An advanced image conversion server and command line tool.
IMPORTANT: For documentation on the old 1.8.x version see here.
- Copyright Notes
- thumbnail generation
- responsive images
- intelligent cropping
- art direction use cases
- adaptive image quality
- no need to pre-generate different image sizes or versions
npm install [-g] converjon
If you want to prevent the fixed dependency versions of the
npm-shrinkwrap.json file to be used on install use
--no-shrinkwrap as a command line argument. See shrinkwrap docs and install docs for more information on this.
Conversion follows Semantic Versioning.
Start the server with
converjon [--config your_config_file] or use the command line utility
converjon-cli to work on local files.
Let's say you have an image at
http://example.org/image.jpg. To get the image through Converjon, put the original URL into the request as a URL encoded parameter:
Another alternative is to specify a file from the server's file system as the source. Instead of
url, you can add the
file parameter for this:
file parameter is URL-encoded. In plain text it is
foobar is the name of a filesystem alias followed by a colon and a relative path. See Configuration: Aliases for more details.
More options are available as GET parameters. All parameters need to be URL encoded.
Several examples are available on the
/demo page which is enabled when starting Converjon with the development config file via
It's recommended to set the server port to
80 in configuration and to run Converjon on a separate subdomain of your site or behind a reverse proxy like Nginx or Varnish.
For information on how to use the command line tool, run
You can either supply a
height or both. If you only specify one dimension, the other one will be derived from the original image's aspect ratio.
If you set both values, the image may be cropped to the new aspect ratio and then resized to the requested pixel dimensions.
Area of Interest
By default images are cropped from the center of the original. You can specify an "area of interest" with the
aoi parameter. The AOI is a rectangle in the following format:
The AOI can also be embedded in the original image's metadata via EXIF or IPTC. The name of this metadata field can be configured and defaults to
aoi. The request parameter overrides the AOI value from the image's metadata.
By default, the AOI in the URL parameters has precedence over the one from the image's metadata. To prefer the embedded AOI from the metadata, set the
prefer_embedded_aoi parameter to any non-empty value.
crop parameter sets the cropping mode. Available modes are:
If an AOI is set, cropping will ensure, that the area is always preserved.
padding_color parameter sets the background color of the padding.
The color should be specified in an HTML URL Encoded format.
If none specified, the default color set in the configuration will be used. see also Converter config
format parameter you can change the format of the image. Supported formats are:
If no specific format is requested, the format of the source image will be used.
quality parameter sets the JPEG quality value. It ranges from
This parameter is ignored, if the requested format is not JPEG.
colors parameter sets the number of colors for GIF compression. It ranges from 2 to 256 (integer).
interlace parameter allows the creation of interlaced images. Supported types of interlacing scheme are:
plane(try this for JPEGs)
A well-known example of interlaced images are progressive JPEGs. You can use this option with PNGs and GIFs as well.
preset parameter allows the automatic usage of a preset of parameters.
New presets may be added in the configuration files as follows:
presets:thumbnail:format: "jpg"quality: 50width: 100hight: 100
So instead of specifying all the parameters in the URL:
you can use the preset:
strip_metadata option removes all metadata (e.g. EXIF, IPTC) from the converted images. This option has no value, it just needs to be present in the URL query parameters.
/status leads to a summary of Converjon's current state and some statistics.
The status page is available as content type
When launching converjon, you can specify one or more configuration files with the
--config option which can be set multiple times to load multiple config files:
converjon --config conf_file1.yml --config conf_file2.json
To load a directory containing one or more config file, use the
--config-dir option. Config files will be added in the order they appear in that directory.
converjon --config-dir /etc/converjon
The default configuration format is YAML but you can also use JSON files.
Every configuration file matches certain image source URLs. If a config file contains a
urls setting, that configuration will only apply to URLs that match at least one of the patterns from that list. Config files without the
url setting apply to all images.
Some configuration are automatically converted:
- "true" (string) is treated as
- "false" (string) is treated as
# this config will only apply to source URLs from localhost or flickrurls:- "*" #this will match URLs on localhost, this is the dev/testing default- "*"
Converjon uses calmcard for string pattern matching. Documentation on how these patterns work can be found there.
This way you can define different settings depending on the source of the requested images.
server: port: 8000 instance_name: null timeout: 20000 send_timeout: 2000 #time that sending the repsonse data may take. access_log_format: "combined" base_url_path: "/" enable_load_test: false
server.port: port for the server to listen on
server.instance_name: the name of this server that will be displayed on the status page
If not set, a random name will be generated.
server.timout: global timeout for incoming requests
server.send_timeout: the time streaming a finished image file into the HTTP response may take. This usually shouldn't need to be changed, except when the servers file system is expected to be exceptionally slow.
server.base_url_path: Normally, Converjon's image URL paths just start with
/, like in
You can set a base path to have better control over the URLs that Converjon uses. If you want the image URLs too like
server.access_log_format: the formatting of access logs:
combined: Apache Combined Log Format (the default)
short: leaves out the date/time information. Use this, if you use other software for log handling that adds timestamps.
download.url_whitelist sets list of URL patterns that are allowed to be requested as image sources.
For example, if you host your source images on
http://myimages.com/images/... you should set the whitelist pattern to
http://myimages.com/images/ to make sure other sources are not allowed.
# this will only allow requests for images from URLs that match these patternsdownload:url_whitelist:- "*"- "*
Calmcard patterns are used for matching by default.
You can also prefix the pattern with
~ ^http://(foo|bar)\.example.org\/.*) to use regular expressions. For most cases, this should not be necessary and is discouraged.
download.timeout sets a timeout after which requests are cancelled, if the source server doesn't respond in time.
Reject Invalid SSL Certificates
true (default) will cause sources to be rejected, if their SSL certificates cannot be validated.
To access the server's local filesystem for source images, you can specify "aliases":
#example from the development configuration files:alias: devbase_file_path: "test/resources/images"fallback_base_file_paths:- "test/resources/photos"- "test/resources/misc"headers:Cache-Control: "max-age=5"
When an image is requested with a
file parameter (
<alias>:<path>) instead of
alias part of that parameter is matched against the configuration files, just like with URLs but only the configs that have that exact alias will be used for that request.
There can only be one alias per config file. If you need multiple aliases, you need to have one config file for each of them. See here on how to load multiple config files.
In a config file with an alias you can set a
base_file_path. This is the directory where your source images are located. It is concatenated with the
path part of the
file parameter to point to the actual file. The
base_file_path can be an absolute path or relative to the working directory of the server.
Additionally there can be multiple
fallback_base_file_paths. If the requested file can't be found in the
base_file_path, Converjon will try to find it in the first fallback path, then the second, and so on.
In addition, you can set HTTP headers that will be sent along with the converted image as if they had come from a source server.
Converjon supports HTTP basic authentication for image sources. If you include authentication credentials in a URL specific configuration file, they will be sent along with every request to URLs that match the URL pattern of that configuration file.
authentication:username: "testuser"password: "testpass"
cache.basepath sets the base directory for the local file cache.
cache.copy_source_file determines, if a source file should be copied, when you're using a local file as source. By
default, files are copied into Converjon's cache directory. Set this to
false to use the original file's location.
You will have to make sure that these files are not changed while Converjon is using them. This could result in corrupted images or failing requests.
The cache directory is not automatically cleaned up and may grow over time.
Converjon can clean up the cache directory automatically. By default this is disabled. It can be enabled with the
garbage_collector config directives:
garbage_collector:enabled: truesource: "cache"target: "immediate"interval: 5000
enabled: turns the garbage collector on or off
source: determines when source files should be cleaned up. Possible values are:
cache: The file will be removed when it's cache lifetime has expired
immediate: The file will be removed as soon as it's no longer in use by any pending request.
- any other value will disable the cleanup for source files
target: same as
sourcebut for the converted target image files
interval: time between garbage collector runs, in milliseconds.
Local source files that were not copied into the cache directory will not be removed by the garbage collector.
process.limit sets the maximum number of child processes that converjon will spawn for converting and analyzing images.
Increasing this will likely increase memory consumption while providing better usage of multiple CPU cores.
converter.padding_color sets the background color that is used, if an image needs padding to fit the requested aspect ratio.
cropping.default_mode sets the default mode for cropping images. Possible options are:
Constraints can be used to limit the possible request parameters, like width and height of images. Every constraint has a
min and a
constraints:quality:min: 1max: 100colors:min: 1max: 256width:min: 1max: 10000height:min: 1max: 10000
There are three logging levels:
debug. Each of them can be directed to either STDOUT, STDERR or into a log file.
logging:error: "stderr" # will log errors to STDERRdebug: "stdout" # will log debug logs to STDOUTaccess: "/var/log/access.log" # will log requests into a log file
To disable a log level, set it to
logging:error: "stderr"debug: false # "false" as a string will also workaccess: "/var/log/access.log"
Logs are not automatically rotated. Use of a tool like
logrotate is recommended.
Execute tests with
npm test. Please note, that you need to have all dependencies installed.
CHANGELOG.md for more information about changes.
Converjon is licensed under the MIT license.
The "sparrow" testing image is © Leon Weidauer, permission to use it for testing is granted.