Hello
i am trying to generate a report and as far as I understand, the command should be something like this:
c:\languageTool>java -jar languagetool-commandline.jar --api -a -l en c:/test/test.properties c:/test/output.txt
but I get following error:
java.lang.IllegalArgumentException: API format makes no sense for automatic application of suggestions
What’s the problem there, please, any suggestion?
You can only use --api (XML output) or -a (automatically apply suggestions), not both, as there can only be one output: XML report or text with suggestions applied.
Hmm, my grammar.xml rules work only if -a is on, I mean, the text is validated only by selected rules, when -a is off, it gets me tons of errors =(
The point of LanguageTool is to find errors, not to correct them automatically. I suggest you forget about the -a option, it’s only useful for very few, very specific cases: when you only have rules that are 100% correct and never introduce new errors.
Ok, Daniel, for now I have the following situation: I opened LT in GUI, unchecked some rules, so grammar.xml has changed, it got default=“off” for some rules, also I added my own rule to disambiguation.xml.
Now I want to run batch check for some folder with my selected rules and if it is possible, get some kind of report file.
Can I do it?
Thank you for your help, I appreciate it 
Hi, Alex.
You will need some batch script to do a batch check. I wrote one for Windows some time ago:
This script reads every file on a folder, converts them to plain text (using Tika library), and sorts the results and shows them in html with a Perl script (there is also a Python version).
You can take it as a guide and adapt it to your needs.
Thanks for your suggestion, jaumeortola!
I see that your manually select and deselect rules. Is it possible to load them using grammar.xml somehow?
You can see all the options of the command-line running “java -jar languagetool-commandline.jar”. In my script the default rules in grammar.xml are used and then some of them are enabled and some of them disabled. You can also use “-eo, --enabledonly” which means “disable all rules except those enabled explicitly in -e”.
$ java -jar languagetool-commandline.jar
Usage: java -jar languagetool-commandline.jar [OPTION]... FILE
FILE plain text file to be checked
Available options:
-r, --recursive work recursively on directory, not on a single file
-c, --encoding ENC character set of the input text, e.g. utf-8 or latin1
-b assume that a single line break marks the end of a paragraph
-l, --language LANG the language code of the text, e.g. en for English, en-GB for British English
--list print all available languages and exit
-adl, --autoDetect auto-detect the language of the input text
-m, --mothertongue LANG the language code of your first language, used to activate false-friend checking
-d, --disable RULES a comma-separated list of rule ids to be disabled (use no spaces between ids)
-e, --enable RULES a comma-separated list of rule ids to be enabled (use no spaces between ids)
-eo, --enabledonly disable all rules except those enabled explicitly in -e
--enablecategories CAT a comma-separated list of category ids to be enabled (use no spaces between ids)
--disablecategories CAT a comma-separated list of category ids to be disabled (use no spaces between ids)
-t, --taggeronly don't check, but only print text analysis (sentences, part-of-speech tags)
-u, --list-unknown also print a summary of words from the input that LanguageTool doesn't know
-b2, --bitext check bilingual texts with a tab-separated input file,
see http://languagetool.wikidot.com/checking-translations-bilingual-texts
--api print results as XML
-p, --profile print performance measurements
-v, --verbose print text analysis (sentences, part-of-speech tags) to STDERR
--version print LanguageTool version number and exit
-a, --apply automatically apply suggestions if available, printing result to STDOUT
--rulefile FILE use an additional grammar file; if the filename contains a known language code,
it is used in addition of standard rules
--falsefriends FILE use external false friend file to be used along with the built-in rules
--bitextrules FILE use external bitext XML rule file (useful only in bitext mode)
--languagemodel DIR a directory with e.g. 'en' sub directory (i.e. a language code) that contains
'1grams'...'3grams' sub directories with Lucene indexes with
ngram occurrence counts; activates the confusion rule if supported
--xmlfilter remove XML/HTML elements from input before checking (this is deprecated)
--line-by-line work on file line by line (for development, e.g. inside an IDE)Hi
I faced another problem. Currently I am using only following categories:
TYPOS,MISC,PLAIN_ENGLISH,GRAMMAR
But I 've got result such as
[error fromy=“0” fromx=“0” toy=“0” tox=“2” ruleId=“UPPERCASE_SENTENCE_START”]
or
[error fromy=“0” fromx=“25” toy=“0” tox=“31” ruleId=“SENTENCE_WHITESPACE”
as far as I understand these rules aren’t applied. but anyway, the check was performed. Also, the rule I added at disambiguation.xml doesn’t work when I try to launch with command line.
Any suggestion, please?
What’s the exact command line you’re running? Does it contain -eo? To get more information about disambiguation rules being applied, you can specify -v.
Hi
Exact command is:
-u -b --api -c utf-8 --enablecategories TYPOS,MISC,PLAIN_ENGLISH,GRAMMAR -l en-Gb