- Changed Tag to an interface and flattened hierarchy
- Changed Description to accept a list of tags and a template
- Allow auto-wiring when constructing tags
phpDocumentor is capable of representing a whole series of types, more
than PHP supports in fact, and this commit represents those types and a
Resolver that is capable of taking a type Expression and resolving that
to a series of Value Objects representing those types.
As part of the re-architecting of phpDocumentor and its Reflection component
we needed to move the type classes from the \phpDocumentor\Descriptor
namespace to \phpDocumentor\Reflection\Types.
In addition to this I have removed the resolution / type creation from the
Collection class into a new Resolver class that is capable of resolving any
type into a Type or Fqsen object.
This is the first commit with the base classes and a test for the Resolver
functionality in subsequent commits we need to add tests for the other
classes and to refactor the collection object.
Contexts are necessary for factories to resolve QSEN into FQSENs based
on partial namespaces and namespace aliases. These provide DocBlocks
with the namespace name and namespace aliases.
The new ContextFactory will enable third parties who don't use
phpDocumentor's Reflection component to construct a Context based on
a class reflector or namespace name (and file contents).
DocBlocks may start with #@+ and #@- to indicate that they are (the start) of a DocBlock
template or the end of a template.
In this commit I have changed the way a DocBlock is parsed to interpret this information
and added tests to show for it. In addition I have added more comments to the Regular
Expression responsible for splitting a DocBlock to show the business rules more clearly.
This is the first step in implementing https://github.com/phpDocumentor/phpDocumentor2/issues/42.
This is especially important when using markdown.
Adding a blank line between text introduces a new paragraph.
Current implementation will result in all text staying in one paragraph
and also other elements like lists and code do not get recognized in
most cases.
Performance improvement in Serializer - wrap length is calculated once during the text portion only. Tags reuse the result. Also no "prefix" adding in tags - the name is simply added before wrapping.
Swapped the indent and indent string options at Serializer's constructor;
Renamed Serializer::setIndentFirstLine() to setIsFirstLineIndented() in accordance with the getter (PHPMD fix);
Line length is now ACTUALLY line length, i.e. it takes the indentation into account, and is applied to tags as well;
Fixed ReturnTag::setContent() to set "types" to NULL;
A lot of doc and CS fixes at Serializer.php.
Changed the list of keywords to be static (for the purpose of reducing memory consumption);
Updated composer.lock with the latest dependencies (PHPUnit in particular).
Simplified @see and @link parsing;
Added comments to the regexes of @method, @example and @source;
Added a test case for what 2812eac046 intended to fix.
Mike van Riel
Colin O'Dell
Kévin Dunglas
Alexander Knöbel
Siad Ardroumli
Carsten Brandt
Mike van Riel
Barry vd. Heuvel
Vasil Rangelov
Jon Cave