iconv on cygwin saves the accents

Anyone knows why iconv saves the accents on cygwin? And if so how can I tell it not to.

[Nifle@cygwin ~]$ echo "ÅÄÖÕŨÉÁ" | iconv -f utf-8 -t ascii//TRANSLIT
A"A"O~O~U'E'A

I want it to behave as it does on my linux servers

[NIfle@linux ~]$ echo "ÅÄÖÕŨÉÁ" | iconv -f utf-8 -t ascii//TRANSLIT
AAOOUEA

cygwin

Nifle

Posted 2016-09-27T14:42:47.267

Reputation: 31 337

As a workaround you may want to pipe it through some perl magic: echo "ÅÄÖÕŨÉÁ" | iconv -f utf-8 -t ascii//TRANSLIT | perl -ne 'foreach (split //) { print "$_" if /\w/; } print "\n";' which suppresses all non-word characters (\w). – mpy – 2017-01-04T18:25:52.743

That wouldn't work because the OP might try to parse something containing the string "Hello" which would be mistakenly converted to Hello – Pat – 2017-01-05T16:57:05.847

Answers

old issue:

read Kobylkin's comments on this thread...

https://sourceware.org/bugzilla/show_bug.cgi?id=2872#c2

also

http://www.yqcomputer.com/422_10096_1.htm

Remember Cygwin < 1.7 does not have locales.

But in your case the local seems to be well interpreted; it is the iconv Transliteration process on Cygwin the one that decides to convert a single character into 2 (or even more) components instead of one; in your example you get the transliterated character itself plus the transliterated character modifier (accent, dieresis, tilde, dot, etc.) if present.

Å   Ä   Ö   Õ   Ũ   É   Á
A  "A  "O  ~O  ~U  'E  'A

Transliteration:
When a character cannot be represented in the target character set, 
it can be approximated through one or several similarly looking characters
(https://www.gnu.org/software/libiconv/)

you can see the libiconv.dll translit.def showing the source of the transliteration table (translit.h) created during the "make" process. That table shows that what Cygwin does is in fact correct.