open-school/backend/tools/validate_svg_quality.py

from __future__ import annotations

import argparse
import re
import sys
import xml.etree.ElementTree as ET
from dataclasses import dataclass
from pathlib import Path


SLASH_FRACTION_RE = re.compile(r"(?<![\w.])\d+\s*/\s*\d+(?![\w.])")
MOJIBAKE_RE = re.compile(r"Ã.|Â.|’|“|â€")
SUSPICIOUS_REPLACEMENT_RE = re.compile(r"[A-Za-zÀ-ÿ]\?|[?]\s*[A-Za-zÀ-ÿ]")
TRANSLATE_RE = re.compile(r"translate\(\s*([-\d.]+)(?:[\s,]+([-\d.]+))?\s*\)")
DATA_BOX_RE = re.compile(r"^\s*([-\d.]+)\s+([-\d.]+)\s+([-\d.]+)\s+([-\d.]+)\s*$")
FONT_SIZE_RE = re.compile(r"font-size\s*:\s*([-\d.]+)px?")
CLASS_STYLE_RE = re.compile(r"\.([A-Za-z0-9_-]+)\s*\{([^}]*)\}")
FORBIDDEN_TEXT = {
    "Ençadrer": "Encadrer",
}


@dataclass
class Issue:
    severity: str
    path: Path
    message: str


@dataclass
class Box:
    kind: str
    label: str
    x1: float
    y1: float
    x2: float
    y2: float

    def padded(self, amount: float) -> "Box":
        return Box(self.kind, self.label, self.x1 - amount, self.y1 - amount, self.x2 + amount, self.y2 + amount)


def local_name(tag: str) -> str:
    return tag.rsplit("}", 1)[-1]


def parse_number(value: str | None, default: float = 0.0) -> float:
    if not value:
        return default
    try:
        return float(value)
    except ValueError:
        return default


def node_text(node: ET.Element) -> str:
    return "".join(node.itertext()).strip()


def class_names(node: ET.Element) -> set[str]:
    return set((node.attrib.get("class") or "").split())


def has_class(node: ET.Element, name: str) -> bool:
    return name in class_names(node)


def parse_styles(root: ET.Element) -> dict[str, dict[str, str]]:
    styles: dict[str, dict[str, str]] = {}
    for node in root.iter():
        if local_name(node.tag) != "style":
            continue
        text = node.text or ""
        for match in CLASS_STYLE_RE.finditer(text):
            declarations: dict[str, str] = {}
            for declaration in match.group(2).split(";"):
                if ":" not in declaration:
                    continue
                key, value = declaration.split(":", 1)
                declarations[key.strip()] = value.strip()
            styles[match.group(1)] = declarations
    return styles


def font_size(node: ET.Element, styles: dict[str, dict[str, str]], default: float = 24.0) -> float:
    direct = parse_number(node.attrib.get("font-size"), 0)
    if direct:
        return direct

    style_attr = node.attrib.get("style") or ""
    style_match = FONT_SIZE_RE.search(style_attr)
    if style_match:
        return parse_number(style_match.group(1), default)

    for class_name in class_names(node):
        value = styles.get(class_name, {}).get("font-size")
        if value:
            return parse_number(value.replace("px", ""), default)
    return default


def line_width(node: ET.Element) -> float:
    return abs(parse_number(node.attrib.get("x2")) - parse_number(node.attrib.get("x1")))


def line_y(node: ET.Element) -> float:
    return (parse_number(node.attrib.get("y1")) + parse_number(node.attrib.get("y2"))) / 2


def text_x(node: ET.Element) -> float:
    return parse_number(node.attrib.get("x"))


def text_y(node: ET.Element) -> float:
    return parse_number(node.attrib.get("y"))


def estimated_text_width(text: str, size: float) -> float:
    # Chiffres scolaires en police sans-serif : estimation volontairement prudente.
    return max(1, len(text.strip())) * size * 0.62


def parse_translate(value: str | None) -> tuple[float, float]:
    if not value:
        return 0.0, 0.0
    match = TRANSLATE_RE.search(value)
    if not match:
        return 0.0, 0.0
    return parse_number(match.group(1)), parse_number(match.group(2), 0.0)


def overlaps(a: Box, b: Box) -> bool:
    return a.x1 < b.x2 and a.x2 > b.x1 and a.y1 < b.y2 and a.y2 > b.y1


def parse_data_box(node: ET.Element, ox: float, oy: float) -> Box | None:
    match = DATA_BOX_RE.match(node.attrib.get("data-box") or "")
    if not match:
        return None
    x, y, width, height = (parse_number(value) for value in match.groups())
    return Box("fraction", node_text(node) or "fraction", ox + x, oy + y, ox + x + width, oy + y + height)


def text_box(node: ET.Element, styles: dict[str, dict[str, str]], ox: float, oy: float) -> Box | None:
    content = node_text(node)
    if not content:
        return None
    size = font_size(node, styles)
    width = estimated_text_width(content, size)
    x = ox + text_x(node)
    y = oy + text_y(node)
    anchor = node.attrib.get("text-anchor", "start")
    if anchor == "middle":
        x1 = x - width / 2
        x2 = x + width / 2
    elif anchor == "end":
        x1 = x - width
        x2 = x
    else:
        x1 = x
        x2 = x + width
    return Box("text", content, x1, y - size * 0.85, x2, y + size * 0.25)


def shape_box(node: ET.Element, ox: float, oy: float) -> Box | None:
    tag = local_name(node.tag)
    classes = class_names(node)
    if tag == "rect":
        if classes.intersection({"bg", "panel", "soft", "card", "white"}):
            return None
        x = ox + parse_number(node.attrib.get("x"))
        y = oy + parse_number(node.attrib.get("y"))
        return Box("shape", "rect", x, y, x + parse_number(node.attrib.get("width")), y + parse_number(node.attrib.get("height")))
    if tag == "circle":
        cx = ox + parse_number(node.attrib.get("cx"))
        cy = oy + parse_number(node.attrib.get("cy"))
        radius = parse_number(node.attrib.get("r"))
        return Box("shape", "circle", cx - radius, cy - radius, cx + radius, cy + radius)
    return None


def collect_layout_boxes(
    node: ET.Element,
    styles: dict[str, dict[str, str]],
    ox: float = 0.0,
    oy: float = 0.0,
    inside_fraction: bool = False,
) -> list[Box]:
    tx, ty = parse_translate(node.attrib.get("transform"))
    ox += tx
    oy += ty

    if local_name(node.tag) == "g" and has_class(node, "fraction-g"):
        box = parse_data_box(node, ox, oy)
        return [box] if box else []

    boxes: list[Box] = []
    if not inside_fraction:
        if local_name(node.tag) == "text":
            box = text_box(node, styles, ox, oy)
            if box:
                boxes.append(box)
        else:
            box = shape_box(node, ox, oy)
            if box:
                boxes.append(box)

    for child in node:
        boxes.extend(collect_layout_boxes(child, styles, ox, oy, inside_fraction))
    return boxes


def validate_layout_collisions(path: Path, root: ET.Element, styles: dict[str, dict[str, str]]) -> list[Issue]:
    issues: list[Issue] = []
    boxes = collect_layout_boxes(root, styles)
    for index, first in enumerate(boxes):
        for second in boxes[index + 1 :]:
            if first.kind != "fraction" and second.kind != "fraction":
                continue
            if overlaps(first, second):
                issues.append(
                    Issue(
                        "error",
                        path,
                        f"Chevauchement autour d'une fraction: {first.kind} {first.label!r} avec {second.kind} {second.label!r}.",
                    )
                )
    return issues


def validate_fraction_group(path: Path, group: ET.Element, styles: dict[str, dict[str, str]]) -> list[Issue]:
    issues: list[Issue] = []
    texts = [node for node in group if local_name(node.tag) == "text"]
    lines = [node for node in group if local_name(node.tag) == "line"]

    if "data-box" not in group.attrib:
        issues.append(Issue("warning", path, "Une fraction-g doit declarer data-box pour les controles anti-chevauchement."))

    if len(texts) < 2 or not lines:
        return [
            Issue(
                "error",
                path,
                "Un groupe fraction-g doit contenir au moins deux textes et une barre horizontale.",
            )
        ]

    numerator = texts[0]
    denominator = texts[-1]
    bar = lines[0]
    num_y = text_y(numerator)
    den_y = text_y(denominator)
    bar_y = line_y(bar)

    if not num_y < bar_y < den_y:
        issues.append(Issue("error", path, "La barre d'une fraction-g n'est pas entre le numérateur et le dénominateur."))

    num_size = font_size(numerator, styles)
    den_size = font_size(denominator, styles)
    min_top_gap = max(8.0, num_size * 0.30)
    min_bottom_gap = max(8.0, den_size * 0.30)

    if bar_y - num_y < min_top_gap:
        issues.append(
            Issue(
                "warning",
                path,
                f"Espace faible entre numérateur et barre ({bar_y - num_y:.1f}px, attendu >= {min_top_gap:.1f}px).",
            )
        )
    if den_y - bar_y < min_bottom_gap:
        issues.append(
            Issue(
                "warning",
                path,
                f"Espace faible entre barre et dénominateur ({den_y - bar_y:.1f}px, attendu >= {min_bottom_gap:.1f}px).",
            )
        )

    expected_width = max(
        estimated_text_width(node_text(numerator), num_size),
        estimated_text_width(node_text(denominator), den_size),
    )
    if line_width(bar) < expected_width * 0.95:
        issues.append(
            Issue(
                "warning",
                path,
                f"Barre de fraction possiblement trop courte ({line_width(bar):.1f}px pour ~{expected_width:.1f}px de texte).",
            )
        )

    center = (parse_number(bar.attrib.get("x1")) + parse_number(bar.attrib.get("x2"))) / 2
    if abs(text_x(numerator) - center) > 4 or abs(text_x(denominator) - center) > 4:
        issues.append(Issue("warning", path, "Numérateur ou dénominateur non centré sur la barre de fraction."))

    return issues


def validate_svg(path: Path) -> list[Issue]:
    issues: list[Issue] = []
    try:
        text = path.read_text(encoding="utf-8")
        root = ET.fromstring(text)
    except Exception as exc:
        return [Issue("error", path, f"SVG illisible: {exc}")]

    if local_name(root.tag) != "svg":
        issues.append(Issue("error", path, "Le fichier n'a pas de racine <svg>."))
        return issues

    if not root.attrib.get("viewBox"):
        issues.append(Issue("warning", path, "SVG sans viewBox : recadrage et responsive moins fiables."))

    styles = parse_styles(root)
    issues.extend(validate_layout_collisions(path, root, styles))

    for node in root.iter():
        if local_name(node.tag) == "text":
            content = node_text(node)
            for forbidden, expected in FORBIDDEN_TEXT.items():
                if forbidden in content:
                    issues.append(
                        Issue(
                            "error",
                            path,
                            f"Texte incorrect détecté: {forbidden!r}. Utiliser {expected!r}.",
                        )
                    )
            if MOJIBAKE_RE.search(content):
                issues.append(
                    Issue(
                        "error",
                        path,
                        f"Encodage texte suspect dans le SVG: {content!r}.",
                    )
                )
            if SUSPICIOUS_REPLACEMENT_RE.search(content):
                issues.append(
                    Issue(
                        "error",
                        path,
                        f"Caractère de remplacement suspect dans le texte SVG: {content!r}.",
                    )
                )
            if SLASH_FRACTION_RE.search(content):
                issues.append(
                    Issue(
                        "error",
                        path,
                        f"Fraction oblique détectée dans un texte SVG: {content!r}. Utiliser une fraction verticale.",
                    )
                )

        if local_name(node.tag) == "g" and has_class(node, "fraction-g"):
            issues.extend(validate_fraction_group(path, node, styles))

    return issues


def iter_svg_paths(paths: list[Path]) -> list[Path]:
    files: list[Path] = []
    for path in paths:
        if path.is_file() and path.suffix.lower() == ".svg":
            files.append(path)
        elif path.is_dir():
            files.extend(sorted(path.rglob("*.svg")))
    return sorted(set(files))


def main() -> int:
    parser = argparse.ArgumentParser(description="Valide les SVG pédagogiques sensibles.")
    parser.add_argument("paths", nargs="+", type=Path, help="Fichiers ou dossiers SVG à contrôler.")
    parser.add_argument("--warnings-as-errors", action="store_true", help="Retourne une erreur si des warnings existent.")
    args = parser.parse_args()

    all_issues: list[Issue] = []
    files = iter_svg_paths(args.paths)
    for path in files:
        all_issues.extend(validate_svg(path))

    for issue in all_issues:
        print(f"{issue.severity.upper()} {issue.path}: {issue.message}")

    errors = [issue for issue in all_issues if issue.severity == "error"]
    warnings = [issue for issue in all_issues if issue.severity == "warning"]
    print(f"SVG contrôlés: {len(files)} | erreurs: {len(errors)} | warnings: {len(warnings)}")

    if errors or (args.warnings_as_errors and warnings):
        return 1
    return 0


if __name__ == "__main__":
    sys.exit(main())