0.35 2014-05-12 - Replaced all uses of File::Slurp with File::Slurp::Tiny. See RT #95479 for the initial bug report and discussion on why File::Slurp should not be used. 0.34 2014-05-02 - More test fixes - relax the date parsing more. 0.33 2014-04-19 - Fixed tests so they pass with the most recent DateTime::Format::Mail (0.400) as well as older versions. 0.32 2014-04-19 - Courriel parsing is now more liberal and will accept an email with mixed line endings (some newline, some CRLF, etc.). Apparently this happens in the wild (of course it does). Reported by Joseph Crotty. RT #94837. 0.31 2013-04-03 [MISC] - Removed the "this is very alpha" warning from the docs. It's really not very alpha any more. 0.30 2012-08-20 [API CHANGES] - Courriel::Builder no longer exports a bcc() subroutine. Creating Bcc headers in emails does not make any sense whatsoever. See https://metacpan.org/module/Email::Sender::Manual::QuickStart#Hey-wheres-my-Bcc-support- for an explanation of why there is no such thing as a Bcc header. 0.29 2012-03-07 [BUG FIXES] - Courriel now allows an email to have an empty subject when parsing, as does Courriel::Builder. Based on a patch from Zbigniew Łukasiak. 0.28 2012-02-01 [BUG FIXES] - Header attribute names need to be looked up case-insensitively, according to the spec. The original casing is preserved in the Courriel::HeaderAttribute object's name(), method, but looking up an attribute by name on a Courriel::Header object now accepts any casing for the name. 0.27 2011-09-24 [ENHANCEMENTS] - All objects which can stringify now support a ->stream_to API as well. This lets you stream their string output to a sub ref, filehandle, or an object with a ->print() method. If you are dealing with large emails, this API may be more efficient than generating a single string in memory. [API CHANGES] - Renamed the Header->as_header_string and HeaderAttribute->as_header_string methods to as_string, so that all objects have the same stringification API. 0.26 2011-09-19 [BUG FIXES] - Add missing prereq for MooseX::Role::Parameterized. Reported by zby. 0.25 2011-09-15 [BUG FIXES] - When a header value contained any non-ASCII content, Courriel would mistakenly MIME encode the whole value. In some cases, this is okay, but for some header fields it is wrong. For example, when a field contains an email address the actual address portion must be unencoded. 0.24 2011-09-13 [ENHANCEMENTS] - The is_binary flag has been replaced with an is_character flag. This flag defaults to false. Suggested by zby. 0.23 2011-09-13 [ENHANCEMENTS] - Courriel->parse now accepts an is_binary flag so you can pass decoded (utf-8) data as the email text. However, it's probably smarter to simply pass binary data to Courriel and let it do the decoding. The issue was brought up by zby. 0.22 2011-09-13 [BUG FIXES] - Require List::MoreUtils 0.28+, as tests fail with earlier versions. - Added missing prereq on File::Slurp. 0.21 2011-09-12 [API CHANGES] - The ->get_attribute() method for Courriel::ContentType and ::Disposition objects was replaced by ->attribute in 0.20. [ENHANCEMENTS] - Added a new ->get_values() method to Courriel::Headers. This lets you get the header string values directly. - Added a new ->attribute_value() for Courriel::Header::ContentType and ::Disposition. This lets you get the attribute's string value, if the attribute exists. Suggested by zby. [BUG FIXES] - The previous release documented a method ->get_attribute() for Courriel::Header::ContentType and ::Disposition objects. This method was removed in 0.20 and replaced with a method named ->attribute(). 0.20 2011-09-11 [API CHANGES] - The way headers are represented has been changed quite a bit internally. Read on for details. All headers are now represented as Courriel::Header objects instead of plain strings. This means that calling $headers->get('Foo') returns objects. To get the value, you need to call ->value on the individual objects. See Courriel::Header for the new class for headers. - As a corrolary, the Courriel::ContentType and Courriel::Disposition classes have been renamed to Courriel::Header::ContentType and Courriel::Header::Disposition. They are both now subclasses of Courriel::Header. The APIs of these classes have remained mostly the same, except for attribute handling (see below). The old classes are still shipped in the distro, but they die when loaded. This avoids leaving behind old modules which no longer work with newer versions of Courriel. Methods which returned the old object now return the new one. The Courriel::Headers class will return the appropriate object when the header is fetched with $headers->get(). - Header attributes for the Content-Type and Content-Disposition object are now represented as Courriel::HeaderAttribute objects. This is necessary because these attributes may contain information about their character set and language, per RFC2231. [ENHANCEMENTS] - Header attributes with character set and language data per RFC2231 are now parsed properly. Reported by zby. RT #70801. - Header attribute parsing is much laxer and will not die on bad attributes syntax (in most cases). It simply stops parsing and ignores bad data. 0.19 2011-08-21 - Fixed some test failures with older versions of Test::More, which stringify objects passed to is() rather than using their overloading. Reported by Maximilian Gaß. RT #70395. - The Courriel::Disposition methods which return DateTime objects now explicitly set the time zone of the objects to UTC before returning it. 0.18 2011-08-20 - More improvements in handling broken email based on jwang's corpus of broken messages. - Courriel tries a lot harder to find a date for the email. It now looks at all the Received headers, and it also uses DateTime::Format::Natural to parse dates if DateTime::Format::Mail fails. - Courriel handles the charset "unknown-8bit" (by ignoring it and treating it as ASCII). 0.17 2011-08-19 - Email parsing is now much more flexible, and throws fewer exceptions. Ultimately, it's likely that Courriel will never throw an exception from parsing, since email in the wild is often broken, and it's better to return something than just give up. Thanks to jwang of grokbase.com for providing a corpus of broken messages to test against. - The Courriel->plain_body_part() and ->html_body_part() methods ignore the case of an attachment's mime type, per RFC 2045. - The Courriel::ContentType->mime_type() method always returns a value in lower case, but it preserves the original casing when constructing a header value. 0.16 2011-07-21 - Add Courriel->from(), ->to(), and ->cc() methods. - Skip another test which may fail on other people's machines, as File::LibMagic does not return consistent results on all machines. 0.15 2011-06-28 - Skip a test which may fail on other people's machines, as File::LibMagic does not return consistent results on all machines. 0.14 2011-06-25 - An email part with non-text content would still be treated as character data rather than binary data, and its Content-Type header would end up with a charset of us-ascii. This has been fixed, and now you must explicitly ask for a charset in the Content-Type header when creating a Courriel::ContentType object. Note that plain and html body parts created via Courriel::Builder are unaffected, since these always default to UTF-8 as the charset. Reported by ico. RT #69085. - Added Courriel::ContentType->has_charset() and Courriel::ContentType->is_binary() methods. 0.13 2011-06-20 - A multipart email built with Courriel::Builder would have a broken Content-Type header, missing the required boundary attribute. Reported by ico. RT #68694. - If you explicitly passed unencoded content when creating a part (which happens when using Courriel::Builder), that content would be modified so it had CRLF line endings. Now, such content is left untouched. Reported by ico. RT #68695. - When passing encoded content to the Courriel::Part::Single constructor, the constructor used to modify the content so it had CRLF line endings. Now it is left alone. - Note that if you're passing raw content to the Single constructor _and_ setting the transfer encoding to 7bit, 8bit, or binary, then you'll end up with a noncomformant email. However, by default all raw content is base64 encoded. It is strongly recommended that you pass raw content and let Courriel take care of the transfer encoding. 0.12 2011-06-07 [BUG FIXES] - The value returned by Courriel::Part::Single->content() was not decoded to Perl's native utf-8 character set. Similarly, when encoding content for transfer, it was not first transformed to raw bytes. 0.11 2011-06-07 [BUG FIXES] - Courriel::Builder now adds a "MIME-Version: 1.0" header to all emails it creates unless you explicitly set this header, which you probably shouldn't, since the only valid value is "1.0". This is a required header for MIME emails per RFC2045. 0.10 2011-06-07 [BUG FIXES] - When parsing an email, the encoding attribute for parts was not set correctly, and ended up defaulting to 8bit for all parts. Now the encoding is set based on the part's Content-Transfer-Encoding header. 0.09 2011-06-06 [NEW FEATURES] - This distro now ships a plugin for Email::Abstract, meaning you can use Courriel with Email::Sender. 0.08 2011-06-06 [API CHANGES] - The content() and encoded_content() methods for Courriel::Part::Single now return strings, rather than references. This makes the typical use simpler. You can still get the reference directly by calling content_ref() and encoded_content_ref(). [BUG FIXES] - Make sure that any Content-ID header set via Courriel::Builder is formatted properly (the id value should be wrapped in angle brackets). 0.07 2011-06-06 - Some tests were failing because File::LibMagic may return slightly different results on different systems (application/x-perl versus text/x-perl). 0.06 2011-06-05 - Add missing test prereq on File::Slurp. - The clone_without_attachments() method could end up setting the wrong encoding when creating a new single part email. 0.05 2011-06-04 - The clone_without_attachments() method would die when calling on an email that contained both plain and html inline parts. 0.04 2011-06-04 - Add missing prereq on File::LibMagic (needed since 0.02). 0.03 2011-06-04 - The encoding parameter is now always set in a part's headers. 0.02 2011-06-04 - Added all_parts_matching() method to Courriel class. - Added clone_without_attachments() method to Courriel class. - The Courriel::Part::Single attribute raw_content has been renamed as encoded_content. - The text_body_part() method is now called plain_body_part(). - The add() and unshift() methods in Courriel::Headers now accept only one value. - Added Courriel::Headers->replace as a convenience method. - A header with attributes (like Content-Type) can escape values inside a quoted string, like value="foo \" bar". Courriel wasn't unescaping these values. - It is now possible to pass a Courriel::Disposition object explicitly when constructing a Courriel::Part::Single object. The docs said this was possible in 0.01, but the parameter was just ignored. - If a part is created with an explicit content_type and/or disposition object, the part's headers will be updated so that the Content-Type and Content-Disposition match the given object. - The boundary for a Multipart part and the boundary in its ContentType's attributes will always be the same. - All classes now use MooseX::StrictConstructor. - Some hacks to make it possible for Courriel to play nice with Email::Abstract. - Fixed some small doc errors. 0.01 2011-05-31 - First release upon an unsuspecting world.