ChatGPT reads Text files
Project description
Text ChatGPT Connector
The Text ChatGPT Connector is a simple script for connecting text files and ChatGPT.
Forked from scrapbox_chatgpt_connector.
Ref: https://scrapbox.io/villagepump/Scrapbox_ChatGPT_Connector
How to install
Install by pip:
$ pip install text_chatgpt_connector
Or you can use repository directly. Install Poetry and run
$ poetry install
$ poetry run tcc ...
You can run `tcc directly if you enter the poetry virtual environment:
$ poetry shell
How to use
Make index
To make index, run:
$ tcc index -i <input_dir>
input_dir
is the directory which has text files (with suffix of .txt
, .md
or .markdown
by default).
This will create pickle file named index.pickle.
If you want to change the name, set -o <output_file>
.
If you run with different input_dir
for the same output_file
,
the index is appended to the previous one.
Ask question
To ask a question, run:
$ tcc ask -q "もっとも大事な問いとは何だろう?"
If you want compare the answer without index, run with -n
(--no_index
) option:
$ tcc ask -n -q "もっとも大事な問いとは何だろう?"
Usage
usage: tcc [-h] [-i INPUT_DIR] [-s INPUT_SUFFIX] [-o OUTPUT_FILE] [-k KEY] [-c CHARACTER_ENCODING] [--chat_model CHAT_MODEL]
[--encoding ENCODING] [--embedding EMBEDDING] [--remain_url] [--keep_spaces] [--block_size BLOCK_SIZE]
[--embed_max_size EMBED_MAX_SIZE] [--max_prompt_size MAX_PROMPT_SIZE] [--return_size RETURN_SIZE] [--prompt PROMPT]
[--bare_prompt BARE_PROMPT] [-q QUESTION] [-n] [-v]
[command]
positional arguments:
command Command (index or ask)
options:
-h, --help show this help message and exit
-i INPUT_DIR, --input_dir INPUT_DIR
Input directory
-s INPUT_SUFFIX, --input_suffix INPUT_SUFFIX
Comma separated suffixes of input files, "txt,md,markdown"
-o OUTPUT_FILE, --output_file OUTPUT_FILE
Output file (pickle), default: "index.pickle"
-k KEY, --key KEY OpenAI API key. If not given, try to get from environment variable: OPENAI_API_KEY.
-c CHARACTER_ENCODING, --character_encoding CHARACTER_ENCODING
Character encoding for input file, default: "utf-8"
--chat_model CHAT_MODEL
Chat model name, default: "gpt-3.5-turbo"
--encoding ENCODING Encoding name for tiktoken, default: "cl100k_base"
--embedding EMBEDDING
Embedding model name, default: "text-embedding-ada-002"
--remain_url Keep URL in the text
--keep_spaces Keep spaces in the text
--block_size BLOCK_SIZE
Block size for embedding, default: 500
--embed_max_size EMBED_MAX_SIZE
Max size for embedding, default: 8150
--max_prompt_size MAX_PROMPT_SIZE
Max size for prompt, default: 4096
--return_size RETURN_SIZE
Return size, default: 250
--prompt PROMPT Prompt template, default: "Read the following text and answer the question. Your reply should be shorter
than 250 characters. ## Text {text} ## Question {question}"
--bare_prompt BARE_PROMPT
Prompt template without index, default: "Read the following text and answer the question. Your reply
should be shorter than 250 characters. ## Text {text} ## Question {question}"
-q QUESTION, --question QUESTION
Question words for ask, default: "What is the most important question?"
-n, --no_index Ask the question directly
-v, --version Show version
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file text_chatgpt_connector-0.0.2.tar.gz
.
File metadata
- Download URL: text_chatgpt_connector-0.0.2.tar.gz
- Upload date:
- Size: 7.5 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.4.0 CPython/3.11.2 Linux/5.15.90.1-microsoft-standard-WSL2
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 2cd2d3de3cb9310507ef82facb0e3e098c0abb76918a919faa8eedbf67aab757 |
|
MD5 | 4f91e6f98210bfd0570adb58e1c738c4 |
|
BLAKE2b-256 | 212a1e5fd4015459a51540f0fbce72b1227ebfd57752e303e0a7c2b43d1c5db1 |
File details
Details for the file text_chatgpt_connector-0.0.2-py3-none-any.whl
.
File metadata
- Download URL: text_chatgpt_connector-0.0.2-py3-none-any.whl
- Upload date:
- Size: 8.4 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.4.0 CPython/3.11.2 Linux/5.15.90.1-microsoft-standard-WSL2
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | f4728cbe9ed90f4fc32119639146d1ef664c1b36b7d6e00ec3c9629e8e883c46 |
|
MD5 | 297ea36b5a1a9028a442af98cccc70b6 |
|
BLAKE2b-256 | c493b48386d1c5b2b63c16b042b0fdbae4eb1367f101c9ecc427731786a7307a |