Add full unicode support to Font.metrics

[ankith26]

Inspired by a conversation in #3326

[ankith26]

To test this PR I wrote a script (albeit with some help from gpt) to test font rendering with a wide range of unicode points and tested that metrics works on all of them.

Code used

import pygame

# Initialize Pygame
pygame.init()

FONT_SIZE = 30

# Set up the screen
WIDTH, HEIGHT = 1000, 800
screen = pygame.display.set_mode((WIDTH, HEIGHT))
pygame.display.set_caption("Pygame Font Metrics Demo (Noto Fonts)")

# Define font-character mapping
noto_font_samples = {
    "Noto Sans": ["A", "é", "ñ", "Ω", "Ж"],
    "Noto Serif": ["A", "ß", "Ж", "Σ", "λ"],
    "Noto Sans CJK SC": ["字", "汉", "日本", "韓國", "學"],
    "Noto Sans Arabic": ["العربية", "ب", "ك", "م", "ن"],
    "Noto Sans Devanagari": ["नमस्ते", "श", "क", "त्र", "ध"],
    "Noto Music": ["𝄞", "𝄢", "𝅘", "𝅘𝅥", "𝆑"],
    "Noto Color Emoji": ["🤖", "🔥", "💡", "🎵", "🚀"],
    "Noto Sans Math": ["∑", "∞", "∂", "∫", "∇"],
    "Noto Sans Old Persian": ["𐎠", "𐎢", "𐎴", "𐎽", "𐎹"],
    "Noto Sans Symbols": ["⚛", "♻", "⚔", "☯", "⚰"],
    "Noto Sans Symbols 2": ["⯑", "⯒", "⯓", "⯔", "⯕"],
}


noto_font_to_script = {
    "Noto Sans CJK SC": ("Hani", pygame.DIRECTION_LTR),
    "Noto Sans Arabic": ("Arab", pygame.DIRECTION_RTL),
    "Noto Sans Devanagari": ("Deva", pygame.DIRECTION_LTR),
    "Noto Sans Old Persian": ("Xpeo", pygame.DIRECTION_LTR),
}


# Function to load a specific Noto font
def load_noto_font(font_name, size):
    """
    Tries to load a font from the system by name.
    """
    font_path = pygame.font.match_font(font_name)
    if not font_path:
        raise FileNotFoundError(font_name)

    font = pygame.font.Font(font_path, size)
    font_script_and_direction = noto_font_to_script.get(font_name)
    if font_script_and_direction:
        font.set_script(font_script_and_direction[0])
        font.set_direction(font_script_and_direction[1])

    return font


# Load fonts
loaded_fonts = {name: load_noto_font(name, FONT_SIZE) for name in noto_font_samples}

# Colors
WHITE = (255, 255, 255)
BLACK = (0, 0, 0)
RED = (255, 0, 0)

# Main loop
running = True
while running:
    screen.fill(WHITE)

    y_offset = 20  # Vertical spacing
    for font_name, chars in noto_font_samples.items():
        font = loaded_fonts.get(font_name)
        # Display font name
        title_surface = pygame.font.Font(None, 30).render(f"{font_name}:", True, BLACK)
        screen.blit(title_surface, (20, y_offset))

        x_offset = 250  # Horizontal spacing
        for char in chars:
            # Render character
            text_surface = font.render(char, True, BLACK)
            screen.blit(text_surface, (x_offset, y_offset))

            # Get font metrics
            metrics = font.metrics(char)
            if metrics:  # Ensure valid metric data
                tot_advance = 0

                for metric in metrics:
                    if not metric:
                        continue

                    min_x, max_x, min_y, max_y, x_advance = metric

                    pygame.draw.rect(
                        screen,
                        RED,
                        (
                            x_offset + tot_advance,
                            y_offset + min_y,
                            max_x - min_x,
                            max_y - min_y,
                        ),
                        1,
                    )
                    tot_advance += x_advance

            x_offset += 150  # Move right for the next character

        y_offset += 70  # Move down for the next font

    # Event handling
    for event in pygame.event.get():
        if event.type == pygame.QUIT:
            running = False

    pygame.display.flip()

pygame.quit()

For me this gives the output

Some issues I see with this results (technically none of these are faults of this PR)

• metrics returns bad min_x for all combinations I tested, resulting all the boxes to have the origin shifted from where they should be.
• Color emoji fonts do not respect the font size (and IDK if it's a freetype issue, harfbuzz issue, or an issue with the font itself)
• Scripts like Arabic (RTL and has character combining) and Hindi (does character combining but is LTR) render properly now that we can use set_script and set_direction but the metrics method still does not respect the proper shaping.

[damusss]

Changes are loyal to the scope of the PR, thanks

[oddbookworm]

LGTM!