Programmatically access document properties for Word 2007 documents

Is there a way in which I can programmatically access the document properties of a Word 2007 document?

I am open to using any language for this, but ideally it might be via a PowerShell script.

My overall aim is to traverse the documents somewhere on a filesystem, parse some document properties from these do开发者_开发知识库cuments, and then collate all of these properties back together into a new Word document.

I essentially want to automatically create a document which is a list of all documents beneath a certain folder of the filesystem; and this list would contain such things as the Title, Abstract and Author document properties; the CreateDate field; etc. for each document.

Check Hey, Scripting Guy! How Can I Retrieve the Custom Properties of a Microsoft Word Document? and maybe Hey, Scripting Guy! How Can I Add Custom Properties to a Microsoft Word Document?. That could help you or at least give an idea ;)

I needed to do this in PowerShell running on a server without MS Office applications installed. The trick, as suggested above, is to peek inside the office file and examine the embedded xml files within.

Here's a function that runs like a cmdlet, meaning you can simply save the script in your PowerShell scripts directory and call the function from any other PowerShell script.

# DocumentOfficePropertiesGet
# Example usage
#   From a PowerShell script:
#       $props = Invoke-Expression "c:\PowerShellScriptFolder\DocumentOfficePropertiesGet.ps1 -DocumentFullPathName ""d:\documents\my excel doc.xlsx"" -OfficeProperties ""dcterms:created;dcterms:modified"""

# Parameters

#    DocumentFullPathName -- full path and name of MS Office document
#    OfficeProperties -- semi-colon delimited string of property names as they
#              appear in the core.xml file. To see these names, rename any
#              MS Office document file to have the extension .zip, then look inside
#              the zip file. In the docProps folder open the core.xml file. The
#              core document properties are nodes under the cp:coreProperties node.

#         Example: dcterms:created;dcterms:modified;cp:lastModifiedBy

# Return value

#   The function returns a hashtable object -- in the above example, $props would contain
#   the name-value pairs for the requested MS Office document properties. In the calling script,
#   to get at the values:

#        $fooProperty = $props.'dcterms:created'
#        $barProperty = $props.'dcterms:modified'

[CmdletBinding()]
    [OutputType([System.Collections.Hashtable])]
    Param
    (
        [Parameter(Position=0,
            Mandatory=$false,
            HelpMessage="Enter the full path name of the document")]
            [ValidateNotNullOrEmpty()]
            [String] $DocumentFullPathName='e:\temp\supplier_List.xlsx',
        [Parameter(Position=1,
            Mandatory=$false,
            HelpMessage="Enter the Office properties semi-colon delimited")]
            [ValidateNotNullOrEmpty()]
            [String] $OfficeProperties='dcterms:created; dcterms:modified ;cp:lastModifiedBy;dc:creator'
    )
# We need the FileSystem assembly
Add-Type -AssemblyName System.IO.Compression.FileSystem

# This function unzips a zip file -- and it works on MS Office files directly: no need to
# rename them from foo.xlsx to foo.zip. It expects the full path name of the zip file
# and the path name for the unzipped files
function Unzip
{
    param([string]$zipfile, [string]$outpath)

    [System.IO.Compression.ZipFile]::ExtractToDirectory($zipfile, $outpath) *>$null
}

# Remove spaces from the OfficeProperties parameter
$OfficeProperties = $OfficeProperties.replace(' ','')

# Compose the name of the folder where we will unzip files
$zipDirectoryName = $env:TEMP + "\" + "TempZip"

# delete the zip directory if present
remove-item $zipDirectoryName -force -recurse -ErrorAction Ignore | out-null

# create the zip directory
New-Item -ItemType directory -Path $zipDirectoryName | out-null

# Unzip the files -- i.e. extract the xml files embedded within the MS Office document
unzip $DocumentFullPathName $zipDirectoryName

# get the docProps\core.xml file as [xml]
$coreXmlName = $zipDirectoryName + "\docProps\core.xml"
[xml]$coreXml = get-content -path $coreXmlName

# create an array of the requested properties
$requiredProperties = $OfficeProperties -split ";"

# create a hashtable to return the values
$docProperties = @{}

# Now look for each requested property
foreach($requiredProperty in $requiredProperties)
{
    # We will be lazy and ignore the namespaces. We need the local name only
    $localName = $requiredProperty -split ":"
    $localName = $localName[1]
    # Use XPath to fetch the node for this property
    $thisNode = $coreXml.coreProperties.SelectSingleNode("*[local-name(.) = '$localName']")
    if($thisNode -eq $null)
    {
        # To the hashtable, add the requested property name and its value -- null in this case
        $docProperties.Add($RequiredProperty, $null)
    }
    else
    {
        # To the hashtable, add the requested property name and its value
        $docProperties.Add($RequiredProperty, $thisNode.innerText)
    }
}

#clean up
remove-item $zipDirectoryName -force -recurse

# return the properties hashtable. To do this, just write the object to the output stream
$docProperties

My guess is that your best bet is VB or C# and the Office Interop Assemblies. I'm unaware of a native way (within Powershell) to do what you want.

That said, if you use VB or C#, you could write a powershell cmdlet to what you are the collation. But at that point, it might be more simple to just write a console app that runs as a scheduled task instead.

I recently learned from watching a DNRTV episode that Office 2007 documents are just zipped XML. Therefore, you can change "Document.docx" to "Document.docx.zip" and see the XML files within. You could probably get the properties via an interop assembly in .NET, but it may be more efficient to just look right into the XML (perhaps with LINQ to XML or some native way I am unaware of).

I wrote up how to do this back in the Monad beta days. It should still work I think.