Pure Lua XPath 2.0 parser

This is part of the speedata Publisher.

See the test file to get get idea of the amount of XPath functionality that is implemented.

Using the library

-- crate a context with variables, namespaces and an XML document
local ctxvalue = {
    namespaces = {
        myns = "http://a.name-space"
    },
    vars = {
        foo = "bar",
        onedotfive = 1.5,
        a = 5,
        ["one-two"] = 12,
    },
    xmldoc = { xmltab },
    sequence = { xmltab }
}
local ctx = lxpath.context:new(ctxvalue)

-- toks is a token list
local toks, msg = lxpath.string_to_tokenlist(str)
if toks == nil then
    print(msg)
    os.exit(-1)
end

-- ef is a function which executes the parsed xpath on a context.
-- you can reuse ef()
local ef, err = lxpath.parse_xpath(toks)
if err ~= nil then
    -- handle error string err
end

local seq, errmsg = ef(ctx)
-- seq is the resulting sequence (a table) of zero or more items.
-- Each item can be a sequence, an element, an attribute, a string or a number.

You can also run one of the convenience functions:

sequence, errormessage = ctx:eval("xpath string")

and

sequence, errormessage = ctx:execute("xpath string")

The difference is that the eval() does not change the context, it only returns the sequence. execute() changes self.

Running the tests

lua lxpath_test.lua

Unicode and UTF8

All input is expected to be in UTF8.

This library is not unicode aware! This means for example upper-case('ä') is not Ä, but ä, since there is no lookup table for unicode.

You can provide your own implementations for string.match and string.find (which might be UTF8 ready) by setting M.stringmatch and M.stringfind.

Limitations

• Work in progress: union/except/intersect and date functions are currently missing
• This library is not unicode aware (see above).
• No schema support.
• Since Lua does not have “real” regular expressions, the functions that expect regular expressions are not implemented (matches(), replace(), tokenize()). You should provide your own implementations of these functions.

You can override the XPath functions.

Registering new XPath functions

You can use the registerFunction() function to add your own definitions:

It expects a table with the following fields:

1. function name
2. namespace
3. function (where the arguments are the context and the provided arguments)
4. minimum number of arguments
5. maximum number of arguments (-1 if arbitrary many arguments allowed)

example:

function fnSubstring(ctx, arg)
    ...
end
registerFunction( { "substring", "http://www.w3.org/2005/xpath-functions", fnSubstring,2, 3 } )

XML representation

Since the xpath library is not parsing the XML file, it has to be represented in a tree like data structure.

Each element (a table) has zero or more children, either a string or another element. The element table has this representation:

{
    [".__name"] = "elementname",
    [".__id"]  = 1,  -- in document order
    [".__type"] = "element",
    [".__local_name"] = "elementname",
    [".__namespace"] = "",
    [".__ns"] = {
        -- some predefined name spaces
        ["myprefix"] = "http://a.name.space",
    },
    [".__attributes"] = {
        ["key"] = "value",
    },
    [1] = "a string for example",
    [2] = { --  a table for an element
        },
    [3] = "perhaps another string",
}

For example the following XML

<data>
    <child attname="attvalue">
        some text
    </child>

    mixed content
</data>

must be encoded in Lua as:

tbl = {
    [".__type"] = "document",
    {
        [1] = {
            [".__name"] = "data",
            [".__id"]  = 1,
            [".__type"] = "element",
            [".__local_name"] = "data",
            [".__namespace"] = "",
            [".__ns"] = {
            },
            [1] = "\n    ",
            [2] = {
                [".__name"] = "child",
                [".__id"]  = 2,
                [".__type"] = "element",
                [".__local_name"] = "child",
                [".__namespace"] = "",
                [".__ns"] = {
                },
                [".__attributes"] = { ["attname"] = "attvalue", },
                [1] = "\n        some text\n    ",
            },
            [3] = "\n\n    mixed content\n",
        },
    },
}