Initial commit: OmniSVG-Lite MVP before live-streaming work
- LM Studio client (httpx-based, OpenAI-compatible) - SVG validator (lxml, whitelist tags, no <script>/<foreignObject>/http refs) - PNG renderer (resvg-py primary, cairosvg fallback - no native cairo dep) - History (SQLite, tracks raw/validated/preview paths) - Gradio UI on 127.0.0.1:8788 with: * mode radio (icon/illustration) * n_candidates slider (default 1) * image upload for image-to-SVG * LM Studio URL/token inputs * model dropdown + refresh button - prompts/ with system_icon.txt, system_illustration.txt, few_shot_examples.txt - docs/spec.md, docs/design.md - 122 unit/integration tests passing
This commit is contained in:
Mavis
+228
@@ -0,0 +1,228 @@
|
||||
"""Валидация SVG-ответов модели.
|
||||
|
||||
Делает две вещи:
|
||||
1) `extract_svg()` — вытаскивает <svg>...</svg> из произвольного текста
|
||||
(модель иногда отвечает с пояснениями или code-fence).
|
||||
2) `validate(svg_text, mode=...)` — строгая проверка по правилам дизайна.
|
||||
3) `validate_svg(svg_text) -> (ok, reason, cleaned_svg)` — обёртка для UI,
|
||||
описанная в задаче: всегда возвращает кортеж, не бросает исключений.
|
||||
|
||||
Правила (соответствуют §5 design.md и §6 spec.md):
|
||||
- well-formed XML (lxml);
|
||||
- ровно один корневой <svg>;
|
||||
- наличие `viewBox` (4 числа);
|
||||
- теги — только из ALLOWED_TAGS;
|
||||
- <script>, <foreignObject>, <image> — запрещены;
|
||||
- http://, https:// — запрещены в href/xlink:href и в `url(...)`;
|
||||
- on* event-handler атрибуты — запрещены (доп. защита по задаче);
|
||||
- размер файла <= лимита mode (32/256 KB).
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import logging
|
||||
import re
|
||||
from dataclasses import dataclass
|
||||
|
||||
from lxml import etree
|
||||
|
||||
|
||||
log = logging.getLogger(__name__)
|
||||
|
||||
|
||||
SVG_NS = "http://www.w3.org/2000/svg"
|
||||
|
||||
ALLOWED_TAGS: set[str] = {
|
||||
"svg",
|
||||
"g",
|
||||
"defs",
|
||||
"symbol",
|
||||
"use",
|
||||
"path",
|
||||
"rect",
|
||||
"circle",
|
||||
"ellipse",
|
||||
"line",
|
||||
"polygon",
|
||||
"polyline",
|
||||
"linearGradient",
|
||||
"radialGradient",
|
||||
"stop",
|
||||
"text",
|
||||
"tspan",
|
||||
"textPath",
|
||||
"title",
|
||||
"desc",
|
||||
"filter",
|
||||
"feGaussianBlur",
|
||||
"feOffset",
|
||||
"feMerge",
|
||||
"feMergeNode",
|
||||
"feColorMatrix",
|
||||
"feBlend",
|
||||
"feFlood",
|
||||
"feComposite",
|
||||
"clipPath",
|
||||
"mask",
|
||||
"pattern",
|
||||
}
|
||||
|
||||
DISALLOWED_TAGS: set[str] = {"script", "foreignObject", "image", "iframe", "style"}
|
||||
|
||||
MAX_BYTES: dict[str, int] = {
|
||||
"icon": 32 * 1024,
|
||||
"illustration": 256 * 1024,
|
||||
}
|
||||
|
||||
|
||||
class ValidatorError(Exception):
|
||||
"""Ошибка валидации с машино-читаемым кодом и человеческим пояснением."""
|
||||
|
||||
def __init__(self, code: str, detail: str = ""):
|
||||
super().__init__(f"{code}: {detail}" if detail else code)
|
||||
self.code = code
|
||||
self.detail = detail
|
||||
|
||||
|
||||
_SVG_RE = re.compile(
|
||||
r"<svg\b[^>]*>.*?</svg\s*>",
|
||||
re.IGNORECASE | re.DOTALL,
|
||||
)
|
||||
|
||||
|
||||
def extract_svg(text: str) -> str:
|
||||
"""Достаёт первое вхождение <svg>...</svg> из произвольного ответа модели."""
|
||||
if not text or "<svg" not in text.lower():
|
||||
raise ValidatorError("not_svg", "в ответе нет <svg>...</svg>")
|
||||
match = _SVG_RE.search(text)
|
||||
if not match:
|
||||
raise ValidatorError("not_svg", "не удалось выделить <svg>...</svg> regex'ом")
|
||||
return match.group(0).strip()
|
||||
|
||||
|
||||
def _local(tag: object) -> str:
|
||||
"""Снимает namespace: {ns}local -> local. Терпимо к non-str (cyfunction)."""
|
||||
s = str(tag)
|
||||
if "}" in s:
|
||||
return s.split("}", 1)[1]
|
||||
return s
|
||||
|
||||
|
||||
def _iter_real_elements(root: etree._Element):
|
||||
"""Итератор только по Element-узлам (пропускает комменты и PI)."""
|
||||
for el in root.iter():
|
||||
# lxml._Element и etree._Element имеют tag как str; cyfunction — это
|
||||
# callable у комментариев/PI. Отфильтруем через isinstance.
|
||||
if isinstance(getattr(el, "tag", None), str):
|
||||
yield el
|
||||
|
||||
|
||||
def _check_viewbox(root: etree._Element) -> None:
|
||||
vb = root.get("viewBox")
|
||||
if not vb:
|
||||
raise ValidatorError("missing_viewbox", "атрибут viewBox отсутствует")
|
||||
try:
|
||||
parts = [float(x) for x in vb.split()]
|
||||
except ValueError as exc:
|
||||
raise ValidatorError("bad_viewbox", f"viewBox='{vb}'") from exc
|
||||
if len(parts) != 4:
|
||||
raise ValidatorError("bad_viewbox", f"viewBox='{vb}' (нужно 4 числа)")
|
||||
|
||||
|
||||
def _check_tags_and_attrs(root: etree._Element) -> None:
|
||||
for el in _iter_real_elements(root):
|
||||
local = _local(el.tag)
|
||||
if local in DISALLOWED_TAGS:
|
||||
raise ValidatorError("disallowed_tag", local)
|
||||
if local not in ALLOWED_TAGS:
|
||||
raise ValidatorError("unknown_tag", local)
|
||||
for attr, val in el.attrib.items():
|
||||
if not val:
|
||||
continue
|
||||
attr_local = _local(attr).lower()
|
||||
if attr_local.startswith("on"):
|
||||
raise ValidatorError("event_handler", f"{attr_local}={val[:80]}")
|
||||
if "http://" in val or "https://" in val:
|
||||
if attr_local.endswith("href") or "url(" in val.lower():
|
||||
raise ValidatorError("external_ref", f"{attr_local}={val[:80]}")
|
||||
|
||||
|
||||
def _ensure_size(svg_text: str, mode: str) -> None:
|
||||
limit = MAX_BYTES.get(mode)
|
||||
if limit is None:
|
||||
return
|
||||
if len(svg_text.encode("utf-8")) > limit:
|
||||
raise ValidatorError(
|
||||
"too_large",
|
||||
f"{len(svg_text.encode('utf-8'))} байт > {limit} байт (mode={mode})",
|
||||
)
|
||||
|
||||
|
||||
def _serialize_clean(root: etree._Element) -> str:
|
||||
if not root.get("xmlns"):
|
||||
new_root = etree.Element(_local(root.tag), nsmap={None: SVG_NS})
|
||||
for k, v in root.attrib.items():
|
||||
new_root.set(_local(k), v)
|
||||
for child in root:
|
||||
new_root.append(child)
|
||||
root = new_root
|
||||
xml_bytes = etree.tostring(
|
||||
root, pretty_print=False, xml_declaration=False, encoding="utf-8"
|
||||
)
|
||||
return xml_bytes.decode("utf-8")
|
||||
|
||||
|
||||
def validate(svg_text: str, *, mode: str = "icon") -> etree._Element:
|
||||
"""Строгая валидация по правилам дизайна. Бросает ValidatorError."""
|
||||
svg_text = extract_svg(svg_text)
|
||||
try:
|
||||
root = etree.fromstring(svg_text.encode("utf-8"))
|
||||
except etree.XMLSyntaxError as exc:
|
||||
raise ValidatorError("malformed_xml", str(exc)) from exc
|
||||
|
||||
if _local(root.tag) != "svg":
|
||||
raise ValidatorError("not_svg", root.tag)
|
||||
|
||||
_check_viewbox(root)
|
||||
_check_tags_and_attrs(root)
|
||||
_ensure_size(svg_text, mode)
|
||||
return root
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class ValidationResult:
|
||||
ok: bool
|
||||
reason: str
|
||||
cleaned_svg: str
|
||||
|
||||
|
||||
def validate_svg(svg_text: str, *, mode: str = "icon") -> tuple[bool, str, str]:
|
||||
"""Удобная обёртка над `validate()`: всегда возвращает кортеж.
|
||||
|
||||
Returns:
|
||||
Кортеж `(ok, reason, cleaned_svg)`.
|
||||
"""
|
||||
if not svg_text or not svg_text.strip():
|
||||
return False, "empty_input", ""
|
||||
try:
|
||||
root = validate(svg_text, mode=mode)
|
||||
except ValidatorError as exc:
|
||||
log.info("validate_svg: reject (%s)", exc.code)
|
||||
return False, exc.code, ""
|
||||
except Exception as exc: # noqa: BLE001
|
||||
log.exception("validate_svg: неожиданная ошибка")
|
||||
return False, f"internal:{type(exc).__name__}", ""
|
||||
return True, "", _serialize_clean(root)
|
||||
|
||||
|
||||
__all__ = [
|
||||
"ALLOWED_TAGS",
|
||||
"DISALLOWED_TAGS",
|
||||
"MAX_BYTES",
|
||||
"SVG_NS",
|
||||
"ValidationResult",
|
||||
"ValidatorError",
|
||||
"extract_svg",
|
||||
"validate",
|
||||
"validate_svg",
|
||||
]
|
||||
Reference in New Issue
Block a user