XPath-like query language for Xigt data

[goodmami]

XPath itself works on the XigtXML, but it doesn't resolve references, nor does it work for potential non-XML serializations of Xigt data. We need functionality like this:

• [x] igt "find Igt objects"
• [x] igt/tier "find tier objects under igts"
• [x] //item "find item objects under the context object (whether corpus, igt, or tier)"
• [x] //item/text() "find the text (in-situ content) of items"
• [x] //item/value() "find the value (resolved content) of items"
• [x] //tier[@type] "find the type attribute of tiers"
• [x] //tier[@type="words"]/item "find the items of tiers with type words"
• [x] //item[../@type="words"] (same as previous)
• [x] igt[metadata//dc:subject/@olac:code="tur"] "find IGTs whose OLAC metadata say they are about Turkish (iso-639-3 code tur)"
• [x] //items[../@type="glosses"][value()="NOM"] "find items in tiers of type glosses whose value is NOM"

So far this is the same as XPath (except for the value() function). But resolving expressions will require a new axis specifier or two (or more). These are only tentative:

• [x] //items[../@type="glosses"]/> "find objects referenced by items in a glosses tier (by default using either the alignment or segmentation attribute)"
• [x] //items[../@type="glosses"]/>@alignment "find objects referenced by items in a glosses tier (via the alignment attribute)"
• [x] //items[../@type="words"]/< "find objects referring to items in a words tier (by default using either the alignment or segmentation attribute)"
• [x] //items[../@type="words"]/<@segmentation "find objects referring to items in a words tier (via the segmentation attribute)"

[goodmami]

Update: referents and referrers are no longer axes, but functions (node tests). This is a little more verbose, but is better in that we can make XQuery functions that do the same thing (so the resulting XPaths would look about the same).

• tier/item/referents()
• //item/referrers("alignment")[../@type="glosses"]
• etc.

EDIT: additional axes like /following-sibling::node have been moved to issue #42