Hi. I’ve created a workbench that reads information from a PDF file, processes that content, then writes it to a database. This works as expected, either when run locally or from within FME Flow. However, it only processes one file at a time. So, to rectify this as a local process, I have created a separate workbench that uses a ‘Directory and File Pathnames’ reader to read PDF files from a named folder, then passes these individually to the original PDF-processing workbench (via a WorkspaceRunner). This, too, works as expected. However, being local, it’s slow.
So, it would be good to replicate this, so that it instead makes its requests to FME Flow, via an FMEFlowJobSubmitter. And...this is where I get stuck, as it requires each PDF file to be uploaded to that server, in sequence. Looking at the documentation, it seems that an HTTPCaller is the preferred way to achieve this, but I’m not sure that I understand how this needs to work in practice. So, has anyone done anything similar? Is there some documentation that covers this type of flowline? I’m using v2025.2.
Regards
Guy Heathcote
Ordnance Survey
Using HTTPCaller to send files to FME Flow
+13- Contributor
Best answer by redgeographics
The PDF file doesn't necessarily have to be uploaded to the server that FME Flow runs on, but FME Flow must be able to reach it. So for example a shared network folder would be ideal.
Alternatively, rather than using the FMEFlowJobSubmitter, you may find it easier to send an API call to run a workspace (which can be sent through the HTTPCaller and allows for uploading source files)
+63- VIP
- Best Answer
The PDF file doesn't necessarily have to be uploaded to the server that FME Flow runs on, but FME Flow must be able to reach it. So for example a shared network folder would be ideal.
Alternatively, rather than using the FMEFlowJobSubmitter, you may find it easier to send an API call to run a workspace (which can be sent through the HTTPCaller and allows for uploading source files)
+13- Contributor
The PDF file doesn't necessarily have to be uploaded to the server that FME Flow runs on, but FME Flow must be able to reach it. So for example a shared network folder would be ideal.
Alternatively, rather than using the FMEFlowJobSubmitter, you may find it easier to send an API call to run a workspace (which can be sent through the HTTPCaller and allows for uploading source files)
Thank you. OK, yes, I could upload the PDF file to a network server. Due to access permissions, this isn’t quite as straightforward as it might be, but it is still an option.
Using an HTTPCaller in place of an FMEFlowJobSubmitter isn’t something I’d considered. I’ll look into that. Cheers for the response.