Froogle Merchant Center merged into GoogleBase

For those folks having issues FTPing from within the mod, try creating/generating the feed file, then manually FTPing with your web browser or FTP client. Instructions on how to manually FTP are on the GoogleBase support site.

http://base.google.com/base/FTP.html
http://base.google.com/support/bin/topic.py?topic=2928

Then at the very least you will know that part is working and will have a record of correct FTP server authentication parameters to experiment within the mod's admin config.

Woody

birdoasis:

I would think php settings are in the php.ini file. however you have to have root access for that.

Perhaps if you can find the command you can put it in your .htaccess file instead.

Anyone with the specific php.ini or .htaccess setting? Just the details please so I can explain to my web host system admin :wink2:

Woody

Forgive me if this has been addressed previously, but there are so many pages.

My googlefroogle feed is working just fine, except it only creates a file with 12 products, not the 153 that are in the store. This started with one of the recent updates to adhere to Google's new standards.

Any ideas on what I need to fix?

Haha, I am having the same issue, and it still has not been resolved.

Zen Cart merchants with Froogle feeds

GoogleBase posted this tidbit on their help site.

Is it enough to simply send the same data as I did for Froogle (name, description, price, product_url, and image_url)?

No. Unlike Froogle, our Google Base format will only be as good as the attributes that you send. Your chances of scoring higher on relevancy will be increased as you send more attributes. We're asking you to send us five additional attributes along with the basic ones you've submitted in the past. Because we think it's important to provide a consistent user experience, we will no longer list items that don't meet a minimal number of attributes by the end of February 2007.

Keep in mind that Andrew's feed generator is a "Froogle feed" generator, supporting old-style tab delimited feed format with an older schema. It is not a GoogleBase XML feed generator (see more below).

Right now when you upload your Froogle feed you are not uploading to Froogle but are uploading to to GoogleBase. GoogleBase maps many of the older Froogle fields (attributes) to newer GoogleBase attributes (when the attribute names have changed). But sometimes that mapping can go haywire.

Note Andrews' latest Google Froogle Feed generator mod v.1.33 currently generates:

title + description + link + price + image_link + expiration_date + label + id + brand+ product_type + condition + currency

That is twelve attributes (not necessarily in above order)

GoogleBase now requires and "supports" all above but two. Currency and label are no longer "supported". It is unclear what "unsupported" means in this context.

See post here: http://googlebase.blogspot.com/2007/02/label-attribute-no-longer-supported.html

From the Google post copied above it appears we all need to add five more attributes to our Froogle feeds. (Unsure if need to add two more than this to also replace currency and label).

Below are 'optional" attributes Google recommends:

Optional attributes
The following attributes are optional, but Google recommends including as many attributes as possible as it will increase your items' "exposure in search results".

* payment_accepted
* payment_notes
* pickup
* price_type
* quantity
* shipping
* tax_percent
* tax_region

And Google says it is best to customize you feed with and custom attributes or at least design your feed around popular "Item Types" (aka Information Types) which Google has assigned pre-set attributes, depending on the product.

Example:
Item Type: Books

• Specific Item Attributes:

Books

* author
* binding
* edition
* genre
* isbn
* pages
* publisher

For more info: http://www.google.com/base/products.html

Pretty much this means Zen Cart shopowners need a revamped mod to support additional attributes. Or hack Andrews latest mod version and hardcode in five more "static" attributes, at least temporarily to keep your Froogle feeds accepted for the short term (this only make sense if you have many products). But understand this will not mean your feed will be published to GoogleBase, only to Froogle (if accepted).

So what we really we need now is a mod designed for GoogleBase which exports XML supporting RSS2+GoogleBase namespace and extended attribute support.

It is possible Andrew's RSS2 mod could be hacked as such, as he has already updated it to support GoogleBase namesapce.

Hoping soon, a white knight developer will appear from the mist and release a GoogleBase for Zen Cart mod supporting the new required schema and optional attributes. And integrate a ZC admin interface so a shop-owner can update the mod output schema whenever Google updates their feed requirements.

Or even better create a "Store Connector" for Zen Cart.

A person can dream can't he?

Woody

If you upload a Froogle feed and observe an error for a specific product, such as:

our bulk upload contains an invalid character for the current encoding setting or contains a disallowed character.

You can see that error doesn't indicate exactly where in your product feed the invalid or disallowed character might reside. (it will tell you the product line number).

So try this:

In your GoogleBase admin (on Google Server) click on My Items tab, click Bulk upload files. If you received an GoogleBase error message you should also see a detail message with a "Change settings" link. Change your encoding to Latin-1 (from UTF-8). Then re-upload your feed.

Changing the encoding worked for me. I'm awaiting now to see how the published items show up on Froogle and GoogleBase and how it might affect search.

Woody

An older version of Andrew's Google Froogle feed mod (v1. 20?) generates old Froogle attribute names in the following format:

product_url name description price image_url category offer_id

Note the above feed format does NOT contain the label attribute and the attribute names are old Froogle attribute names, not the more recent GoogleBase attribute names. And with only seven attributes the feed from this older version feed format does not meet the new GoogleBase feed guidelines.

However I just used the older mod to generate a Froogle feed from Zen Cart and my uploaded feed was completely accepted by GoogleBase and all items are now published and searchable.

So it appears that Google is still mapping the old Froogle attributes to the newer GoogleBase attribute.

So if you're having issues with more recent googlefroogle mod you might consider trying an older version, at least temporarily. YMMV.

Woody

Woodymon:

Anyone with the specific php.ini or .htaccess setting? Just the details please so I can explain to my web host system admin :wink2:

Woody

Woody,

This setting worked for me:

max_execution_time = 300
max_input_time = 90
memory_limit = 32M

Please keep in mind that we only have about 12,000 products so it might not work for others.

Basically what's happening is the new GoogleFroogle generator is using IH handler image now and the script needs to allow for IH2 to create all images needed if the image doesn't exist in bmz_cache. In our case because not all product pages has been visited prior to using the generator, the mod was timing out creating the need image and point to the IH2 url for image instead of the old way where it just uses url/images/large.

BlessIsaacola:

This setting worked for me:...
Hey that is major helpful info. Thanks!

BlessIsaacola:

Basically what's happening is the new GoogleFroogle generator is using IH handler image now and the script needs to allow for IH2 to create all images needed if the image doesn't exist in bmz_cache.

Question: I was of understanding Andrew updated the mod for IH2 compatibility and the generated feed's image_link attribute referred to the large image files in the "standard" /images directory and that the feed does not include links to the small and medium images (which as you explain are created on the fly and thrown into bmzcache for futre reference). At least thats how I understand it. And my feeds image_link url's alwasy refer to the large images in /images

Or is that anytme the large image is referred to then that auto prompts the generation of the other two associated images?

GoogleBase/Froogle wants the large images.

Woody

Woodymon:

Hey that is major helpful info. Thanks!

Question: I was of understanding Andrew updated the mod for IH2 compatibility and the generated feed's image_link attribute referred to the large image files in the "standard" /images directory and that the feed does not include links to the small and medium images (which as you explain are created on the fly and thrown into bmzcache for futre reference). At least thats how I understand it. And my feeds image_link url's alwasy refer to the large images in /images

Or is that anytme the large image is referred to then that auto prompts the generation of the other two associated images?

GoogleBase/Froogle wants the large images.

Woody

Prior to the last version Andrew released (1.3.3) the image url generated was url/images/large/filename which of course doesn't exist if you have IH2 installed. In the past you have to make a code change to generate the right path when using IH2.

The lastest mod generates url using IH2 url if installed. The only issue with this is small, medium and large images are ALL generated on the fly and stored in bmz_cache folders. IH2 doesn't use image/medium or image/large folders at all.

Woody,

Another interesting thing is I actually played around with```
max_execution_time = 300

We are with a new host with a little fast processor and memory and perhaps that accounted for not needing it to be at 300 anymore to generate all the products.

I also noticed that as I ran googlefroogle, the more products created the more images are added to my bmz_cache folder. So the max_execution_time more than any other setting is the key to this thing.

Here is the statistics from the completed task:

Froogle Data Feeder v.1.3.3 07.02.2007 20:45 started 2007/03/20 20:03:58
Feed file - DELETED FOR SECURITY REASON
Processing: Feed - Yes, Upload - No
Froogle File Complete In 123.431367 Seconds 10043 Records

As you can see it took 123.43 seconds to generate the file. If I would have set my limit to less than 124 this wouldn't have worked.

I have just tried this on a testing machine and I am getting the wrong prices in my feed file.

I am getting my price including tax with tax added to it again. If I view my product in the product/catalogue admin and resubmit it to the site with no changes to the price I get the right price back again.

I cannot do this for 1000+ product though. Any one else had this issue?

thanks

MG

BlessIsaacola:

The lastest mod generates url using IH2 url if installed. The only issue with this is small, medium and large images are ALL generated on the fly and stored in bmz_cache folders. IH2 doesn't use image/medium or image/large folders at all.

It appears I'm not understanding the process of rendering large images with IH2.

I have latest version of IH2 and v.133 version of Andrew's mod installed on ZC 1.37. And in my feed the image_link urls's for each of my products correctly refers to my large images in /images

My large images don't appear to be generated on the fly. I must ask why would they need to be "generated" since the images are the original large images uploaded to the server. On my product_info _display page when I open the large image pop-up and click properties and check the path it shows the image resides in /images.

Are you using IH2 watermark/overlay function or employing Zen Lightbox and that may requires a re-rendering of the large image?

Just trying to gain a better understanding of the process here. Thanks for your quality feedback.

Woody

Woodymon:

It appears I'm not understanding the process of rendering large images with IH2.

I have latest version of IH2 and v.133 version of Andrew's mod installed on ZC 1.37. And in my feed the image_link urls's for each of my products correctly refers to my large images in /images

My large images don't appear to be generated on the fly. I must ask why would they need to be "generated" since the images are the original large images uploaded to the server. On my product_info _display page when I open the large image pop-up and click properties and check the path it shows the image resides in /images.

Are you using IH2 watermark/overlay function or employing Zen Lightbox and that may requires a re-rendering of the large image?

Just trying to gain a better understanding of the process here. Thanks for your quality feedback.

Woody

Well, you enlightened me Woody! I have always used watermark on our large image for two reasons: 1. for those stealing our images without paying attention they help us by advertising our website and two to burn our url into our customers mind when they are viewing the large image :)

As result, our large image is generated on the fly and stored in bmz_cache. Of course I have always thought this how IH2 works not knowing it works that way because of my selection to watermark large image.

I temporarily removed watermark and now my large images are also referenced as url/image/filename. The funny thing is some of my products still is referencing the bmz_cache images when using googlefroogle and some are accurately reflecting url/images/filename. I wonder if I need to clear cache before regenerating googlefroogle to make sure it's still not behaving as though I have watermark on large images.

It really doesn't make sense how the Large image works because I have always thought that IH compresses and resize the large images as well. For example I upload very high quality image (sometimes 1200 x 800) to my image folder but my setting for large image on Zen Cart admin side is 600 X 400. If I know that IH2 is simply going to display my large image of 1200 x 800 there is no point uploading such a huge file to the server. I could have simply processed all images to be 600 X 400, compress it and upload it.

I am afraid I will leave the images alone for now because I do not have the time to reprocess 20,000 images.

Thanks for clearing this up.

Woody,

I don't know how to explain what's happening technically but I will just tell you what happened :D

After I change my setting in Admin to remove watermark I ran GoogleFroogle and it didn't complete the process, generated 1MB worth of data. Ran it the 2nd time and it generated 3MB, Ran it the 3rd time it generated 8.01MB which is all the products but the process did not finish. Ran it the 4th time and it generated 8.01MB but this time it finished and displayed all the statistics associated with generating the file.

It's almost like each time it didn't finish, it was able to use certain information from previous session. I still have a feeling it's something to do with generating the image url. If the script is generating images based on whether it's image/large or bmz_cache folder that's some data crunching especially when you have over 10,000 plus products (in my case).

I suppose I could have increased my max_execution_time and it would have ran fine the first time.

After disabling watermarks, did you manually clear the bmz_cache before generating the feed?
(just for testing purposes as you still want the watermarks after you get this resolved).

Woodymon:

After disabling watermarks, did you manually clear the bmz_cache before generating the feed?
(just for testing purposes as you still want the watermarks after you get this resolved).

Nope I did not clear it but googlefroogle works fine in either case as long as the max_execution_time is enough so I am cool! I just found another mod that needed the max_execution_time set to 300 or unlimited.

I'm attempting to install the feeder.

But when I run the updates via Zen Cart admin or Phpadmin I get a mysql error that a dublicate field exist, or a whole series of errors.

This is version 1.3.7

Any ideas?

birdoasis:

when I run the updates via Zen Cart admin or Phpadmin I get a mysql error that a dublicate field exist, or a whole series of errors.
Without seeing at least a few of the errors hard to tell. But likely you have a prefix on your database tables. There is already much in this thread (and other mod threads) about this subject and possible workarounds.

Woody