/*
Title: XPath Quick Reference
License:
- Version 3.13c by Titan <http://www.autohotkey.net/~Titan/#xpath>
- GNU General Public License 3.0 or higher <http://www.gnu.org/licenses/gpl-3.0.txt>
*/
/*
Function: xpath
Selects nodes and attributes from a standard xpath expression.
Parameters:
doc - an xml object returned by xpath_load()
step - a valid XPath 2.0 expression
set - (optional) text content to replace the current selection
Returns:
The XML source of the selection.
If the content was set to be modified the previous value will be returned.
Remarks:
Since multiple return values are seperated with a comma,
any found within the text content will be entity escaped as ,.
To get or set the text content of a node use the text function,
e.g. /root/node/text(); this behaviour is always assumed within predicates.
For performance reasons count() and last() do not update with the creation of new nodes
so you may need to alter the results accordingly.
*/
xpath(ByRef doc, step, set = "") {
static sc, scb = "" ; use EOT (\x04) as delimiter
If step contains %scb% ; i.e. illegal character
Return
sid := scb . &doc . ":" ; should be unique identifier for current XML object
If (InStr(step, "select:") == 1) { ; for quick selection
stsl = 1
StringTrimLeft, step, step, 7
}
If (InStr(step, "/") == 1)
str = 1 ; root selected
Else {
StringGetPos, p, sc, %sid% ; retrieve previous path
If ErrorLevel = 1
t = /
Else StringMid, t, sc, p += StrLen(sid) + 1, InStr(sc, scb, "", p + 2) - p
step = %t%/%step%
}
; normalize path e.g. /root/node/../child/. becomes /root/child:
step := RegExReplace(step, "(?<=^|/)(?:[^\[/@]+/\.{2}|\.)(?:/|$)|^.+(?=//)")
If (str == 1 or stsl == 1) { ; if not relative path and no select:
; remove last node and trailing attributes and functions:
xpr := RegExReplace(step, (str == 1 and stsl != 1 ? "/(?:\w+:)?\w+(?:\[[^\]]*])?" : "")
. "(?:/@.*?|/\w+\(\))*$")
StringReplace, xpr, xpr, [-1], [1], All ; -1 become just 1
StringReplace, xpr, xpr, [+1], [last()], All ; +1 becomes last()
StringGetPos, p, sc, %sid%
If ErrorLevel = 1
sc = %sc%%sid%%xpr%%scb% ; add record or update as necessary:
Else sc := SubStr(sc, 1, p += StrLen(sid)) . xpr . SubStr(sc, InStr(sc, scb, "", p))
}
; for unions call each operand seperately and join results:
If (InStr(step, "|")) {
StringSplit, s, step, |, `n`t `r
Loop, %s0%
res .= xpath(doc, s%A_Index%, set) . ","
Return, SubStr(res, 1, -1)
}
Else If (InStr(step, "//") == 1) { ; for wildcard selectors use regex searching mode:
StringTrimLeft, step, step, 2
re = 1
rew = 1
xp = /(?:(?:\w+:)?\w+)*/
}
NumPut(160, doc, 0, "UChar") ; unmask variable
; resolve xpath components to: absolute node path, attributes and predicates (if any)
Loop, Parse, step, /
{
s = %A_LoopField%
If (InStr(s, "*") == 1) ; regex mode for wildcards
{
re = 1
s := "(?:\w+:)?\w+" . SubStr(s, 2)
}
Else If (InStr(s, "@") == 1) { ; if current step is attribute:
StringTrimLeft, atr, s, 1
Continue
}
StringGetPos, p, s, [ ; look for predicate opening [...]
If ErrorLevel = 0
{
If s contains [+1],[-1] ; for child nodes record creation instructions in a list
{
If A_Index = 2 ; root node creation
{
StringLeft, t, s, p
t = <%t%/:: ></%t%/>
If (InStr(s, "+1")) {
If doc =
doc = .
doc = %doc%%t%
}
Else doc = .%t%%doc%
StringLeft, s, s, InStr(s, "[") - 1
}
Else {
nw = %nw%%s%
Continue
}
}
Else { ; i.e. for conditional predicates
If (InStr(s, "last()")) { ; finding last node
StringLeft, t, s, p
t := "<" . SubStr(xp, 2) . t . "/:: "
If re ; with regex:
{
os = 0
Loop
If (!os := RegExMatch(doc, t, "", 1 + os)) {
t = %A_Index%
Break
}
t--
}
Else { ; otherwise using StringReplace
StringReplace, doc, doc, %t%, %t%, UseErrorLevel
t = %ErrorLevel%
}
If (RegExMatch(s, "i)last\(\)\s*\-\s*(\d+)", a)) ; i.e. [last() - 2]
StringReplace, s, s, %a%, % t - a1
Else StringReplace, s, s, last(), %t%
}
; concat. the predicate to the list against the current absolute path:
ax = %ax%%xp%%s%
StringLeft, s, s, p
}
}
Else If (InStr(s, "()")) { ; if step is function, add to list
fn = %fn%+%s% ; use + prefix to prevent overlapping naming conflicts
Continue
}
; finally, if step is not any of the above, assume it's the name of a child node
xp = %xp%%s%/ ; ... and add to list, forming the absolute path
}
If (xp == "" or xp == "/") ; i.e. error cases
Return
; remove initial root selector (/) as this is how the parser works by default:
StringTrimLeft, xp, xp, 1
StringTrimRight, ax, ax, 1
StringTrimRight, nw, nw, 1
ct = 0 ; counter
os = 0 ; offset for main loop starts at zero
Loop {
; find offset of next element, and its closing tag offset:
If re
os := RegExMatch(doc, "<" . xp . ":: ", "", 1 + os)
, osx := RegExMatch(doc, "</" . xp . ">", rem, os) + StrLen(rem)
Else {
StringGetPos, osx, doc, </%xp%>, , os := InStr(doc, "<" . xp . ":: ", true, 1 + os)
osx += 4 + StrLen(xp)
}
If os = 0 ; stop looping when no more tags are found
Break
; predicate parser:
If ax !=
{
sk = 0
Loop, Parse, ax, ] ; for each predicate
{
; split components to: (1) path, (2) selector, (3) operator, (4) quotation char, (5) operand
If (!RegExMatch(A_LoopField, "/?(.*?)\[(.+?)(?:\s*([<>!=]{1,2})\s*(['""])?(.+)(?(4)\4))?\s*$", a))
Continue
a1 = %a1%/
If re
RegExMatch(rem, "(?<=^</)" . a1, a1)
If a2 is integer ; i.e. match only certain index
{
StringGetPos, t, a1, /, R2
StringMid, t, a1, 1, t + 1
t := InStr(SubStr(doc, 1, os), "<" . t . ":: ", true, 0)
; extract parent node:
StringMid, sub, doc, t, InStr(doc, ">", "", os) - t
xpf := "<" . a1 . ":: "
; get index of current element within parent node:
StringReplace, sub, sub, %xpf%, %xpf%, UseErrorLevel
If a2 != %ErrorLevel%
sk = 1
Continue
}
StringReplace, xp, xp, /, /, UseErrorLevel
t = %ErrorLevel%
StringReplace, a1, a1, /, /, UseErrorLevel
; extract result for deep analysis
If t = %ErrorLevel% ; i.e. /root/node[child='test']
StringMid, sub, doc, os, osx - os
Else StringMid, sub, doc
, t := InStr(SubStr(doc, 1, os), "<" . a1 . ":: ", true, 0)
, InStr(doc, "</" . a1 . ">", true, t) + 1
If a2 = position()
sub = %i%
Else If (InStr(a2, "@") == 1) ; when selector is an attribute:
RegExMatch(SubStr(sub, 1, InStr(sub, ">"))
, a3 == "" ? "\b" . SubStr(a2, 2) . "=([""'])[^\1]*?\1"
: "\b(?<=" . SubStr(a2, 2) . "=([""']))[^\1]+?(?=\1)", sub)
Else ; otherwise child node:
{
If a2 = . ; if selector is current node don't append to path:
a2 = /
Else a2 = %a2%/
StringMid, sub, sub
, t := InStr(sub, ">", "", InStr(sub, "<" . a1 . a2 . ":: ", true) + 1) + 1
, InStr(sub, "</" . a1 . a2 . ">", true) - t
}
; dynamic mini expression evaluator:
sk += !(a3 == "" ? (sub != "")
: a3 == "=" ? sub == a5
: a3 == "!=" ? sub != a5
: a3 == ">" ? sub > a5
: a3 == ">=" ? sub >= a5
: a3 == "<" ? sub < a5
: a3 == "<=" ? sub <= a5)
}
If sk != 0 ; if conditions were not met for this result, skip it
Continue
}
If nw != ; for node creation
{
If re
nwp := SubStr(rem, 3, -1)
Else nwp = %xp%
Loop, Parse, nw, ]
{
StringLeft, nwn, A_LoopField, InStr(A_LoopField, "[") - 1
nwn = %nwn%/
nwt = <%nwp%%nwn%:: ></%nwp%%nwn%>
If (t := InStr(A_LoopField, "-1")
? InStr(doc, ">", "", InStr(doc, "<" . nwp . ":: ", true, os) + 1) + 1
: InStr(doc, "</" . nwp . ">", true, os))
os := t
StringLen, osx, nwt
osx += os
doc := SubStr(doc, 1, os - 1) . nwt . SubStr(doc, os)
nwp = %nwp%%nwn%
}
StringLen, t, nwp
If (InStr(fn, "+text()") and atr == "")
os += t + 5, osx -= t + 3
}
If atr !=
{
; extract attribute offsets, with surrounding declaration if text() is not used:
If (t := RegExMatch(SubStr(doc, os, InStr(doc, ">", "", os) - os), InStr(fn, "+text()")
? "(?<=\b" . atr . "=([""']))[^\1]*?(?=\1)"
: "\b" . atr . "=([""'])[^\1]*?\1", rem))
os += t - 1, osx := os + StrLen(rem)
Else { ; create attribute
os := InStr(doc, ">", "", os + 1)
, doc := SubStr(doc, 1, os - 1) . " " . atr . "=""""" . SubStr(doc, os)
, osx := ++os + StrLen(atr) + 3
If (InStr(fn, "+text()"))
osx := os += StrLen(atr) + 2
}
}
Else If (InStr(fn, "+text()") and nw == "") ; for text content:
os := InStr(doc, ">", "", os) + 1, osx := InStr(doc, re ? rem : "</" . xp . ">", true, os)
If InStr(fn, "+index-of()") ; get index rather than content
sub = %A_Index%
Else StringMid, sub, doc, os, osx - os ; extract result
If (InStr(fn, "+count()")) ; increment counter if count() function is used
ct++
Else res = %res%%sub%, ; ... and concat to list
If (set != "" or InStr(fn, "+remove()")) ; modify or remove...
setb = %setb%%os%.%osx%| ; mark for modification
}
If setb !=
{
If (InStr(set, "node:") == 1) {
set := SubStr(xpath(doc, SubStr(set, 6) . "/rawsrc()"), 2)
StringReplace, set, set, <, <%xp%, All
StringReplace, set, set, <%xp%/, </%xp%, All
NumPut(160, doc, 0, "UChar")
}
StringTrimRight, setb, setb, 1
Loop, Parse, setb, |
{
StringSplit, setp, A_LoopField, .
StringLen, t, xp
If (InStr(fn, "+append()"))
setp2 := setp1 := setp2 - t - 3
Else If (InStr(fn, "+prepend()"))
setp2 := setp1 := InStr(doc, ">", "", setp1) + 2
doc := SubStr(doc, 1, setp1 - 1) . set . SubStr(doc, setp2) ; dissect then insert new value
}
}
If (InStr(fn, "+count()"))
res = %ct%, ; trailing char since SubStr is used below
nsid := scb . &doc . ":" ; update sid as necessary
If nsid != %sid%
StringReplace, sc, sc, %sid%, %nsid%
NumPut(0, doc, 0, "UChar") ; remask variable to prevent external editing
StringTrimRight, res, res, 1
If (InStr(fn, "+rawsrc()")) {
StringTrimLeft, t, xpr, 1
StringReplace, res, res, <%t%/, <, All
StringReplace, res, res, </%t%, <, All
StringReplace, res, res, `,, `,%scb%, All
Return, scb . res
}
; remove trailing comma and absolute paths from result before returning:
Return, RegExReplace(res, "S)(?<=<)(\/)?(?:(\w+)\/)+(?(1)|:: )", "$1$2")
}
/*
Function: xpath_save
Saves an XML document to file or returns the source.
Parameters:
doc - an xml object returned by xpath_load()
src - (optional) a path to a file where the XML document should be saved;
if the file already exists it will be replaced
Returns:
False if there was an error in saving the document, true otherwise.
If the src parameter is left blank the source code of the document is returned instead.
*/
xpath_save(ByRef doc, src = "") {
xml := RegExReplace(SubStr(doc, 2), "S)(?<=<)(\/)?(?:(\w+)\/)+(?(1)|:: )", "$1$2") ; remove metadata
xml := RegExReplace(xml, "<([\w:]+)([^>]*)><\/\1>", "<$1$2 />") ; fuse empty nodes
;xml := RegExReplace(xml, " (?=(?:\w+:)?\w+=['""])") ; remove prepending whitespace on attributes
xml := RegExReplace(xml, "^\s+|\s+$") ; remove start and leading whitespace
If InStr(xml, "<?xml") != 1 ; add processor instruction if there isn't any:
xml = <?xml version="1.0" encoding="iso-8859-1"?>`n%xml%
StringReplace, xml, xml, `r, , All ; normalize linefeeds:
StringReplace, xml, xml, `n, `r`n, All
sp := " "
StringLen, sl, sp
s =
VarSetCapacity(sxml, StrLen(xml) * 1.1)
Loop, Parse, xml, <, `n`t `r
{
If A_LoopField =
Continue
If (sb := InStr(A_LoopField, "/") == 1)
StringTrimRight, s, s, sl
sxml = %sxml%`n%s%<%A_LoopField%
If sb
StringTrimRight, s, s, sl
If (InStr(A_LoopField, "?") != 1 and InStr(A_LoopField, "!") != 1
and !InStr(A_LoopField, "/>"))
s .= sp
}
StringTrimLeft, sxml, sxml, 1
sxml := RegExReplace(sxml, "(\n(?:" . sp . ")*<((?:\w+:)?\w+\b)[^<]+?)\n(?:"
. sp . ")*</\2>", "$1</$2>")
If src = ; if save path not specified return the XML document:
Return, sxml
FileDelete, %src% ; delete existing file
FileAppend, %sxml%, %src% ; create new one
Return, ErrorLevel ; return errors, if any
}
/*
Function: xpath_load
Loads an XML document.
Parameters:
doc - a reference to the loaded XML file as a variable, to be used in other functions
src - (optional) the document to load, this can be a file name or a string,
if omitted the first paramter is used as the source
Returns:
False if there was an error in loading the document, true otherwise.
*/
xpath_load(ByRef doc, src = "") {
If src = ; if source is empty assume the out variable is the one to be loaded
src = %doc%
Else If FileExist(src) ; otherwise read from file (if it exists)
FileRead, src, %src%
If src not contains <,>
Return, false
; combined expressions slightly improve performance:
src := RegExReplace(src, "<((?:\w+:)?\w+\b)([^>]*)\/\s*>", "<$1$2></$1>") ; defuse nodes
, VarSetCapacity(doc, VarSetCapacity(xml, StrLen(src) * 1.5) * 1.1) ; pre-allocate enough space
Loop, Parse, src, < ; for each opening tag:
{
If (A_Index == 2 and InStr(A_LoopField, "?xml") == 1)
Continue
Else If (InStr(A_LoopField, "?") == 1) ; ignore all other processor instructions
xml = %xml%<%A_LoopField%
Else If (InStr(A_LoopField, "![CDATA[") == 1) { ; escape entities in CDATA sections
cdata := SubStr(A_LoopField, 9, -3)
StringReplace, cdata, cdata, ", ", All
StringReplace, cdata, cdata, &, &, All
StringReplace, cdata, cdata, ', ', All
StringReplace, cdata, cdata, <, <, All
StringReplace, cdata, cdata, >, >, All
xml = %xml%%cdata%
}
Else If (!pos := RegExMatch(A_LoopField, "^\/?(?:\w+:)?\w+", tag)) ; if this isn't a valid tag:
{
If A_LoopField is not space
xml = %xml%<%A_LoopField% ; convert to escaped entity value
}
Else {
StringMid, ex, A_LoopField, pos + StrLen(tag) ; get tag name
If InStr(tag, "/") = 1 { ; if this is a closing tag:
xml = %xml%</%pre%%ex% ; close tag
StringGetPos, pos, pre, /, R2
StringLeft, pre, pre, pos + 1
}
Else {
pre = %pre%%tag%/
xml = %xml%<%pre%:: %ex%
}
}
}
StringReplace, doc, xml, `,, ,, All ; entity escape commas (which are used as array delimiters)
NumPut(0, doc := " " . doc, 0, "UChar") ; mask variable from text display with nullbyte
Return, true ; assume sucessful load by this point
}