I want to perform Key Phrase detection using AWS Comprehend.
Is there any way to load data directly from an S3 URI instead of manually loading data from S3 and passing it to the SDK?
Yes.
For Amazon Comprehend, there are usually 3 ways to do the same action:
Synchronous action for one document e.g. DetectKeyPhrases
Synchronous action for multiple documents e.g. BatchDetectKeyPhrases
Asynchronous action for multiple documents e.g. StartKeyPhrasesDetectionJob
Most, if not all, of the time the synchronous actions take in Text or TextList directly & the asynchronous operations allow you to specific an S3 URI.
For detecting key phrases, this would be the StartKeyPhrasesDetectionJob, which takes in S3Uri for input data as well as output data.
All of these operations are available in the AWS SDK for Java v2 so feel free to refer to the SDK documentation for getting started.
Related
In the scenario of listing all versions of an object using its key as a prefix:
import boto3
bucket = 'bucket name'
key = 'key'
s3 = boto3.resource('s3')
versions = s3.Bucket(bucket).object_versions.filter(Prefix=key)
for version in versions:
obj = version.get()
print(obj.get('VersionId'), obj.get('ContentLength'), obj.get('LastModified'))
Do I get charged only for listing the objects that are matching the prefix?
If so, is each object/version listed treated as a separate list request?
No, each object/version listed is not treated as a separate list request. You're only paying for the API requests to S3 (at something like $0.005 per 1000 API requests). A single API request will return many (up to 1000) objects/versions that match the indicated prefix. The prefix filtering itself happens server-side in S3.
The way to get a handle on this is to understand that AWS SDK calls ultimately result in API requests to AWS service endpoints e.g. S3 APIs. What you need to do is work out how your SDK client requests map to the underlying API requests to determine what is likely happening.
If your request is a simple 'list objects in my bucket' case, the boto3 SDK is going to make one or more ListObjectsV2 API calls. I say "or more" because the SDK may need to make more than one API request because API requests typically yield a maximum number of results (e.g. 1000 objects in a ListObjectsV2 response). If there are 2500 objects in the bucket, for example, then three ListObjectsV2 requests would need to be made to the S3 API.
If your request is 'list objects in my bucket with a given prefix', then you need to know what capabilities are present on the ListObjectsV2 API call. Importantly, prefix is one of the parameters. This is how you know that S3 itself is doing the filtering on your supplied prefix (where you have indicated .filter(Prefix=key) in your code). If this were not a feature of the underlying S3 API, then your SDK (boto3 etc.) would be the one doing the filtering on prefix and that would be a much more expensive and vastly slower operation, because the SDK would have to list all objects, potentially resulting in many more LIST requests, and filter them client-side. Note: the ListObjectVersions API is similar to ListObjectsV2 in this regard and both support prefix.
Also, note that VersionId, Size, and LastModifed are all attributes that appear in the ListObjectVersions response, so no further API requests are needed to fetch this information.
So, in your case, assuming that there are fewer than 1000 object versions that match your indicated prefix, I believe that this equates to one S3 API request to ListObjectVersions (and this is considered a LIST request rather than a GET request for billing afaik, even though it is a GET HTTP request to https://mybucket.s3.amazonaws.com/?versions under the covers).
I'm trying to use Go to send objects in a S3 bucket to Textract and collect the response.
I'm using the aws go sdk package and able to connect to my S3 bucket and list all the objects contained within. So far so good. I now need to be able to send one of those objects (a .pdf file) to Textract and collect the response(s).
The AWS Go SDK content for interacting with Textract seem to be quite extensive but I cannot find a good example for how to do this.
I would be very grateful for a sample or advice on how to do this.
To start a job, you invoke StartDocumentTextDetection, using a DocumentLocation to specify the file, and you specify a SNS topic where Textract will publish a notification when it has finished to process your job.
You have now two possibilities:
Subscribe to the SNS topic, and when you receive a message retrieve the result
Create a lambda function triggered by the SNS topic, which retrieves the result.
The second option is IMO better 'cause it use less computation time (doesn't run until the job hasn't finished).
To retrieve the job, you use GetDocumentTextDetection
If anyone else reaches this site searching for an answer:
I understood the documentation as if I could just call the StartDocumentAnalysis function through the textract SDK but in fact what was missing is the fact that you need to create a new Session first and do the calls based on the session:
https://docs.aws.amazon.com/sdk-for-go/api/service/textract/#New
We're currently processing daily mobile app log data with AWS lambda and posting it into redshift. The lambda structures the data but it is essentially raw. The next step is to do some actual processing of the log data into sessions etc, for reporting purposes. The final step is to have something do feature engineering, and then use the data for model training.
The steps are
Structure the raw data for storage
Sessionize the data for reporting
Feature engineering for modeling
For step 2, I am looking at using Quicksight and/or Kibana to create reporting dashboard. But the typical stack as I understand it is to do the log processing with logstash, then have it go to elasticsreach and finally to Kibana/Quicksight. Since we're already handling the initial log processing through lambda, is it possible to skip this step and pass it directly into elasticsearch? If so where does this happen - in the lambda function or from redshift after it has been stored in a table? Or can elasticsearch just read it from the same s3 where I'm posting the data for ingestion into a redshift table?
Elasticsearch uses JSON to perform all operations. For example, to add a document to an index, you use a PUT operation (copied from docs):
PUT twitter/_doc/1
{
"user" : "kimchy",
"post_date" : "2009-11-15T14:12:12",
"message" : "trying out Elasticsearch"
}
Logstash exists to collect log messages, transform them into JSON, and make these PUT requests. However, anything that produces correctly-formatted JSON and can perform an HTTP PUT will work. If you already invoke Lambdas to transform your S3 content, then you should be able to adapt them to write JSON to Elasticsearch. I'd use separate Lambdas for Redshift and Elasticsearch, simply to improve manageability.
Performance tip: you're probably processing lots of records at a time, in which case the bulk API will be more efficient than individual PUTs. However, there is a limit on the size of a request, so you'll need to batch your input.
Also: you don't say whether you're using an AWS Elasticsearch cluster or self-managed. If the former you'll also have to deal with authenticated requests, or use an IP-based access policy on the cluster. You don't say what language your Lambdas are written in, but if it's Python you can use the aws-requests-auth library to make authenticated requests.
I've written a function in Python that uploads a folder and its content to S3. Now I would like S3 to generate an event (so I can send it to a lambda function). S3 allows to generate events only at file level, in fact folders on s3 are just a visualization layer, which means that S3 has no internal representation for folders, keys with the same root are simply grouped together. That said, as for now I've come up with three approaches that revolves around the idea of a 'poison pill'.
Send a special file at the end of the folder upload process, the creation of which sends an event to lambda that can open the file to read custom directives to act on. Seems that this approach is quite flexible, however it poses serious concerns security-wise (I know that ACLs are in place for this reason but I'm not quite sure if it's enough), and generates some overhead while downloading/uploading/deleting the file from/to local memory.
Map an event to the target lambdas and fire it directly. The difference in approaches is simply that in this case I'm not really creating a file on S3, I'm just making S3 believe so. I would use CloudWatch to fire custom S3-object-created events with the name of the folder for lambda to pick up. This approach feels a little more hacky than the other two, plus when I did my research on the matter it seemed like it shouldn't be possible to generate "mock" events on AWS (i.e. Trigger S3 create event). To my understanding however, the function put_events should do the trick.
Using SQS would allow to put the folder name into an SQS task that can be later consumed by lambda. This has some advantages over the other two approaches, since SQS has now a LIFO variant that allows for exactly-once-delivery, failures reprocessing (via dead letters queue), etc, however this generates a non-trivial amount of complexity compared to the other approaches.
At this point I'm trying to opt for the most 'correct' approach, and
in order to do so I'm trying to weight pros and cons to make an informed decision, which led me to some questions:
Is there another way I'm missing out to proceed that does not involve client notification ? (all the aforementioned approaches rely on the client sending the notification in one way or another, which is not very "cloudy")?
Is there a substantial difference between approaches 2 and 3, considering that both rely on sending the information in and out of a stream (CloudWatch and SQS respectively)?
Have you consider using the prefix option of S3 bucket event, I tested it and it worked fine. In my S3 bucket I created two folder test1 and test2. On s3 event I added prefix test1 with that in place every time put/copy operation happen on bucket lambda is trigger.
I think your question nets down to "how can I trigger a Lambda function after I have uploaded a folder full of files to S3?"
Unless you have some information a priori server-side that you can use to determine when the folder upload has completed, the client is going to have to tell you.
Options I would consider:
change your client to publish a message to SNS or to SQS upon the completion of uploading to S3. That message can then trigger your Lambda function.
after the last file has been uploaded to folder images/dogs/, upload a zero-sized object whose key is the same as the folder (images/dogs/). This is a 'sentinel file'. Use an S3 event trigger with suffix of / to detect the upload of that 'folder' object and trigger your Lambda.
I prefer the 1st option. It achieves the end goal without resulting in extraneous S3 objects. With SNS you can also configure multiple downstream processes in response to the ‘finished upload’ message (a fan out) if needed.
I am seeking advice on what's the best way to design this -
Use Case
I want to put multiple files into S3. Once all files are successfully saved, I want to trigger a lambda function to do some other work.
Naive Approach
The way I am approaching this is by saving a record in Dynamo that contains a unique identifier and the total number of records I will be uploading along with the keys that should exist in S3.
A basic implementation would be to take my existing lambda function which is invoked anytime my S3 bucket is written into, and have it check manually whether all the other files been saved.
The Lambda function would know (look in Dynamo to determine what we're looking for) and query S3 to see if the other files are in. If so, use SNS to trigger my other lambda that will do the other work.
Edit: Another approach is have my client program that puts the files in S3 be responsible for directly invoking the other lambda function, since technically it knows when all the files have been uploaded. The issue with this approach is that I do not want this to be the responsibility of the client program... I want the client program to not care. As soon as it has uploaded the files, it should be able to just exit out.
Thoughts
I don't think this is a good idea. Mainly because Lambda functions should be lightweight, and polling the database from within the Lambda function to get the S3 keys of all the uploaded files and then checking in S3 if they are there - doing this each time seems ghetto and very repetitive.
What's the better approach? I was thinking something like using SWF but am not sure if that's overkill for my solution or if it will even let me do what I want. The documentation doesn't show real "examples" either. It's just a discussion without much of a step by step guide (perhaps I'm looking in the wrong spot).
Edit In response to mbaird's suggestions below-
Option 1 (SNS) This is what I will go with. It's simple and doesn't really violate the Single Responsibility Principal. That is, the client uploads the files and sends a notification (via SNS) that its work is done.
Option 2 (Dynamo streams) So this is essentially another "implementation" of Option 1. The client makes a service call, which in this case, results in a table update vs. a SNS notification (Option 1). This update would trigger the Lambda function, as opposed to notification. Not a bad solution, but I prefer using SNS for communication rather than relying on a database's capability (in this case Dynamo streams) to call a Lambda function.
In any case, I'm using AWS technologies and have coupling with their offering (Lambda functions, SNS, etc.) but I feel relying on something like Dynamo streams is making it an even tighter coupling. Not really a huge concern for my use case but still feels dirty ;D
Option 3 with S3 triggers My concern here is the possibility of race conditions. For example, if multiple files are being uploaded by the client simultaneously (think of several async uploads fired off at once with varying file sizes), what if two files happen to finish uploading at around the same time, and two or more Lambda functions (or whatever implementations we use) query Dynamo and gets back N as the completed uploads (instead of N and N+1)? Now even though the final result should be N+2, each one would add 1 to N. Nooooooooooo!
So Option 1 wins.
If you don't want the client program responsible for invoking the Lambda function directly, then would it be OK if it did something a bit more generic?
Option 1: (SNS) What if it simply notified an SNS topic that it had completed a batch of S3 uploads? You could subscribe your Lambda function to that SNS topic.
Option 2: (DynamoDB Streams) What if it simply updated the DynamoDB record with something like an attribute record.allFilesUploaded = true. You could have your Lambda function trigger off the DynamoDB stream. Since you are already creating a DynamoDB record via the client, this seems like a very simple way to mark the batch of uploads as complete without having to code in knowledge about what needs to happen next. The Lambda function could then check the "allFilesUploaded" attribute instead of having to go to S3 for a file listing every time it is called.
Alternatively, don't insert the DynamoDB record until all files have finished uploading, then your Lambda function could just trigger off new records being created.
Option 3: (continuing to use S3 triggers) If the client program can't be changed from how it works today, then instead of listing all the S3 files and comparing them to the list in DynamoDB each time a new file appears, simply update the DynamoDB record via an atomic counter. Then compare the result value against the size of the file list. Once the values are the same you know all the files have been uploaded. The down side to this is that you need to provision enough capacity on your DynamoDB table to handle all the updates, which is going to increase your costs.
Also, I agree with you that SWF is overkill for this task.