PipeJinja2

The PipeJinja2 operator is a powerful utility for rendering Jinja2 templates. It allows you to dynamically generate text by combining data from your pipeline's working memory with a template. This is ideal for creating formatted reports, HTML content, or constructing complex, multi-part prompts for LLMs.

How it works

PipeJinja2 takes all the data currently in the WorkingMemory and uses it as the context for rendering a Jinja2 template. The resulting text is then saved back to the working memory as a new Text output.

The template itself can be provided in one of two ways: 1. By Name: Referring to a template file that has been loaded into Pipelex's template provider. This is the most common and maintainable method. 2. Inline: Providing the template as a multi-line string directly in the .toml file.

Templating Context

The Jinja2 template has access to all the "stuffs" currently in the working memory. You can access them by the names they were given in previous pipeline steps. For example, if a previous step produced an output named user_profile, you can access its attributes in the template like {{ user_profile.name }} or {{ user_profile.email }}.

Configuration

PipeJinja2 is configured in your pipeline's .toml file.

TOML Parameters

Parameter	Type	Description	Required
PipeJinja2	string	A descriptive name for the pipe's function.	Yes
output	string	The concept for the output. Defaults to native.Text.	No
jinja2_name	string	The name of a pre-loaded template file.	Yes (or jinja2)
jinja2	string	An inline Jinja2 template string.	Yes (or jinja2_name)
extra_context	table (dict)	A table of key-value pairs to add to the rendering context, making them available as variables in the template.	No

Example: Generating a report from a template file

This example assumes you have a user profile object and a list of activities in the working memory, and a template file named weekly_report.md has been registered.

weekly_report.md template file:

# Weekly Report for {{ user.name }}

Hello {{ user.first_name }},

Here is a summary of your activity this week:

{% for activity in activities %}
- {{ activity.timestamp }}: {{ activity.description }}
{% else %}
- No activity logged this week.
{% endfor %}

Report generated on: {{ report_date }}

Pipeline TOML definition:

[pipe.generate_weekly_report]
PipeJinja2 = "Generate a formatted weekly report for a user"
output = "WeeklyReportText"
jinja2_name = "weekly_report.md"
extra_context = { report_date = "2023-10-27" }

In this scenario: - PipeJinja2 will load the weekly_report.md template. - It will use the user and activities objects from the working memory. - It will add report_date to the context from the extra_context table. - The rendered markdown text will be saved as the WeeklyReportText concept.