So, I could get neither xCopy nor BlueDog's DavCopy to work so I had to revert to brute force.

The problem is that UNC paths can use goofy characters like &, #, ~, ,, and . and URL paths can't.  So when, in some file share, you've got a file named "Snakes & Arrows.doc" when you try to upload that to a doc library you get the error "The Parameter is Incorrect."

We also run into this error when a file type is on the Firbidden File Types list in Central Admin and when we get a ~AAA9999.tmp file like when Word or Excel is auto-saving a draft, although I'm not sure why one would be left on the file share unless Word exits abnormally.

Now, we had over 15,000 documents in 2700 doc libraries.  So we went to scrubbing file name using a combination of the DOS Dir command, Notepad, Excel and DOS batch files.

First, for example, to find all the files with an & in the filename, we mapped a drive, L:, to the target file share and ran this from a command prompt:

DIR L:*&*.* /S /B /N >c:and.txt

That creates a file in the root of your c: drive named and.txt and it will list all the files and all the folders on the L: drive with an & in the file name.

Open the text file and copy and paste the content into Column A of an Excel worksheet.

Now, you have to use Excel string magic to remove one layer of directories labels at a time. So, in Column B, use this formula:

=IF(ISERROR(FIND("",A1,1)),A1,REPLACE(A1,1,FIND("",A1,1),""))

Then repeat that formula in subsequent columns to the right until you're left with a column with only the file names; this will depend on how deep your folders nest.  Copy the column of file names and use Paste Special | Values to paste the values, not the formulas, into a column of a new worksheet.  Then, on this worksheet, execute a find and replace, replacing the & with the string "and".

Copy the modified column and paste it back on the original worksheet in the next available column; if your original directory had folders as many as five layers deep, Column H would be good.  So, now, you've got the original DIR output in column A and the scrubbed file name in column H. Move over to column I and use this formula.

=CONCATENATE("REN """,A1,""" """,H1,"""")

This formula creates a text string that looks something like this:

REN "L:MyFolder01MyFolder02Snakes & Arrows.doc" "Snakes and Arrows.doc"

Now, copy the cells on row 1 and paste them all the way down to the end of your list. (Don't bugger your column H, the one with the new file names.)  Then copy Row I, the one with all the REN statements and paste them into Notepad and save the file as and.bat.  Find it in Windows Explorer and double click on it.  If you want, you can run it from a command line and append a >andout.txt; that will give you a nice output file for figuring out what you've done.

One thing about this process is this:  If you've got an & in a folder name, the folder will get renamed first and then the files will return a "file not found" error.  You'll have through the whole process again.  Of course, you could fight through the spreadsheet and make the folder name changes last but that would be burdensome.

And, finally, once you've scrubbed away all the &'s you'll have to find the #'s, the ,'s and any other goofy character.  The ~'s are particularly difficult to find since I don't think DOS will find them; perhaps if we knew the appropriate escape method to actually find a file with a ~ in the file name we could. Finally, the dots, i.e. ., are only a problem when there's two of them at the end in front of the file extension like this: My Firm Inc..doc.

So we migrated 17,000 documents with all of these issues in just 8 about hours. I was dragging folders from one Explorer window to another where I had a drive mapped to my doc library.  The folders would take as much as 10 minutes to copy and, often, they would crap out because of a .tmp file or some file I had missed; the ~ files were real hard to find before they errored.  I learned that, typically, an Explorer click and drag copy routine will copy the files first and then the folders in reverse alphabetical order.  Regradless, I would often have to open a third Explorer window to search for the problem file without making the error boxes go away.  Fix the file name, or move it to a separate "problem" folder, and run the click and drag again.

So, at least for this robot, brute force wins out.

hth

-robot

So we've got about 150,000 legal documents to migrate into MOSS from a network file share.  Of course, the customer wants the document library to look just like the file share so I think, great, I'll just drag and drop the folders into Explorer View.

Well, no such luck.  First, Explorer View is a big fat Page Not Found error on my host.  I started the web client like Joel said yesterday and everything but no explorer view.

So I just create a network place for the doc library and I go to copy folders and, of course, it barfs everytime there's a special character in the file or folder name.

So that's got me looking for a better way to get these document migrated, something that will move the files without blowing up every time it hits an ampersand.

Since I'm an old robot (me and HAL9000 were school buds) I recall the first pc command I ever learned, dir, and it does alot more than make text fly by your command prompt. In fact, if you assume the network file share is mapped to L: then this one might be handy here:

L:>dir /s /b /-n >c:Files.txt

This will provide an alpahbetical list of all files in all folders in the network share.  Now, all that remains is copying the text into Excel and writing some string magic in excel to write a copy command that can get batched into 2500 MakeDir commands and 15,000 copy commands.  The MakeDir will acually create the folders in the document library.  The copy will copy the old file with the old file name to the new folder with a "URLHappy" file name, i.e. one with no characters that are prohibited in URLs.

Let's I bet this works. The only question I have is this…

Which is easier:

• Writing an executable to do it.
• Writing VBA in Excel to do it.
• Just doing it in Excel?

-robot

Of course we all loved the Explorer View in SharePoint 2003.

But, sometimes it doesn't work.  And sometimes click-and-drag doesn't work.

Well, our new best friend Mark helps with some of this nonsense.  On his site, he describes the fact that the Explorer View requires the Web Client Service to be running.  You get to this service in your Services .mmc which is an Administrative Tool under your control panel.  Be sure the service is running and, just like that, your Explorer View springs to life.

hth

-robot

OK People, we now have two new best friends.  If you are "All SharePoint All the Time" skip the first one covered in the next three paragraphs.

First of all, does anyone know who Chris Squire is?  He plays bass guitar for Yes, this robot's favoritest band of all time. At his website you can order a personally autographed updated multimedia "Fish Out of Water" cddvd for $50 that includes a remastered version of his '70's solo album and a number of videos and other extras.

So, I ordered mine and it involved a number of eMails back and forth regarding format and paypal and they all were "from" Chris Squire.  So I thought, sure and then, finally, there was a question, "what would you like the autograph to say" and so I refered to a deep-cut phrase "42nd scream-down" from Tormato, the band's 1978 effort.

So the eMail comes back and he gives me two paragraphs about the bands lyrics and what they meant to him and how much he appreciated my indications of appreciation.  It really had to be him because, and of this I am certain, no robot or other human could have conceived of those words.  So, now, Chris is my best friend.  But I digress.

Our other new best friend is Ethan.

He hinted that the solution to my install woes might be found in two key activities:

1. Build out the web sites first; one for portal, one for SSP and one for MySites.  Give them host virtual folders.  Use port 80 for all of them but give them unique host headers and .Net 2.0 support.
2. Use a local hosts file to support demonstration of appropriate IIS routing prior to the MOSS install.

So this worked like money.  There was none of the ugliness of creating IIS web sites during the initial Central Administration tasks effort, no confusion over which server and port and application and app pool. Each of the three sites had an IIS web site standing ready to support it.  Created the AppPools on the fly and drive safe, thanks for coming.

So, using his hints, the table of content on my implementation plan goes like this:

Pre-Install Prep:

1. Hardware
2. Database Server
3. Install Account
4. Service Account
5. Outgoing and Incoming EMail Accounts
6. Portal Administrators

Install:

1. Create Accounts
2. Install .Net 2.0 and 3.0
3. Secure License Key
4. Create Host Virtual Folders for IIS Web Sites (Here, Ethan creates siblings to wwwroot for MySite and SSP. I created subfolders in wwwroot; an esoteric difference.)
5. Modify HOSTS file to support the three host headers used in the three web sites.
6. Create the new web sites in IIS configuring them to recognize the different Host Headers.
7. Set the web sites to run .Net 2.0
8. Run the CD.
9. Set the eMail settings.
10. Create the Web Applications.
11. Assign Services to Servers.
12. Create the SSP.
13. Create the Portal.

EEEK Dorks!!!

Post Install:

1. eMail your HOSTS file to your DNS guy
2. Take the rest of the week off.

Frigging Munney!

I would be glad to share more details on this with anyone that will leave a polite comment.

Big thanks to Ethan and Chris and if you order the cddvd, be sure to tell him the 42nd Sream-down Tormato guy sent you.

Have a Sunny SharePoint Day!

-robot 

I've been chided for inadequate planning.  So, today, I plan to install 2007 again.

You guys that know me know that I don't see how that's inadequate.

Ok, maybe just a little more background.

In the past, I've gotten the software installed countless times.  Then I was on hiatus for a few weeks and, then, many of my previous blog posts were lost to the ether.  So right now, I'm just reminded how quickly I can get so stupid.

Here's what's happening, I get the software installed to where we get the central administration tasks.  And I've always tried to follow them explicitly.  But when I get to the Shared Service Provider part, it always wants to use my primary port 80 web site for the SSP.  When I go into "Create new web application" the machine starts to act weird, taking forever and then leaving me on some strange page, I forget the details.

So, our new best friend Ethan, dropped me a note that suggest working all of the IIS website stuff first.  His suggestion involves creating the IIS web sites for SSP and MySites, differentiating them by host headers, modifying the HOSTS file to support the host header names, and creating application pools and sibling folders to the wwwroot folder to virtualize their contents.  Then, when you're staring at Central Admin and you go to create your SSP, there's already a perfectly functional web site there to use.

Now, I've always liked to leave Default Web Site alone so that you could always prove IIS was okay dokey just by pulling up the Under Construction page in wwwroot.  However, bowing to the color of expertise, I'm going to try to follow his plan explicitly.

So, with that plan, off I go.  Wish me luck and I'll get you an update.

-robot

I was recently asked by a manager with money to spend before the end of the budgeting quarter: "What third-party products can we buy today that will help us with our operation."  The part he didn't say was "that is currently planned three to six months out."

The only one I had used was CorasWorks and that was 2003.  While that was a great product, I've never used their 2007 products and I'm not sure they aren't suffering from the Microsoft disease where their "niche" gets rolled into the product sucking all the money with it.

I saw the Bamboo toolset and it helped with AD integration which is a real value ad as far as I was concerned.  I recall that 2003 and AD were kind of like the Hatfields and McCoys.  For those of you that care, SPS2003 was the Hatfields.

Now, I've seen the toolset from IDevFactory.com.  It seems interesting since it's "feature based" and as our 2007 environments get more complex, I imagine we'll be working hard to differentiate the word "feature" from "application" in the minds of our customers.

I'd love to hear anyone's comments about the IDevFactory toolset.  In the meantime, I've got to get one of my customers to let me install it and see if I can break it.

I hope everyone is having a Sunny SharePoint Day!

-robot  

Here's what I've been able to recover from some of my previous InfoPath entries that got ate in the crash:

More Frustration with the InfoPath Form Server

An Item with the Same Key has Already been Added.

SharePoint RSS Generator

WSS v.3 and Team Foundation Server

MOSS 2007 Search