dotty.tools.dotc.transform.PatternMatcher.Translator
PatternExpander
Related Doc:
package Translator
If elementArity is...
If elementArity is... 0: A perfect match between extractor and the fixed patterns. If there is a star pattern it will match any sequence. > 0: There are more patterns than products. There will have to be a sequence which can populate at least <elementArity> patterns. < 0: There are more products than patterns: compile time error.
An 'extractor' can be a case class or an unapply or unapplySeq method.
An 'extractor' can be a case class or an unapply or unapplySeq method. Decoding what it is that they extract takes place before we arrive here, so that this class can concentrate only on the relationship between patterns and types.
In a case class, the class is the unextracted type and the fixed and repeated types are derived from its constructor parameters.
In an unapply, this is reversed: the parameter to the unapply is the unextracted type, and the other types are derived based on the return type of the unapply method.
In other words, this case class and unapply are encoded the same:
case class Foo(x: Int, y: Int, zs: Char*) def unapplySeq(x: Foo): Option[(Int, Int, Seq[Char])]
Both are Extractor(Foo, Int :: Int :: Nil, Repeated(Seq[Char], Char, Char*))
The type in its unextracted form
The non-sequence types which are extracted
The sequence type which is extracted
It's not optimal that we're carrying both sequence and repeated type here, but the implementation requires more unraveling before it can be avoided.
It's not optimal that we're carrying both sequence and repeated type here, but the implementation requires more unraveling before it can be avoided.
sequenceType is Seq[T], elementType is T, repeatedType is T*.
An extractor returns: F1, F2, ..., Fi, opt[Seq[E] or E*] A case matches: P1, P2, ..., Pj, opt[Seq[E]] Put together: P1/F1, P2/F2, ... Pi/Fi, Pi+1/E, Pi+2/E, ... Pj/E, opt[Seq[E]]
Here Pm/Fi is the last pattern to match the fixed arity section.
prodArity: the value of i, i.e. the number of non-sequence types in the extractor nonStarArity: the value of j, i.e. the number of non-star patterns in the case definition elementArity: j - i, i.e. the number of non-star patterns which must match sequence elements starArity: 1 or 0 based on whether there is a star (sequence-absorbing) pattern totalArity: nonStarArity + starArity, i.e. the number of patterns in the case definition
Note that prodArity is a function only of the extractor, and nonStar/star/totalArity are all functions of the patterns. The key value for aligning and typing the patterns is elementArity, as it is derived from both sets of information.