ooxml: tokenise.xsl annotate

annotate tokenise.xsl @ 55:05cf88c20cc5

sample I/O pair, with range annotations

author	Henry S. Thompson <ht@markup.co.uk>
date	Tue, 30 May 2017 09:52:49 +0100
parents	191c95187e87
children

rev	line source
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	1 <?xml version='1.0'?>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	2 <!DOCTYPE xsl:stylesheet SYSTEM "../../../lib/xml/xsl.dtd" >
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	3 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="3.0" xmlns:s="http://schemas.openxmlformats.org/spreadsheetml/2006/main" xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:e="http://markup.co.uk/excel" exclude-result-prefixes="xs s e xf" xmlns="http://markup.co.uk/excel" xmlns:xf="http://www.w3.org/2005/xpath-functions"> <xsl:param name="sheet-number"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	4 <xsl:param name="xlDir"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	5
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	6 <xsl:variable name="pat1">("[^"]")\|(\{[^}]+})\|(,)\|([^=\-+/();:,.$<>^!]+(?:\.[^=\-+/();:,.$<>^!]+)\()\|([)])\|(^=\|\()\|((?:(?:'[^']+')\|(?:\[[0-9]+\][^!])\|(?:[a-zA-Z_][a-zA-Z0-9._])!))\|(\$?[A-Z]+\$?[0-9]+)\|([a-zA-Z_\\][a-zA-Z0-9._]*)\|(.)</xsl:variable>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	7 <xsl:variable name="pat2">("[^"]")\|(\{[^}]+})\|(,)\|([^=\-+/();:,.$<>^!]+(?:\.[^=\-+/();:,.$<>^!]+)\()\|([)])\|(^=\|\()\|((?:'[^']+'!)\|(?:[\[0-9A-Za-z_][^=\-+/();:,.$<>^!]!))\|(\$?[A-Z]+\$?[0-9]+)\|([a-zA-Z_\\][a-zA-Z0-9._]*)\|(.)</xsl:variable>
4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	8 <xsl:param name="pat" select="$pat2"/><!-- xsl:param for refinement debugging by passing in the pattern -->
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	9
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	10 <xsl:variable name="workbook" select="document(concat($xlDir,'/workbook.xml'))/*"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	11 <xsl:variable name="sheet-name" select="$workbook/s:sheets/s:sheet[@sheetId=$sheet-number]/@name"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	12
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	13 <xsl:function name="e:lookup" as="xs:string*">
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	14 <xsl:param name="name" as="xs:string" required="yes"/>
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	15 <xsl:value-of select="string($workbook/s:definedNames/s:definedName[@name=$name])"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	16 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	17
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	18 <xsl:function name="e:tokenise" as="element()">
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	19 <!-- Tokenise a formula, recursively wrt variables
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	20 Output is composed of e:* as follows:
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	21 b: Boolean constant
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	22 c: A list (function parameter) separator
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	23 e: An external (variable, cell or range) reference
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	24 l: The beginning of the formula or an opening paren
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	25 m: A constant matrix
40 ac6d1ca099f7 simplified rect output, refs (not refs2) working again Henry S. Thompson <ht@markup.co.uk> parents: 39 diff changeset	26 o: A function name followed by an opening parenthesis
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	27 p: A close-paren
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	28 q: A text (delimited by double quotes)
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	29 r: A range reference (two children, either e or s or u (unsupported))
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	30 s: A single-cell reference
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	31 v: A variable name [should only occur inside e]
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	32 x: Amalgamated single characters not matched by anything else
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	33 -->
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	34 <xsl:param name="formula" as="xs:string" required="yes"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	35 <!-- The row and column number of the cell whence the formula came -->
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	36 <xsl:param name="row" required="yes" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	37 <xsl:param name="col" required="yes" as="xs:integer"/>
40 ac6d1ca099f7 simplified rect output, refs (not refs2) working again Henry S. Thompson <ht@markup.co.uk> parents: 39 diff changeset	38 <xsl:if test="false()"><xsl:message>tok: <xsl:value-of select="$formula"/></xsl:message></xsl:if>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	39 <xsl:sequence select="
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	40 let $tokens := analyze-string($formula,$pat)/xf:match/xf:group
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	41 return e:tok1($tokens,count($tokens),1,$row,$col,())"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	42 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	43
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	44 <xsl:function name="e:tok1" as="element()">
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	45 <xsl:param name="tokens" as="element(xf:group)*" required="yes"/>
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	46 <xsl:param name="n" required="yes" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	47 <xsl:param name="i" required="yes" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	48 <xsl:param name="row" required="yes" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	49 <xsl:param name="col" required="yes" as="xs:integer"/>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	50 <xsl:param name="soFar" as="element()"/>
4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	51 <xsl:variable name="last" select="$soFar[count($soFar)]"></xsl:variable>
53 8000dbed09df better, now failing on [1]\!'AES,DIVIDEND' Henry S. Thompson <ht@markup.co.uk> parents: 52 diff changeset	52 <xsl:if test="false()"><xsl:message>tok1@<xsl:value-of select="$row"/>,<xsl:value-of select="$col"/>: <xsl:value-of select="$n"/>\|<xsl:value-of select="$i"/>\|<xsl:value-of select="if ($i=1) then string-join($tokens,'\') else string-join(($tokens[$i],':',$tokens[$i]/@nr,',',$tokens[$i+1],':',$tokens[$i+1]/@nr,',',$tokens[$i+2],':',$tokens[$i+2]/@nr),'')"/>\|<xsl:value-of select="if ($last instance of element()) then name($last) else 'bogus'"/>\|<xsl:value-of select="string($last)"/></xsl:message></xsl:if>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	53 <!-- Tail-recursion incrementing index $i into $tokens -->
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	54 <xsl:sequence select="
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	55 if ($i gt $n)
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	56 then $soFar
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	57 else
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	58 let $next := e:expand($tokens,$i,true(),$row,$col),
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	59 $j := $next?1,
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	60 $res := $next?2 return
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	61 e:tok1($tokens,$n,$j,$row,$col,($soFar,$res))"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	62 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	63
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	64 <xsl:function name="e:expand" as="array(*)">
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	65 <xsl:param name="tokens" required="yes" as="element(xf:group)*"/>
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	66 <xsl:param name="i" required="yes" as="xs:integer"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	67 <xsl:param name="local" required="yes" as="xs:boolean"/>
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	68 <xsl:param name="row" required="yes" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	69 <xsl:param name="col" required="yes" as="xs:integer"/>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	70 <!-- Consume one or more tokens, returns new index and an element -->
40 ac6d1ca099f7 simplified rect output, refs (not refs2) working again Henry S. Thompson <ht@markup.co.uk> parents: 39 diff changeset	71 <xsl:if test="false()"><xsl:message>exp: <xsl:value-of select="$tokens[$i]/@nr"/>:<xsl:value-of select="$tokens[$i]"/>,<xsl:value-of select="$tokens[$i+1]"/></xsl:message></xsl:if>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	72 <xsl:sequence select="
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	73 let $t := $tokens[$i],
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	74 $r := $tokens[$i + 1] return
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	75 if ($t/@nr=1) then e:exp1($i,'q',string($t))
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	76 else if ($t/@nr=2) then e:exp1($i,'m',string($t))
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	77 else if ($t/@nr=3) then e:exp1($i,'c',',')
40 ac6d1ca099f7 simplified rect output, refs (not refs2) working again Henry S. Thompson <ht@markup.co.uk> parents: 39 diff changeset	78 else if ($t/@nr=4) then e:exp1($i,'o',string($t))
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	79 else if ($t/@nr=5) then e:exp1($i,'p',')')
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	80 else if ($t/@nr=6) then e:exp1($i,'l',string($t))
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	81 else if ($t/@nr=7)
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	82 then let $xref := substring-before($t,'!') return
4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	83 if ($xref=('[0]',$sheet-name))
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	84 then (: it's a local reference after all :)
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	85 e:expand($tokens,$i+1,true(),$row,$col)
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	86 else let $ext := if ($tokens[$i+1]="'")
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	87 then (: odd case, see notes wrt enron/kenneth_lay :)
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	88 e:oddx($tokens,$i+2,$tokens[$i+1])
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	89 else
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	90 e:expand($tokens,$i+1,false(),$row,$col) return
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	91 [$ext?1,e:external($xref,$ext?2)]
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	92 else if ($t/@nr=10) then e:misc($tokens,$i+1,string($t))
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	93 else if ($r[@nr=10 and .=':'])
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	94 then (: a range, takes priority :)
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	95 [$i+3,e:range($tokens,$i,$local,$row,$col)]
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	96 else if ($t/@nr=8) then [$i+1,e:single(string($t),$row,$col)]
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	97 else if ($t/@nr=9)
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	98 then if (matches($t,'^(true\|false)$','i'))
4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	99 then e:exp1($i,upper-case($t),'b')
4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	100 else (: a variable name, I think :) if ($local)
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	101 then let $sub := e:tokenise(e:lookup(string($t)),$row,$col) return
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	102 [$i+1,$sub]
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	103 else (: can't expand :) e:exp1($i,'v',string($t))
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	104 else (: shouldn't ever get here :) ()"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	105 </xsl:function>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	106
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	107 <xsl:function name="e:misc">
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	108 <xsl:param name="tokens" as="element(xf:group)*"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	109 <xsl:param name="i" as="xs:integer"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	110 <xsl:param name="soFar" as="xs:string"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	111 <!-- Amalgamate bits of type 10 and wrap in <x> -->
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	112 <xsl:variable name="content" select="e:amalgamate($tokens,$i,$soFar,10,'')"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	113 <xsl:variable name="x"><x><xsl:value-of select="$content?2"/></x></xsl:variable>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	114 <xsl:sequence select="[$content?1,$x/*]"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	115 </xsl:function>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	116
54 191c95187e87 working now on enron1k, uli1/sheet1 Henry S. Thompson <ht@markup.co.uk> parents: 53 diff changeset	117 <xsl:function name="e:oddx">
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	118 <xsl:param name="tokens" as="element(xf:group)*"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	119 <xsl:param name="i" as="xs:integer"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	120 <xsl:param name="soFar" as="xs:string"/><!-- Will always be a lonely ' -->
54 191c95187e87 working now on enron1k, uli1/sheet1 Henry S. Thompson <ht@markup.co.uk> parents: 53 diff changeset	121 <!-- Amalgamate bits of type 3, 9 and 10 as <v> for wrapping in <e> -->
191c95187e87 working now on enron1k, uli1/sheet1 Henry S. Thompson <ht@markup.co.uk> parents: 53 diff changeset	122 <xsl:variable name="content" select="e:amalgamate($tokens,$i,$soFar,(10,9,3),$soFar)"/>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	123 <xsl:variable name="x"><v><xsl:value-of select="
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	124 if ($tokens[$content?1]=$soFar)
53 8000dbed09df better, now failing on [1]\!'AES,DIVIDEND' Henry S. Thompson <ht@markup.co.uk> parents: 52 diff changeset	125 then concat($content?2,$soFar)
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	126 else e:shouldnt('oddx',$tokens[$content?1])"/></v></xsl:variable>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	127 <xsl:sequence select="[$content?1+1,$x/*]"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	128 </xsl:function>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	129
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	130 <xsl:function name="e:amalgamate" as="array(*)">
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	131 <xsl:param name="tokens" as="element(xf:group)*"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	132 <xsl:param name="i" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	133 <xsl:param name="soFar" as="xs:string"/>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	134 <xsl:param name="nrs" as="xs:integer*"/>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	135 <xsl:param name="stop" as="xs:string"/>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	136 <xsl:choose>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	137 <xsl:when test="$tokens[$i]/@nr=$nrs and $tokens[$i]!=$stop">
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	138 <xsl:sequence select="e:amalgamate($tokens,$i+1,concat($soFar,
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	139 string($tokens[$i])),
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	140 $nrs,$stop)"/>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	141 </xsl:when>
4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	142 <xsl:otherwise>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	143 <xsl:if test="false()"><xsl:message>amal <xsl:value-of select="$nrs"/>: <xsl:value-of select="$soFar"/></xsl:message></xsl:if>
9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	144 <xsl:sequence select="[$i,$soFar]"/>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	145 </xsl:otherwise>
52 9bb415e0adc9 try to fix error processin odd REUTER\|IDN\!'.SPX' external ref Henry S. Thompson <ht@markup.co.uk> parents: 40 diff changeset	146 </xsl:choose>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	147 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	148
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	149 <xsl:function name="e:exp1" as="array(*)">
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	150 <xsl:param name="i" as="xs:integer"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	151 <xsl:param name="name" as="xs:string"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	152 <xsl:param name="val" as="xs:string"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	153 <xsl:variable name="elt">
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	154 <xsl:element name="{$name}" namespace="http://markup.co.uk/excel">
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	155 <xsl:value-of select="$val"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	156 </xsl:element>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	157 </xsl:variable>
40 ac6d1ca099f7 simplified rect output, refs (not refs2) working again Henry S. Thompson <ht@markup.co.uk> parents: 39 diff changeset	158 <xsl:if test="false()"><xsl:message>exp1: <xsl:value-of select="$name"/>\|<xsl:value-of select="$val"/>\|<xsl:value-of select="name($elt/*)"/></xsl:message></xsl:if>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	159 <xsl:sequence select="[$i+1,$elt/*]"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	160 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	161
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	162 <xsl:function name="e:single" as="element(e:s)">
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	163 <!-- I'm _guessing_ that external doesn't matter, i.e. that you
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	164 can copy an external relative ref and have it change -->
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	165 <xsl:param name="val" as="xs:string"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	166 <xsl:param name="row" as="xs:integer"/>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	167 <xsl:param name="col" as="xs:integer"/>
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	168 <xsl:sequence select="e:cr($val,$row,$col)"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	169 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	170
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	171 <xsl:function name="e:range" as="element(e:r)">
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	172 <xsl:param name="tokens" as="element(xf:group)*"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	173 <xsl:param name="i" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	174 <xsl:param name="local" as="xs:boolean"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	175 <xsl:param name="row" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	176 <xsl:param name="col" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	177 <xsl:variable name="l" select="$tokens[$i]"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	178 <xsl:variable name="r" select="$tokens[$i+2]"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	179 <xsl:variable name="lv" select="e:rPart($l,$local,$row,$col)"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	180 <xsl:variable name="rv" select="e:rPart($r,$local,$row,$col)"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	181 <r>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	182 <xsl:copy-of select="($lv,$rv)"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	183 </r>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	184 </xsl:function>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	185
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	186 <xsl:function name="e:rPart" as="element(*)">
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	187 <xsl:param name="g" as="element(xf:group)"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	188 <xsl:param name="local" as="xs:boolean"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	189 <xsl:param name="row" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	190 <xsl:param name="col" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	191 <xsl:sequence select="
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	192 if ($g/@nr=8) then e:single(string($g),$row,$col)
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	193 else if ($g/@nr=9)
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	194 then if ($local)
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	195 then let $tokens := e:tokenise(e:lookup(string($g)),
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	196 $row,$col) return
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	197 if (count($tokens)=1 and
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	198 $tokens[local-name()='s' or
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	199 (local-name()='e' and $tokens/e:s)])
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	200 then $tokens
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	201 else e:badRP(string($g),$g/@nr,$local,$row,$col,$tokens)
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	202 else e:var(string($g))
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	203 else e:badRP(string($g),$g/@nr,$local,$row,$col,())"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	204 </xsl:function>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	205
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	206 <xsl:function name="e:var" as="element(e:v)">
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	207 <xsl:param name="name" as="xs:string"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	208 <v><xsl:value-of select="$name"/></v>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	209 </xsl:function>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	210 <xsl:function name="e:badRP" as="element(e:u)">
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	211 <xsl:param name="s" as="xs:string"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	212 <xsl:param name="t" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	213 <xsl:param name="local" as="xs:boolean"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	214 <xsl:param name="row" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	215 <xsl:param name="col" as="xs:integer"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	216 <xsl:param name="toks" as="element()"/>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	217 <xsl:message>Bad range part in <xsl:value-of select="concat(e:n2a($col),$row)"/>: <xsl:value-of select="$s"/> of type <xsl:value-of select="$t"/> (<xsl:value-of select="if ($local) then 'local' else 'external'"/>: (<xsl:value-of select="string-join($toks,',')"/>)</xsl:message>
38 468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	218 <u r="{concat(e:n2a($col),$row)}" s="{$s}" t="{$t}" local="{$local}">
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	219 <xsl:value-of select="string-join($toks,',')"/>
468a6cf8bf0b big change wrt formulae compiles, crashes Henry S. Thompson <ht@markup.co.uk> parents: 37 diff changeset	220 </u>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	221 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	222
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	223 <xsl:function name="e:external" as="element(e:e)">
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	224 <xsl:param name="xref" as="xs:string" required="yes"/>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	225 <xsl:param name="ref" as="element(*)" required="yes"/>
39 4c6a341e75da big rework works on sample2, w/o refs processing Henry S. Thompson <ht@markup.co.uk> parents: 38 diff changeset	226 <e x="{$xref}"><xsl:sequence select="$ref"/></e>
37 ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	227 </xsl:function>
ac3cd8de7a10 towards big rework of tokenisation Henry S. Thompson <ht@markup.co.uk> parents: diff changeset	228 </xsl:stylesheet>

Mercurial > hg > ooxml

annotate tokenise.xsl @ 55:05cf88c20cc5