svg correction

backend/tools/validate_svg_quality.py

@@ -10,7 +10,9 @@ 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 = {
@@ -25,6 +27,19 @@ class Issue:
     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]
 
@@ -105,11 +120,122 @@ def estimated_text_width(text: str, size: float) -> float:
     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(
@@ -187,6 +313,7 @@ def validate_svg(path: Path) -> list[Issue]:
         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":
@@ -208,6 +335,14 @@ def validate_svg(path: Path) -> list[Issue]:
                         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(