Skipped records in CSV Import?


Would someone know what might cause the CSV Import plugin to skip a few - always the same few - records out of a large number of records imported? The plugin is the latest version, Omeka is version 2.4.

I’ve a CSV with 46 rows. Each row/record has a number of metadata columns plus a number of files (one XML file and a number of jpg’s), each file in its own column. The files are all located on a server in the same directory, and the address in each column is of the form http://path/to/file.jpg. When importing, I map the file columns to the appropriate metadata fields and select “files” in the second dropdown. Each time, 42 records are imported fine. Four records, always the same four, are skipped (skipped records, specifically, as opposed to skipped rows). If I attempt to import from a different CSV containing only these four records, the result is the same.

I have narrowed the problem down to the files, as when I do not select “files” for any of the file fields, the records are imported successfully (without the files), but otherwise I’m puzzled. There are no differences as to file sizes between these four and the other 42 records. The file names do not have special characters. The rows for these four records in the CSV file are identical to the other 42 rows - and couldn’t really be different as the CSV is not created manually but produced with an XSLT stylesheet, which picks selected data from the XML files. Five columns are reserved for files, but the records have a varying number of files, so there are blank cells in the file columns. I don’t see how this could be the problem with the four skipped records, however, as it’s the same situation with the 42 records that do get imported successfully.

Does anyone have any idea as to what else I might try here?

… and just now I noticed, checking the error log, that it was my own mishap: the file url’s/file names in the CSV are identical as the XSLT stylesheet creates them in a specific way, but the names of the actual files for these four records are slightly different from those of the other 42 records, so CSV Import naturally cannot find them.