Jump to content

Module:SLAXML

From Meta, a Wikimedia project coordination wiki
Module documentation

SLAXML is a pure-Lua SAX-like streaming XML parser. It is more robust than many (simpler) pattern-based parsers that exist, properly supporting code like <expr test="5 > 7" />, CDATA nodes, comments, namespaces, and processing instructions.

It is currently not a truly valid XML parser, however, as it allows certain XML that is syntactically-invalid (not well-formed) to be parsed without reporting an error.

Features

[edit]
  • Pure Lua in a single file (two files if you use the DOM parser).
  • Streaming parser does a single pass through the input and reports what it sees along the way.
  • Supports processing instructions (<?foo bar?>).
  • Supports comments (<!-- hello world -->).
  • Supports CDATA sections (<![CDATA[ whoa <xml> & other content as text ]]>).
  • Supports namespaces, resolving prefixes to the proper namespace URI (<foo xmlns="bar"> and <wrap xmlns:bar="bar"><bar:kittens/></wrap>).
  • Supports unescaped greater-than symbols in attribute content (a common failing for simpler pattern-based parsers).
  • Unescapes named XML entities (< > & " ') and numeric entities (e.g. &#10;) in attributes and text nodes (but—properly—not in comments or CDATA). Properly handles edge cases like &amp;.
  • Optionally ignore whitespace-only text nodes (as appear when indenting XML markup).
  • Includes a DOM parser that is both a convenient way to pull in XML to use as well as a nice example of using the streaming parser.
  • Does not add any keys to the global namespace.

Usage

[edit]
local SLAXML = require 'slaxml'

local myxml = io.open('my.xml'):read()

-- Specify as many/few of these as you like
parser = SLAXML:parser{
  startElement = function(name,nsURI)       end, -- When "<foo" or <x:foo is seen
  attribute    = function(name,value,nsURI) end, -- attribute found on current element
  closeElement = function(name,nsURI)       end, -- When "</foo>" or </x:foo> or "/>" is seen
  text         = function(text)             end, -- text and CDATA nodes
  comment      = function(content)          end, -- comments
  pi           = function(target,content)   end, -- processing instructions e.g. "<?yes mon?>"
}

-- Ignore whitespace-only text nodes and strip leading/trailing whitespace from text
-- (does not strip leading/trailing whitespace from CDATA)
parser:parse(myxml,{stripWhitespace=true})

If you just want to see if it will parse your document correctly, you can simply do:

local SLAXML = require 'slaxml'
SLAXML:parse(myxml)

which will cause SLAXML to use its built-in callbacks that print the results as seen.

DOM Builder

[edit]

If you simply want to build tables from your XML, you can alternatively:

local SLAXML = require 'slaxdom' -- also requires slaxml.lua; be sure to copy both files
local doc = SLAXML:dom(myxml)

The returned table is a 'document' comprising tables for elements, attributes, text nodes, comments, and processing instructions. See the following documentation for what each supports.

DOM Table Features

[edit]
  • Document - the root table returned from the SLAXML:dom() method.
    • doc.type : the string "document"
    • doc.name : the string "#doc"
    • doc.kids : an array table of child processing instructions, the root element, and comment nodes.
    • doc.root : the root element for the document
  • Element
    • someEl.type : the string "element"
    • someEl.name : the string name of the element (without any namespace prefix)
    • someEl.nsURI : the namespace URI for this element; nil if no namespace is applied
    • someEl.attr : a table of attributes, indexed by name and index
      • local value = someEl.attr['attribute-name'] : any namespace prefix of the attribute is not part of the name
      • local someAttr = someEl.attr[1] : an single attribute table (see below); useful for iterating all attributes of an element, or for disambiguating attributes with the same name in different namespaces
    • someEl.kids : an array table of child elements, text nodes, comment nodes, and processing instructions
    • someEl.el : an array table of child elements only
    • someEl.parent : reference to the parent element or document table
  • Attribute
    • someAttr.type : the string "attribute"
    • someAttr.name : the name of the attribute (without any namespace prefix)
    • someAttr.value : the string value of the attribute (with XML and numeric entities unescaped)
    • someEl.nsURI : the namespace URI for the attribute; nil if no namespace is applied
    • someEl.parent : reference to the parent element table
  • Text - for both CDATA and normal text nodes
    • someText.type : the string "text"
    • someText.name : the string "#text"
    • someText.value : the string content of the text node (with XML and numeric entities unescaped for non-CDATA elements)
    • someText.parent : reference to the parent element table
  • Comment
    • someComment.type : the string "comment"
    • someComment.name : the string "#comment"
    • someComment.value : the string content of the attribute
    • someComment.parent : reference to the parent element or document table
  • Processing Instruction
    • someComment.type : the string "pi"
    • someComment.name : the string name of the PI, e.g. <?foo …?> has a name of "foo"
    • someComment.value : the string content of the PI, i.e. everything but the name
    • someComment.parent : reference to the parent element or document table

Finding Text for a DOM Element

[edit]

The following function can be used to calculate the "inner text" for an element:

function elementText(el)
  local pieces = {}
  for _,n in ipairs(el.kids) do
    if n.type=='element' then pieces[#pieces+1] = elementText(n)
    elseif n.type=='text' then pieces[#pieces+1] = n.value
    end
  end
  return table.concat(pieces)
end

local xml  = [[<p>Hello <em>you crazy <b>World</b></em>!</p>>]]
local para = SLAXML:dom(xml).root
print(elementText(para)) --> "Hello you crazy World!""

A Simpler DOM

[edit]

If you want the DOM tables to be simpler-to-serialize you can supply the simple option via:

local dom = SLAXML:dom(myXML,{ simple=true })

In this case no table will have a parent attribute, elements will not have the el collection, and the attr collection will be a simple array (without values accessible directly via attribute name). In short, the output will be a strict hierarchy with no internal references to other tables, and all data represented in exactly one spot.

Known Limitations / TODO

[edit]
  • Does not require or enforce well-formed XML. Certain syntax errors are silently ignored and consumed. For example:
    • foo="yes & no" is seen as a valid attribute
    • <root><child> invokes two startElement() calls but no closeElement() calls
    • <foo></bar> invokes startElement("foo") followed by closeElement("bar")
    • <foo> 5 < 6 </foo> is seen as valid text contents
  • No support for custom entity expansion other than the standard XML entities (< > " ' &) and numeric ASCII entities (e.g. )
  • XML Declarations (<?xml version="1.x"?>) are incorrectly reported as Processing Instructions
  • No support for DTDs
  • No support for extended (Unicode) characters in element/attribute names
  • No support for charset
  • No support for XInclude

History

[edit]
v0.5.2 2013-Nov-7
  • Lua 5.2 compatible
  • Parser now errors if it finishes without finding a root element, or if there are unclosed elements at the end.
  • (Proper element pairing is not enforced by the parser, but is—as in previous releases—enforced by the DOM builder.)
v0.5.1 2013-Feb-18
  • <foo xmlns="bar"> now directly generates startElement("foo","bar") with no post callback for namespace required.
v0.5 2013-Feb-18
  • Use the local SLAXML=require 'slaxml' pattern to prevent any pollution of the global namespace.
v0.4.3 2013-Feb-17
  • Bugfix to allow empty attributes, i.e. foo=""
  • closeElement no longer includes namespace prefix in the name, includes the nsURI
v0.4 2013-Feb-16
  • DOM adds .parent references
  • SLAXML.ignoreWhitespace is now :parse(xml,{stripWhitespace=true})
  • "simple" mode for DOM parsing
v0.3 2013-Feb-15
  • Support namespaces for elements and attributes
    • <foo xmlns="barURI"> will call startElement("foo",nil) followed by namespace("barURI") (and then attribute("xmlns","barURI",nil)); you must apply the namespace to your element after creation.
    • Child elements without a namespace prefix that inherit a namespace will receive startElement("child","barURI")
    • <xy:foo> will call startElement("foo","uri-for-xy")
    • <foo xy:bar="yay"> will call attribute("bar","yay","uri-for-xy")
    • Runtime errors are generated for any namespace prefix that cannot be resolved
  • Add (optional) DOM parser that validates hierarchy and supports namespaces
v0.2 2013-Feb-15
  • Supports expanding numeric entities e.g. " -> "
  • Utility functions are local to parsing (not spamming the global namespace)
v0.1 2013-Feb-7
  • Option to ignore whitespace-only text nodes
  • Supports unescaped > in attributes
  • Supports CDATA
  • Supports Comments
  • Supports Processing Instructions

License

[edit]

Copyright (c) 2013 Gavin Kistner

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

--[=====================================================================[
v0.5.2 Copyright © 2013 Gavin Kistner <!@phrogz.net>; MIT Licensed
See http://github.com/Phrogz/SLAXML for details.
--]=====================================================================]
local SLAXML = {
	VERSION = "0.5.2",
	_call = {
		pi = function(target,content)
			mw.log(string.format("<?%s %s?>",target,content))
		end,
		comment = function(content)
			mw.log(string.format("<!-- %s -->",content))
		end,
		startElement = function(name,nsURI)
			mw.log(string.format("<%s%s>",name,nsURI and (" ("..nsURI..")") or ""))
		end,
		attribute = function(name,value,nsURI)
			mw.log(string.format("  %s=%q%s",name,value,nsURI and (" ("..nsURI..")") or ""))
		end,
		text = function(text)
			mw.log(string.format("  text: %q",text))
		end,
		closeElement = function(name,nsURI)
			mw.log(string.format("</%s>",name))
		end,
	}
}

function SLAXML:parser(callbacks)
	return { _call=callbacks or self._call, parse=SLAXML.parse }
end

function SLAXML:parse(xml,options)
	if not options then options = { stripWhitespace=false } end

	-- Cache references for maximum speed
	local find, sub, gsub, char, push, pop = string.find, string.sub, string.gsub, string.char, table.insert, table.remove
	local first, last, match1, match2, match3, pos2, nsURI
	local unpack = unpack or table.unpack
	local pos = 1
	local state = "text"
	local textStart = 1
	local currentElement={}
	local currentAttributes={}
	local currentAttributeCt
	local nsStack = {}

	local entityMap  = { ["lt"]="<", ["gt"]=">", ["amp"]="&", ["quot"]='"', ["apos"]="'" }
	local entitySwap = function(orig,n,s) return entityMap[s] or n=="#" and char(s) or orig end
	local function unescape(str) return gsub( str, '(&(#?)([%d%a]+);)', entitySwap ) end
	local anyElement = false

	local function finishText()
		if first>textStart and self._call.text then
			local text = sub(xml,textStart,first-1)
			if options.stripWhitespace then
				text = gsub(text,'^%s+','')
				text = gsub(text,'%s+$','')
				if #text==0 then text=nil end
			end
			if text then self._call.text(unescape(text)) end
		end
	end

	local function findPI()
		first, last, match1, match2 = find( xml, '^<%?([:%a_][:%w_.-]*) ?(.-)%?>', pos )
		if first then
			finishText()
			if self._call.pi then self._call.pi(match1,match2) end
			pos = last+1
			textStart = pos
			return true
		end
	end

	local function findComment()
		first, last, match1 = find( xml, '^<!%-%-(.-)%-%->', pos )
		if first then
			finishText()
			if self._call.comment then self._call.comment(match1) end
			pos = last+1
			textStart = pos
			return true
		end
	end

	local function nsForPrefix(prefix)
		for i=#nsStack,1,-1 do if nsStack[i][prefix] then return nsStack[i][prefix] end end
		error(("Cannot find namespace for prefix %s"):format(prefix))
	end

	local function startElement()
		anyElement = true
		first, last, match1 = find( xml, '^<([%a_][%w_.-]*)', pos )
		if first then
			currentElement[2] = nil
			finishText()
			pos = last+1
			first,last,match2 = find(xml, '^:([%a_][%w_.-]*)', pos )
			if first then
				currentElement[1] = match2
				currentElement[2] = nsForPrefix(match1)
				match1 = match2
				pos = last+1
			else
				currentElement[1] = match1
				for i=#nsStack,1,-1 do if nsStack[i]['!'] then currentElement[2] = nsStack[i]['!']; break end end
			end
			currentAttributeCt = 0
			push(nsStack,{})
			return true
		end
	end

	local function findAttribute()
		first, last, match1 = find( xml, '^%s+([:%a_][:%w_.-]*)%s*=%s*', pos )
		if first then
			pos2 = last+1
			first, last, match2 = find( xml, '^"([^<"]*)"', pos2 ) -- FIXME: disallow non-entity ampersands
			if first then
				pos = last+1
				match2 = unescape(match2)
			else
				first, last, match2 = find( xml, "^'([^<']*)'", pos2 ) -- FIXME: disallow non-entity ampersands
				if first then
					pos = last+1
					match2 = unescape(match2)
				end
			end
		end
		if match1 and match2 then
			local currentAttribute = {match1,match2}
			local prefix,name = string.match(match1,'^([^:]+):([^:]+)$')
			if prefix then
				if prefix=='xmlns' then
					nsStack[#nsStack][name] = match2
				else
					currentAttribute[1] = name
					currentAttribute[3] = nsForPrefix(prefix)
				end
			else
				if match1=='xmlns' then
					nsStack[#nsStack]['!'] = match2
					currentElement[2] = match2
				end
			end
			currentAttributeCt = currentAttributeCt + 1
			currentAttributes[currentAttributeCt] = currentAttribute
			return true
		end
	end

	local function findCDATA()
		first, last, match1 = find( xml, '^<!%[CDATA%[(.-)%]%]>', pos )
		if first then
			finishText()
			if self._call.text then self._call.text(match1) end
			pos = last+1
			textStart = pos
			return true
		end
	end

	local function closeElement()
		first, last, match1 = find( xml, '^%s*(/?)>', pos )
		if first then
			state = "text"
			pos = last+1
			textStart = pos

			if self._call.startElement then self._call.startElement(unpack(currentElement)) end
			if self._call.attribute then for i=1,currentAttributeCt do self._call.attribute(unpack(currentAttributes[i])) end end

			if match1=="/" then
				pop(nsStack)
				if self._call.closeElement then self._call.closeElement(unpack(currentElement)) end
			end
			return true
		end
	end

	local function findElementClose()
		first, last, match1, match2 = find( xml, '^</([%a_][%w_.-]*)%s*>', pos )
		if first then
			nsURI = nil
			for i=#nsStack,1,-1 do if nsStack[i]['!'] then nsURI = nsStack[i]['!']; break end end
		else
			first, last, match2, match1 = find( xml, '^</([%a_][%w_.-]*):([%a_][%w_.-]*)%s*>', pos )
			if first then nsURI = nsForPrefix(match2) end
		end
		if first then
			finishText()
			if self._call.closeElement then self._call.closeElement(match1,nsURI) end
			pos = last+1
			textStart = pos
			pop(nsStack)
			return true
		end
	end

	while pos<#xml do
		if state=="text" then
			if not (findPI() or findComment() or findCDATA() or findElementClose()) then		
				if startElement() then
					state = "attributes"
				else
					first, last = find( xml, '^[^<]+', pos )
					pos = (first and last or pos) + 1
				end
			end
		elseif state=="attributes" then
			if not findAttribute() then
				if not closeElement() then
					error("Was in an element and couldn't find attributes or the close.")
				end
			end
		end
	end

	if not anyElement then error("Parsing did not discover any elements") end
	if #nsStack > 0 then error("Parsing ended with unclosed elements") end
end

return SLAXML