Spaces:
Sleeping
Sleeping
Commit
·
547d83c
1
Parent(s):
2b02c75
add create dataset script
Browse files- create_argilla_dataset.py +88 -0
- requirements.txt +3 -0
- template.jinja +12 -0
create_argilla_dataset.py
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import random
|
| 2 |
+
import urllib.parse
|
| 3 |
+
from http import client
|
| 4 |
+
|
| 5 |
+
import argilla as rg
|
| 6 |
+
import requests
|
| 7 |
+
from jinja2 import Template
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
def generate_html(image_urls, app_url):
|
| 11 |
+
"""none"""
|
| 12 |
+
template_str = """
|
| 13 |
+
<html>
|
| 14 |
+
<body>
|
| 15 |
+
<iframe src="{{ full_url }}" width="400" height="400" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
|
| 16 |
+
</body>
|
| 17 |
+
</html>
|
| 18 |
+
"""
|
| 19 |
+
base_url = f"{app_url}/?"
|
| 20 |
+
encoded_urls = [
|
| 21 |
+
"img{}={}".format(i + 1, urllib.parse.quote(url, safe=""))
|
| 22 |
+
for i, url in enumerate(image_urls)
|
| 23 |
+
]
|
| 24 |
+
full_url = base_url + "&".join(encoded_urls)
|
| 25 |
+
|
| 26 |
+
template = Template(template_str)
|
| 27 |
+
return template.render(full_url=full_url)
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
def fetch_data(max_images, name="horse"):
|
| 31 |
+
base_url = "https://datasets-server.huggingface.co/rows"
|
| 32 |
+
params = {
|
| 33 |
+
"dataset": "gigant/horse2zebra",
|
| 34 |
+
"config": name,
|
| 35 |
+
"split": "train",
|
| 36 |
+
"offset": 0,
|
| 37 |
+
"length": max_images,
|
| 38 |
+
}
|
| 39 |
+
|
| 40 |
+
response = requests.get(base_url, params=params)
|
| 41 |
+
|
| 42 |
+
if response.status_code == 200:
|
| 43 |
+
data = response.json()
|
| 44 |
+
return [row["row"]["image"]["src"] for row in data["rows"]]
|
| 45 |
+
else:
|
| 46 |
+
print(f"Failed to fetch data. Status code: {response.status_code}")
|
| 47 |
+
return None
|
| 48 |
+
|
| 49 |
+
|
| 50 |
+
def log_records(dataset, app_url, max_images=10):
|
| 51 |
+
horse_urls = fetch_data(max_images=max_images, name="horse")
|
| 52 |
+
zebra_urls = fetch_data(max_images=max_images, name="zebra")
|
| 53 |
+
|
| 54 |
+
records = []
|
| 55 |
+
for horse_url, zebra_url in zip(horse_urls, zebra_urls):
|
| 56 |
+
markdown_str = generate_html(image_urls=[horse_url, zebra_url], app_url=app_url)
|
| 57 |
+
record = rg.Record(fields={"gradio_app": markdown_str})
|
| 58 |
+
records.append(record)
|
| 59 |
+
|
| 60 |
+
dataset.records.log(records)
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
def create_dataset(api_url="http://localhost:6900", api_key="owner.apikey"):
|
| 64 |
+
client = rg.Argilla(api_url=api_url, api_key=api_key)
|
| 65 |
+
|
| 66 |
+
# Create a dataset
|
| 67 |
+
settings = rg.Settings(
|
| 68 |
+
fields=[rg.TextField(name="gradio_app")],
|
| 69 |
+
questions=[
|
| 70 |
+
rg.LabelQuestion(
|
| 71 |
+
name="is_zebra",
|
| 72 |
+
description="Is this a zebra?",
|
| 73 |
+
labels=["true", "false"],
|
| 74 |
+
)
|
| 75 |
+
],
|
| 76 |
+
)
|
| 77 |
+
dataset = rg.Dataset(
|
| 78 |
+
name=f"horse2zebra_{random.randint(0,100)}", settings=settings, client=client
|
| 79 |
+
)
|
| 80 |
+
dataset.create()
|
| 81 |
+
return dataset
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
if __name__ == "__main__":
|
| 85 |
+
dataset = create_dataset(api_url="https://burtenshaw-gradio-field.hf.space")
|
| 86 |
+
log_records(
|
| 87 |
+
dataset, max_images=10, app_url="https://burtenshaw-gradio-field-app.hf.space"
|
| 88 |
+
)
|
requirements.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Jinja2
|
| 2 |
+
datasets
|
| 3 |
+
gradio
|
template.jinja
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% set base_url = "http://127.0.0.1:7860/?" %}
|
| 2 |
+
{% set encoded_urls = [] %}
|
| 3 |
+
|
| 4 |
+
{% for i, url in enumerate(image_urls) %}
|
| 5 |
+
{% set encoded_urls = encoded_urls + [ "img" ~ loop.index ~ "=" ~ url | urlencode ] %}
|
| 6 |
+
{% endfor %}
|
| 7 |
+
|
| 8 |
+
{% set full_url = base_url + encoded_urls | join("&") %}
|
| 9 |
+
|
| 10 |
+
# Image Gallery
|
| 11 |
+
|
| 12 |
+
<iframe src="{{ full_url }}" width="1000" height="1000" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
|