False positives on bibcodes

[bperel]

Hello,

Perhaps an edge case, but bibcodes which are used to reference literature references about astronomical data, are counted as errors since they can contain dots, for instance 2002AJ....123..549P.

I found a regular expression to match these bibcodes :

\d{4}[A-Za-z\.\&]{5}[\w\.]{4}[ELPQ-Z\.][\d\.]{4}[A-Z]

[bperel]

I'm willing to create the rule myself but even after reading the Development overview page I'm not sure if this counts as a "rule" : I just don't want the "2 consecutive dots" error to occur. Does it mean that the method org.languagetool.rules.DoublePunctuationRule#match should be altered?

[dpelle]

@bperel wrote:

Does it mean that the method org.languagetool.rules.DoublePunctuationRule#match should be altered?

You probably don't need to change any Java code to add a rule. In general, adding a rule to grammar.xml is enough.

The audience of bibcodes seem too specialized to be added to LanguageTool I suspect. Well it could be added on the condition that it almost never cause false errors. Or do you want to add a rule just for your own build of LT?

You found a regexp to match valid bibcodes, but keep in mind that LT detects errors by specifying erroneous patterns. So what would quality as a pattern for probable bibcode which is not a valid bibcode?

[bperel]

I'm not sure if the audience is too specialized actually: I ran LanguageTool on 500 Wikipedia articles, and 10% of the errors that it found were "Too consecutive dots" due to bibcodes. I suspect that it is mostly in the "References" section of articles that these codes appear though.

I wouldn't mind integrating the rule in upstream LanguageTool if it's useful to people, but like you say the issue is that the regex that I have is for matches, not errors. I don't think that it would be useful to have a bibcode-checker rule (now that would be specialized!), that's why I suggested integrating it into the "two consecutive dots" rule : basically, if the double dots are surrounded by characters forming a bibcode, then don't throw an error. If you believe that it's too specialized, then I will add it to my fork only.