If the multipart contains only text data, it might be possible to use
those processors Andy suggested. However, if we support multipart
natively, it can be done much easier and safer. I created a JIRA for
this improvement before, but haven't been able to find a time to
implement it myself.
If anyone interested in tackling this enhancement, it'd be greatly appreciated.
https://issues.apache.org/jira/browse/NIFI-3469

Thanks,
Koji

On Tue, Feb 28, 2017 at 4:59 AM, Andy LoPresto <[email protected]> wrote:
> Absolutely. Use some combination of SplitText, SplitContent, SplitJSON, and
> ExtractText to split the content into multiple flowfiles on regular
> expressions, new lines, and/or byte sequences and then RouteOnContent to
> send the flowfiles which reference a file to a processor to retrieve/parse
> the file contents.
>
> Andy LoPresto
> [email protected]
> [email protected]
> PGP Fingerprint: 70EC B3E5 98A6 5A3F D3C4  BACE 3C6E F65B 2F7D EF69
>
> On Feb 27, 2017, at 2:08 PM, mohammed shambakey <[email protected]>
> wrote:
>
> This is an example of the output file from "handlehttprequest":
>
> "
> -----------------------------831805444208556434014280960^M
> Content-Disposition: form-data; name="userid"^M
> ^M
> jpl_user^M
> -----------------------------831805444208556434014280960^M
> Content-Disposition: form-data; name="scripts"; filename="a.py"^M
> Content-Type: text/x-python^M
> ^M
> ^M
> -----------------------------831805444208556434014280960^M
> Content-Disposition: form-data; name="scripts"; filename="b.py"^M
> Content-Type: text/x-python^M
> ^M
> ^M
> -----------------------------831805444208556434014280960^M
> Content-Disposition: form-data; name="scripts"; filename="conf.json"^M
> Content-Type: application/json^M
> ^M
> {
> "userid":"jpl_user_1",
> "scripts":["a.py","b.py"]
> }
> ^M
> -----------------------------831805444208556434014280960--^M
> "
>
> Can I extract different components of this file without script at server
> side?
>
> Regards
>
> On Mon, Feb 27, 2017 at 10:52 AM, Andy LoPresto <[email protected]>
> wrote:
>>
>> Mohammed,
>>
>> So from what I understand you have a separate web application which
>> accepts a variety of data (a combination of form fields and file uploads)
>> and submits that (I’m assuming via POST) to NiFi. I’m not sure what problem
>> you had extracting these “components” from the output flowfile from the
>> HandleHTTPRequest processor. Can you change your web interface to send
>> separate POST requests to NiFi and logically separate these components from
>> the beginning? Can you modify the format of the request (i.e. make it JSON,
>> etc.) so it is properly structured for NiFi to parse and operate on this
>> data?
>>
>> NiFi treats the content of the flowfile the same once it is in the system
>> regardless of the source processor, so you should be able to split, extract,
>> or manipulate the contents of a single flowfile containing multiple form
>> elements and file bodies.
>>
>> If you are referencing NiFi’s Input Port components as a destination for
>> an arbitrary HTTP request, this will not work. The input ports are designed
>> to work with NiFi’s Site to Site protocol (either over raw sockets or
>> HTTP(S)).
>>
>>
>> Andy LoPresto
>> [email protected]
>> [email protected]
>> PGP Fingerprint: 70EC B3E5 98A6 5A3F D3C4  BACE 3C6E F65B 2F7D EF69
>>
>> On Feb 27, 2017, at 11:09 AM, mohammed shambakey <[email protected]>
>> wrote:
>>
>> Thanks Andy
>>
>> I have a user web interface that send some form information (e.g., user
>> id), as well as upload multiple files, to a remote NIFI instance. When using
>> "handlhttprequest" processor, it results in a single file that contains
>> "form inputs" and bodies of uploaded files.
>>
>> I want to extract individual components of this file (i.e., download all
>> uploaded files and extract other form information (e.g., userid) in a JSON
>> or other text file. I had a similar problem before and I learnt that
>> "httphandlerequest" has a problem to extract individual compoents from, so I
>> have to write my own scripts to extract these components (form inputs and
>> uploaded files).
>>
>> As NIFI input ports does not face this problem of "handlehttprequest", I
>> though if I can send form inputs and upload files directly to "NIFI input
>> port" at remote site, then I don't have to write my own scripts to extract
>> individual components.
>>
>> Regards
>>
>> On Mon, Feb 27, 2017 at 10:02 AM, Andy LoPresto <[email protected]>
>> wrote:
>>>
>>> Mohammed,
>>>
>>> I’m not sure I understand your question. NiFi has a REST API which
>>> supports the UI and allows scripts/other applications to make requests of
>>> the platform. This runs on the input port specified in nifi.properties.
>>>
>>> If you want to submit form data and uploaded files as data to be ingested
>>> by NiFi, HandleHTTPRequest or (less likely) ListenHTTP are the processors
>>> you would want to use.
>>>
>>> If you can clarify your use case, we should be able to offer more
>>> assistance.
>>>
>>> Andy LoPresto
>>> [email protected]
>>> [email protected]
>>> PGP Fingerprint: 70EC B3E5 98A6 5A3F D3C4  BACE 3C6E F65B 2F7D EF69
>>>
>>> On Feb 27, 2017, at 9:15 AM, mohammed shambakey <[email protected]>
>>> wrote:
>>>
>>> Hi
>>>
>>> Is it possible to send http request (including form data and uploaded
>>> files) to NIFI input port instead of "HandleHTTPRequest" processor?
>>>
>>> Regards
>>>
>>> --
>>> Mohammed
>>>
>>>
>>
>>
>>
>> --
>> Mohammed
>>
>>
>
>
>
> --
> Mohammed
>
>

Reply via email to