Read pdf object from S3

I am trying to create a lambda function that will access a pdf form uploaded to s3 and strip out the data entered into the form and send it elsewhere.

I am able to do this when I can download the file locally. So the below script works and allows me to read the data from the pdf into my pandas dataframe.:

import PyPDF2 as pypdf
import pandas as pd

s3 = boto3.resource('s3')
s3.meta.client.download_file(bucket_name, asset_key, './target.pdf')

pdfobject = open("./target.pdf", 'rb')
pdf = pypdf.PdfFileReader(pdfobject)
data = pdf.getFormTextFields()

pdf_df = pd.DataFrame(data, columns=get_cols(data), index=[0])

But with lambda I cannot save the file locally because I get a "read only filesystem" error.

I have tried using the s3.get_object() method like below:

s3_response_object= s3.get_object(
    Bucket='pdf-forms-bucket',
    Key='target.pdf',
)

pdf_bytes = s3_response_object['Body'].read()

But I have no idea how to convert the resulting bytes into an object that can be parsed with PyDF2. The output that I need and that PyDF2 will produce is like below:

{'form1[0].#subform[0].nameandmail[0]': 'Burt Lancaster',
 'form1[0].#subform[0].mailaddress[0]': '675 Creighton Ave, Washington DC',
 'form1[0].#subform[0].Principal[0]': 'David St. Hubbins',
 'Principal[1]': None,
 'form1[0].#subform[0].Principal[2]': 'Bart Simpson',
 'Principal[3]': None}

So in summary, I need o be able to read a pdf with fillable forms, into memory and parse it without downloading the file because my lambda function environment won't allow local temp files.

Solution

Edit (March 2024)

There seems to have been a change in the boto3 library, here is the latest solution

from io import BytesIO
import boto3
from PyPDF2 import PdfReader

s3 = boto3.client("s3")
pdf_file = s3.get_object(Bucket="***", Key="***")[
    "Body"
].read()
reader = PdfReader(BytesIO(pdf_file))

for page in reader.pages:
    print(f"Text: {page.extract_text()}")

Previous Answer:

Solved, this does the trick:

import boto3
from PyPDF2 import PdfFileReader
from io import BytesIO

bucket_name ="pdf-forms-bucket"
item_name = "form.pdf"


s3 = boto3.resource('s3')
obj = s3.Object(bucket_name, item_name)
fs = obj.get()['Body'].read()
pdf = PdfFileReader(BytesIO(fs))

data = pdf.getFormTextFields()