Command line

You can also run PlantUML using the following command :

java -jar plantuml.jar file1 file2 file3

This will look for @startXXX into file1, file2 and file3. For each diagram, a .png file will be created.

For processing a whole directory, you can use:

java -jar plantuml.jar "c:/directory1" "c:/directory2"

This command will search for @startXXX and @endXXX into .c, .h, .cpp, .txt, .pu, .tex, .html, .htm or .java files of the c:/directory1 and c:/directory2 directories.

Wildcards

You can also use wildcards :

For a single character, use ?
For zero or more characters, use *
For zero or more characters, (including / or \), use a double **

So to process any .cpp files in all directories starting by dummy :

java -jar plantuml.jar "dummy*/*.cpp"

And to process any .cpp files in all directories starting by dummy, and theirs subdirectories :

java -jar plantuml.jar "dummy*/**.cpp"

Excluded files

You can exlude some files from the process using the -x option:

java -jar plantuml.jar -x "**/common/**" -x "**/test/Test*" "dummy*/**/*.cpp"

Output Directory

You can specify an output directory for all images using the -o switch:

java -jar plantuml.jar -o "c:/outputPng" "c:/directory2"

If you recurse into several directory, there is a slight difference if you provide an absolute or a relative path for this output directory.

absolute path: An unique output directory will be used, with all images generated inside
relative path: Several output directories will be used across the file tree.

Configuration File

You can also provide a configuration file which will be included before each diagram:

java -jar plantuml.jar -config "./config.cfg" dir1

Metadata

After all preprocessing (includes etc), PlantUML saves the diagram's source code in the generated PNG Metadata in the form of encoded text.

It is possible to retrieve this source with the -metadata option. This means that the PNG is almost "editable": you can post it on a corporate wiki where you cannot install plugins, and someone in the future can update the diagram by getting the metadata, editing and re-uploading again. Also, the diagram is stand-alone.
Conversely, the -checkmetadata option checks whether the target PNG has the same source and if there are no changes, doesn't regenerate the PNG, thus saving all processing time. This allows you to run PlantUML on a whole folder (or tree with the -recursive option) incrementally.

Sounds like magic! No, merely clever engineering :-)

Example:

  java -jar plantuml.jar -metadata diagram.png > diagram.puml

Unfortunately this option works only with local files. It doesn't work with -pipe so you cannot fetch a URL with eg curl and feed the PNG to PlantUML.

However, the Plantuml server has a similar feature, where it can get a PNG from a URL and extract its metadata.

Exit code

When there are some errors in diagrams the command returns an error (-1) exit code. But even if some diagrams contain some errors, all diagrams are generated, which can be time consuming for large project.

You can use the -failfast flag to change this behavior to stop diagram generations as soon as one error occurs. In that case, some diagrams will be generated, and some will not.

There is also a -failfast2 flag that does a first checking pass. If some error is present, no diagram will be generated at all. In case of error, -failfast2 runs even faster than -failfast, which may be useful for huge project.

Standard Input & Output

Using the -pipe option, you can easily use PlantUML in your scripts.

With this option, a diagram description is received through standard input and the png file is generated to standard output. No file is written on the local file system.

Example:

cat somefile.puml | java -jar plantuml.jar -pipe > somefile.png

The --pipemap option can be used to generate PNG map data (hyperlink rectangles) for use in HTML, eg:

cat somefile.puml | java -jar plantuml.jar -pipemap > somefile.map

The map file looks like this:

<map id="plantuml_map" name="plantuml_map">
<area shape="rect" id="id1" href="http://plantuml.com" title="http://plantuml.com"
      alt="" coords="1,8,88,44"/>
</map>

Note: Also take a look at -pipedelimitor and -pipeNoStderr to implement proper multiplexing of several PNG in a stream (in case the puml file contains multiple diagrams), and error handling.

Help

You can have a help message by launching :

java -jar plantuml.jar -help

This will output:

Usage: java -jar plantuml.jar [options] -gui
	(to execute the GUI)
	or java -jar plantuml.jar [options] [file/dir] [file/dir] [file/dir]
	(to process files or directories)

You can use the following wildcards in files/dirs:
	*	means any characters but '\'
	?	one and only one character but '\'
	**	means any characters (used to recurse through directories)

where options include:
	-DVAR1=value                   Set a preprocessing variable as if '!define VAR1 value' were used
	-I/path/to/file                Include file as if '!include file' was used
	-I/path/to/*.puml              Include files with pattern
	-Sparam1=value                 Set a skin parameter as if 'skinparam param1 value' were used
	-author[s]                     Print information about PlantUML authors
	-charset xxx                   Use a specific charset (default is windows-1252).
	                               Use "utf-8" for extra international chars
	-checkmetadata                 Skip PNG files that don't need to be regenerated
	-checkonly                     Check the syntax of files without generating images
	-checkversion                  Check if a newer version is available for download
	-computeurl|-encodeurl         Compute the encoded URL of a PlantUML source file
	-config "file"                 Include a file before each diagram (same as -I ??)
	-cypher                        Cypher texts of diagrams so that you can share them
	-decodeurl                     Retrieve the PlantUML source from an encoded URL
	-disablestats                  Disable statistics computation (default)
	-duration                      Print the duration of complete diagrams processing
	-[exclude|x] pattern           Exclude files that match the provided pattern
	-enablestats                   Enable statistics computation
	-encodesprite 4|8|16[z] "file" Encode a sprite at gray level (z for compression) from an image
	-extractstdlib                 Extract PlantUML Standard Library into stdlib folder
	-failfast                      Stop processing as soon as a syntax error in diagram occurs
	-failfast2                     Do a first syntax check before processing files, to fail even faster
	-filename \"example.puml\"     Override %filename% variable
	-graphvizdot "exe"             Specify the graphviz "dot" executable.
	                               If needed, you can setup the environment variable GRAPHVIZ_DOT.
	-gui                           Run the graphical user interface
	-h[elp]                        Display this help message
	-htmlstats                     Output general statistics in file plantuml-stats.html
	-keepfiles                     Do NOT delete temporary files after processing
	-language                      Print the list of PlantUML keywords
	-loopstats                     Continuously print statistics about usage
	-metadata                      Retrieve PlantUML sources from PNG images
	-nbthread N                    Use N threads for processing
	-nbthread auto                 Use 4 threads for processing
	-nosuggestengine               Disable the suggest engine when errors in diagrams
	-o[utput] "dir"                Generate images in the specified directory
	-overwrite                     Allow overwriting of read-only files
	-p[ipe]                        Use stdin for PlantUML source and stdout for PNG/SVG/EPS generation
	-pattern                       Print the list of Regular Expression used by PlantUML
	-pipeNoStderr                  When used with -pipe, print errors to stdout (no image is generated in that case).
	                               The first line is "ERROR" and it respects -pipedelimitor
	-pipedelimitor "xxx"           When used with -pipe, append "xxx" after every image
	-pipeimageindex N              Generate only the N-th image when used with -pipe
	-pipemap                       Generate PNG map data of hyperlink rectangles, for use in HTML
	-preproc                       Output the preprocessed text of diagrams
	-printfonts                    Print fonts available on your system
	-progress                      Display a textual progress bar at the console
	-quiet                         Do NOT print error messages to the console
	-r[ecurse]                     Recurse through directories
	-realtimestats                 Generate statistics on the fly rather than at the end
	-splash                        Display a splash screen with some progress bar
	-stdlib                        Print standard library info
	-syntax                        Report any syntax error from standard input without generating images
	-teps                          Generate images using EPS format
	-testdot                       Test the installation of graphviz
	-thtml                         Generate HTML file for class diagram
	-tlatex                        Generate images using LaTeX/Tikz format
	-tlatex:nopreamble             Generate images using LaTeX/Tikz format without preamble
	-tpdf                          Generate images using PDF format
	-tpng                          Generate images using PNG format (default)
	-tscxml                        Generate SCXML file for state diagram
	-tsvg                          Generate images using SVG format
	-ttxt                          Generate images with ASCII art
	-tutxt                         Generate images with ASCII art using Unicode characters
	-tvdx                          Generate images using VDX format
	-txmi                          Generate XMI file for class diagram
	-v[erbose]                     Have log information
	-version                       Display information about PlantUML and Java versions
	-xmlstats                      Output general statistics in file plantuml-stats.xml