Update

[TBrewerton]

Hi Wilson,

Could we please have an update on the update? possibly an eta?

Regards,

Tim

[Wilson Logan]

Hi Tim,

The sit-rep is that I need to call in reinforcements in the shape of one of my smartest friends (CTO of a large insurance company).

I'm hoping to see him on Friday. What we find at that point will determine the way forward.

I'll let you know what happens.

Cheers,

Wilson.

[sonex293]

Wilson,

I hope you caught up with your friend on Friday.  So.....how bad is it?

--
Michael

[Wilson Logan]

Well its bad... real bad.

But I did catch up with my friend and we are working on it.

Basically the parser uses regular expressions to sift out the From, Date, Subject, etc fields. These expressions have worked very well for a long time without any maintenance. This is because they are fiendishly complex and are written to deal with a large number of eventualities. Early users of PGO will remember that we were patching it sometimes twice a month for format changes. I actually thought that Yahoo had been quiet. No, it's just that the regular expressions have been catching all the changes. Up until now.

They were written by Alexander Nechiporenko, who was on the PGO build team and maintained it untill a couple of years ago. Now he's gone and neither I nor Matti are familiar enough with regular expressions to quickly modify them as required.

My friend has more experience and I'm hoping he'll be able to get a couple of hours free and just hammer through them. Not a lot has happened this Friday - Monday because its the Easter holiday here and I'm not asking him to work through his holiday.

And that's where we're up to.

Cheers,

Wilson.

[MyNameIsScott]

Parsing is fiendishly difficult to automate, especially on a moving target.

Where does PGOffline get the file to be parsed?



Scott

[Wilson Logan]

Hi Scott,

We're just using the raw HTML from the Yahoo messages.

Are you familiar with regular expressions?

Cheers,

Wilson.

[MyNameIsScott]

By name only.

The last code I wrote were some dBase II procedures to calculate principal and interest splits on consumer loan repayments.

A few days after going live we found out I left out record pointers so the history reports weren't coming out accurately. Two nights of eyeballing fixed it, but it was enough to remind me to stick with the thing I was trained for.

My recent experience is in parsing text lines in Excel to create journal entries for QuickBooks. Most things laid out purely for human eyes are tough subjects for what you are trying to do. Even Acrobat web capture can't render frames right much of the time. Many of the blog sites and Wikipedia probably don't figure it is in their best interests to make that easy.

I had some clients here that did some heavy duty Access/VB coding not long ago, though. Some of them are still around.

Scott