https://github.com/macrat/llawk
AWK-like simple text processor but working with LLM (Large Language Model).
https://github.com/macrat/llawk
Last synced: 12 months ago
JSON representation
AWK-like simple text processor but working with LLM (Large Language Model).
- Host: GitHub
- URL: https://github.com/macrat/llawk
- Owner: macrat
- License: mit
- Created: 2025-01-16T09:31:23.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-04-27T13:47:32.000Z (about 1 year ago)
- Last Synced: 2025-04-27T14:33:27.820Z (about 1 year ago)
- Language: Go
- Size: 22.5 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# llawk
AWK-like simple text processor but working with LLM (Large Language Model).
It allows you to easily perform text conversion, extraction, translation, and any other text processing tasks.
For example:
```shell
$ ls -l | llawk -f json "Extract file names and sizes"
{
"files": [
{
"name": "LICENSE",
"size": 1079
},
{
"name": "README.md",
"size": 3358
},
.
.
.
]
}
```
## Supported LLMs
- OpenAI
- GPT-4.1
- GPT-4.1-mini
- GPT-4.1-nano
- GPT-4o
- GPT-4o-mini
- o3
- o4-mini
- Google
- gemini-2.0-flash
- gemini-2.0-flash-lite
- gemini-1.5-pro
- Anthropic
- claude-3.7-sonnet
- claude-3.5-haiku
- Ollama
- any models which supports chat completion
## Installation
To install `llawk`, use the following command:
```shell
$ go install github.com/llawk/llawk@latest
```
## Usage
llawk takes an instruction from the command line, and an input text from stdin or a file.
```shell
$ echo hello | llawk "Translate to Japanese"
^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^
Input Instruction
$ llawk -i input.txt "Translate to Japanese"
^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^
InputFile Instruction
```
It supports a JSON output format using the `-f json` option:
```shell
$ ls -l | llawk -f json "Extract file names and sizes"
```
You can also specify a JSON Schema for the output:
```shell
$ echo hello | llawk -f '{"type":"object","properties":{"japanese":{"type":"string"},"french":{"type":"string"}}}' "Translate to Japanese"
```
You can specify the model using the `-m` option or the `LLAWK_MODEL` environment variable:
```shell
$ echo hello | llawk -m gemini-2.0-flash "Translate to Japanese"
```
To see the list of supported models, run `llawk -m list`.
### OpenAI models
OpenAI models require an API key to access.
Please set `OPENAI_API_KEY` environment variable before using OpenAI models.
```shell
$ export OPENAI_API_KEY=sk-xxxxxx
$ echo hello | llawk -m gpt-4o "Translate to Japanese"
```
You can specify the OpenAI API organization using the `OPENAI_ORG_ID` environment variable.
### Google models
Google models require an API key to access.
Please set `GEMINI_API_KEY` environment variable before using Google models.
```shell
$ export GEMINI_API_KEY=xxxxxx
$ echo hello | llawk -m gemini-2.0-flash "Translate to Japanese"
```
### Anthropic models
Anthropic models require an API key to access.
Please set `ANTHROPIC_API_KEY` environment variable before using Anthropic models.
```shell
$ export ANTHROPIC_API_KEY=xxxxxx
$ echo hello | llawk -m claude-3.5-haiku "Translate to Japanese"
```
### Ollama models
You can use Ollama models without any special settings because Ollama does not require an API key.
Any model supported by Ollama that offers chat completion can be used.
```shell
$ echo hello | llawk -m ollama:llama3.2 "Translate to Japanese"
$ echo hello | llawk -m ollama:hf.co/elyza/Llama-3-ELYZA-JP-8B-GGUF "Translate to Japanese"
```
If you want to connect to a specific Ollama server, you can specify the server URL with the `OLLAMA_HOST` environment variable.
```shell
$ export OLLAMA_HOST=https://192.168.1.123:11434
$ echo hello | llawk -m ollama:llama3.2 "Translate to Japanese"
```