AskYourPDF is a cutting-edge solution revolutionising document interactions. Our API empowers developers with the ability to programmatically extract valuable information from PDF files and leverage it to create custom chatbots. By providing seamless access to the content within PDF documents, developers can build powerful applications that enable users to navigate, search, and engage with data effortlessly. This comprehensive guide will walk you through the API's features, pricing, authentication methods, and usage guidelines, opening up endless possibilities to enhance productivity and knowledge retrieval for your applications. Harness the potential of AskYourPDF API and embark on an innovative journey of intelligent document analysis and interactive experiences.
Uploading A Document
To upload a document, choose between generating a document ID on our website or using the API endpoint. We support various document formats, including '.pdf', '.txt', '.ppt', '.pptx', '.csv', '.epub', and '.rtf'. Additionally, you can upload PDFs using links. Moreover, the API enables interaction with any blog post or website by providing the link to the post.
Authentication
Authentication is required for all our API endpoints. To access them, you must generate API keys on your AskYourPDF account. These API keys need to be set in your request header as "x-api-key." It is essential to treat your API key as a secret and ensure its protection.
headers: {"x-api-key":"ask_xxxxx" }
1. Adding Document via URL
Download Pdf
Download a PDF file from a URL and save it to the database.
Args:
user: The user who is uploading the file.
url: The URL of the PDF file to download.
Returns:
dict: The document ID of the downloaded file.
Endpoint for initiating a chat with a document. This endpoint requires API Key Header authentication.
Chat With Doc Endpoint
Chat with a single document.
Args:
doc_id: The document id.
model_name: The model name.
messages: The list of messages.
temperature: The model temperature max value is 1.5 and min value is 0.1.
stream: Whether to stream the response or not.
user: The current user.
doc_id (string, required): The ID of the document for the conversation.
Query Parameters:
model_name: Available values are CLAUDE1, GPT3 AND GPT4(GPT4 Credits Required). Default is GPT3.
stream (boolean, optional): Flag for streaming. Default is false.
Request body:
sender (required): The sender of the chat message. The sender should be User
message (required): The chat message content.
The request body to ask a single question.
[ {"sender":"User","message":"What does this document say?" }]
To ask a follow-up question and provide the model with context you can send your previous questions and responses in the following format.
[ { "sender":"user","message":"What does the document say?" }, {"sender":"bot","message":"The document consists of words in different languages expressing gratitude" }, {"sender":"user","message":"What is the word expressing gratitude in Spanish?" }]
Response:
The response when the query parameter stream = False
200 Successful Response
question: The question posed by the user, with sender, message, and type.
sender
message
type
answer: The answer provided by the AI, with sender, message, and type.
sender
message
type
created (string): The time the chat was created.
{"question": {"sender":"user","message":"What does this document say?","type":"question" },"answer": {"sender":"bot","message":"This document talks about AI","type":"response" },"created":"2023-07-20T11:14:55.928Z"}
By setting the query parameter stream = True, you can receive the response as a stream of token. An example of the response can be seen below.
page (integer, optional): The page to return. Defaults to 1.
page_size (integer, optional): The number of documents to return per page. Defaults to 10.
Response:
200 Successful Response
total_pages (Integer): The total number of pages available for querying
documents (Array): A list of document details belonging to the user.
name
doc_id
summary
language
pages
shareable
date_time
{"total_pages":1,"documents": [ {"name":"Nature.pdf","doc_id":"6e60e87c-6154-4dff-8e62-ff10d8ed16dd","summary":"This is a document about nature","language":"en","pages":10,"shareable":true,"date_time":"2023-07-21T06:18:18.891Z" } ]}
Examples:
cURL
curl -X GET
-H 'x-api-key: YOUR_API_KEY'
'https://api.askyourpdf.com/v1/api/documents?page=1&page_size=10'
doc_id (string, required): The ID of the document for the conversation.
Response:
200 Successful Response
name
doc_id
summary
language
pages
shareable
date_time
{"name":"Nature.pdf","doc_id":"6e60e87c-6154-4dff-8e62-ff10d8ed16dd","summary":"This is a document about nature","language":"en","pages":10,"shareable":true,"date_time":"2023-07-21T06:18:18.891Z" }
Examples:
cURL
curl -X GET
-H 'x-api-key: YOUR_API_KEY'
'https://api.askyourpdf.com/v1/api/documents/{YOUR_DOC_ID}'
stream (boolean, optional): Flag for streaming. Default is False.
Request body:
doc_id (String, required): The ID of the document to summarise.
length (String, required): How long the summary should be, with options AUTO, LONG and SHORT.
format (String, required): The arrangement of the summary with options AUTO, PARAGRAPH, BULLET and PAPER.
markdown (boolean, required): Flag for determining if the response should be in markdown. Default is false.
prompt (String, required): Additional instructions to be given for summarising the document.
The request body to summarise a document
{"doc_id":"6e60e87c-6154-4dff-8e62-ff10d8ed16dd","length":"AUTO","format":"AUTO","markdown":true,"prompt":"Summarise the following context"}
Response:
200 Successful Response
summary (String): The summary of the document given based on the instructions from the request.
date_time (String): The time the summary was created.
prompt (String): The set of additional instructions given from the request body.
language (String): The language in which the summary is given.
document_name (String): The name of the document.
pages (Integer): The number of pages of the document.
{"summary":"This is a document that talks about nature","date_time":"2023-10-01T12:42:50.675Z","prompt":"Summarise the following context","language":"en","document_name":"Nature.pdf","pages":10}
By setting the query parameter, stream = true, you can receive the response as a stream of token. An example of the response can be seen below.
Endpoint for initiating a chat with multiple documents. This endpoint requires API Key Header authentication.
Chat With Multiple Documents
Chat with multiple document.
Args:
knowledge_base: The knowledge base request payload containing the list of documents and messages.
model_name: The model name.
temperature: The model temperature max value is 1.5 and min value is 0.1.
stream: Whether to stream the response or not.
user: The current user.
model_name: Available values are CLAUDE1, GPT3 AND GPT4(GPT4 Credits Required). Default is GPT3.
stream (boolean, optional): Flag for streaming. Default is False.
Request body:
documents (Array, required): A list of documents from which a response will be generated. Takes in multiple document IDs.
messages (Array, required): A list of messages to send.
sender (String, required): The sender of the chat message. The sender should be User
message (String, required): The chat message content.
The request body to ask a single question.
{"documents": ["6e60e87c-6154-4dff-8e62-ff10d8ed16dd","7f71f98d-7265-5egg-9f73-gg21e9fe27ee" ],"messages": [ {"sender":"User","message":"What is the common theme of the documents?" } ]}
To ask a follow-up question and provide the model with context, you can send your previous questions and responses in the following format.
{"documents": ["6e60e87c-6154-4dff-8e62-ff10d8ed16dd","7f71f98d-7265-5egg-9f73-gg21e9fe27ee" ],"messages": [ {"sender":"user","message":"What is the common theme of the documents?" }, {"sender":"bot","message":"These documents talk about words and their synonyms across various languages." }, {"sender":"user","message":"What are the words for expressing gratitude in Spanish?" } ]}
Response:
The response when the query parameter stream = False
200 Successful Response
question: The question posed by the user, with sender, message, and type.
sender (String)
message (String)
type (String)
answer: The answer provided by the AI, with sender, message, and type.
sender (String)
message (String)
type (String)
created (String): The time the chat was created.
{"question": {"sender":"user","message":"What is the common theme of the documents?","type":"question" },"answer": {"sender":"bot","message":"These documents talk about words and their synonyms across various languages.","type":"response" },"created":"2023-07-20T11:14:55.928Z"}
By setting the query parameter stream = True, you can receive the response as a stream of token. An example of the response can be seen below.
Chunk: These
Chunk: documents
Chunk: talk
Chunk: about
Chunk: words
Chunk: and
Chunk: their
Chunk: synonyms
Chunk: across
Chunk: various
Chunk: languages.
Example when stream = False:
cURL
curl -X 'POST' \
'https://api.askyourpdf.com/v1/api/knowledge_base_chat?model_name=GPT3' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-d '{
"documents": [
"6e60e87c-6154-4dff-8e62-ff10d8ed16dd",
"7f71f98d-7265-5egg-9f73-gg21e9fe27ee"
],
"messages": [
{
"sender": "user",
"message": "What is the common theme of the documents?"
}
]
}'
Python
import requests
import json
headers = {
'Content-Type': 'application/json',
'x-api-key': 'YOUR_API_KEY'
}
data = {
"documents": [
"6e60e87c-6154-4dff-8e62-ff10d8ed16dd",
"7f71f98d-7265-5egg-9f73-gg21e9fe27ee"
],
"messages": [
{
"sender": "user",
"message": "What is the common theme of the documents?"
}
]
}
response = requests.post(
'https://api.askyourpdf.com/v1/api/knowledge_base_chat?model_name=GPT3',
headers=headers, data=json.dumps(data))
if response.status_code == 200:
print(response.json())
else:
print('Error:', response.status_code)
Javascript
const axios = require('axios');
const headers = {
'Content-Type': 'application/json',
'x-api-key': 'YOUR_API_KEY'
};
const data = {
"documents": [
"6e60e87c-6154-4dff-8e62-ff10d8ed16dd",
"7f71f98d-7265-5egg-9f73-gg21e9fe27ee"
],
"messages": [
{
"sender": "user",
"message": "What is the common theme of the documents?"
}
]
};
axios.post('https://api.askyourpdf.com/v1/api/knowledge_base_chat?model_name=GPT3',
data, {headers: headers})
.then((response) => {
if (response.status === 200) {
console.log(response.data);
} else {
console.log('Error:', response.status);
}
})
.catch((error) => {
console.error(error);
});
Example when stream = True:
Python
import requests
import json
headers = {
'Content-Type': 'application/json',
'x-api-key': 'YOUR_API_KEY'
}
data = {
"documents": [
"6e60e87c-6154-4dff-8e62-ff10d8ed16dd",
"7f71f98d-7265-5egg-9f73-gg21e9fe27ee"
],
"messages": [
{
"sender": "user",
"message": "What is the common theme of the documents?"
}
]
}
try:
response = requests.post(
'https://api.askyourpdf.com/v1/api/knowledge_base_chat?stream=True',
headers=headers, data=json.dumps(data))
response.raise_for_status()
for chunk in response.iter_content(chunk_size=24):
chunk_str = chunk.decode('utf-8')
print("Chunk:", chunk_str)
except requests.exceptions.RequestException as error:
print("Error:", error)