point in each. The first column, as produced by awk, is 2,4,6,8,10. These are interpreted as the IDs of the curves to be plotted.

If we're plotting vnlog data (https://www.github.com/dkogan/vnlog) then we can get the curve IDs from the vnlog header. Vnlog is a trivial data format where lines starting with # are comments and the first comment contains column labels. If we have such data, feedgnuplot --vnlog can interpret these column labels if the vnlog perl modules are available.

The --autolegend option adds a legend using the given IDs to label the curves. The IDs need not be numbers; generic strings are accepted. As many points as desired can appear on a single line. --domain can be used in conjunction with --dataid or --vnlog.

Multi-value style support

Depending on how gnuplot is plotting the data, more than one value may be needed to represent the range of a single point. Basic 2D plots have 2 numbers representing each point: 1 domain and 1 range. But if plotting with --circles, for instance, then there's an extra range value: the radius. Many other gnuplot styles require more data: errorbars, variable colors (with points palette), variable sizes (with points ps variable), labels and so on. The feedgnuplot tool itself does not know about all these intricacies, but they can still be used, by specifying the specific style with --style, and specifying how many values are needed for each point with any of --rangesizeall, --tuplesizeall, --rangesize, --tuplesize. These options are required only for styles not explicitly supported by feedgnuplot; supported styles do the right thing automatically.

Specific example: if making a 2d plot of y error bars, the exact format can be queried by running gnuplot and invoking help yerrorbars. This tells us that there's a 3-column form: x y ydelta and a 4-column form: x y ylow yhigh. With 2d plots feedgnuplot will always output the 1-value domain x, so the rangesize is 2 and 3 respectively. Thus the following are equivalent:

$ echo '1 2 0.3
        2 3 0.4
        3 4 0.5' | feedgnuplot --domain --rangesizeall 2 --with 'yerrorbars'

$ echo '1 2 0.3
        2 3 0.4
        3 4 0.5' | feedgnuplot --domain --tuplesizeall 3 --with 'yerrorbars'

$ echo '1 2 1.7 2.3
        2 3 2.6 3.4
        3 4 3.5 4.5' | feedgnuplot --domain --rangesizeall 3 --with 'yerrorbars'

3D data

To plot 3D data, pass in --3d. --domain MUST be given when plotting 3D data to avoid domain ambiguity. If 3D data is being plotted, there are by definition 2 domain values instead of one (Z as a function of X and Y instead of Y as a function of X). Thus the first 2 values on each line are interpreted as the domain instead of just 1. The rest of the processing happens the same way as before.

Time/date data

If the input data domain is a time/date, this can be interpreted with --timefmt. This option takes a single argument: the format to use to parse the data. The format is documented in 'set timefmt' in gnuplot, although the common flags that strftime understands are generally supported. The backslash sequences in the format are not supported, so if you want a tab, put in a tab instead of \t. Whitespace in the format is supported. When this flag is given, some other options act a little bit differently:

• --xlen and --binwidth are integers in seconds

• --xmin and --xmax must use the format passed in to --timefmt

Using this option changes both the way the input is parsed and the way the x-axis tics are labelled. Gnuplot tries to be intelligent in this labelling, but it doesn't always do what the user wants. The labelling can be controlled with the gnuplot set format command, which takes the same type of format string as --timefmt. Example:

$ sar 1 -1 |
  awk '$1 ~ /..:..:../ && $8 ~/^[0-9\.]*$/ {print $1,$8; fflush()}' |
  feedgnuplot --stream --domain
               --lines --timefmt '%H:%M:%S'
               --set 'format x "%H:%M:%S"'

This plots the 'idle' CPU consumption against time.

Note that while gnuplot supports the time/date on any axis, feedgnuplot currently supports it only as the x-axis domain. This may change in the future.

'using' expressions

We just described how feedgnuplot parses its input data. When passing this data to gnuplot, each curve is sent independently. The domain appears in the leading columns followed by --rangesize columns to complete each row. Without --domain, feedgnuplot explicitly writes out sequential integers. gnuplot then knows how many values it has for each point, and it knows which style we're using, so it's able to interpret the data appropriately, and to make the correct plot.

As an example, if gnuplot is passed 2 columns of data, and it is plotting with points, it will use column 1 for the x coordinate and column 2 for the y coordinate. This is the default behavior, but the meaning of each column can be controlled via a using expression in gnuplot (not feedgnuplot; keep reading). The default is sequential integers, so this example uses using 1:2 by default. We can flip the meaning of the columns by passing using 2:1. Arbitrary expressions may be specified by enclosing each field in (), and using $ to denote each data column. So to use the 2nd column as the x coordinate and the sum of the two columns as the y coordinate, using 2:($1+$2) is passed. Furthermore, the number of columns can vary. For instance gnuplot can read the same two columns of data, but produce a plot with the extra column encoding the sum as the color: using 1:2:($1+$2) with points palette. Please see the gnuplot documentation for lots of detail.

That's how gnuplot works. Most of the time, feedgnuplot doesn't pass any using expressions at all, and gnuplot does the default thing. But if we want to do something fancy, feedgnuplot supports --using curveID expression and --usingall expression. So we can plot a parabola:

seq 100 | feedgnuplot --lines --usingall '1:($2*$2)'

This is powerful, but there are some things to keep in mind:

• --using overrides whatever using expression feedgnuplot was going to pass. feedgnuplot passes a using expression only if --histogram or --timefmt or --xticlabels are given. So if --using is given together with any of these, the user must take care to do the right thing (whatever that means at that time).

• The --tuplesize controls the data passed to feedgnuplot and the data then passed to gnuplot. It does not directly control how gnuplot eventually interprets the data: --using does that. So for instance we can plot color-coded points:

  seq 10 | feedgnuplot --with 'points pt 7 palette' --usingall '1:2:2'

  Here feedgnuplot read 1 column of data. It defauled to --tuplesize 2, so it passed 2 columns of data to gnuplot. gnuplot then produced 3 values for each point, and plotted them as indicated with the points palette style.

• You always need a column of data to generate a curve. You might want to use a using expression to plot a time series and its cumulative integral. The using expression can compute the integral, but you must pass in the data twice; once for each curve to plot:

  seq 100 |                           \
    awk '{print $1,$1}' |             \
    feedgnuplot                       \
      --cmds 'sum=0'                  \
      --cmds 'accum(x) = (sum=sum+x)' \
      --using 1 '1:(accum($2))'       \
      --lines --y2 1

Real-time streaming data

To plot real-time data, pass in the --stream [refreshperiod] option. Data will then be plotted as it is received. The plot will be updated every refreshperiod seconds. If the period isn't specified, a 1Hz refresh rate is used. To refresh at specific intervals indicated by the data, set the refreshperiod to 0 or to 'trigger'. The plot will then only be refreshed when a data line 'replot' is received. This 'replot' command works in both triggered and timed modes, but in triggered mode, it's the only way to replot. Look in "Special data commands" for more information.

To plot only the most recent data (instead of all the data), --xlen windowsize can be given. This will create an constantly-updating, scrolling view of the recent past. windowsize should be replaced by the desired length of the domain window to plot, in domain units (passed-in values if --domain or line numbers otherwise). If the domain is a time/date via --timefmt, then windowsize is and integer in seconds. If we're plotting a histogram, then --xlen causes a histogram over a moving window to be computed. The subtlely here is that with a histogram you don't actually see the domain since only the range is analyzed. But the domain is still there, and can be utilized with --xlen. With --xlen we can plot only histograms or only non-histograms.

Special data commands

If we are reading streaming data, the input stream can contain special commands in addition to the raw data. Feedgnuplot looks for these at the start of every input line. If a command is detected, the rest of the line is discarded. These commands are

replot

This command refreshes the plot right now, instead of waiting for the next refresh time indicated by the timer. This command works in addition to the timed refresh, as indicated by --stream [refreshperiod].

clear

This command clears out the current data in the plot. The plotting process continues, however, to any data following the clear.

exit

This command causes feedgnuplot to exit.

Hardcopy output

The script is able to produce hardcopy output with --hardcopy outputfile. The output type can be inferred from the filename, if .ps, .eps, .pdf, .svg, .png or .gp is requested. If any other file type is requested, --terminal must be passed in to tell gnuplot how to make the plot. If --terminal is passed in, then the --hardcopy argument only provides the output filename.

The .gp output is special. Instead of asking gnuplot to plot to a particular terminal, writing to a .gp simply dumps a self-executable gnuplot script into the given file. This is similar to what --dump does, but writes to a file, and makes sure that the file can be self-executing.

Self-plotting data files

This script can be used to enable self-plotting data files. There are several ways of doing this: with a shebang (#!) or with inline perl data.

Self-plotting data with a #!

A self-plotting, executable data file data is formatted as

$ cat data
#!/usr/bin/feedgnuplot --lines --points
2 1
4 4
6 9
8 16
10 25
12 36
14 49
16 64
18 81
20 100
22 121
24 144
26 169
28 196
30 225

This is the shebang (#!) line followed by the data, formatted as before. The data file can be plotted simply with

$ ./data

The caveats here are that on Linux the whole #! line is limited to 127 characters and that the full path to feedgnuplot must be given. The 127 character limit is a serious limitation, but this can likely be resolved with a kernel patch. I have only tried on Linux 2.6.

Self-plotting data with gnuplot

Running feedgnuplot --hardcopy plotdata.gp .... will create a self-executable gnuplot script in plotdata.gp

Self-plotting data with perl inline data

Perl supports storing data and code in the same file. This can also be used to create self-plotting files:

$ cat plotdata.pl
#!/usr/bin/perl
use strict;
use warnings;

open PLOT, "| feedgnuplot --lines --points" or die "Couldn't open plotting pipe";
while( <DATA> )
{
  my @xy = split;
  print PLOT "@xy\n";
}
__DATA__
2 1
4 4
6 9
8 16
10 25
12 36
14 49
16 64
18 81
20 100
22 121
24 144
26 169
28 196
30 225

This is especially useful if the logged data is not in a format directly supported by feedgnuplot. Raw data can be stored after the __DATA__ directive, with a small perl script to manipulate the data into a useable format and send it to the plotter.

ARGUMENTS

• --[no]domain

  If enabled, the first element of each line is the domain variable. If not, the point index is used

• --[no]dataid

  If enabled, each data point is preceded by the ID of the data set that point corresponds to. This ID is interpreted as a string, NOT as just a number. If not enabled, the order of the point is used.

  As an example, if line 3 of the input is "0 9 1 20" then

  • --nodomain --nodataid would parse the 4 numbers as points in 4 different curves at x=3

  • --domain --nodataid would parse the 4 numbers as points in 3 different curves at x=0. Here, 0 is the x-variable and 9,1,20 are the data values

  • --nodomain --dataid would parse the 4 numbers as points in 2 different curves at x=3. Here 0 and 1 are the data IDs and 9 and 20 are the data values

  • --domain --dataid would parse the 4 numbers as a single point at x=0. Here 9 is the data ID and 1 is the data value. 20 is an extra value, so it is ignored. If another value followed 20, we'd get another point in curve ID 20

• --vnlog

  Vnlog is a trivial data format where lines starting with # are comments and the first comment contains column labels. Some tools for working with such data are available from the vnlog project: https://www.github.com/dkogan/vnlog. With the vnlog perl modules installed, we can read the vnlog column headers with feedgnuplot --vnlog. This replaces --dataid, and we can do all the normal things with these headers. For instance feedgnuplot --vnlog --autolegend will generate plot legends for each column in the vnlog, using the vnlog column label in the legend.

• --[no]3d

  Do [not] plot in 3D. This only makes sense with --domain. Each domain here is an (x,y) tuple

• --timefmt [format]

  Interpret the X data as a time/date, parsed with the given format

• --colormap

  This is a legacy option used to who a colormapped xy plot. It does:

  - Adds palette to --curvestyleall

  - Adds 1 to the default --tuplesize (if --tuplesizeall is not given

  - Uses --zmin, --zmax to set the colorbar range

  It's clearer to set the relevant options explicitly, but --colormap still exists for compatibility

• --stream [period]

  Plot the data as it comes in, in realtime. If period is given, replot every period seconds. If no period is given, replot at 1Hz. If the period is given as 0 or 'trigger', replot only when the incoming data dictates this. See the "Real-time streaming data" section of the man page.

• --[no]lines

  Do [not] draw lines to connect consecutive points

• --[no]points

  Do [not] draw points

• --circles

  Plot with circles. This requires a radius be specified for each point. Automatically sets the --rangesize/--tuplesize. Not supported for 3d plots.

• --title xxx

  Set the title of the plot

• --legend curveID legend

  Set the label for a curve plot. Use this option multiple times for multiple curves. With --dataid, curveID is the ID. Otherwise, it's the index of the curve, starting at 0

• --autolegend

  Use the curve IDs for the legend. Titles given with --legend override these

• --xlen xxx

  When using --stream, sets the size of the x-window to plot. Omit this or set it to 0 to plot ALL the data. Does not make sense with 3d plots. Implies --monotonic. If we're plotting a histogram, then --xlen causes a histogram over a moving window to be computed. The subtlely here is that with a histogram you don't actually see the domain since only the range is analyzed. But the domain is still there, and can be utilized with --xlen. With --xlen we can plot only histograms or only non-histograms.

• --xmin/xmax/x2min/x2max/ymin/ymax/y2min/y2max/zmin/zmax xxx

  Set the range for the given axis. These x-axis bounds are ignored in a streaming plot. The x2/y2-axis bounds do not apply in 3d plots. The z-axis bounds apply only to 3d plots or colormaps. Note that there is no --xrange to set both sides at once or --xinv to flip the axis around: anything more than the basics supported in this option is clearly obtainable by talking to gnuplot, for instance --set 'xrange [20:10]' to set the given inverted bounds.

• --xlabel/x2label/ylabel/y2label/zlabel/cblabel xxx

  Label the given axis. The x2/y2-axis labels do not apply to 3d plots while the z-axis label applies only to 3d plots. The "cblabel" applies to the colorbar, if there is one.

• --x2/--y2/--x1y2/--x2y1/--x2y2 xxx

  By default data is plotted against the x1 and y1 axes (the left and bottom one respectively). If we want a particular curve plotted against a different axis, we can specify that with these options. You pass --AXIS ID where AXIS defines the axis (x2 or y2 or x1y2 or x2y1 or x2y2) and the ID is the curve ID. --x2 is a synonym for --x2y1 and --y2 is a synonym for --x1y2. The curve ID is an ordered 0-based index or a specific ID if --dataid or --vnlog. None of these apply to 3d plots. Can be passed multiple times for different curve IDs, multiple IDs can be passed in as a comma-separated list. By default the curves plotted against the various axes aren not drawn in any differentiated way: the viewer of the resulting plot has to be told which is which via an axes label, legend, colors, etc. Prior to version 1.25 of feedgnuplot the curves plotted on the y2 axis were drawn with a thicker line. This is no longer the case, but that behavior can be brought back by passing something like

  --y2 curveid --style curveid 'linewidth 3'

• --histogram curveID

  Set up a this specific curve to plot a histogram. The bin width is given with the --binwidth option (assumed 1.0 if omitted). If a drawing style is not specified for this curve (--curvestyle) or all curves (--with, --curvestyleall) then the default histogram style is set: filled boxes with borders. This is what the user generally wants. This works with --domain and/or --stream, but in those cases the x-value is used only to cull old data because of --xlen or --monotonic. I.e. the domain values are not drawn in any way. Can be passed multiple times, or passed a comma- separated list

• --xticlabels

  If given, the x-axis tic labels are not numerical, but are read from the data. This changes the interpretation of the input data: with --domain, each line begins with x label ..... Without --domain, each line begins with label .... Clearly, the labels may not contain whitespace. This does not affect the tuple size. This makes sense only without --3d. Please see the guide (https://github.com/dkogan/feedgnuplot/blob/master/guide/guide.org) for usage examples.

• --binwidth width

  The width of bins when making histograms. This setting applies to ALL histograms in the plot. Defaults to 1.0 if not given.

• --histstyle style

  Normally, histograms are generated with the 'smooth frequency' gnuplot style. --histstyle can be used to select different smooth settings (see the gnuplot help smooth page for more info). Allowed values are 'frequency' (the default), 'fnormal' (available in very recent gnuplots), 'unique', 'cumulative' and 'cnormal'. 'fnormal' is a normalized histogram. 'unique' indicates whether a bin has at least one item in it: instead of counting the items, it'll always report 0 or 1. 'cumulative' is the integral of the 'frequency' histogram. 'cnormal' is like 'cumulative', but rescaled to end up at 1.0.

• --style curveID style

  Additional styles per curve. With --dataid, curveID is the ID. Otherwise, it's the index of the curve, starting at 0. curveID can be a comma-separated list of IDs to which the given style should apply. Use this option multiple times for multiple curves. --styleall does not apply to curves that have a --style.

• --curvestyle curveID

  Synonym for --style

• --styleall xxx

  Additional styles for all curves that have no --style. This is overridden by any applicable --style. Exclusive with --with.

• --curvestyleall xxx

  Synonym for --styleall

• --with xxx

  Same as --styleall, but prefixed with "with". Thus

  --with boxes

  is equivalent to

  --styleall 'with boxes'

  Exclusive with --styleall.

• --every curveID factor

  Decimates the input. Instead of plotting every point in the given curve, plot one point per factor. This is useful to quickly process huge datasets. For instance, to plot 1% of the data, pass a factor of 100.

• --everyall factor

  Decimates the input. This works exactly like --every, except it applies to all the curves.

• --using curveID expression

  Specifies a using expression to micromanage the plot. This is a powerful option that allows gnuplot to interpret the input data in arbitrary ways. A using expression tells gnuplot how to map the input columns of data to tuples expected by the plotting style. Please see the "'using' expressions" section above for more detail.

• --usingall expression

  Global "using" expressions. This works exactly like --using, except it applies to all the curves.

• --cmds xxx

  Additional commands to pass on to gnuplot verbatim. These could contain extra global styles for instance. Can be passed multiple times.

• --extracmds xxx

  Synonym for --cmds xxx

• --set xxx

  Additional 'set' commands to pass on to gnuplot verbatim. --set 'a b c' will result in gnuplot seeing a set a b c command. Can be passed multiple times.

• --unset xxx

  Additional 'unset' commands to pass on to gnuplot verbatim. --unset 'a b c' will result in gnuplot seeing a unset a b c command. Can be passed multiple times.

• --image filename

  Overlays the data on top of a raster image given in filename. This is passed through to gnuplot via --equation, and is not interpreted by feedgnuplot other than checking for existence. Usually images have their origin at the top-left corner, while plots have it in the bottom-left corner instead. Thus if the y-axis extents are not specified (--ymin, --ymax, --set 'yrange ...') this option will also flip around the y axis to make the image appear properly. Since this option is just a passthrough to gnuplot, finer control can be achieved by passing in --equation and --set yrange ... directly.

• --equation xxx

  Gnuplot can plot both data and symbolic equations. feedgnuplot generally plots data, but with this option can plot symbolic equations also. This is generally intended to augment data plots, since for equation-only plots you don't need feedgnuplot. --equation can be passed multiple times for multiple equations. The given strings are passed to gnuplot directly without anything added or removed, so styling and such should be applied in the string. A basic example:

  seq 100 | awk '{print $1/10, $1/100}' |
    feedgnuplot --with 'lines lw 3' --domain --ymax 1
                --equation 'sin(x)/x' --equation 'cos(x)/x with lines lw 4'

  Here I plot the incoming data (points along a line) with the given style (a line with thickness 3), and I plot two damped sinusoids on the same plot. The sinusoids are not affected by feedgnuplot styling, so their styles are set separately, as in this example. More complicated example:

  seq 360 | perl -nE '$th=$_/360 * 3.14*2; $c=cos($th); $s=sin($th); say "$c $s"' |
    feedgnuplot --domain --square
                --set parametric --set "trange [0:2*3.14]" --equation "sin(t),cos(t)"

  Here the data I generate is points along the unit circle. I plot these as points, and I also plot a true circle as a parametric equation.

• --equation-below xxx

  Synonym for --equation. These are rendered below all the other data.

• --equation-above xxx

  Like --equation, but is rendered on top of all the other data.

• --square

  Plot data with aspect ratio 1. For 3D plots, this controls the aspect ratio for all 3 axes

• --square-xy

  For 3D plots, set square aspect ratio for ONLY the x,y axes

• --hardcopy xxx

  If not streaming, output to a file specified here. Format inferred from filename, unless specified by --terminal. If --terminal is given, --hardcopy sets only the output filename.

• --terminal xxx

  String passed to 'set terminal'. No attempts are made to validate this. --hardcopy sets this to some sensible defaults if --hardcopy is set to a filename ending in .png, .pdf, .ps, .eps or .svg. If any other file type is desired, use both --hardcopy and --terminal

• --maxcurves N

  The maximum allowed number of curves. This is 100 by default, but can be reset with this option. This exists purely to prevent perl from allocating all of the system's memory when reading bogus data

• --monotonic

  If --domain is given, checks to make sure that the x-coordinate in the input data is monotonically increasing. If a given x-variable is in the past, all data currently cached for this curve is purged. Without --monotonic, all data is kept. Does not make sense with 3d plots. No --monotonic by default. The data is replotted before being purged. This is useful in streaming plots where the incoming data represents multiple iterations of the same process (repeated simulations of the same period in time, for instance).

• --rangesize curveID N

  The options --rangesizeall and --rangesize set the number of values are needed to represent each point being plotted (see "Multi-value style support" above). These options are only needed if unknown styles are used, with --styleall or --with for instance.

  --rangesize is used to set how many values are needed to represent the range of a point for a particular curve. This overrides any defaults that may exist for this curve only.

  With --dataid, curveID is the ID. Otherwise, it's the index of the curve, starting at 0. curveID can be a comma-separated list of IDs to which the given rangesize should apply.

• --tuplesize curveID N

  Very similar to --rangesize, but instead of specifying the range only, this specifies the whole tuple. For instance if we're plotting circles, the tuplesize is 3: x,y,radius. In a 2D plot there's a 1-dimensional domain: x, so the rangesize is 2: y,radius. This dimensionality can be given either way.

• --rangesizeall N

  Like --rangesize, but applies to all the curves.

• --tuplesizeall N

  Like --tuplesize, but applies to all the curves.

• --dump

  Instead of printing to gnuplot, print to STDOUT. Very useful for debugging. It is possible to send the output produced this way to gnuplot directly.

• --exit

  This controls what happens when the input data is exhausted, or when some part of the feedgnuplot pipeline is killed. This option does different things depending on whether --stream is active, so read this closely.

  With interactive gnuplot terminals (qt, x11, wxt), the plot windows live in a separate process from the main gnuplot process. It is thus possible for the main gnuplot process to exit, while leaving the plot windows up (a caveat is that such decapitated windows aren't interactive). There are 3 possible states of the polotting pipeline:

  Alive: feedgnuplot, gnuplot alive, plot window process alive, no shell prompt (shell busy with feedgnuplot)

  Half-alive: feedgnuplot, gnuplot dead, plot window process alive (but non-interactive), shell prompt available

  Dead: feedgnuplot, gnuplot dead, plot window process dead, shell prompt available

  The possibilities are:

  No --stream, all data read in

  no --exit (default)

  Alive. Need to Ctrl-C to get back into the shell

  --exit

  Half-alive. Non-interactive prompt up, and the shell accepts new commands. Without --stream the goal is to show a plot, so a Dead state would not be useful.

  --stream, all data read in or the feedgnuplot process terminated

  no --exit (default)

  Alive. Need to Ctrl-C to get back into the shell. This means that when making live plots, the first Ctrl-C kills the data feeding process, but leaves the final plot up for inspection. A second Ctrl-C kills feedgnuplot as well.

  --exit

  Dead. No plot is shown, and the shell accepts new commands. With --stream the goal is to show a plot as the data comes in, which we have been doing. Now that we're done, we can clean up everything.

  Note that one usually invokes feedgnuplot as a part of a shell pipeline:

  $ write_data | feedgnuplot

  If the user terminates this pipeline with ^C, then all the processes in the pipeline receive SIGINT. This normally kills feedgnuplot and all its gnuplot children, and we let this happen unless --stream and no --exit. If --stream and no --exit, then we ignore the first ^C. The data feeder dies, and we behave as if the input data was exhausted. A second ^C kills us also.

• --geometry

  Specifies the size, position of the plot window. This applies only to the x11 gnuplot terminal, and has no effect otherwise. To control the window size for any other terminal, ask for the terminal explicitly, with the options specifying the size. For instance --terminal 'qt size 1024,768'

• --version

  Print the version and exit

RECIPES

For a tutorial and a gallery please see the guide at https://github.com/dkogan/feedgnuplot/blob/master/guide/guide.org

Basic plotting of piped data

$ seq 5 | awk '{print 2*$1, $1*$1}'
2 1
4 4
6 9
8 16
10 25

$ seq 5 | awk '{print 2*$1, $1*$1}' |
  feedgnuplot --lines --points --legend 0 "data 0" --title "Test plot" --y2 1

Realtime plot of network throughput

Looks at wlan0 on Linux.

$ while true; do sleep 1; cat /proc/net/dev; done |
  gawk '/wlan0/ {if(b) {print $2-b; fflush()} b=$2}' |
  feedgnuplot --lines --stream --xlen 10 --ylabel 'Bytes/sec' --xlabel seconds

Realtime plot of battery charge in respect to time

Uses the result of the acpi command.

$ while true; do acpi; sleep 15; done |
  perl -nE 'BEGIN{ $| = 1; } /([0-9]*)%/; say join(" ", time(), $1);' |
  feedgnuplot --stream --ymin 0 --ymax 100 --lines --domain --xlabel 'Time' --timefmt '%s' --ylabel "Battery charge (%)"

Realtime plot of temperatures in an IBM Thinkpad

Uses /proc/acpi/ibm/thermal, which reports temperatures at various locations in a Thinkpad.

$ while true; do cat /proc/acpi/ibm/thermal | awk '{$1=""; print}' ; sleep 1; done |
  feedgnuplot --stream --xlen 100 --lines --autolegend --ymax 100 --ymin 20 --ylabel 'Temperature (deg C)'

Plotting a histogram of file sizes in a directory, granular to 10MB

$ ls -l | awk '{print $5/1e6}' |
  feedgnuplot --histogram 0
    --binwidth 10
    --ymin 0 --xlabel 'File size (MB)' --ylabel Frequency

Plotting a live histogram of the ping round-trip times for the past 20 seconds

$ ping -D 8.8.8.8 |
  perl -anE 'BEGIN { $| = 1; }
             $F[0] =~ s/[\[\]]//g or next;
             $F[7] =~ s/.*=//g    or next;
             say "$F[0] $F[7]"' |
  feedgnuplot --stream --domain --histogram 0 --binwidth 10 \
              --xlabel 'Ping round-trip time (s)'  \
              --ylabel Frequency --xlen 20

Plotting points on top of an existing image

This can be done with --image:

$ < features_xy.data
  feedgnuplot --points --domain --image "image.png"

or with --equation:

$ < features_xy.data
  feedgnuplot --points --domain
    --equation '"image.png" binary filetype=auto flipy with rgbimage'
    --set 'yrange [:] reverse'

The --image invocation is a convenience wrapper for the --equation version. Finer control is available with --equation.

Here an existing image is given to gnuplot verbatim, and data to plot on top of it is interpreted by feedgnuplot as usual. flipy is useful here because usually the y axis points up, but when looking at images, this is usually reversed: the origin is the top-left pixel.

ACKNOWLEDGEMENT

This program is originally based on the driveGnuPlots.pl script from Thanassis Tsiodras. It is available from his site at http://users.softlab.ece.ntua.gr/~ttsiod/gnuplotStreaming.html

REPOSITORY

https://github.com/dkogan/feedgnuplot

AUTHOR

Dima Kogan, <dima@secretsauce.net>

LICENSE AND COPYRIGHT

Copyright 2011-2021 Dima Kogan.

This program is free software; you can redistribute it and/or modify it under the terms of either: the GNU General Public License as published by the Free Software Foundation; or the Artistic License.

See http://dev.perl.org/licenses/ for more information.