Skip to content

Latest commit

 

History

History
67 lines (50 loc) · 2.58 KB

CaptionsToPromptList.md

File metadata and controls

67 lines (50 loc) · 2.58 KB

Captions To Prompt List Node

The Captions To Prompt List node is designed to combine multiple caption text files from a directory into a single consolidated list. This node is particularly useful when working with datasets that contain multiple caption files and you need to merge them into a single prompt list.

Parameters

Required Inputs

  1. directory_path (STRING)
    • The path to the directory containing caption text files
    • Default value: "/path/to/dataset"
    • Should point to a directory containing .txt files with captions

Optional Inputs

  1. reload (BOOLEAN)
    • Controls when the node should reload and reprocess the files
    • Default value: False
    • Two modes:
      • "if file changed" (True): Reloads when file contents change
      • "if value changed" (False): Reloads only when the directory path changes

Functionality

The node performs the following operations:

  1. Walks through the specified directory and its subdirectories
  2. Reads all .txt files found in the directory structure
  3. Combines all captions from these files into a single list
  4. Generates an output filename based on the input directory name

File Processing Rules

  • Only processes files with .txt extension
  • Ignores .png files silently
  • Logs when it ignores other file types
  • Handles UTF-8 encoded text files
  • Processes each caption file line by line

Outputs

The node returns two values:

  1. combined_captions (STRING)

    • All captions combined into a single string
    • Each caption is separated by a newline
    • Maintains the original order of captions within each file

  2. output_filename (STRING)

    • Generated filename for the combined captions
    • Created by taking the last directory name from the path and adding .txt extension
    • Example: If path is "/data/my_dataset", output filename will be "my_dataset.txt"

Important Notes

  1. Error Handling:

    • The node will attempt to process each .txt file independently
    • If an error occurs while processing a specific file, it will:
      • Print an error message with the file path and error details
      • Continue processing remaining files
    • UTF-8 encoding is required for text files

  2. Reload Behavior:

    • When reload is False: Only reprocesses when directory path changes
    • When reload is True: Uses MD5 hashing to detect changes in file contents
    • Hash calculation includes both filenames and file contents
    • Ensures efficient processing by only reloading when necessary

  3. The node is categorized under "custom/text" in the ComfyUI interface