"""Deterministic conversion of a static HTML site into Nuxt pages that render identically.
The problem this solves: a prototype from an AI tool (index.html, about.html, assets/css, assets/js ...) has
to become a real Nuxt app, and asking a model to retype eleven pages into components loses fidelity (or, as
seen in practice, the model serves the files as they are). This module does the mechanical part exactly.
A *faithful shell* per page:
- ``legacy/pages/.html`` keeps the page's body markup verbatim, minus the ``",
"",
"",
" ",
"",
"",
]
return "\n".join(lines)
def render_redirect_page(name: str) -> str:
"""``/about`` -> ``/about.html``: one canonical URL, and the page's scripts always see a file name."""
return "\n".join([
"",
"",
"",
"",
])
@dataclass
class SiteConversion:
files: dict[str, str] = field(default_factory=dict)
pages: list[dict[str, Any]] = field(default_factory=list)
skipped: list[dict[str, str]] = field(default_factory=list)
warnings: list[str] = field(default_factory=list)
@property
def converted(self) -> bool:
return bool(self.pages)
def _is_text_asset(path: str) -> bool:
lowered = path.lower()
return lowered.endswith(_TEXT_ASSET_SUFFIXES) and (
lowered.startswith(_ASSET_PREFIXES) or "/" not in lowered
)
def convert_static_site(
files: dict[str, str], *, existing_routes: frozenset[str] | set[str] = frozenset()
) -> SiteConversion:
"""Convert the root-level ``*.html`` pages of a static site and carry over its text assets.
Pages in sub-folders, unsafe names and routes that already exist in the project are reported, not converted:
relative URLs in a nested page would resolve against a different folder, and an existing route belongs to
the project."""
out = SiteConversion()
taken = set(existing_routes)
for path in sorted(files):
if not path.lower().endswith((".html", ".htm")):
continue
stem = path.rsplit(".", 1)[0]
if "/" in path:
out.skipped.append({"path": path, "reason": "nested page: only root-level pages are converted"})
continue
if not _SAFE_NAME.fullmatch(stem.lower()) or stem != stem.lower():
out.skipped.append({"path": path, "reason": "name is not a safe route name"})
continue
route = route_for(stem)
html_route = "/" if stem == "index" else f"/{stem}.html"
if route in taken or html_route in taken:
out.skipped.append({"path": path, "reason": f"the project already has a {route} route"})
continue
page = convert_page(files[path], stem)
taken.update({route, html_route})
out.files[f"legacy/pages/{stem}.html"] = page.body_html + "\n"
if stem == "index":
out.files["pages/index.vue"] = render_vue_page(page)
else:
out.files[f"pages/{stem}.html.vue"] = render_vue_page(page)
out.files[f"pages/{stem}.vue"] = render_redirect_page(stem)
out.pages.append(
{"source": path, "route": html_route, "redirect_from": None if stem == "index" else route,
"alias": "/index.html" if stem == "index" else None, "scripts": len(page.body_scripts),
"warnings": page.warnings}
)
out.warnings.extend(f"{path}: {w}" for w in page.warnings)
if any(p["scripts"] for p in out.pages):
out.files[LEGACY_SCRIPTS_COMPOSABLE_PATH] = LEGACY_SCRIPTS_COMPOSABLE
for path in sorted(files):
if _is_text_asset(path) and path not in out.files:
out.files[f"public/{path}"] = files[path]
return out