This is a pre-production deployment of Warehouse. Changes made here affect the production instance of PyPI (
Help us improve Python packaging - Donate today!

Do you speak London?

Project Description
# Do you speak London?

Command line tool for naturla language identification, also known as langID. Currently supporting 4 languages only, English, Spanish, Portuguese and Arabic:

`$ python Can you tell if this is in English?`

You can let dysl read from custom training data with your own set of languages.

`$ python --corpus /PATH/TO/TRAINING/DATA ¿Puede usted decir si esto es en español?`

You also can use the CLI to add more samples to your training data.

`$ python --corpus /PATH/TO/TRAINING/DATA --lang en Here you are some text in English to be add to your training data`

To show the currently installed version:

`$ python --version`

For listing of all CLI options:

`$ python --help`

## Use in Code

You can also use dysl's LangID within your code

from dysl.langid import LangID

l = LangID()

text = u'hello, world'

lang = l.classify(text)
print text, 'Language:', lang
## Use your own Training Data

As stated earlier, we currently support English, Spanish, Portuguese and Arabic out of the box, however, this doesn't stop you from using your own data to train dysl. Thus, you can even use it to classify your choice of fictional languages such as George Orwell's Newspeak, Valyrian Languages, and The Klingon language if you have the needed data.

For compatibility with [](, we require the training data to be in the following format:

> To train a model, we require multiple corpora of monolingual documents. Each document should be a single file, and each file should be in a 2-deep folder hierarchy, with language nested within domain. For example, we may have a number of English files:

`./corpus/domain1/en/File1.txt ./corpus/domainX/en/001-file.xml`

We, however do not rely on domains as they do in, so you can use a single domain if you want to.

In brief, if you want to classify Valyrian and Klingon for example, your corpus should look somehow as follows:



As you can see, domain and filename can be anything, just the folders containing the example files should be named after the languages you want to classify.

Finally, you need to specify the path to the training example in your code.

from dysl.langid import LangID

l = LangID()

text = u'hello, world'

lang = l.classify(text)
print text, 'Language:', lang
## Adding New Samples to Your Training Data

You can add new training samples to your custom training-set.
You do that on two stages.

To add a new samples:

`l.add_training_sample(text=u'tlhIngan Hol Dajatlha?', lang='Klingon')`

`l.add_training_sample(text=u'jIyajbe', lang='Klingon')`

Then to save changes to disk:



By default, a folder name generated from the current timestamp, `/batchTSYYMMDDHHMMSS` is used for the domain, and `file.txt` is used for sample files in language folders. To specify the domain folder name and file name, you can use the following command instead:

`l.save_training_samples(domain='MyDomain', filename='MyFile.txt')`

Notice that saving new sample data only works when using custom training-set,
adding samples to builtin training-set is not permitted.

# Retrain When Modified

Let's say you are going to build a daemon that calls uses dysl. Normally, you would train dysl when starting then use that data for any further language classifications.

What if some other process, or human, updates the training dataset during that? How will your daemon know that it neads to retrain again?

Well, in LangID, there is a method, `is_training_modified()`, that can tell you if the training data was modified after it was first trained on it or not:


It returns `True` if the data was modified, and `False` otherwise. Of course, when using the builtin dataset, it will always return `False` as it not allowed to be modified.

## Unknown Languages (_Experimantal_)

By default, we try to classify a given text as one of the languages dysl is trained on. However, by setting `unk=True` we allow dysl to return 'unk' when the given text is not in any of the languages it has seen before.

`l = LangID(unk=True)`


# Contacts

+ Project Owner: [Tarek Amr](
Release History

Release History

This version
History Node


History Node


History Node


History Node


Download Files

Download Files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
dysl-0.13.tar.gz (1.8 MB) Copy SHA256 Checksum SHA256 Source Oct 30, 2014

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting