Pre-translate technical terms in lmh translate

[tkw1536]

Currently, lmh translate generated-subring.en.tex de translates

Let $R$ be a \trefi[ring]{ring}.

into

\ttl{Let} $R$ \ttl{be a} \mtrefi[ring?ring]{\ttl{ring}}.

This can be improved given that ring.en.tex contains \defi{ring} and ring.en.tex contains \def[ring]{Ring}, which induces the translation mapping ring-->Ring, with which we can improve the translation to

\ttl{Let} $R$ \ttl{be a} \mtrefi[ring?ring]{Ring}.

and thus reduce the translation effort by a third.

I think that the pre-translation should be done in collaboration between between lmh and MMT, where MMT computes the translation mapping for a given glossary component (we kwow all the \gimport) and lmh applies it. So @m-iancu should help, maybe Enxhell as well.

[tkw1536]

@m-iancu I belive this is what we were talking about today. I have implemented this on dev, please test.

[tkw1536]

@eluzhnica Please also take a look. You should get the exact syntax via "lmh translate --help"

[m-iancu]

Latest version of dev crashes on lmh translate defeq.en.tex al with:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/local/lib/python2.7/dist-packages/lmh_core/__init__.py", line 289, in main
    run_lmh(install)
  File "/usr/local/lib/python2.7/dist-packages/lmh_core/__init__.py", line 88, in run_lmh
    s.main()
  File "/var/data/localmh/bin/lmh", line 11, in main
    lmh.run(sys.argv[1:])
  File "/var/data/localmh/bin/../lmh/__init__.py", line 128, in run
    if main(argv):
  File "/var/data/localmh/bin/../lmh/__init__.py", line 122, in main
    return submods[args.action].do(args)
  File "/var/data/localmh/bin/../lmh/commands/translate.py", line 82, in do
    ret = transmod(ofn[0], ofn[1], lang, pre_terms = args.terms) and ret
  File "/var/data/localmh/bin/../lmh/lib/modules/translate.py", line 217, in transmod
    content = re.sub(r"\\ttl\{([^\}]*)\}", replacer2, content)
  File "/usr/lib/python2.7/re.py", line 151, in sub
    return _compile(pattern, flags).sub(repl, string, count)
  File "/var/data/localmh/bin/../lmh/lib/modules/translate.py", line 215, in replacer2
    return "\\ttl{"+term+"}"
TypeError: cannot concatenate 'str' and 'tuple' objects

lmh seems to have crashed with <type 'exceptions.TypeError'>
a report will be generated in 

lmh translate defeq.en.tex al --terms test.json with a dummy json does the same thing.

[m-iancu]

Doesn't crash for me anymore so that's good. Didn't test any further though (I didn't have a good json file to use).

[eluzhnica]

Doesn't crash anymore. In most of the cases it works, but it looks like it's not working for terms that contain "-" in them. For instance, in defeq.en.tex, definitional-equation should be translated to definitionale-Gleichung in German. Here is a test json.

{"en" : {"de" : {"structure" : "%DStruktur", "equal" : "gleich", "definiens" : "Definiens", "definitional-equation" : "definitionale-Gleichung", "approximately-equal" : "annährend-gleich", "definiendum" : "Definiendum", "component" : "Komponenten"}}, "de" : {"en" : {"Komponenten" : "component", "Definiens" : "definiens", "annährend-gleich" : "approximately-equal", "Definiendum" : "definiendum", "gleich" : "equal", "definitionale-Gleichung" : "definitional-equation", "%DStruktur" : "structure"}}}

[m-iancu]

@eluzhnica, Unrelated to your report but the %D from e.g. from Struktur shouldn't be there. You should be careful how you print the delimiters to get not the internal encoding but the content that they encode (i.e. use .text instead of .toString).