[文本处理] 求助批处理从一堆文本文件中提取关键词写到csv中

for f in *.xml; do awk -v RS='</item>' -F '\n' '/Soul Crystal/{for(i=1;i<=NF;i++){if($i~/Soul Crystal/){gsub(".*id=\"|\".*","",$(i+1));gsub(".*id=\"|\".*","",$(NF-1));print $(i+1)","$(NF-1)}}}' "$f" > "$f.csv"; doneCOPY

@echo off &setlocal enabledelayedexpansion
findstr /inc:"Soul Crystal" /inc:"</item>" "test.xml" 2>nul>tmp.0
(for /f "tokens=1 delims=:" %%k in (tmp.0) do (
	if not defined v1 (call :getV "test.xml" %%k v1) else (
		set/a "n=%%k-2"
		(call :getV "test.xml" !n! v2)
		echo,!v1!,!v2!
		set "v1="
	)
))>"test.csv"
del /q tmp.0
endlocal &pause &exit/b
:getV //%1=文件名 %2=行号 %3=返回值变量名
	for /f "usebackq skip=%~2 delims=" %%s in ("%~1") do (
		for /f tokens^=1-4^delims^=^" %%1 in ("%%~s") do (set "%~3=%%4")
		exit/b
	) & exit/bCOPY

@echo off
cd /d "%~dp0"
(for %%i in (*.xml) do (
	setlocal
	for /f "useback delims=" %%a in ("%%i") do (
		set str=%%a
		setlocal enabledelayedexpansion
		set "str1=!str:Soul Crystal=!"
		set "str2=!str:	</item>=!"
		if "!str!" neq "!str1!" (
			endlocal
			set m=1
		) else if defined m (
			endlocal
			for /f tokens^=4delims^=^" %%x in ("%%a") do (
				set n=%%x
			)
			set m=
		) else if "!str!" neq "!str2!" (
			endlocal
			if defined n (
				setlocal enabledelayedexpansion
				for /f tokens^=4delims^=^" %%x in ("!_str!") do (
					echo !n!,%%x
				)
				endlocal
				set n=
			)
		) else (
			endlocal
		)
		set _str=%%a
	)
	endlocal
))>data.csv
pauseCOPY

awk -v"OFS=," "/^\t+<item>$/,/^\t+<\/item>$/{if(/Soul Crystal/)exist_id=1;if(/^\t+<ingredient count=\"1\" id=\"[0-9]+\" \/>$/)if(exist_id)A=gensub(/^\t+<ingredient count=\"1\" id=\"([0-9]+)\" \/>$/,\"\\1\",\"g\",$0);if(/^\t+<production count=\"1\" id=\"[0-9]+\" \/>$/)if(exist_id)print A,gensub(/^\t+<production count=\"1\" id=\"([0-9]+)\" \/>$/,\"\\1\",\"g\",$0);if(/^\t+<\/item>$/)exist_id=0}" *.xml>out.csvCOPY

@if(0)==(0) echo off
dir /b *.xml | cscript -nologo -e:jscript "%~f0" > result.csv
pause & exit
@end
 
var fso = new ActiveXObject('Scripting.FileSystemObject');
 
while(!WSH.StdIn.AtEndOfStream) {
    var f = WSH.StdIn.ReadLine();
    var s = fso.OpenTextfile(f, 1).ReadAll();
    var m = s.match(/<item>(?:(?!<\/?item>)[\s\S])+<\/item>/g);
    for(var i=0; i<m.length; i++) {
        var m1 = m[i].match(/\bSoul Crystal\b.+\n(?:(?!\bid=).)+\bid="([^"]+)"/);
        var m2 = m[i].match(/\bproduction\b(?:(?!\bid=).)+\bid="([^"]+)"/);
        if(m1 && m2) WSH.Echo(m1[1] + ',' + m2[1]);
    }
}COPY

@if(0)==(0) echo off
dir /b *.xml | cscript -nologo -e:jscript "%~f0" > result.csv
pause & exit
@end
 
var xml = new ActiveXObject('Microsoft.XMLDOM');
var reg = /\bSoul Crystal\b/;
 
while(!WSH.StdIn.AtEndOfStream) {
    var f = WSH.StdIn.ReadLine();
    getXmlData(f);
}
 
function getXmlData(fileName) {
    xml.load(fileName);
    var nodes = xml.selectNodes('//item');
    for(var i=0; i<nodes.length; i++) {
        var s2 = nodes[i].lastChild.getAttribute('id');
        var childs = nodes[i].childNodes;
        for(var j=0; j<childs.length; j++) {
            if(childs[j].nodeName == '#comment' && reg.test(childs[j].text)) {
                var s1 = childs[j+1].getAttribute('id');
                WSH.Echo(s1 + ',' + s2);
                break;
            }
        }
    }
}COPY

@echo off&chcp 936
echo 进行中,请耐心等待
set "outfile=___output.csv"
 
cd.>"%outfile%"
for %%b in ("*.xml") do (
setlocal
echo,file,%%~b>>%outfile%
(for /f "skip=1 delims=" %%c in ('findstr /n .* "%%~b"') do (
set line1=
set /p line1=
set line2=%%c
 
if not defined item (
setlocal enabledelayedexpansion
if "!line1:<item>=!" neq "!line1!" (
for %%- in (1) do endlocal&set item=%%-) else endlocal
) else (
 
setlocal enabledelayedexpansion
if "!line1:Soul Crystal=!" neq "!line1!" (
set line3=!line2:* =!&set line3=!line3:~0,-2!
2>nul set /a !line3:id=,id!
for %%- in ("!id!") do endlocal&set o=%%~-&set match=1
if "!p!" == "!q!" endlocal
) else endlocal
 
if defined match (
setlocal enabledelayedexpansion
if "!line2:</item>=!" neq "!line2!" (
set line3=!line1:* =!&set line3=!line3:~0,-2!
2>nul set /a !line3:id=,id!
echo,!o!,!id!
endlocal&set o=&set match=&set item=
) else endlocal
)
 
)
))<"%%~b">>"%outfile%"
endlocal
)
echo,&echo 完成&pauseCOPY