Back to LanguageTool Homepage - Privacy - Imprint

Paid work possible?


(Ruud Baars) #1

Since I am not a programmer, and no-one is programming specifically for Dutch, i am wondering if there is a way to get someone to make an extension in a paid way.

To give the Dutch LT a boost, I am looking for a function that can easily signal a flag for a word, where the flag is checked in a rule. This will require a data file, and adjustment of the DTD, I think.
The data file could look like:

reservatie;MF;reservering

The rule would have
<token flag="MF">
and
<Message><flagmatch> is not as common as <suggestion><flagsuggestion></suggestion>. Consider replacing it.

I am planning to use the rule for differences between Belgian and Dutch Dutch, suggest easier words and more.

Since there is a change in the DTD involved, I guess there should be more language interested in this function.

The other option I could apply is to generate ordinary rules from a data file.


(Daniel Naber) #2

Does this rule depend on the context? Would the rules have just this one token? In that case, it’s probably better to write a Java rule that reads the pairs from an external file. Would the tag then even be needed?


(Ruud Baars) #3

No, there is no context. The flag is meant to be able to have multiple rules with each a message on its own. Like one for a more frequent word, one for an easier word, one for a ‘nicer’ word etc.
For now, I assumed one token. But if multiple tokens is not difficult, it would be even better. It would make it possible to treat word groups as well.

An option to avoid a change in the DTD could be to have almost identical Java rules, one for every rule. Then the flag would not even be necessary.


(Knorr) #4

I think for this use case we can use a modified version of German OldSpellingRule which works with “alt_neu.csv”: In German, this rule detects words that were valid only with the pre-reform spelling and suggests the now valid variant. (e.g., Amboß -> Amboss)


(Ruud Baars) #5

I’ll check this routine out. Maybe it is possible. I could do a bit of aletration myself, but integration is out of my comfort zone.


(Ruud Baars) #6

I checked them out, and it looked good.I only wondered why there where 3 Java files needed for that rule, when CoherencyRule only seems to need one.

But if you could copy the german speller rule to Dutch, changing the names to something like AlternativeWordRule, and have them called, I can do the internal polishing, like altering the texts, var names etc.
The only entry in the data file could be:
rijwiel;fiets
Which is a real rule entry.


(Knorr) #7

Hi @Ruud_Baars !
I have added and activated the new rule.
Please translate the strings in PreferredWordRule, PreferredWordRuleWithSuggestion, PreferredWordData.
Then, you can add new pairs to preferredwords.csv


(Ruud Baars) #8

Great! I will get to it tomorrow.


(Ruud Baars) #9

I updated and pushed it. There is one thing that bothers me: it appears to have a Spelling error label, which is not correct; both forms are actually correct. That does not have consequence, apart from the error colour, right? Nevertheless, if it is easy to alter, that would be best.
Would you please have a look at this?