I'm using R and the textstem package in Windows (10, 64-bit), which relies on koRpus. I spotted what appears to be two bugs:
1) In the shell(sys.tt.call) for when it's not unix.OS, the translate = TRUE converts the /'s in the regex substitution in TT.filter.command's "| perl -pe ..." to become backslashes, which breaks perl. The path to TreeTagger by that time contains double backslashes due to normalizePath called earlier, so using translate=FALSE keeps the substition's /'s as is and the path is still fine.
2) The line 'cat(paste(tknz.results, collapse = "\n"), "\n", file = tknz.tempfile)' requires a sep="" or else there's a hidden space appended to the last line/token, resulting in TreeTagger's tagging that last token as .
I am using R from a Cygwin environment, but I believe these errors would occur from a normal Windows environment.
Hi,
I'm using R and the textstem package in Windows (10, 64-bit), which relies on koRpus. I spotted what appears to be two bugs: 1) In the shell(sys.tt.call) for when it's not unix.OS, the translate = TRUE converts the /'s in the regex substitution in TT.filter.command's "| perl -pe ..." to become backslashes, which breaks perl. The path to TreeTagger by that time contains double backslashes due to normalizePath called earlier, so using translate=FALSE keeps the substition's /'s as is and the path is still fine. 2) The line 'cat(paste(tknz.results, collapse = "\n"), "\n", file = tknz.tempfile)' requires a sep="" or else there's a hidden space appended to the last line/token, resulting in TreeTagger's tagging that last token as.
I am using R from a Cygwin environment, but I believe these errors would occur from a normal Windows environment.
Thanks!
Best, Jay