…block

Otherwise function prototypes will be totally ignored and our variables not
parsed correctly. This is also reason for why perl throw warning message:

Name "Encode::MIME::Header::STRICT_DECODE" used only once: possible typo

There are no functional changes.

…alid mime words

…E words

Broken email clients (like Thunderbird 38) sometimes do not encode spaces
and so they appear in MIME words. To make sure that Encode::MIME::Header
will be able to decode emails generated by Thunderbird allow spaces in non
strict mode of decoder.

Fix functions which calculate length of encoded MIME word, formula for
base64 was incorrect. And do not call bytes::length() function because
using it is wrong approach. Rather encode string to UTF-8 and after that
count bytes with standard length().

It acts same as find_encoding() but search only for object with mime name.

Address https://rt.cpan.org/Public/Bug/Display.html?id=114034
* Use Encode::find_mime_encoding() for retrieving encoding object needed for inner string
* Always use strict UTF-8 encoder and decoder for inner strings
* Decoding inner strings is not called with flag Encode::FB_PERLQQ anymore
* Propagate check flags from MIME encoder/decoder to inner string encoder/decoder

Previous behaviour with Encode::FB_PERLQQ can be achieved by passing
Encode::FB_PERLQQ check flag to decode().

Implemented is also modification of input string (when requested by flags).
Now implementation of encoder now should match behaviour described in
Encode module documentation.

…ctions directly

…ort for check flags

Now decoder fully respect check flags and so it does not croak anymore when
it is called without any flag (or Encode::FB_DEFAULT flag) and when in
input string is invalid or incorrect charset. It means that implementation
of MIME decoder should match behaviour described in Encode module
documentation.

Invalid and unknown MIME words are not touched and they stay in decoded
output as were on input.

Previous behaviour (croak on unknown charset) can be achieved by passing
Encode::FB_CROAK check flag to decode().

Perl versions prior to 5.9.4 have recursive Regular expressions Engine
which cause stack overflow.

…FC 1766)

Basically I rewrote whole module so add me as author of it.

Add examples and describe what is module doing.

…er match Encode terminology

…ded-word as 76 bytes

Module MIME::Base64 has hardcoded its own limit to maximum 76 bytes per
string which cannot be changed. Longer strings are first split and then
each part encoded separately. Therefore for correct functionality of
Encode::MIME::Header it is not possible to have larger MIME encoded-words
as 76 bytes.