# Extract text from file

### What's it for?

* The client can send files in docx, pdf, rtf, txt or any other format;
* The bot will be able to process the file, or pass the result from the file to other action steps.

### Creating an action to extract text from a file

{% stepper %}
{% step %}

### Go to the “Workflows” section and click the “+ Add” button

<figure><img src="/files/lTi8pHB1GpdOA4cSrSI6" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### We choose "Extract text from file"

<figure><img src="/files/ojkEZX3FxXw5bIIbxDtB" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### Go to Settings

<figure><img src="/files/SBEvdDyOHgHt7xWgxHVT" alt=""><figcaption></figcaption></figure>

* **Action name —** a display name by which you can easily identify the required tool in the general list
* **+ Add Step —** you can trigger multiple webhooks or perform multiple other functions within a single action. For each new webhook, you must add an additional step using the "+ Add Step" button.
* ⚙️(*Bot response rules) —* with this setting, we have the option to choose which step's response will be passed to the bot - from the first, the last, or from all.
* **Argument in which the file will be transferred** — is a field for selecting the argument in which the file will be passed to the function.
* **Export mode —** here you can choose one of two options for extracting text from files:
  1. HTML extraction
  2. Markdown extraction
* **Put the received text into a variable —** enable the option so that after calling the webhook, the received text can be placed in a variable and passed to another function if necessary.
* **Scanned Text Reading Mode** — you can choose one of three modes for reading scanned text:
  1. Do not use text recognition
  2. Use recognition if the file is a PDF and contains at least one image
  3. Always recognize text on images
     {% endstep %}

{% step %}

### **Trigger conditions**

<figure><img src="/files/0gjaJAWcLeXUKBgwrTs0" alt=""><figcaption></figcaption></figure>

{% hint style="info" %}
To automatically generate a **Function Name**, fill in the **Action Name** field and press the **auto-generation button** to the right of the function name field.
{% endhint %}

* **Call by LLM model** — the function is called by the LLM model through instructions written to the bot.
* **Trigger call** — the function is called based on a pre-defined event when the specified conditions are met.
* **Function Name** — should describe the main idea of this action. Only English alphabet characters and underscores (\_) are allowed.
* **Function Description** — a description of what this action will do.
  {% endstep %}

{% step %}

### **Action variables**

<figure><img src="/files/o58OBWegppww1hvFr078" alt=""><figcaption></figcaption></figure>

* **Argument** - since the argument for the function any File by design, in the **Argument** field we write **`file`**.
* **Argument Description** - this field is optional and serves as a description to avoid confusion if there are multiple arguments.

{% hint style="warning" %}
Ensure that the **argument** is written in English and without spaces. You can use English alphabet characters and underscores (`_`).
{% endhint %}
{% endstep %}

{% step %}

### Additional Settings

<figure><img src="/files/dVkoQnDD0jv1ZSYj0xLC" alt=""><figcaption></figcaption></figure>

* **Stop dialog on call** — after calling the function to "Extract text from file", the dialogue will stop and the bot will stop responding.
* **Do not respond on call** — after calling the "Extract text from file" function, the dialogue will not be stopped, but no response will be sent to the dialogue.
* **Notify on error** — if a function call fails, you will receive a notification.
* **Stop on error** — if the function call fails, the dialog will stop.
* **Error Message** — what text will be returned to the bot in case of an error. We recommend writing something like "An error occurred"
  {% endstep %}
  {% endstepper %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.suvvy.ai/en/workflows/extract-text-from-file.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.