Context Enricher

Modules concerning the context enrichers used in Gen AI applications.

`MetadataContextEnricher(metadata_fields, position=MetadataPosition.PREFIX, separator='\n---\n', field_template='- {field}: {value}', skip_empty=True, binary_handling=BinaryHandlingStrategy.BASE64)`

Bases: BaseContextEnricher

A metadata context enricher that adds metadata to the chunk content.

This enricher formats metadata fields into a string and appends it to the chunk content based on the specified position (prefix or suffix).

Attributes:

Name	Type	Description
`metadata_fields`	`list[str]`	List of metadata fields to include in the enriched content.
`position`	`MetadataPosition`	Position of the metadata in the content. Valid values are defined in the MetadataPosition enum: - PREFIX: Metadata block is placed before content - SUFFIX: Metadata block is placed after content
`separator`	`str`	Separator between the metadata and the content.
`field_template`	`str`	Template for formatting each metadata field.
`skip_empty`	`bool`	Whether to skip fields with empty values.
`binary_handling`	`BinaryHandlingStrategy`	Strategy for handling binary data. Valid values are defined in the BinaryHandlingStrategy: - BASE64: Binary data is converted to base64 (default) - HEX: Binary data is converted to hexadecimal - NONE: Binary data is not included in the metadata block

Initialize the metadata context enricher.

Parameters:

Name	Type	Description	Default
`metadata_fields`	`list[str]`	List of metadata field names to include.	required
`position`	`MetadataPosition`	Where to place metadata block. Valid values are defined in the MetadataPosition enum: 1. "prefix": Metadata block is placed before content 2. "suffix": Metadata block is placed after content	`PREFIX`
`separator`	`str`	String to separate metadata from content.	`'\n---\n'`
`field_template`	`str`	Template for formatting each metadata field. Available fields: - {field}: Field name - {value}: Field value	`'- {field}: {value}'`
`skip_empty`	`bool`	Whether to skip empty fields.	`True`
`binary_handling`	`BinaryHandlingStrategy`	Strategy for handling binary data. Valid values are defined in the BinaryHandlingStrategy: 1. "base64": Binary data is converted to base64 (default) 2. "hex": Binary data is converted to hexadecimal 3. "none": Binary data is not included in the metadata block	`BASE64`

`TranscriptMode`

Bases: StrEnum

Modes for transcript rendering in video context enrichment.

`VideoContextEnricher(transcript=TranscriptMode.PLAIN, group_by=None, format_timestamps=_format_seconds_to_time)`

Bases: BaseContextEnricher

A context enricher specialized for video content.

This strategy transforms raw video metadata into a structured, human-readable format ideal for LM context. It supports two main operational modes:

Single Mode: Processes a single chunk containing a full video summary and multiple segments.
Grouped Mode: Aggregates multiple chunks (e.g., individual segments) sharing the same identifier (like file_id) into a single comprehensive chunk to avoid header duplication.

Output Format

Video: [video_<file_id>]
    Summary: <video_summary>

    [<MM:SS-MM:SS>]
    Caption: <description of the visual content>
    Transcript: <spoken words in the segment>

Key Features

Consolidation: Use group_by to merge segment-level chunks.
Dynamic Formatting: Toggles transcripts and normalizes timestamps (e.g., 70s -> 01:10).

Examples:

enricher = VideoContextEnricher(group_by="file_id")
enriched_chunks = await enricher.enrich(chunks)
print("\n".join(enriched_chunks[0].additional_context))
# Output:
# Video: [my_video.mp4]
# Summary: A chef preparing a meal.
#
# [00:00-00:05]
# Caption: The chef chops onions.
# Transcript: First, we dice the onions finely.

Attributes:

Name	Type	Description
`transcript`	`TranscriptMode`	Controls transcript rendering.
`group_by`	`str \| tuple[str, ...] \| None`	Metadata key(s) used to identify chunks belonging to the same video.
`format_timestamps`	`Callable[[float \| int], str]`	Converts seconds to `MM:SS` / `HH:MM:SS`.

Initializes the Video Context Enricher.

Parameters:

Name	Type	Description	Default
`transcript`	`TranscriptMode`	Controls transcript rendering. 1. `TranscriptMode.PLAIN`: includes transcript text. 2. `TranscriptMode.TIMESTAMPS`: prefixes each line with the segment timestamp. 3. `TranscriptMode.NONE`: omits transcripts entirely. Defaults to TranscriptMode.PLAIN.	`PLAIN`
`group_by`	`str \| tuple[str, ...] \| None`	Metadata key(s) to use for grouping chunks. Crucial for consolidating individual segment chunks into one video summary. 1. `None`: No grouping; each chunk enriched independently. 2. `"file_id"`: Merge chunks sharing the same `file_id`. 3. `"video_id"`: Merge chunks sharing the same `video_id`. 4. `"source"`: Merge chunks sharing the same `source` path. 5. `("source_type", "file_id")`: Merge chunks sharing the same combination of `source_type` and `file_id` (composite key). Video chunks missing any grouping key (or with empty/falsy values) are enriched as singletons. Non-video chunks pass through unchanged.	`None`
`format_timestamps`	`Callable[[float \| int], str]`	A callable that converts seconds to a formatted string. Defaults to `_format_seconds_to_time` (MM:SS / HH:MM:SS).	`_format_seconds_to_time`