Merge remote-tracking branch 'hf/main'

.agents/skills/extract-image-data/SKILL.md

@@ -0,0 +1,261 @@
+---
+name: extract-image-data
+description: Extract structured data from document images through the electblake/image-data-extractor Gradio Space API. Use when calling its template selection, template generation, or image extraction endpoints from Python, JavaScript, or cURL.
+---
+
+# Extract Image Data (API)
+
+This skill describes how to use the electblake/image-data-extractor Gradio Space programmatically.
+
+## API Endpoints
+
+### `/select_structured_json_template`
+
+**Parameters:**
+
+- `key` [Dropdown]: `Literal['Basic receipt', 'Advanced receipt', 'Basic bank statement', 'Advanced bank statement', 'Invoice with line items', 'Advanced global record']`, default: `Invoice with line items`
+
+**Returns:**
+
+- `Structured JSON template` [Textbox]: `str`
+
+**Python:**
+
+```python
+from gradio_client import Client
+
+client = Client("electblake/image-data-extractor")
+result = client.predict(
+	key="Invoice with line items",
+	api_name="/select_structured_json_template",
+)
+print(result)
+```
+
+**JavaScript:**
+
+```javascript
+import { Client } from "@gradio/client";
+
+const client = await Client.connect("electblake/image-data-extractor");
+const result = await client.predict("/select_structured_json_template", {
+		key: "Invoice with line items",
+});
+
+console.log(result.data);
+```
+
+**cURL:**
+
+```bash
+curl -X POST https://electblake-image-data-extractor.hf.space/call/v2/select_structured_json_template -s -H "Content-Type: application/json" \
+  -d '{"key": "Invoice with line items"}' \
+  | awk -F'"' '{ print $4}' \
+  | read EVENT_ID; curl -N https://electblake-image-data-extractor.hf.space/call/select_structured_json_template/$EVENT_ID
+```
+
+### `/generate_template`
+
+**Parameters:**
+
+- `image` [Image]: `filepath` (required)
+
+**Returns:**
+
+- `Structured JSON template` [Textbox]: `str`
+
+**Python:**
+
+```python
+from gradio_client import Client, handle_file
+
+client = Client("electblake/image-data-extractor")
+result = client.predict(
+	image=handle_file('https://raw.githubusercontent.com/gradio-app/gradio/main/test/test_files/bus.png'),
+	api_name="/generate_template",
+)
+print(result)
+```
+
+**JavaScript:**
+
+```javascript
+import { Client } from "@gradio/client";
+
+const response_0 = await fetch("https://raw.githubusercontent.com/gradio-app/gradio/main/test/test_files/bus.png");
+const exampleImage = await response_0.blob();
+
+const client = await Client.connect("electblake/image-data-extractor");
+const result = await client.predict("/generate_template", {
+		image: exampleImage,
+});
+
+console.log(result.data);
+```
+
+**cURL:**
+
+```bash
+FILE_PATH=$(curl -s -X POST https://electblake-image-data-extractor.hf.space/upload -F 'files=@/path/to/your/file' | tr -d '[]" ')
+
+curl -X POST https://electblake-image-data-extractor.hf.space/call/v2/generate_template -s -H "Content-Type: application/json" \
+  -d '{"image": {"path": "'$FILE_PATH'", "meta": {"_type": "gradio.FileData"}}}' \
+  | awk -F'"' '{ print $4}' \
+  | read EVENT_ID; curl -N https://electblake-image-data-extractor.hf.space/call/generate_template/$EVENT_ID
+```
+
+### `/extract`
+
+**Parameters:**
+
+- `image` [Image]: `filepath` (required)
+- `text` [Textbox]: `str` (required)
+- `template` [Textbox]: `str`, default: `{
+  "document_type": "verbatim-string",
+  "vendor": {
+    "name": "verbatim-string",
+    "address": "verbatim-string",
+    "email": "email-address",
+    "phone": "phone-number"
+  },
+  "invoice_number": "verbatim-string",
+  "invoice_date": "verbatim-string",
+  "line_items": [
+    {
+      "description": "verbatim-string",
+      "category": ["product", "service", "shipping", "tax", "other"],
+      "quantity": "integer",
+      "unit_price": "number",
+      "unit": "unit-code",
+      "amount": "number"
+    }
+  ],
+  "subtotal": "number",
+  "tax": "number",
+  "total": "number",
+  "currency": "currency"
+}`
+- `enable_thinking` [Checkbox]: `bool`, default: `False`
+
+**Returns:**
+
+- `Structured JSON` [Json]: `str | float | bool | list | dict`
+
+**Python:**
+
+```python
+from gradio_client import Client, handle_file
+
+client = Client("electblake/image-data-extractor")
+result = client.predict(
+	image=handle_file('https://raw.githubusercontent.com/gradio-app/gradio/main/test/test_files/bus.png'),
+	text="Hello!!",
+	template="{
+  "document_type": "verbatim-string",
+  "vendor": {
+    "name": "verbatim-string",
+    "address": "verbatim-string",
+    "email": "email-address",
+    "phone": "phone-number"
+  },
+  "invoice_number": "verbatim-string",
+  "invoice_date": "verbatim-string",
+  "line_items": [
+    {
+      "description": "verbatim-string",
+      "category": ["product", "service", "shipping", "tax", "other"],
+      "quantity": "integer",
+      "unit_price": "number",
+      "unit": "unit-code",
+      "amount": "number"
+    }
+  ],
+  "subtotal": "number",
+  "tax": "number",
+  "total": "number",
+  "currency": "currency"
+}",
+	enable_thinking=False,
+	api_name="/extract",
+)
+print(result)
+```
+
+**JavaScript:**
+
+```javascript
+import { Client } from "@gradio/client";
+
+const response_0 = await fetch("https://raw.githubusercontent.com/gradio-app/gradio/main/test/test_files/bus.png");
+const exampleImage = await response_0.blob();
+
+const client = await Client.connect("electblake/image-data-extractor");
+const result = await client.predict("/extract", {
+		image: exampleImage,
+		text: "Hello!!",
+		template: "{
+  "document_type": "verbatim-string",
+  "vendor": {
+    "name": "verbatim-string",
+    "address": "verbatim-string",
+    "email": "email-address",
+    "phone": "phone-number"
+  },
+  "invoice_number": "verbatim-string",
+  "invoice_date": "verbatim-string",
+  "line_items": [
+    {
+      "description": "verbatim-string",
+      "category": ["product", "service", "shipping", "tax", "other"],
+      "quantity": "integer",
+      "unit_price": "number",
+      "unit": "unit-code",
+      "amount": "number"
+    }
+  ],
+  "subtotal": "number",
+  "tax": "number",
+  "total": "number",
+  "currency": "currency"
+}",
+		enable_thinking: false,
+});
+
+console.log(result.data);
+```
+
+**cURL:**
+
+```bash
+FILE_PATH=$(curl -s -X POST https://electblake-image-data-extractor.hf.space/upload -F 'files=@/path/to/your/file' | tr -d '[]" ')
+
+curl -X POST https://electblake-image-data-extractor.hf.space/call/v2/extract -s -H "Content-Type: application/json" \
+  -d '{"image": {"path": "'$FILE_PATH'", "meta": {"_type": "gradio.FileData"}}, "text": "Hello!!", "template": "{
+  "document_type": "verbatim-string",
+  "vendor": {
+    "name": "verbatim-string",
+    "address": "verbatim-string",
+    "email": "email-address",
+    "phone": "phone-number"
+  },
+  "invoice_number": "verbatim-string",
+  "invoice_date": "verbatim-string",
+  "line_items": [
+    {
+      "description": "verbatim-string",
+      "category": ["product", "service", "shipping", "tax", "other"],
+      "quantity": "integer",
+      "unit_price": "number",
+      "unit": "unit-code",
+      "amount": "number"
+    }
+  ],
+  "subtotal": "number",
+  "tax": "number",
+  "total": "number",
+  "currency": "currency"
+}", "enable_thinking": false}' \
+  | awk -F'"' '{ print $4}' \
+  | read EVENT_ID; curl -N https://electblake-image-data-extractor.hf.space/call/extract/$EVENT_ID
+```
+

.agents/skills/extract-image-data/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Extract Image Data (API)"
+  short_description: "Extract structured data from document images"
+  default_prompt: "Use $extract-image-data to extract structured data from a document image through the API."

app.py

@@ -1,3 +1,4 @@
+import zipfile
 from pathlib import Path
 
 import gradio as gr
@@ -29,6 +30,21 @@ structured_json_templates = {
 default_structured_json_template = "Invoice with line items"
 
 
+def download_agentskill(include_gradio_skills):
+    with zipfile.ZipFile(
+        "extract-image-data.zip", "w", compression=zipfile.ZIP_DEFLATED
+    ) as archive:
+        for path in Path(".agents/skills/extract-image-data").rglob("*"):
+            archive.write(path, path.relative_to(".agents/skills"))
+        if include_gradio_skills:
+            for skill_folder in (
+                Path(".agents/skills/gradio"),
+                Path(".agents/skills/hf-gradio"),
+            ):
+                for path in skill_folder.rglob("*"):
+                    archive.write(path, path.relative_to(".agents/skills"))
+    return "extract-image-data.zip"
+
 with gr.Blocks(title="Image Data Extractor") as demo:
     gr.Markdown(
         "# Image Data Extractor\n"
@@ -51,7 +67,9 @@ with gr.Blocks(title="Image Data Extractor") as demo:
                 info="Choose a starting structure, then edit it below.",
             )
             generate_template_button = gr.Button("Generate template from Image")
-            with gr.Accordion("Structured JSON template", open=False):
+            with gr.Accordion(
+                "Structured JSON template", open=False
+            ) as template_accordion:
                 template = gr.Textbox(
                     label="Structured JSON template",
                     value=structured_json_templates[default_structured_json_template],
@@ -65,6 +83,20 @@ with gr.Blocks(title="Image Data Extractor") as demo:
                 open=True,
                 show_indices=True,
             )
+            gr.Markdown(
+                "## Install the agent skill\n"
+                "Download the ZIP, extract it, and place its skill folders in your "
+                "project's `.agents/skills/` directory. Agents should read the "
+                "`SKILL.md` and use it whenever a task requires structured extraction "
+                "from an image. Keep **Include Gradio Skills** checked to include the "
+                "supporting `gradio` and `hf-gradio` skills."
+            )
+            include_gradio_skills = gr.Checkbox(
+                label="Include Gradio Skills",
+                value=True,
+            )
+            download_agentskill_button = gr.Button("Download agent skill")
+            download_agentskill_file = gr.File(label="Agent skill ZIP")
 
     gr.Examples(
         examples=[
@@ -93,7 +125,19 @@ with gr.Blocks(title="Image Data Extractor") as demo:
         api_name="select_structured_json_template",
     )
 
+    download_agentskill_button.click(
+        download_agentskill,
+        inputs=include_gradio_skills,
+        outputs=download_agentskill_file,
+        api_name="download_agentskill",
+    )
+
     generate_template_button.click(
+        lambda: gr.Accordion(open=True),
+        outputs=template_accordion,
+        api_visibility="private",
+        show_progress="hidden",
+    ).then(
         generate_template,
         inputs=image,
         outputs=template,