> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bigdata.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Inline Attribution Formatter

Chat answers contain inline attributions and uses
`DefaultFormatter` by default.

Example:

```python theme={null}
# Create a new chat
chat = bigdata.chat.new("Pfizer company analysis", formatter=DefaultFormatter())

# Ask a question
response = chat.ask("Evaluate the experience and reputation of the management team of Pfizer")
```

Output:

```Output theme={null}
## Pfizer's Management Team

### CEO and Chairman: Dr. Albert Bourla
- Dr. Bourla has been the Chairman and CEO of Pfizer since 2019 `:ref[13]`  
- He is a Doctor of Veterinary Medicine and holds a Ph.D. from the Veterinary School of Aristotle University `:ref[13]`  
- In 2022, he was named the Genesis Prize Laureate, often referred to as the "Jewish Nobel Prize" `:ref[13]`  
...
```

`DefaultFormatter` adds inline attributes as a
`` `:ref[index]` `` where `index` is an index of
inline attribution element in `sources` property in
`ChatInteraction`

There are two formatters available at the moment and it is also possible
to [create custom formatter](#create-a-custom-formatter)

1. DefaultFormatter
2. MarkdownLinkFormatter

# How to configure it

You can configure the inline attribution formatter at different levels:

* When chat is created
* When getting existing chat
* When getting list of chats
* When making `chat.ask` request

## When chat is created

```python theme={null}
from bigdata_client.models.chat import MarkdownLinkFormatter

chat = bigdata.chat.new("Pfizer company analysis", formatter=MarkdownLinkFormatter())

# all following chat.ask calls will use formatter we set when creating chat
response = chat.ask("Evaluate the experience and reputation of the management team of Pfizer")
print(response.answer)
```

## When getting existing chat

```python theme={null}
from bigdata_client.models.chat import MarkdownLinkFormatter

# Get existing chat
chat_id = chat.id

chat = bigdata.chat.get(chat_id, formatter=MarkdownLinkFormatter())

# All chat interactions will be formatted using provided formatter
for interaction in chat.interactions:
    # answer will be formatted with formatter
    print(interaction.answer)

# All following chat.ask calls will use formatter we set when fetching chat
response = chat.ask("Evaluate the experience and reputation of the management team of Pfizer")
print(response.answer)
```

## When getting list of available chats

```python theme={null}
from bigdata_client.models.chat import MarkdownLinkFormatter

# Get list of chats
chats = bigdata.chat.list(formatter=MarkdownLinkFormatter())

# All chat interactions will be formatted using provided formatter, let's focus only in our first chat
for interaction in chats[0].interactions:
    # answer will be formatted with formatter
    print(interaction.answer)

# All following chat.ask calls will use formatter we set when fetching chat list
```

## When making `chat.ask` request

In case it's necessary to chage formatter only for one
`chat.ask` call there is a possibility:

```python theme={null}
# Create a new chat
chat = bigdata.chat.new("Pfizer company analysis")

# This specific call to chat.ask will use MarkdownLinkFormatter
response = chat.ask(
    "Evaluate the experience and reputation of the management team of Pfizer",
    formatter=MarkdownLinkFormatter()
)
print(response.answer)

# All following chat.ask calls will use formatter we set when creating chat
response = chat.ask("Evaluate the experience and reputation of the management team of Pfizer")
print(response.answer)
```

# Create a custom formatter

It is possible to create custom inline attribution formatter.

All inline attribution formatters extend abstract class
`InlineAttributionFormatter`

It is necessary to implement `format` method.

Here an example how to create custom formatter:

```python theme={null}
from bigdata_client.models.chat import InlineAttributionFormatter, ChatSource

class HtmlLinkFormatter(InlineAttributionFormatter):
    def format(self, index: int, attribution: ChatSource) -> str:
        a_text = attribution.headline
        a_href = attribution.url
        return f'<a href="{a_href}">{a_text}</a>'

# Create a new chat using custom formatter
chat = bigdata.chat.new("Pfizer company analysis", formatter=HtmlLinkFormatter())
response = chat.ask("Evaluate the experience and reputation of the management team of Pfizer")
print(response.answer)
```

Output:

```Output theme={null}
## Pfizer's Management Team

Pfizer's management team is led by a highly experienced and respected group of executives:

### Albert Bourla, Chairman and Chief Executive Officer
- Joined Pfizer in 1993 and has held various leadership roles, including Group President of Pfizer's Innovative Health business <a href="https://www.linkedin.com/in/albert-bourla">Albert Bourla - Chief Executive Officer - Pfizer</a> 
- Became CEO in 2019 and was named the Genesis Prize Laureate in 2022 <a href="https://www.linkedin.com/in/albert-bourla">Albert Bourla - Chief Executive Officer - Pfizer</a> 
```