Powershell 选择 HTML 文本

我有以下 powershell 脚本来打开多台计算机上的多个 HTML 文件以提取许可信息。寻找一些帮助提取内部文本元素<td> where id=2.

寻找提取：

产品、序列号、产品密钥、请求代码

ForEach ($system in (Get-Content C:\temp\computers.txt)) {
  $folder = "\\$system\c`$\ProgramData\Autodesk\AdLM\"
  Get-ChildItem $folder *.html  |
  ForEach-Object {
    # Get current filename without .html
    $c = $_.BaseName
    # Create new Internet Explorer Object
    $ie=New-Object -ComObject InternetExplorer.Application
    # $_ (dollar underscore) 'THIS' token. Typically refers to the item inside a foreach loop. Open current HTML file.
    $ie.Navigate("$_")
    while ($ie.busy -eq $true) {
      Start-Sleep -Milliseconds 500
    }

    $doc=$ie.Document
    $tables=$doc.getElementsByTagName('table')
     foreach ($table in $tables) {
        if ($div.innertext -match "Product: ") 
        if ($div.innertext -match "Serial number: ")
        if ($div.innertext -match "Product Key: ")
        if ($div.innertext -match "Request code: ")
    }
    $elements.innerText | ForEach-Object { Add-Content -path c:\temp\results.csv "$c,$system,$para1" }
  }
}

HTML 文件示例

<table width="100%" cellspacing=0 cellpadding=0>
    <tr><td align=left id=1 width=15%>Product:</td><td align=left id=2 width=35%><!--PREVINFO_START-->Autodesk Building Design Suite Premium 2016<!--PREVINFO_END--></td><td align=left id=1>&nbsp;</td><td align=left id=2>&nbsp;</td></tr>
    <tr><td align=left id=1 width=15%>Serial number:</td><td align=left id=2><!--PREVINFO_START-->999-9999999<!--PREVINFO_END--></td></tr>
</table>
<table width="100%" cellspacing=0 cellpadding=0>
    <tr><td align=left id=1 width=15%>Product key:</td><td align=left id=2>424242</td><td align=left id=1>&nbsp;</td><td align=left id=2>&nbsp;</td></tr>
</table>
<table width="100%" cellspacing=0 cellpadding=0>
    <tr><td align=left id=1 width=15%>Request code:</td><td align=left id=3 style='word-break:break-all'>7777 7777 7777 7777</td></tr>
    <tr><td align=left id=1>&nbsp;</td><td align=left id=3 style='word-break:break-all'>7777 7777 7777 7777</td></tr>
</table>

我的解决方案，

# Script to extract Autocad Licensing informatoin from HTML files
# 1. Add the computers you want to access into computers.txt 
# 2. Download the Script and execute.
# 3. Results will be saved to the $file variable (default is c:\tools\results.csv)

#File Paths
$file="C:\tools\results.csv";
$computerfile="C:\tools\computers.txt";
$computerfoldername="\c`$\ProgramData\Autodesk\AdLM\";

# Create empty computers array
$computers = @() 


ForEach ($system in (Get-Content $computerfile)) {
# Append computers with \\computer\foldername
$computers += "\\" + $system + $computerfoldername
}
 write-host "=========================================================" -foreground "red"
 write-host "MULTI-HTML FILE PARSING AND APPENDING TO CSV SCRIPT "
 write-host "Script Executing at "  (Get-Date).ToString()
 #folders in which html files are kept



write-host "Folders To Parse"
for ($i = 0; $i -lt $computers.Count ; $i++) {
  write-host $computers[$i]

}

write-host "Initializing Script";

# Create new Internet Explorer Object
$ie = new-object -com "InternetExplorer.Application"

# Iterate through each \\computer\folder in the list
for ($i = 0; $i -lt $computers.Count ; $i++) {

write-host ""
write-host "=========================================="
write-host ""
write-host "COMPUTER/FOLDER" + $computers[$i] -foreground "yellow"
write-host ""

# Get HTML files within current \\computer\folder
Get-ChildItem $computers[$i] *.html  |
ForEach-Object {
# For each HTML file within current \\computer\folder
 $innerfile = $computers[$i] + $_.BaseName + $_.Extension
 write-host "=========================================="


write-host "Fetching" + $innerfile
# The easiest way to accomodate for slowness of IE
Start-Sleep -Seconds 1
# open current HTML file
$ie.Navigate($innerfile)
# The easiest way to accomodate for slowness of IE
Start-Sleep -Seconds 1
$doc = $ie.Document

write-host "IE Initiated for file " + $innerfile ;
# Get all <tr> elements in teh HTML file.
$trs=$doc.getElementsByTagName('tr')
$requestcodeflag=0;

foreach ($table in $trs) {

        $machine=$innerfile;
        # Product Key    
        Try {
            if ($table.innertext.Substring(0,8) -eq "Product:") {
                write-host "Parsed Product ID " + $table.innertext.Substring(8)
                $product=$table.innertext.Substring(8);
            }
        }
        Catch{}


        # Serial number
        Try {
             if ($table.innertext.Substring(0,14) -eq "Serial number:") {
                write-host "Parsed Serial Number  " + $table.innertext.Substring(14)
                $serialno=$table.innertext.Substring(14);
            }
        }
        Catch{}



        # Product Key
        Try {
            if ($table.innertext.Substring(0,12) -eq "Product key:") {
                write-host "Parsed Product Key  " + $table.innertext.Substring(12)
                $productkey=$table.innertext.Substring(12);
            }
        }
        Catch{}


        # Request code 2
        if($requestcodeflag -eq "1") {
        Try {
            $requestcodeflag=2;
            $requestcode= $requestcode + " | " + $table.innertext;
            write-host "Parsed 2nd Request Code  " +  $table.innertext
            }
        Catch{}
        }


        # Request code 1st
        Try{
            if ($table.innertext.Substring(0,13) -eq "Request code:") {
                write-host "Parsed Request Code  " +  $table.innertext.Substring(13)
                $requestcode=$table.innertext.Substring(13);
                $requestcodeflag=1;
            }
        }
        Catch{}


}



write-host "Writing to CSV file " + $file
$NewLine = "{0},{1},{2},{3},{4}" -f $machine,$product,$serialno,$productkey,$requestcode
$NewLine | add-content -path $file
}
}
$ie.Quit()
write-host "=========================================="
write-host "Data is  appended to csv file" + $file
write-host "Executed Thank You "
write-host "=========================================="
write-host ""