xpath tests

[amitmurthy]

add tests for xpath, possibly based on http://msdn.microsoft.com/en-us/library/ms256086.aspx

[amitmurthy]

cc: @vtjnash

Some not-so-trivial xpath queries are failing. Will be good to add some standard tests for the same.

[vtjnash]

Sounds good, will do.

Can you give some examples of xpath queries that are failing for you? There is a class of unimplemented functionality (e.g. parenthesized expressions, many functions, namespaces, returning objects of types other than nodes), but there is also likely still some legit bugs.

[amitmurthy]

From the above xml

julia> pd["@specialty"]
ERROR: no method convert(Type{ParsedData},ASCIIString)
 in push! at array.jl:663
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:798
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:516
 in getindex at /home/amitm/.julia/LibExpat/src/xpath.jl:898

julia> pd["/@specialty"]
0-element ParsedData Array

julia> pd["//@specialty"]
ERROR: no method convert(Type{ParsedData},ASCIIString)
 in push! at array.jl:663
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:798
 in xpath_descendant at /home/amitm/.julia/LibExpat/src/xpath.jl:891
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:847
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:818
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:516
 in getindex at /home/amitm/.julia/LibExpat/src/xpath.jl:898

julia> pd["/bookstore@specialty"]
0-element ParsedData Array

julia> pd["/bookstore/@specialty"]
ERROR: no method convert(Type{ParsedData},ASCIIString)
 in push! at array.jl:663
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:798
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:806
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:840
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:818
 in xpath at /home/amitm/.julia/LibExpat/src/xpath.jl:516
 in getindex at /home/amitm/.julia/LibExpat/src/xpath.jl:898

The below returns 2 objects, should return only one:

julia> pd["/bookstore/book[1]/author/award"]
2-element ParsedData Array:
 <award>Trenton Literary Review Honorable Mention</award>
 <award>Pulitzer</award> 

cannot get the text anyways:

julia> pd["/bookstore/book[1]/author/award/text()"]
0-element ParsedData Array

julia> 

I have not done extensive testing, was just thinking of using this interface and ran into problems.

[vtjnash]

Some of these are intentionally not supported currently (retrieving attributes via pd["@specialty"] or text text()), so there would only be one possible return type
These ones show correct behavior (pd["/bookstore@specialty"] pd["/@specialty"])

This one seems wrong: pd["/bookstore/book[1]/author/award"]

[amitmurthy]

(pd["/bookstore@specialty"] pd["/@specialty"]) are returning an empty array.

[amitmurthy]

How does one get the text using xpath then? pd["/bookstore/magazine/price#string"] does not work either.

[amitmurthy]

It will also be good to list the xpath subset supported more explicitly in the README.

[vtjnash]

This one pd["/bookstore/book[1]/author/award"] succeeds for me:

julia> pd["/bookstore/book[1]/author/award"]
1-element ParsedData Array:
 <award>Trenton Literary Review Honorable Mention</award>

julia> pd["/bookstore/book[3]/author/award"]
1-element ParsedData Array:
 <award>Pulitzer</award>

I'll add an API to support the other node types.