Drupal Association members fund grants that make connections all over the world.
I am using Feeds plus Feed Tamper to create nodes on my website from an RSS feed from another website.
My website is running the current 6x versions of Drupal, Feeds, and Feed Tamper. The other website is running the current 6x versions of Drupal and Views. Both websites run numerous other third-party modules besides those listed.
I am using Feed Tamper to process the taxonomy terms within the RSS feed. There can be several taxonomy terms per feed item. Feed Tamper converts the list of terms into an array that it supplies to Feeds. This is working.
I am having problems with the feeds, in that when an item present in a feed has already been used to create a new node on my website, Feeds generates duplicate nodes. It is not making any difference how I set things up in the feed importer - I get duplicates when I specify "don't update existing nodes," "replace existing nodes," or "update existing nodes."
In the mapping section for the feed importer, I have the "guid" field set as the "unique field." I am working with a custom content type created via CCK, and "guid" is one of the custom fields. I am getting duplicates even though I have the "unique field" box checked for the "guid" field.
When I look at my website's database within phymyadmin, I see some interesting results. For the custom type, "external news," there is a "content_type_external_news" table. It has all of the custom fields for that type except for one - the GUID. There is a separate table, "content_field_guid," that has the values of the GUID field for the "external news" custom type plus an additional custom type - "events."
When I look at the values for the GUID field, it is null for every record in the "content_field_guid" table - regardless of whether the GUID applies to an event node or to an external news node. So I am wondering if this is part of the problem - that Feeds is expecting the GUID field to be in the same table as the rest of the custom fields. [Table "content_type_external_news" instead of "content_field_guid."]
But there is something even stranger happening. Someone else created a feed importer for the events. Her feed importer does not create duplicate event nodes when her feed importer is run. To troubleshoot, I created an exact copy of her feed importer from scratch so that I could see if there was some difference between her importer and mine. My importer creates duplicate event nodes, even though it is processing the same feed and feed items.
I exported the configurations for each feed importer to text files. I then ran the "Beyond Compare" program to display the differences between the two files - in other words, the differences between the two feed importers' configurations. The only differences shown were in the feed importers' names and id's. Otherwise, they were an exact match.
Additionally, the accounts used to create the two feed importers are configured the same. Both have the same two role memberships, one of which grants every possible permission.
There is one other difference that might be significant. In the "feeds_node_item" table, I see that the "guid" field has been filled for four events. [There are four events in the feed.] Otherwise, for all other records the "guid" field is null.