From Apertium
Revision as of 19:00, 18 December 2013 by Sushain (talk | contribs) (add to usage and installation)
Jump to navigation Jump to search

Apertium-APy stands for "Apertium API in Python". It's a simple apertium API server written in python, meant as a drop-in replacement for ScaleMT. It is currently found in the svn under trunk/apertium-tools/apertium-apy, where servlet.py is basically its entirety. This is meant for front ends like the simple one in trunk/apertium-tools/simple-html (where index.html is the main deal).


First, compile and install apertium/lttoolbox/apertium-lex-tools, and compile your language pairs. See Minimal_installation_from_SVN for how to do this. Then

svn co https://svn.code.sf.net/p/apertium/svn/trunk/apertium-tools/apertium-apy
cd apertium-apy
export APERTIUMPATH="/path/to/apertium/svn/trunk"
./servlet.py "$APERTIUMPATH"

Optional arguments include:

  • --langNamesDB: path to database of localized language names
  • -port --port: port to run server on (2737 by default)
  • --ssl: path to SSL certificate


URL Parameters Example
/listPairs - List available pairs None
$ curl 'http://localhost:2737/listPairs'
/list - List available mode information
  • q: type of information to list
    • pairs (alias for /listPairs)
    • analyzers/analysers
    • generators
    • taggers/disambiguators
$ curl 'http://localhost:2737/list?q=analyzers'
$ curl 'http://localhost:2737/list?q=generators'
$ curl 'http://localhost:2737/list?q=taggers'
/translate - Translate text
  • langpair: language pair to use for translation
  • q: text to translate
$ curl 'http://localhost:2737/translate?langpair=kaz|tat&q=Сен+бардың+ба?'
/analyze - Morphologically analyze text
  • mode: language to use for analysis
  • q: text to analyze
$ curl --data "mode=kaz&q=Сен+бардың+ба?" http://localhost:2737/analyze
[["Сен/сен<v><tv><imp><p2><sg>/сен<prn><pers><p2><sg><nom>","Сен "],
["бардың ба/бар<adj><subst><gen>+ма<qst>/бар<v><iv><ifi><p2><sg>+ма<qst>","бардың ба"],
/generate - Generate surface forms from text
  • mode: language to use for generation
  • q: text to generate
$ curl --data "mode=kaz&q=^сен<v><tv><imp><p2><sg>$+^сен<v><tv><imp><p2><pl>$" http://localhost:2737/generate
[["сен ","^сен<v><tv><imp><p2><sg>$ "],["сеніңдер","^сен<v><tv><imp><p2><pl>$"]]
/perWord - Perform morphological tasks per word
  • language: language to use for tasks
  • modes: morphological tasks to perform on text
    • tagger/disambig
    • biltrans
    • translate
    • biltrans+morph (in any order)
    • translate+tagger (in any order)
    • morph+tagger/morph+disambig (in any order)
  • q: text to perform tasks on


== Threading ==
Currently it uses TCPServer inheriting ThreadingMixIn. A lock on translateNULFlush (which has to have at most one thread per pipeline) ensures that part stays single-threaded (to avoid Alice getting Bob's text).

===Try it out===

Try testing with e.g.
    export APERTIUMPATH="/path/to/svn/trunk"
    python3 servlet "$APERTIUMPATH" 2737 &
    curl -s --data-urlencode 'langpair=nb|nn' --data-urlencode \
    'q@/tmp/reallybigfile' 'http://localhost:2737/translate' >/tmp/output &
    curl 'http://localhost:2737/translate?langpair=nb|nn&q=men+ikke+den'
    curl 'http://localhost:2737/translate?langpair=nb|nn&q=men+ikke+den'
    curl 'http://localhost:2737/translate?langpair=nb|nn&q=men+ikke+den'
And see how the last three (after a slight wait) start outputting before the first request is done.

===Morphological Analysis and Generation===

To analyze text, send a POST or GET request to <code>/analyze</code> with parameters <code>mode</code> and <code>q</code> set. For example: 

    $ curl --data "mode=kaz&q=Сен+бардың+ба?" http://localhost:2737/analyze
    [["Сен/сен<v><tv><imp><p2><sg>/сен<prn><pers><p2><sg><nom>","Сен "],["бардың ба/бар<adj><subst><gen>+ма<qst>/бар<v><iv><ifi><p2><sg>+ма<qst>","бардың ба"],["?/?<sent>","?"],["./.<sent>",".\n"]]

The JSON response will consist of a list of lists each of form <code>[analysis with following non-analyzed text*, original input token]</code>. To receive a list of valid analyzer modes, send a request to <code>/listAnalyzers</code>.

To generate surface forms from an analysis, send a POST or GET request to <code>/generate</code> with parameters <code>mode</code> and <code>q</code> set. For example: 

    $ curl --data "mode=kaz&q=^сен<v><tv><imp><p2><sg>$+^сен<v><tv><imp><p2><pl>$" http://localhost:2737/generate
    [["сен ","^сен<v><tv><imp><p2><sg>$ "],["сеніңдер","^сен<v><tv><imp><p2><pl>$"]]

The JSON response will consist of a list of lists each of form <code>[generated form with following non-analyzed text*, original lexical unit input]</code>. To receive a list of valid generator modes, send a request to <code>/listGenerators</code>.

* e.g. whitespace, superblanks

To test with a self-signed signature:
openssl req -new -x509 -keyout server.pem -out server.pem -days 365 -nodes

Then run with --ssl server.pem, and test with https and the -k argument to curl (-k means curl accepts self-signed or even slightly "lying" signatures):

curl -k --data "mode=kaz-tat&q=Сен+бардың+ба?" https://localhost:2737/analyze

If you have a signed signature for e.g. apache, it's likely to be split into two files, one .key and one .crt. You can cat them together into one to use with servlet.py:

cat server.key server.crt > server.keycrt

Now you should be able to use curl without -k for the domain which the certificate is signed for:

curl --data "mode=kaz-tat&q=Сен+бардың+ба?" https://oohlookatmeimencrypted.com:2737/analyze

Remember to open port 2737 to your server.


  • It should be possible to set a time-out for translation threads, so if a translation is taking too long, it gets killed and the queue moves along.
  • It should use one lock per pipeline, since we don't need to wait for mk-en just because sme-nob is running.
  • http://stackoverflow.com/a/487281/69663 recommends select/polling over threading (http://docs.python.org/3.3/library/socketserver.html for more on the differences) but requires either lots of manually written dispatching code (http://pymotw.com/2/select/) or a framework like Twisted.
  • some language pairs still don't work (sme-nob?)
  • hfst-proc -g doesn't work with null-flushing (or?)