Extract all HTML Objects from a Web Page

Demonstrates how to download a Web page (at a URL) and extract all HTML objects. Eg. images, links, CSS files, JavaScript files, etc.

Download Chilkat MHT ActiveX

CREATE PROCEDURE ChilkatSample
AS
BEGIN
    DECLARE @hr int
    DECLARE @sTmp0 nvarchar(4000)
    DECLARE @mht int
    EXEC @hr = sp_OACreate 'Chilkat.Mht', @mht OUT
    IF @hr <> 0
    BEGIN
        PRINT 'Failed to create ActiveX component'
        RETURN
    END

    DECLARE @success int

    EXEC sp_OAMethod @mht, 'UnlockComponent', @success OUT, 'Anything for 30-day trial'
    IF @success <> 1
      BEGIN
        EXEC sp_OAGetProperty @mht, 'LastErrorText', @sTmp0 OUT

        PRINT @sTmp0
        RETURN
      END

    --  Download a URL into an in-memory MHT web archive contained
    --  in a string variable:
    DECLARE @mhtDoc nvarchar(4000)

    EXEC sp_OAMethod @mht, 'GetMHT', @mhtDoc OUT, 'http://www.gopackaging.com/'
    IF @mhtDoc Is NULL 
      BEGIN
        EXEC sp_OAGetProperty @mht, 'LastErrorText', @sTmp0 OUT

        PRINT @sTmp0
        RETURN
      END

    --  Now extract the HTML and embedded objects:
    DECLARE @unpackDir nvarchar(4000)

    SELECT @unpackDir = '/Users/chilkat/temp/'
    DECLARE @htmlFilename nvarchar(4000)

    SELECT @htmlFilename = 'gopackaging.html'
    DECLARE @partsSubdir nvarchar(4000)

    SELECT @partsSubdir = 'objects'
    --  Extract to /Users/chilkat/temp/gopackaging.html.
    --  images and other embedded objects are placed in
    --  /Users/chilkat/temp/objects.  Directories are automatically
    --  created if they don't already exist.
    EXEC sp_OAMethod @mht, 'UnpackMHTString', @success OUT, @mhtDoc, @unpackDir, @htmlFilename, @partsSubdir
    IF @success <> 1
      BEGIN
        EXEC sp_OAGetProperty @mht, 'LastErrorText', @sTmp0 OUT

        PRINT @sTmp0
      END
    ELSE
      BEGIN
        PRINT 'Unpacked!'
      END

END
GO