| """Demo dataset generation for trying the explorer locally.""" |
|
|
| from __future__ import annotations |
|
|
| import json |
| from pathlib import Path |
|
|
| from PIL import Image, ImageDraw |
|
|
|
|
| def create_demo_dataset(output_dir: Path) -> Path: |
| """Create a tiny MathVision-like dataset with simple generated images.""" |
|
|
| image_dir = output_dir / "images" |
| image_dir.mkdir(parents=True, exist_ok=True) |
|
|
| records = [ |
| { |
| "id": "demo-red-squares", |
| "question": "How many red squares are visible?", |
| "answer": "4", |
| "image": "images/red-squares.png", |
| "options": ["3", "4", "5"], |
| "subject": "counting", |
| "level": 1, |
| "solution": "Count the four red square tiles.", |
| }, |
| { |
| "id": "demo-red-squares-small", |
| "question": "How many small red squares are visible?", |
| "answer": "5", |
| "image": "images/red-squares-small.png", |
| "options": ["4", "5", "6"], |
| "subject": "counting", |
| "level": 2, |
| "solution": "The red tiles form a group of five.", |
| }, |
| { |
| "id": "demo-blue-triangles", |
| "question": "How many blue triangles are visible?", |
| "answer": "3", |
| "image": "images/blue-triangles.png", |
| "options": ["2", "3", "4"], |
| "subject": "geometry", |
| "level": 1, |
| "solution": "There are three separate blue triangles.", |
| }, |
| { |
| "id": "demo-blue-pyramids", |
| "question": "Which shape appears repeatedly?", |
| "answer": "triangle", |
| "image": "images/blue-pyramids.png", |
| "options": ["circle", "triangle", "square"], |
| "subject": "geometry", |
| "level": 2, |
| "solution": "The repeated blue shapes are triangles.", |
| }, |
| { |
| "id": "demo-red-grid", |
| "question": "Which tile color dominates the grid?", |
| "answer": "red", |
| "image": "images/red-grid.png", |
| "options": ["red", "blue", "green"], |
| "subject": "pattern", |
| "level": 2, |
| "solution": "Most grid cells are red.", |
| }, |
| { |
| "id": "demo-green-grid", |
| "question": "Which tile color dominates this grid?", |
| "answer": "green", |
| "image": "images/green-grid.png", |
| "options": ["red", "blue", "green"], |
| "subject": "pattern", |
| "level": 2, |
| "solution": "Green appears in most grid cells.", |
| }, |
| { |
| "id": "demo-number-line", |
| "question": "Which point is closest to 4?", |
| "answer": "C", |
| "image": "images/number-line.png", |
| "options": ["A", "B", "C"], |
| "subject": "algebra", |
| "level": 1, |
| "solution": "Point C is drawn nearest to the tick labeled 4.", |
| }, |
| { |
| "id": "demo-clock", |
| "question": "Which hour does the short hand point to?", |
| "answer": "3", |
| "image": "images/clock.png", |
| "options": ["2", "3", "4"], |
| "subject": "measurement", |
| "level": 1, |
| "solution": "The shorter hand points toward 3.", |
| }, |
| ] |
|
|
| _draw_red_squares(image_dir / "red-squares.png") |
| _draw_red_squares_small(image_dir / "red-squares-small.png") |
| _draw_blue_triangles(image_dir / "blue-triangles.png") |
| _draw_blue_pyramids(image_dir / "blue-pyramids.png") |
| _draw_red_grid(image_dir / "red-grid.png") |
| _draw_green_grid(image_dir / "green-grid.png") |
| _draw_number_line(image_dir / "number-line.png") |
| _draw_clock(image_dir / "clock.png") |
|
|
| jsonl_path = output_dir / "demo.jsonl" |
| with jsonl_path.open("w", encoding="utf-8") as jsonl_file: |
| for record in records: |
| jsonl_file.write(json.dumps(record, sort_keys=True)) |
| jsonl_file.write("\n") |
| return jsonl_path |
|
|
|
|
| def _new_canvas() -> Image.Image: |
| return Image.new("RGB", (420, 280), color=(248, 250, 252)) |
|
|
|
|
| def _draw_red_squares(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| for x, y in [(80, 60), (170, 60), (80, 150), (170, 150)]: |
| draw.rectangle((x, y, x + 58, y + 58), fill=(220, 38, 38), outline=(127, 29, 29), width=3) |
| image.save(path) |
|
|
|
|
| def _draw_blue_triangles(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| triangles = [ |
| [(90, 190), (130, 80), (170, 190)], |
| [(190, 190), (230, 80), (270, 190)], |
| [(290, 190), (330, 80), (370, 190)], |
| ] |
| for triangle in triangles: |
| draw.polygon(triangle, fill=(37, 99, 235), outline=(30, 64, 175)) |
| image.save(path) |
|
|
|
|
| def _draw_red_squares_small(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| for x, y in [(78, 54), (148, 54), (218, 54), (112, 134), (184, 134)]: |
| draw.rectangle((x, y, x + 46, y + 46), fill=(239, 68, 68), outline=(127, 29, 29), width=3) |
| image.save(path) |
|
|
|
|
| def _draw_blue_pyramids(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| for x, y, size in [(82, 178, 52), (162, 178, 68), (262, 178, 82)]: |
| draw.polygon( |
| [(x, y), (x + size // 2, y - size), (x + size, y)], |
| fill=(59, 130, 246), |
| outline=(30, 64, 175), |
| ) |
| image.save(path) |
|
|
|
|
| def _draw_red_grid(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| colors = [ |
| (220, 38, 38), |
| (220, 38, 38), |
| (22, 163, 74), |
| (220, 38, 38), |
| (37, 99, 235), |
| (220, 38, 38), |
| ] |
| for index, color in enumerate(colors): |
| row, column = divmod(index, 3) |
| x = 92 + column * 82 |
| y = 64 + row * 82 |
| draw.rectangle((x, y, x + 64, y + 64), fill=color, outline=(15, 23, 42), width=2) |
| image.save(path) |
|
|
|
|
| def _draw_green_grid(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| colors = [ |
| (22, 163, 74), |
| (22, 163, 74), |
| (220, 38, 38), |
| (22, 163, 74), |
| (37, 99, 235), |
| (22, 163, 74), |
| ] |
| for index, color in enumerate(colors): |
| row, column = divmod(index, 3) |
| x = 92 + column * 82 |
| y = 64 + row * 82 |
| draw.rectangle((x, y, x + 64, y + 64), fill=color, outline=(15, 23, 42), width=2) |
| image.save(path) |
|
|
|
|
| def _draw_number_line(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| draw.line((62, 148, 358, 148), fill=(15, 23, 42), width=4) |
| for index in range(6): |
| x = 62 + index * 59 |
| draw.line((x, 134, x, 162), fill=(15, 23, 42), width=3) |
| draw.text((x - 5, 170), str(index), fill=(15, 23, 42)) |
| points = [ |
| ("A", 174, (37, 99, 235)), |
| ("B", 246, (22, 163, 74)), |
| ("C", 296, (220, 38, 38)), |
| ] |
| for label, x, color in points: |
| draw.ellipse((x - 9, 108, x + 9, 126), fill=color) |
| draw.text((x - 5, 86), label, fill=(15, 23, 42)) |
| image.save(path) |
|
|
|
|
| def _draw_clock(path: Path) -> None: |
| image = _new_canvas() |
| draw = ImageDraw.Draw(image) |
| center = (210, 140) |
| draw.ellipse((100, 30, 320, 250), fill=(255, 255, 255), outline=(15, 23, 42), width=4) |
| for label, xy in [("12", (199, 48)), ("3", (290, 132)), ("6", (205, 220)), ("9", (120, 132))]: |
| draw.text(xy, label, fill=(15, 23, 42)) |
| draw.line((center[0], center[1], 282, 140), fill=(220, 38, 38), width=6) |
| draw.line((center[0], center[1], 210, 68), fill=(37, 99, 235), width=4) |
| draw.ellipse((202, 132, 218, 148), fill=(15, 23, 42)) |
| image.save(path) |
|
|
