コーディングなしで自動的にPDFからテキストを抽出する方法！

Build your automated pdf to text extractor in a minute! Perfect solution for those who deal with large amounts of incoming CVs or other documents! Extract data from documents using remote workflow on Latenode.com!

‍
In this article you’ll see how to create an automated no code workflow that will extract data from the PDF files on your Google Drive using API and Chat GPT, and filling the Google Sheet form with the data.

Also, you can get the template of that scenario that you can copy and use for yourself for free! 

You can upgrade this template or use it not only for pdf text extraction.

Scenario overview

Let’s take a look at the whole pdf scraper scenario first, and then break each step down.

‍

How does this pdf extract scenario work? Once in an hour it takes all the CVs in pdf format from the Google Drive, converts to the txt files using HTTP-request, then chatGPT assistant Extracts the needed data and fills out the Google Spreadsheet.

Main steps of scenario:

1. Schedule Trigger. Activates the scenario at a set period of time.
2. Google Drive. Here we have 2 nodes: Find file and Download File.
3. Javascript. We’ll use Javascript nodes with AI copilot for file formatting throughout the scenario. First, I asked AI to give me code that converts files from Google Drive to base64 format.
4. HTTP request. Use API to extract text from pdf using tried party services API docs.
5. ChatGPT.  ChatGPT assistant extracts the data from the text.
6. Google Sheets. Fills out the sheet with that information,

Scenario breakdown

And here’s a step-by-step instruction for those who want to learn the process. Remember about free to use and ready to go templates at the end of the article.

Preparations

For this scenario, you’ll need to have the ability to use Open AI Assistants and any API converter.

OpenAI assistant

それは簡単です。OpenAIのアカウントにログインして アシスタントにアクセスし、右上の "create "ボタンをクリックします。アシスタントの設定パネルが表示されます。

ここで必要なのは

1. アシスタントに名前をつける
2. どのように回答してほしいか、文脈を指示してください。添付ファイルを使うように指示することをお勧めします。
3. モデルを選ぶ
4. 使用可能なツールを選択し、アシスタントが使用できるファイルを追加します。ファイルの追加を可能にするために、検索をオンにすることを忘れないでください。

Copy the assistant’s ID (you see it under the Name column), then go to API keys, create one and save it.

PDF to TXT converter

Use any you like. I took ConvertAPI because it has a free trial and provides a lot of info.

‍

Create scenario

Now we switch to Latenode.com. Here, we have to create the pdf to text conversion scenario: (You don't need to create it from scratch, just copy the template at the end of the article)

スケジュール・トリガー

Click “Add node” in the scenario tab, choose “Schedule” from the list. Click on the node to set it up. Specify the interval and the timezone, save the changes. I also added a trigger on Run once, just for convenience.

Google Drive nodes

Click “Add node”, search for Google drive folder in actions tab, and choose “Find file” node.
To make it work you have to log into gmail account to get access token, choose drive and Search name. In this case, I want to extract data from files that have CV in their name.

‍

Next add “Download file” from the same action folder of Google drive.

Use id from the results of the previous node, and then click run node once to save the changes and make the data flow through the scenario. You’ll get the file in the output.

Javascript #1: Convert to base64

That is the code the AI gave me. You can take it here(#1). Replace const fileContentPath with your object from the previous node.

HTTPリクエスト

Find HTTP request in the list of actions. To understand how to create it, visit ConvertAPI documentation. Here we can take info about setting up the request.

Here’s how my pdf to txt conversion request looks like.

‍

I use an object from Find file Google Drive node to specify the name of the downloaded file, and file content in base64 from the Javascript node. And add Content-Type=application/json pair in Headers.

Run node once to get the file from the HTTP request.

Javascript #2

Another code node, #2 here.

This time, I asked AI to extract text from the txt file.

ChatGPT nodes

We’ll face 3 GPT nodes here:

• Create Thread
• Create Message
• クリエイト・ラン

 Each node performs an action with OpenAI.

First we create a thread, or conversation with the GPT Assistant

Insert your Open AI API key, that’s it! Run node once and get the id of the created thread in the output.

Create message:

Here you need an API key again. In the thread ID field put the result of the previous node. You’ll see it in the helper window after you click on the input field.

In “Message content” give some additional instructions if you want and put the filecontent from the last Javascript node. Automated pdf extractor is one step closer!

クリエイト・ラン

This node gets the pdf scraper reply.

Specify the node just like your OpenAI assistant and use GPT assistant ID.

Javascript #3

Here we use JS node the last time, to make 3 separate json objects out of the Assistants reply.

Here’s the example, just put your data in content const.

グーグル シート

Let’s put this data somewhere, Google Sheets is a good option for this pdf data extract scenario on latenode.com.

Log into gmail account once more to get access token, choose the drive and the sheet, and put jsons into the fields, save the scenario and click run once to run or deploy the scenario to activate the schedule trigger.

Results

After a successful scenario execution this workflow will extract the text from pdf file on your Google Drive and put it into your Google Spreadsheet.

That is how to create a pdf extractor with no code on latenode.

As I promised, here’s the template of this workflow. Just copy it and follow this guide article to set it up.

There’s a video about it, don’t bother reading!

If you want somebody to help you, check out our Discord channel, we have some devs in it ready to assist!

‍