Compare commits
1 Commits
claude/iss
...
fix/868
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
0a814f5bef |
@@ -1,194 +1,354 @@
|
||||
[
|
||||
{
|
||||
"id": "screenshot_github_home",
|
||||
"url": "https://github.githubassets.com/images/modules/logos_page/GitHub-Mark.png",
|
||||
"url": "test_images/screenshot_github_home.png",
|
||||
"category": "screenshot",
|
||||
"expected_keywords": ["github", "logo", "mark"],
|
||||
"expected_keywords": [
|
||||
"github",
|
||||
"logo",
|
||||
"mark"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "diagram_mermaid_flow",
|
||||
"url": "https://mermaid.ink/img/pako:eNpdkE9PwzAMxb-K5VOl7gc7sAOIIDuAw9gptnRaSJLSJttQStmXs9LCH-ymBOI1ef_42U6cUSae4IkDxbAAWtB6siSZXVhjQTlgl1nigHg5fRBOzSfebopROCu_cytObSfgLSE1ANOeZWkO2IH5upZxYot8m1hqAdpD_63WRl0xdUG1jdl9kPiOb_EWk2JBtPaiKkF4eVIYgO0EtkW-RSgC4gJ6HJYRG1UNdN0HNVd0Bftjj7X8P92qPj-F8l8T3w",
|
||||
"url": "test_images/diagram_mermaid_flow.png",
|
||||
"category": "diagram",
|
||||
"expected_keywords": ["flow", "diagram", "process"],
|
||||
"expected_keywords": [
|
||||
"flow",
|
||||
"diagram",
|
||||
"process"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_1",
|
||||
"url": "https://picsum.photos/seed/vision1/400/300",
|
||||
"url": "test_images/photo_random_1.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_2",
|
||||
"url": "https://picsum.photos/seed/vision2/400/300",
|
||||
"url": "test_images/photo_random_2.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "chart_simple_bar",
|
||||
"url": "https://quickchart.io/chart?c={type:'bar',data:{labels:['Q1','Q2','Q3','Q4'],datasets:[{label:'Revenue',data:[100,150,200,250]}]}}",
|
||||
"url": "test_images/chart_simple_bar.png",
|
||||
"category": "chart",
|
||||
"expected_keywords": ["bar", "chart", "revenue"],
|
||||
"expected_keywords": [
|
||||
"bar",
|
||||
"chart",
|
||||
"revenue"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "chart_pie",
|
||||
"url": "https://quickchart.io/chart?c={type:'pie',data:{labels:['A','B','C'],datasets:[{data:[30,50,20]}]}}",
|
||||
"url": "test_images/chart_pie.png",
|
||||
"category": "chart",
|
||||
"expected_keywords": ["pie", "chart", "percentage"],
|
||||
"expected_keywords": [
|
||||
"pie",
|
||||
"chart",
|
||||
"percentage"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "diagram_org_chart",
|
||||
"url": "https://mermaid.ink/img/pako:eNpdkE9PwzAMxb-K5VOl7gc7sAOIIDuAw9gptnRaSJLSJttQStmXs9LCH-ymBOI1ef_42U6cUSae4IkDxbAAWtB6iuyIWyrLgXLALrPEAfFy-iCcmk-83RSjcFZ-51ac2k7AW0JqAKY9y9IcsAPzdS3jxBb5NrHUAraH_lutjbpi6oJqG7P7IPEd3-ItJsWCaO1FVYLw8qQwANsJbIt8i1AExAX0OCwjNqoa6LoPaq7oCvbHHmv5f7pVfX4K5b8mvg",
|
||||
"url": "test_images/diagram_org_chart.png",
|
||||
"category": "diagram",
|
||||
"expected_keywords": ["organization", "hierarchy", "chart"],
|
||||
"expected_keywords": [
|
||||
"organization",
|
||||
"hierarchy",
|
||||
"chart"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "screenshot_terminal",
|
||||
"url": "https://raw.githubusercontent.com/nicehash/nicehash-quick-start/main/images/nicehash-terminal.png",
|
||||
"url": "test_images/screenshot_terminal.png",
|
||||
"category": "screenshot",
|
||||
"expected_keywords": ["terminal", "command", "output"],
|
||||
"expected_keywords": [
|
||||
"terminal",
|
||||
"command",
|
||||
"output"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_3",
|
||||
"url": "https://picsum.photos/seed/vision3/400/300",
|
||||
"url": "test_images/photo_random_3.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "chart_line",
|
||||
"url": "https://quickchart.io/chart?c={type:'line',data:{labels:['Jan','Feb','Mar','Apr'],datasets:[{label:'Temperature',data:[5,8,12,18]}]}}",
|
||||
"url": "test_images/chart_line.png",
|
||||
"category": "chart",
|
||||
"expected_keywords": ["line", "chart", "temperature"],
|
||||
"expected_keywords": [
|
||||
"line",
|
||||
"chart",
|
||||
"temperature"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "diagram_sequence",
|
||||
"url": "https://mermaid.ink/img/pako:eNpdkE9PwzAMxb-K5VOl7gc7sAOIIDuAw9gptnRaSJLSJttQStmXs9LCH-ymBOI1ef_42U6cUSae4IkDxbAAWtB6iuyIWyrLgXLALrPEAfFy-iCcmk-83RSjcFZ-51ac2k7AW0JqAKY9y9IcsAPzdS3jxBb5NrHUAraH_lutjbpi6oJqG7P7IPEd3-ItJsWCaO1FVYLw8qQwANsJbIt8i1AExAX0OCwjNqoa6LoPaq7oCvbHHmv5f7pVfX4K5b8mvg",
|
||||
"url": "test_images/diagram_sequence.png",
|
||||
"category": "diagram",
|
||||
"expected_keywords": ["sequence", "interaction", "message"],
|
||||
"expected_keywords": [
|
||||
"sequence",
|
||||
"interaction",
|
||||
"message"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_4",
|
||||
"url": "https://picsum.photos/seed/vision4/400/300",
|
||||
"url": "test_images/photo_random_4.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "screenshot_webpage",
|
||||
"url": "https://github.githubassets.com/images/modules/site/social-cards.png",
|
||||
"url": "test_images/screenshot_webpage.png",
|
||||
"category": "screenshot",
|
||||
"expected_keywords": ["github", "page", "web"],
|
||||
"expected_keywords": [
|
||||
"github",
|
||||
"page",
|
||||
"web"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "chart_radar",
|
||||
"url": "https://quickchart.io/chart?c={type:'radar',data:{labels:['Speed','Power','Defense','Magic'],datasets:[{label:'Hero',data:[80,60,70,90]}]}}",
|
||||
"url": "test_images/chart_radar.png",
|
||||
"category": "chart",
|
||||
"expected_keywords": ["radar", "chart", "skill"],
|
||||
"expected_keywords": [
|
||||
"radar",
|
||||
"chart",
|
||||
"skill"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_5",
|
||||
"url": "https://picsum.photos/seed/vision5/400/300",
|
||||
"url": "test_images/photo_random_5.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "diagram_class",
|
||||
"url": "https://mermaid.ink/img/pako:eNpdkE9PwzAMxb-K5VOl7gc7sAOIIDuAw9gptnRaSJLSJttQStmXs9LCH-ymBOI1ef_42U6cUSae4IkDxbAAWtB6iuyIWyrLgXLALrPEAfFy-iCcmk-83RSjcFZ-51ac2k7AW0JqAKY9y9IcsAPzdS3jxBb5NrHUAraH_lutjbpi6oJqG7P7IPEd3-ItJsWCaO1FVYLw8qQwANsJbIt8i1AExAX0OCwjNqoa6LoPaq7oCvbHHmv5f7pVfX4K5b8mvg",
|
||||
"url": "test_images/diagram_class.png",
|
||||
"category": "diagram",
|
||||
"expected_keywords": ["class", "object", "attribute"],
|
||||
"expected_keywords": [
|
||||
"class",
|
||||
"object",
|
||||
"attribute"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "chart_doughnut",
|
||||
"url": "https://quickchart.io/chart?c={type:'doughnut',data:{labels:['Desktop','Mobile','Tablet'],datasets:[{data:[60,30,10]}]}}",
|
||||
"url": "test_images/chart_doughnut.png",
|
||||
"category": "chart",
|
||||
"expected_keywords": ["doughnut", "chart", "device"],
|
||||
"expected_keywords": [
|
||||
"doughnut",
|
||||
"chart",
|
||||
"device"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_6",
|
||||
"url": "https://picsum.photos/seed/vision6/400/300",
|
||||
"url": "test_images/photo_random_6.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "screenshot_error",
|
||||
"url": "https://http.cat/404.jpg",
|
||||
"url": "test_images/screenshot_error.png",
|
||||
"category": "screenshot",
|
||||
"expected_keywords": ["404", "error", "cat"],
|
||||
"expected_keywords": [
|
||||
"404",
|
||||
"error",
|
||||
"cat"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "diagram_network",
|
||||
"url": "https://mermaid.ink/img/pako:eNpdkE9PwzAMxb-K5VOl7gc7sAOIIDuAw9gptnRaSJLSJttQStmXs9LCH-ymBOI1ef_42U6cUSae4IkDxbAAWtB6iuyIWyrLgXLALrPEAfFy-iCcmk-83RSjcFZ-51ac2k7AW0JqAKY9y9IcsAPzdS3jxBb5NrHUAraH_lutjbpi6oJqG7P7IPEd3-ItJsWCaO1FVYLw8qQwANsJbIt8i1AExAX0OCwjNqoa6LoPaq7oCvbHHmv5f7pVfX4K5b8mvg",
|
||||
"url": "test_images/diagram_network.png",
|
||||
"category": "diagram",
|
||||
"expected_keywords": ["network", "node", "connection"],
|
||||
"expected_keywords": [
|
||||
"network",
|
||||
"node",
|
||||
"connection"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_7",
|
||||
"url": "https://picsum.photos/seed/vision7/400/300",
|
||||
"url": "test_images/photo_random_7.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "chart_stacked_bar",
|
||||
"url": "https://quickchart.io/chart?c={type:'bar',data:{labels:['2022','2023','2024'],datasets:[{label:'Cloud',data:[100,150,200]},{label:'On-prem',data:[200,180,160]}]},options:{scales:{x:{stacked:true},y:{stacked:true}}}}",
|
||||
"url": "test_images/chart_stacked_bar.png",
|
||||
"category": "chart",
|
||||
"expected_keywords": ["stacked", "bar", "chart"],
|
||||
"expected_keywords": [
|
||||
"stacked",
|
||||
"bar",
|
||||
"chart"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 50, "min_sentences": 2, "has_numbers": true}
|
||||
"expected_structure": {
|
||||
"min_length": 50,
|
||||
"min_sentences": 2,
|
||||
"has_numbers": true
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "screenshot_dashboard",
|
||||
"url": "https://github.githubassets.com/images/modules/site/features-code-search.png",
|
||||
"url": "test_images/screenshot_dashboard.png",
|
||||
"category": "screenshot",
|
||||
"expected_keywords": ["search", "code", "feature"],
|
||||
"expected_keywords": [
|
||||
"search",
|
||||
"code",
|
||||
"feature"
|
||||
],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "photo_random_8",
|
||||
"url": "https://picsum.photos/seed/vision8/400/300",
|
||||
"url": "test_images/photo_random_8.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"ground_truth_ocr": "",
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1, "has_numbers": false}
|
||||
"expected_structure": {
|
||||
"min_length": 30,
|
||||
"min_sentences": 1,
|
||||
"has_numbers": false
|
||||
}
|
||||
}
|
||||
]
|
||||
|
||||
BIN
benchmarks/test_images/chart_doughnut.png
Normal file
|
After Width: | Height: | Size: 4.4 KiB |
BIN
benchmarks/test_images/chart_line.png
Normal file
|
After Width: | Height: | Size: 4.1 KiB |
BIN
benchmarks/test_images/chart_pie.png
Normal file
|
After Width: | Height: | Size: 4.0 KiB |
BIN
benchmarks/test_images/chart_radar.png
Normal file
|
After Width: | Height: | Size: 3.5 KiB |
BIN
benchmarks/test_images/chart_simple_bar.png
Normal file
|
After Width: | Height: | Size: 4.2 KiB |
BIN
benchmarks/test_images/chart_stacked_bar.png
Normal file
|
After Width: | Height: | Size: 5.0 KiB |
BIN
benchmarks/test_images/diagram_class.png
Normal file
|
After Width: | Height: | Size: 4.6 KiB |
BIN
benchmarks/test_images/diagram_mermaid_flow.png
Normal file
|
After Width: | Height: | Size: 4.8 KiB |
BIN
benchmarks/test_images/diagram_network.png
Normal file
|
After Width: | Height: | Size: 5.0 KiB |
BIN
benchmarks/test_images/diagram_org_chart.png
Normal file
|
After Width: | Height: | Size: 5.1 KiB |
BIN
benchmarks/test_images/diagram_sequence.png
Normal file
|
After Width: | Height: | Size: 5.2 KiB |
BIN
benchmarks/test_images/photo_random_1.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/photo_random_2.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/photo_random_3.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/photo_random_4.png
Normal file
|
After Width: | Height: | Size: 2.9 KiB |
BIN
benchmarks/test_images/photo_random_5.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/photo_random_6.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/photo_random_7.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/photo_random_8.png
Normal file
|
After Width: | Height: | Size: 3.0 KiB |
BIN
benchmarks/test_images/screenshot_dashboard.png
Normal file
|
After Width: | Height: | Size: 7.1 KiB |
BIN
benchmarks/test_images/screenshot_error.png
Normal file
|
After Width: | Height: | Size: 6.2 KiB |
BIN
benchmarks/test_images/screenshot_github_home.png
Normal file
|
After Width: | Height: | Size: 7.1 KiB |
BIN
benchmarks/test_images/screenshot_terminal.png
Normal file
|
After Width: | Height: | Size: 7.1 KiB |
BIN
benchmarks/test_images/screenshot_webpage.png
Normal file
|
After Width: | Height: | Size: 7.2 KiB |
@@ -11,17 +11,19 @@ Usage:
|
||||
|
||||
# Single image test
|
||||
python benchmarks/vision_benchmark.py --url https://example.com/image.png
|
||||
python benchmarks/vision_benchmark.py --url benchmarks/test_images/photo_random_1.png
|
||||
|
||||
# Generate test report
|
||||
python benchmarks/vision_benchmark.py --images benchmarks/test_images.json --output benchmarks/vision_results.json
|
||||
|
||||
Test image dataset: benchmarks/test_images.json (50-100 diverse images)
|
||||
Test image dataset: benchmarks/test_images.json (committed local fixtures under benchmarks/test_images/)
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import asyncio
|
||||
import base64
|
||||
import json
|
||||
import mimetypes
|
||||
import os
|
||||
import statistics
|
||||
import sys
|
||||
@@ -67,6 +69,28 @@ EVAL_PROMPTS = {
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
|
||||
def _is_remote_image_source(image_source: str) -> bool:
|
||||
return image_source.startswith(("http://", "https://", "data:", "file://"))
|
||||
|
||||
|
||||
def _image_source_to_payload_url(image_source: str) -> str:
|
||||
"""Convert local image paths into data URLs; keep remote URLs unchanged."""
|
||||
if image_source.startswith(("http://", "https://", "data:")):
|
||||
return image_source
|
||||
|
||||
resolved = image_source[len("file://"):] if image_source.startswith("file://") else image_source
|
||||
local_path = Path(os.path.expanduser(resolved)).resolve()
|
||||
if not local_path.is_file():
|
||||
return image_source
|
||||
|
||||
mime_type, _ = mimetypes.guess_type(str(local_path))
|
||||
if not mime_type:
|
||||
mime_type = "application/octet-stream"
|
||||
|
||||
encoded = base64.b64encode(local_path.read_bytes()).decode("ascii")
|
||||
return f"data:{mime_type};base64,{encoded}"
|
||||
|
||||
|
||||
async def analyze_with_model(
|
||||
image_url: str,
|
||||
prompt: str,
|
||||
@@ -84,6 +108,8 @@ async def analyze_with_model(
|
||||
"""
|
||||
import httpx
|
||||
|
||||
image_payload_url = _image_source_to_payload_url(image_url)
|
||||
|
||||
provider = model_config["provider"]
|
||||
model_id = model_config["model_id"]
|
||||
|
||||
@@ -93,7 +119,7 @@ async def analyze_with_model(
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "text", "text": prompt},
|
||||
{"type": "image_url", "image_url": {"url": image_url}},
|
||||
{"type": "image_url", "image_url": {"url": image_payload_url}},
|
||||
],
|
||||
}
|
||||
]
|
||||
@@ -570,8 +596,18 @@ def generate_sample_dataset() -> List[dict]:
|
||||
|
||||
def load_dataset(path: str) -> List[dict]:
|
||||
"""Load test dataset from JSON file."""
|
||||
with open(path) as f:
|
||||
return json.load(f)
|
||||
dataset_path = Path(path).resolve()
|
||||
with open(dataset_path) as f:
|
||||
dataset = json.load(f)
|
||||
|
||||
base_dir = dataset_path.parent
|
||||
for image in dataset:
|
||||
image_url = image.get("url")
|
||||
if not image_url or _is_remote_image_source(image_url):
|
||||
continue
|
||||
image["url"] = str((base_dir / image_url).resolve())
|
||||
|
||||
return dataset
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
@@ -582,7 +618,7 @@ def load_dataset(path: str) -> List[dict]:
|
||||
async def main():
|
||||
parser = argparse.ArgumentParser(description="Vision Benchmark Suite (Issue #817)")
|
||||
parser.add_argument("--images", help="Path to test images JSON file")
|
||||
parser.add_argument("--url", help="Single image URL to test")
|
||||
parser.add_argument("--url", help="Single image URL or local file path to test")
|
||||
parser.add_argument("--category", default="photo", help="Category for single URL")
|
||||
parser.add_argument("--output", default=None, help="Output JSON file")
|
||||
parser.add_argument("--runs", type=int, default=1, help="Runs per model per image")
|
||||
|
||||
@@ -211,43 +211,6 @@ class PluginContext:
|
||||
}
|
||||
logger.debug("Plugin %s registered CLI command: %s", self.manifest.name, name)
|
||||
|
||||
# -- memory provider registration ----------------------------------------
|
||||
|
||||
def register_memory_provider(self, provider) -> None:
|
||||
"""Register a memory provider supplied by this plugin.
|
||||
|
||||
The provider must be an instance of ``agent.memory_provider.MemoryProvider``.
|
||||
Only one plugin-registered memory provider is accepted; a second
|
||||
attempt is rejected with a warning.
|
||||
|
||||
The registered provider is retrievable via
|
||||
``get_plugin_memory_provider()`` and is picked up by ``run_agent.py``
|
||||
when ``memory.provider`` in *config.yaml* matches the provider's
|
||||
``name`` property.
|
||||
"""
|
||||
from agent.memory_provider import MemoryProvider
|
||||
|
||||
if not isinstance(provider, MemoryProvider):
|
||||
logger.warning(
|
||||
"Plugin '%s' tried to register a memory provider that does not "
|
||||
"inherit from MemoryProvider. Ignoring.",
|
||||
self.manifest.name,
|
||||
)
|
||||
return
|
||||
if self._manager._plugin_memory_provider is not None:
|
||||
logger.warning(
|
||||
"Plugin '%s' tried to register a memory provider, but one is "
|
||||
"already registered by another plugin. Only one plugin-supplied "
|
||||
"memory provider is allowed at a time.",
|
||||
self.manifest.name,
|
||||
)
|
||||
return
|
||||
self._manager._plugin_memory_provider = provider
|
||||
logger.info(
|
||||
"Plugin '%s' registered memory provider: %s",
|
||||
self.manifest.name, provider.name,
|
||||
)
|
||||
|
||||
# -- context engine registration -----------------------------------------
|
||||
|
||||
def register_context_engine(self, engine) -> None:
|
||||
@@ -360,7 +323,6 @@ class PluginManager:
|
||||
self._plugin_tool_names: Set[str] = set()
|
||||
self._cli_commands: Dict[str, dict] = {}
|
||||
self._context_engine = None # Set by a plugin via register_context_engine()
|
||||
self._plugin_memory_provider = None # Set by a plugin via register_memory_provider()
|
||||
self._discovered: bool = False
|
||||
self._cli_ref = None # Set by CLI after plugin discovery
|
||||
# Plugin skill registry: qualified name → metadata dict.
|
||||
@@ -737,11 +699,6 @@ def get_plugin_context_engine():
|
||||
return get_plugin_manager()._context_engine
|
||||
|
||||
|
||||
def get_plugin_memory_provider():
|
||||
"""Return the plugin-registered memory provider, or None."""
|
||||
return get_plugin_manager()._plugin_memory_provider
|
||||
|
||||
|
||||
def get_plugin_toolsets() -> List[tuple]:
|
||||
"""Return plugin toolsets as ``(key, label, description)`` tuples.
|
||||
|
||||
|
||||
12
run_agent.py
@@ -1193,18 +1193,6 @@ class AIAgent:
|
||||
from plugins.memory import load_memory_provider as _load_mem
|
||||
self._memory_manager = _MemoryManager()
|
||||
_mp = _load_mem(_mem_provider_name)
|
||||
# Fall back to a user plugin that called register_memory_provider()
|
||||
if _mp is None:
|
||||
try:
|
||||
from hermes_cli.plugins import get_plugin_memory_provider as _gpm
|
||||
_candidate = _gpm()
|
||||
if _candidate and _candidate.name == _mem_provider_name:
|
||||
_mp = _candidate
|
||||
except Exception as _gpm_err:
|
||||
logger.debug(
|
||||
"get_plugin_memory_provider() failed during fallback lookup: %s",
|
||||
_gpm_err,
|
||||
)
|
||||
if _mp and _mp.is_available():
|
||||
self._memory_manager.add_provider(_mp)
|
||||
if self._memory_manager.providers:
|
||||
|
||||
@@ -19,7 +19,6 @@ from hermes_cli.plugins import (
|
||||
PluginManifest,
|
||||
get_plugin_manager,
|
||||
get_pre_tool_call_block_message,
|
||||
get_plugin_memory_provider,
|
||||
discover_plugins,
|
||||
invoke_hook,
|
||||
)
|
||||
@@ -610,105 +609,3 @@ class TestPreLlmCallTargetRouting:
|
||||
# in PluginContext (hermes_cli/plugins.py). The tests referenced _plugin_commands,
|
||||
# commands_registered, get_plugin_command_handler, and GATEWAY_KNOWN_COMMANDS
|
||||
# integration — all of which are unimplemented features.
|
||||
|
||||
|
||||
# ── TestRegisterMemoryProvider ─────────────────────────────────────────────
|
||||
|
||||
|
||||
class TestRegisterMemoryProvider:
|
||||
"""Regression tests for PluginContext.register_memory_provider() — issue #990.
|
||||
|
||||
The MemPalace plugin (and any user plugin following the developer guide)
|
||||
calls ``ctx.register_memory_provider(provider)`` inside ``register(ctx)``.
|
||||
Before the fix, PluginContext had no such method and the plugin failed to
|
||||
load with: 'PluginContext' object has no attribute 'register_memory_provider'.
|
||||
"""
|
||||
|
||||
def _make_memory_plugin(self, plugins_dir: "Path", name: str) -> None:
|
||||
"""Write a minimal user plugin that registers a stub MemoryProvider."""
|
||||
from agent.memory_provider import MemoryProvider
|
||||
|
||||
plugin_dir = plugins_dir / name
|
||||
plugin_dir.mkdir(parents=True, exist_ok=True)
|
||||
(plugin_dir / "plugin.yaml").write_text(
|
||||
f"name: {name}\nversion: 0.1.0\ndescription: Stub memory plugin\n"
|
||||
)
|
||||
# The register() body imports and calls register_memory_provider — this
|
||||
# is the exact pattern documented in memory-provider-plugin.md and used
|
||||
# by third-party plugins such as MemPalace.
|
||||
(plugin_dir / "__init__.py").write_text(
|
||||
"from agent.memory_provider import MemoryProvider\n"
|
||||
"\n"
|
||||
"class _StubProvider(MemoryProvider):\n"
|
||||
" @property\n"
|
||||
f" def name(self): return '{name}'\n"
|
||||
" def is_available(self): return True\n"
|
||||
" def initialize(self, session_id, **kw): pass\n"
|
||||
" def get_tool_schemas(self): return []\n"
|
||||
"\n"
|
||||
"def register(ctx):\n"
|
||||
" ctx.register_memory_provider(_StubProvider())\n"
|
||||
)
|
||||
|
||||
def test_register_memory_provider_succeeds(self, tmp_path, monkeypatch):
|
||||
"""A user plugin calling register_memory_provider() loads without error."""
|
||||
plugins_dir = tmp_path / "hermes_test" / "plugins"
|
||||
self._make_memory_plugin(plugins_dir, "mempalace")
|
||||
monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes_test"))
|
||||
|
||||
mgr = PluginManager()
|
||||
mgr.discover_and_load()
|
||||
|
||||
assert "mempalace" in mgr._plugins
|
||||
assert mgr._plugins["mempalace"].enabled, (
|
||||
mgr._plugins["mempalace"].error
|
||||
)
|
||||
|
||||
def test_plugin_memory_provider_stored(self, tmp_path, monkeypatch):
|
||||
"""The provider instance is accessible via get_plugin_memory_provider()."""
|
||||
import hermes_cli.plugins as plugins_mod
|
||||
|
||||
plugins_dir = tmp_path / "hermes_test" / "plugins"
|
||||
self._make_memory_plugin(plugins_dir, "mempalace")
|
||||
monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes_test"))
|
||||
|
||||
mgr = PluginManager()
|
||||
# Swap the singleton so get_plugin_memory_provider() sees our manager
|
||||
monkeypatch.setattr(plugins_mod, "_plugin_manager", mgr)
|
||||
mgr.discover_and_load()
|
||||
|
||||
provider = get_plugin_memory_provider()
|
||||
assert provider is not None
|
||||
assert provider.name == "mempalace"
|
||||
|
||||
def test_second_registration_rejected(self, tmp_path, monkeypatch):
|
||||
"""Only one plugin-registered memory provider is accepted."""
|
||||
plugins_dir = tmp_path / "hermes_test" / "plugins"
|
||||
self._make_memory_plugin(plugins_dir, "first_provider")
|
||||
self._make_memory_plugin(plugins_dir, "second_provider")
|
||||
monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes_test"))
|
||||
|
||||
mgr = PluginManager()
|
||||
mgr.discover_and_load()
|
||||
|
||||
# The manager should hold exactly one provider
|
||||
assert mgr._plugin_memory_provider is not None
|
||||
assert mgr._plugin_memory_provider.name in {"first_provider", "second_provider"}
|
||||
|
||||
def test_non_provider_rejected(self, tmp_path, monkeypatch):
|
||||
"""Passing a non-MemoryProvider object logs a warning and is ignored."""
|
||||
plugins_dir = tmp_path / "hermes_test" / "plugins"
|
||||
plugin_dir = plugins_dir / "bad_provider"
|
||||
plugin_dir.mkdir(parents=True, exist_ok=True)
|
||||
(plugin_dir / "plugin.yaml").write_text("name: bad_provider\n")
|
||||
(plugin_dir / "__init__.py").write_text(
|
||||
"def register(ctx):\n"
|
||||
" ctx.register_memory_provider('not-a-provider')\n"
|
||||
)
|
||||
monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes_test"))
|
||||
|
||||
mgr = PluginManager()
|
||||
mgr.discover_and_load()
|
||||
|
||||
# Plugin still loads (warning only), but no provider is stored
|
||||
assert mgr._plugin_memory_provider is None
|
||||
|
||||
@@ -11,12 +11,14 @@ import pytest
|
||||
sys.path.insert(0, str(Path(__file__).parent.parent / "benchmarks"))
|
||||
|
||||
from vision_benchmark import (
|
||||
analyze_with_model,
|
||||
compute_ocr_accuracy,
|
||||
compute_description_completeness,
|
||||
compute_structural_accuracy,
|
||||
aggregate_results,
|
||||
to_markdown,
|
||||
generate_sample_dataset,
|
||||
load_dataset,
|
||||
MODELS,
|
||||
EVAL_PROMPTS,
|
||||
)
|
||||
@@ -197,6 +199,71 @@ class TestMarkdown:
|
||||
|
||||
|
||||
class TestDataset:
|
||||
def test_repo_dataset_uses_local_image_paths(self):
|
||||
dataset_path = Path(__file__).parent.parent / "benchmarks" / "test_images.json"
|
||||
dataset = json.loads(dataset_path.read_text())
|
||||
|
||||
assert dataset, "benchmark dataset should not be empty"
|
||||
assert all(not entry["url"].startswith(("http://", "https://")) for entry in dataset)
|
||||
|
||||
def test_load_dataset_resolves_relative_local_paths(self, tmp_path):
|
||||
images_dir = tmp_path / "images"
|
||||
images_dir.mkdir()
|
||||
image_path = images_dir / "sample.png"
|
||||
image_path.write_bytes(b"png-bytes")
|
||||
|
||||
dataset_path = tmp_path / "dataset.json"
|
||||
dataset_path.write_text(json.dumps([
|
||||
{
|
||||
"id": "sample",
|
||||
"url": "images/sample.png",
|
||||
"category": "photo",
|
||||
"expected_keywords": [],
|
||||
"expected_structure": {"min_length": 30, "min_sentences": 1},
|
||||
}
|
||||
]))
|
||||
|
||||
loaded = load_dataset(str(dataset_path))
|
||||
|
||||
assert loaded[0]["url"] == str(image_path.resolve())
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_analyze_with_model_encodes_local_file_as_data_url(self, tmp_path, monkeypatch):
|
||||
image_path = tmp_path / "tiny.png"
|
||||
image_path.write_bytes(
|
||||
bytes.fromhex(
|
||||
"89504E470D0A1A0A"
|
||||
"0000000D49484452000000010000000108060000001F15C489"
|
||||
"0000000D49444154789C6360000002000154A24F5D00000000"
|
||||
"49454E44AE426082"
|
||||
)
|
||||
)
|
||||
|
||||
fake_response = MagicMock()
|
||||
fake_response.raise_for_status.return_value = None
|
||||
fake_response.json.return_value = {
|
||||
"choices": [{"message": {"content": "Looks like a tiny image."}}],
|
||||
"usage": {"prompt_tokens": 1, "completion_tokens": 2, "total_tokens": 3},
|
||||
}
|
||||
|
||||
fake_client = MagicMock()
|
||||
fake_client.post = AsyncMock(return_value=fake_response)
|
||||
fake_ctx = MagicMock()
|
||||
fake_ctx.__aenter__ = AsyncMock(return_value=fake_client)
|
||||
fake_ctx.__aexit__ = AsyncMock(return_value=None)
|
||||
|
||||
monkeypatch.setenv("OPENROUTER_API_KEY", "test-key")
|
||||
with patch("httpx.AsyncClient", return_value=fake_ctx):
|
||||
result = await analyze_with_model(
|
||||
str(image_path),
|
||||
"Describe this image",
|
||||
{"provider": "openrouter", "model_id": "fake/model"},
|
||||
)
|
||||
|
||||
assert result["success"] is True
|
||||
sent_url = fake_client.post.await_args.kwargs["json"]["messages"][0]["content"][1]["image_url"]["url"]
|
||||
assert sent_url.startswith("data:image/png;base64,")
|
||||
|
||||
def test_sample_dataset_has_entries(self):
|
||||
dataset = generate_sample_dataset()
|
||||
assert len(dataset) >= 4
|
||||
|
||||