This page contains some frequently asked questions about PyEnchant, along with some questions which haven’t technically been asked but which are probably of interest to many readers.
You can use the issue reporting tools on the github project page .
Like the underlying Enchant library, PyEnchant is made available under the GNU LGPL with a special exemption allowing linking with proprietary spellchecking plugins.
There are several projects to produce Python bindings for the aspell spellchecking library, including aspell-python and pyaspell. The myspell-python package offers a Python wrapper for the MySpell engine. It is also possible to invoke a command-line program such as ispell, as shown in this ASPN Python Recipie.
Most of the solutions listed above are tied to a single spellchecking platform, such as aspell or MySpell. By contrast, Enchant supports multiple spellchecking platforms. A good discussion of why this is an advantage can be found on the Enchant website under the heading “Enchant and Multiple Backends”.
Different backends can be used for different languages, depending on which does a better job
Integration with the user’s “native” spellchecker, whatever that may be
This flexibility is transparent to the application programmer
As explained above, PyEnchant is available under the GNU LGPL. This may mean it can be used in some projects where other libraries (such as GPL-licensed libraries) cannot.
The Enchant API is also generally simpler than that provided by other spellchecking solutions. This can be an advantage or disadvantage depending on the needs of your program.
As of version 1.1, PyEnchant supports input of both standard Python strings and unicode strings. It follows the standard Python practice of returning unicode strings when unicode strings are given, and standard strings when standard strings are given.
As of version 1.5, PyEnchant is compatible with the new string infrastructure of Python 3, and uses it unicode strings consistently throughout. It does not accept bytestrings in Python 3.
Unfortunately, the author (a native English speaker) does not have a great deal of experience with the use of unicode. Any help with testing the support for unicode input/output would be greatly appreciated.
The simple answer is: whichever works best for your language of choice. The source distribution contains a tool to conduct a Provider Shootout given an appropriate set of test data, which can be used to determine which provider performs the best for your language.
Ideally, this choice should be made by the system administrator when enchant is installed. One of the premises of the Enchant library is to relieve the user from making such low-level choices.
See the Provider Ordering section of the tutorial.
If you are using
nuspel providers, then Enchant will
<user_config_dir>/<provider> for additional dictionaries.
user_config_dir is set by a call to
ENCHANT_CONFIG_DIRenvironment variable is set, it will return its value
Otherwise, it will call g_get_user_config_dir() which will return something like ~/.config/enchant.