$ pip3 install logplot
In order to plot interesting graphs from custom logs, logplot
needs to know what log entries are considered interesting.
This can be done by defining entries
and special_entries
sections in the config file.
The main purpose of entries
is to plot trends, i.e. connected points. This is useful e.g. for plotting state machine states.
special_entries
is for plotting single, unconnected points, such as errors.
Each entry in entries
and special_entries
should have the following:
identifier
: string to be searched from log entriesvalue
: y-axis value in the plot
Ideally, one entry in entries
should have initial_state: true
. This is used for separating plotted series from each other.
You can define specific marker style for each entry in special_entries
, refer to matplotlib
docs for available options.
In addition to the entries
and special-entries
sections, user can override default settings (see src/logplot/default_conf.yaml) in general
section.
For example, it's the place where you can define the command to be used for opening the log in your favorite editor.
See doc/example/user_conf.yaml and corresponding log file (doc/example/log.txt) for full example.
$ logplot path/to/log -c path/to/conf/file
For example, this:
$ logplot doc/examples/log.txt -c doc/examples/user_conf.yaml
yields this:
Clicking a data point in the plot opens the log file in the correct line number in the editor defined in the configuration file.
If the log_open_cmd
is not configured, the log will be opened with a default program defined by the OS.
Note that log_open_cmd
has to configured to be able to open the log in the correct line number.
logplot
will automatically identify different trends in the plotted entries.
You can toggle the visibility of these trends by clicking items in the legend.
Here's an example of a bit bigger log which contains 100k entries, 13 different states, and 15 different trends.
$ logplot doc/examples/log_huge.txt -c doc/examples/user_conf_huge.yaml
And the same with three most occuring trends filtered out.