Update app.py
Browse files
app.py
CHANGED
@@ -4,10 +4,8 @@ import matplotlib.pyplot as plt
|
|
4 |
from pathlib import Path
|
5 |
from huggingface_hub import HfApi, Repository
|
6 |
|
7 |
-
|
8 |
-
|
9 |
-
"font.size": 10,
|
10 |
-
})
|
11 |
|
12 |
# Global DataFrame
|
13 |
df = pd.DataFrame()
|
@@ -17,21 +15,19 @@ def upload_csv(file):
|
|
17 |
df = pd.read_csv(file.name)
|
18 |
if "text" not in df.columns or "label" not in df.columns:
|
19 |
return (
|
20 |
-
None, # table
|
21 |
"❌ CSV must contain 'text' and 'label' columns.",
|
22 |
-
gr.update(visible=False), #
|
23 |
-
gr.update(visible=False), #
|
24 |
-
gr.update(visible=False), #
|
25 |
-
gr.update(visible=False), # push accordion
|
26 |
)
|
27 |
df["label"] = df["label"].fillna("")
|
28 |
return (
|
29 |
-
df[["text","label"]],
|
30 |
-
"✅
|
31 |
-
gr.update(visible=True),
|
32 |
-
gr.update(visible=True),
|
33 |
-
gr.update(visible=True),
|
34 |
-
gr.update(visible=True),
|
35 |
)
|
36 |
|
37 |
def save_changes(table):
|
@@ -45,14 +41,13 @@ def download_csv():
|
|
45 |
df.to_csv(path, index=False)
|
46 |
return path
|
47 |
|
48 |
-
def
|
49 |
global df
|
50 |
counts = df["label"].value_counts().sort_values(ascending=False)
|
51 |
labels, values = counts.index.tolist(), counts.values.tolist()
|
52 |
-
|
53 |
fig, (ax_table, ax_bar) = plt.subplots(
|
54 |
ncols=2,
|
55 |
-
gridspec_kw={"width_ratios":
|
56 |
figsize=(8, max(2, len(labels)*0.4)),
|
57 |
tight_layout=True
|
58 |
)
|
@@ -66,11 +61,11 @@ def create_distribution_figure():
|
|
66 |
ax_bar.invert_yaxis(); ax_bar.set_xlabel("Count")
|
67 |
return fig
|
68 |
|
69 |
-
def
|
70 |
-
fig =
|
71 |
-
|
72 |
-
fig.savefig(
|
73 |
-
return fig,
|
74 |
|
75 |
def push_to_hub(repo_name, hf_token):
|
76 |
global df
|
@@ -89,53 +84,67 @@ def push_to_hub(repo_name, hf_token):
|
|
89 |
use_auth_token=hf_token
|
90 |
)
|
91 |
df.to_csv(local_dir/"data.csv", index=False)
|
92 |
-
repo.push_to_hub(commit_message="📑
|
93 |
-
return f"🚀 Pushed to
|
94 |
except Exception as e:
|
95 |
return f"❌ Push failed: {e}"
|
96 |
|
97 |
-
with gr.Blocks(
|
98 |
-
gr.Markdown("## 🏷️ Label It! Text Annotation Tool
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
# Step 1: Upload
|
103 |
with gr.Row():
|
104 |
-
|
105 |
upload_btn = gr.Button("Upload")
|
106 |
-
|
107 |
-
#
|
108 |
table = gr.Dataframe(headers=["text","label"], interactive=True, visible=False)
|
109 |
status = gr.Textbox(label="Status", interactive=False)
|
110 |
-
|
111 |
-
# Step 2
|
112 |
with gr.Row(visible=False) as action_row:
|
113 |
-
save_btn
|
114 |
-
download_btn
|
115 |
-
visualize_btn= gr.Button("📊 Visualize
|
116 |
-
|
|
|
|
|
|
|
117 |
chart_plot = gr.Plot(label="Label Distribution")
|
118 |
-
download_chart_out = gr.File(label="📥
|
119 |
-
|
120 |
-
# Push
|
121 |
-
|
122 |
-
with
|
123 |
-
repo_in
|
124 |
-
token_in
|
125 |
-
push_btn
|
126 |
push_status = gr.Textbox(label="Push Status", interactive=False)
|
127 |
-
|
128 |
-
#
|
129 |
upload_btn.click(
|
130 |
upload_csv,
|
131 |
-
inputs=
|
132 |
-
outputs=[table, status,
|
133 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
134 |
)
|
135 |
-
save_btn.click(save_changes, inputs=table, outputs=status)
|
136 |
-
download_btn.click(download_csv, outputs=download_csv_out)
|
137 |
-
visualize_btn.click(visualize_and_download_chart,
|
138 |
-
outputs=[chart_plot, download_chart_out])
|
139 |
-
push_btn.click(push_to_hub, inputs=[repo_in, token_in], outputs=push_status)
|
140 |
|
141 |
app.launch()
|
|
|
4 |
from pathlib import Path
|
5 |
from huggingface_hub import HfApi, Repository
|
6 |
|
7 |
+
# Matplotlib style
|
8 |
+
plt.rcParams.update({"font.family":"sans-serif", "font.size":10})
|
|
|
|
|
9 |
|
10 |
# Global DataFrame
|
11 |
df = pd.DataFrame()
|
|
|
15 |
df = pd.read_csv(file.name)
|
16 |
if "text" not in df.columns or "label" not in df.columns:
|
17 |
return (
|
18 |
+
None, # hide table
|
19 |
"❌ CSV must contain 'text' and 'label' columns.",
|
20 |
+
gr.update(visible=False), # hide actions
|
21 |
+
gr.update(visible=False), # hide outputs
|
22 |
+
gr.update(visible=False), # hide push accordion
|
|
|
23 |
)
|
24 |
df["label"] = df["label"].fillna("")
|
25 |
return (
|
26 |
+
df[["text","label"]], # show table
|
27 |
+
"✅ Uploaded! Edit below or use the buttons.",
|
28 |
+
gr.update(visible=True), # show action row
|
29 |
+
gr.update(visible=True), # show output row
|
30 |
+
gr.update(visible=True), # show push accordion
|
|
|
31 |
)
|
32 |
|
33 |
def save_changes(table):
|
|
|
41 |
df.to_csv(path, index=False)
|
42 |
return path
|
43 |
|
44 |
+
def make_figure():
|
45 |
global df
|
46 |
counts = df["label"].value_counts().sort_values(ascending=False)
|
47 |
labels, values = counts.index.tolist(), counts.values.tolist()
|
|
|
48 |
fig, (ax_table, ax_bar) = plt.subplots(
|
49 |
ncols=2,
|
50 |
+
gridspec_kw={"width_ratios":[1,2]},
|
51 |
figsize=(8, max(2, len(labels)*0.4)),
|
52 |
tight_layout=True
|
53 |
)
|
|
|
61 |
ax_bar.invert_yaxis(); ax_bar.set_xlabel("Count")
|
62 |
return fig
|
63 |
|
64 |
+
def visualize_and_download():
|
65 |
+
fig = make_figure()
|
66 |
+
png_path = "label_distribution.png"
|
67 |
+
fig.savefig(png_path, dpi=150, bbox_inches="tight")
|
68 |
+
return fig, png_path
|
69 |
|
70 |
def push_to_hub(repo_name, hf_token):
|
71 |
global df
|
|
|
84 |
use_auth_token=hf_token
|
85 |
)
|
86 |
df.to_csv(local_dir/"data.csv", index=False)
|
87 |
+
repo.push_to_hub(commit_message="📑 Updated data")
|
88 |
+
return f"🚀 Pushed to datasets/{repo_name}"
|
89 |
except Exception as e:
|
90 |
return f"❌ Push failed: {e}"
|
91 |
|
92 |
+
with gr.Blocks() as app:
|
93 |
+
gr.Markdown("## 🏷️ Label It! Text Annotation Tool")
|
94 |
+
gr.Markdown("Upload a `.csv` (columns: **text**, **label**), then annotate, export, visualize, or push.")
|
95 |
+
|
|
|
96 |
# Step 1: Upload
|
97 |
with gr.Row():
|
98 |
+
csv_input = gr.File(label="📁 Upload CSV", file_types=[".csv"])
|
99 |
upload_btn = gr.Button("Upload")
|
100 |
+
|
101 |
+
# Table + Status
|
102 |
table = gr.Dataframe(headers=["text","label"], interactive=True, visible=False)
|
103 |
status = gr.Textbox(label="Status", interactive=False)
|
104 |
+
|
105 |
+
# Step 2: Actions (hidden initially)
|
106 |
with gr.Row(visible=False) as action_row:
|
107 |
+
save_btn = gr.Button("💾 Save")
|
108 |
+
download_btn = gr.Button("⬇️ Download CSV")
|
109 |
+
visualize_btn = gr.Button("📊 Visualize")
|
110 |
+
|
111 |
+
# Outputs row (hidden initially)
|
112 |
+
with gr.Row(visible=False) as output_row:
|
113 |
+
download_csv_out = gr.File(label="📥 CSV File")
|
114 |
chart_plot = gr.Plot(label="Label Distribution")
|
115 |
+
download_chart_out = gr.File(label="📥 Chart PNG")
|
116 |
+
|
117 |
+
# Push section
|
118 |
+
push_accordion = gr.Accordion("📦 Push to Hugging Face Hub", open=False, visible=False)
|
119 |
+
with push_accordion:
|
120 |
+
repo_in = gr.Textbox(label="Repo (username/dataset)")
|
121 |
+
token_in = gr.Textbox(label="🔑 HF Token", type="password")
|
122 |
+
push_btn = gr.Button("🚀 Push")
|
123 |
push_status = gr.Textbox(label="Push Status", interactive=False)
|
124 |
+
|
125 |
+
# Bind events
|
126 |
upload_btn.click(
|
127 |
upload_csv,
|
128 |
+
inputs=csv_input,
|
129 |
+
outputs=[table, status, action_row, output_row, push_accordion]
|
130 |
+
)
|
131 |
+
save_btn.click(
|
132 |
+
save_changes,
|
133 |
+
inputs=table,
|
134 |
+
outputs=status
|
135 |
+
)
|
136 |
+
download_btn.click(
|
137 |
+
download_csv,
|
138 |
+
outputs=download_csv_out
|
139 |
+
)
|
140 |
+
visualize_btn.click(
|
141 |
+
visualize_and_download,
|
142 |
+
outputs=[chart_plot, download_chart_out]
|
143 |
+
)
|
144 |
+
push_btn.click(
|
145 |
+
push_to_hub,
|
146 |
+
inputs=[repo_in, token_in],
|
147 |
+
outputs=push_status
|
148 |
)
|
|
|
|
|
|
|
|
|
|
|
149 |
|
150 |
app.launch()
|