core/validator.py

"""
core/validator.py

Two validators:
  validate(spec)           — legacy single-component spec (unchanged)
  validate_page_spec(spec) — new page spec
"""
from __future__ import annotations

import logging
import re


logger = logging.getLogger(__name__)


SUPPORTED_COMPONENTS = {
    "Slider", "Textbox", "Image", "Dropdown", "Checkbox",
    "Button", "Audio", "File", "Plot", "Dataframe", "Number",
    "HTML", "Radio",
}

FALLBACK_SPEC = {
    "component":        "Textbox",
    "props":            {"label": "Output", "placeholder": "Result will appear here"},
    "layout":           "vertical",
    "connected_output": None,
}

ALLOWED_PROPS: dict[str, set[str]] = {
    "Slider":    {"label", "minimum", "maximum", "step", "value", "info"},
    "Textbox":   {"label", "placeholder", "value", "lines", "max_lines", "info"},
    "Image":     {"label", "type", "height", "width", "info"},
    "Dropdown":  {"label", "choices", "value", "multiselect", "info"},
    "Checkbox":  {"label", "value", "info"},
    "Button":    {"value", "variant", "size", "icon"},
    "Audio":     {"label", "type", "format", "info"},
    "File":      {"label", "file_count", "file_types", "info"},
    "Plot":      {"label", "format"},
    "Dataframe": {"label", "headers", "datatype", "row_count", "col_count", "info"},
    "Number":    {"label", "value", "minimum", "maximum", "step", "info"},
    "HTML":      {"html_template", "css"},
    "Radio":     {"label", "choices", "value", "info"},
}

_CSS_FORBIDDEN = re.compile(r"url\s*\(|@import|fetch\s*\(|<script", re.IGNORECASE)


def validate(spec: dict) -> tuple[bool, dict]:
    """Validate a single-component spec dict. Returns (ok, validated_spec_or_fallback)."""
    if not isinstance(spec, dict):
        logger.warning("spec is not a dict — using fallback")
        return False, FALLBACK_SPEC.copy()

    component = spec.get("component", "")
    if component not in SUPPORTED_COMPONENTS:
        logger.warning("Unknown component %r — using fallback", component)
        return False, FALLBACK_SPEC.copy()

    if "props" not in spec or not isinstance(spec.get("props"), dict):
        logger.warning("Missing or invalid props for %r — using fallback", component)
        return False, FALLBACK_SPEC.copy()

    allowed   = ALLOWED_PROPS.get(component, set())
    filtered  = {k: v for k, v in spec["props"].items() if k in allowed}

    clean_spec = {
        "component":        component,
        "props":            filtered,
        "layout":           spec.get("layout", "vertical"),
        "connected_output": spec.get("connected_output", None),
    }
    return True, clean_spec



_VALID_LAYOUTS      = {"centered", "sidebar", "dashboard", "split", "fullpage", "tabbed"}
_VALID_SECTION_TYPES = {"hero", "form", "result", "chart", "table", "gallery",
                        "nav", "footer", "custom"}
_VALID_ROLES        = {"input", "output", "control", "display"}
_VALID_ACTIONS      = {"new", "edit", "add", "remove", "replace", "wire", "restyle"}


def validate_page_spec(spec: dict) -> tuple[bool, list[str]]:
    """
    Validate a page spec.

    Returns (ok, errors) where errors is a list of human-readable strings.
    ok is True only when errors is empty.

    Checks:
    - page.sections is a non-empty list
    - every section has id, type, components list
    - every component has id, component, role
    - no duplicate ids across all sections and all components
    - every glue_fn trigger references a real section+component id
    - every glue_fn output references a real section+component id
    - html_shell is non-empty for non-custom sections
    - CSS fields do not contain url(), @import, fetch(), or <script
    """
    errors: list[str] = []

    page = spec.get("page", spec)
    if not isinstance(page, dict):
        return False, ["page is not a dict"]

    # layout
    layout = page.get("layout", "centered")
    if layout not in _VALID_LAYOUTS:
        errors.append(f"Unknown layout {layout!r}. Valid: {sorted(_VALID_LAYOUTS)}")

    # sections
    sections = page.get("sections", [])
    if not sections:
        errors.append("page.sections is empty — at least one section required")
        return False, errors

    all_section_ids:   set[str] = set()
    all_component_ids: set[str] = set()

    for sec in sections:
        sec_id   = sec.get("id", "")
        sec_type = sec.get("type", "")

        if not sec_id:
            errors.append("A section is missing its 'id' field")
        elif sec_id in all_section_ids:
            errors.append(f"Duplicate section id: {sec_id!r}")
        else:
            all_section_ids.add(sec_id)

        if sec_type not in _VALID_SECTION_TYPES:
            errors.append(f"Section {sec_id!r} has unknown type {sec_type!r}")

        shell = sec.get("html_shell", "")
        if sec_type != "custom" and not shell.strip():
            errors.append(f"Section {sec_id!r} (type={sec_type!r}) has empty html_shell")

        sec_css = sec.get("css", "")
        if sec_css and _CSS_FORBIDDEN.search(sec_css):
            errors.append(f"Section {sec_id!r} CSS contains forbidden pattern")

        components = sec.get("components", [])
        if not isinstance(components, list):
            errors.append(f"Section {sec_id!r} components is not a list")
            continue

        for comp in components:
            comp_id   = comp.get("id", "")
            comp_type = comp.get("component", "")
            role      = comp.get("role", "")

            if not comp_id:
                errors.append(f"A component in section {sec_id!r} is missing 'id'")
            elif comp_id in all_component_ids:
                errors.append(f"Duplicate component id: {comp_id!r}")
            else:
                all_component_ids.add(comp_id)

            if comp_type not in SUPPORTED_COMPONENTS:
                errors.append(
                    f"Component {comp_id!r} in section {sec_id!r} has "
                    f"unsupported type {comp_type!r}"
                )

            if role not in _VALID_ROLES:
                errors.append(
                    f"Component {comp_id!r} has invalid role {role!r}. "
                    f"Valid: {sorted(_VALID_ROLES)}"
                )

    # global CSS
    global_css = page.get("global_css", "")
    if global_css and _CSS_FORBIDDEN.search(global_css):
        errors.append("global_css contains forbidden pattern (url/import/fetch/script)")

    # glue_fns
    glue_fns = page.get("glue_fns", [])
    if not isinstance(glue_fns, list):
        errors.append("page.glue_fns is not a list")
        glue_fns = []

    for i, fn in enumerate(glue_fns):
        if not isinstance(fn, dict):
            errors.append(f"glue_fns[{i}] is not a dict (got {type(fn).__name__!r}) — skipping")
            continue

        fn_id   = fn.get("id", "<unknown>")
        trigger = fn.get("trigger", {})
        if not isinstance(trigger, dict):
            errors.append(f"glue_fn {fn_id!r} trigger is not a dict")
            continue
        t_sec   = trigger.get("section",   "")
        t_comp  = trigger.get("component", "")

        if t_sec not in all_section_ids:
            errors.append(f"glue_fn {fn_id!r} trigger section {t_sec!r} not found")
        if t_comp not in all_component_ids:
            errors.append(f"glue_fn {fn_id!r} trigger component {t_comp!r} not found")

        for ref in fn.get("inputs", []) + fn.get("outputs", []):
            if not isinstance(ref, dict):
                continue
            ref_sec  = ref.get("section",   "")
            ref_comp = ref.get("component", "")
            if ref_sec and ref_sec not in all_section_ids:
                errors.append(
                    f"glue_fn {fn_id!r} references unknown section {ref_sec!r}"
                )
            if ref_comp and ref_comp not in all_component_ids:
                errors.append(
                    f"glue_fn {fn_id!r} references unknown component {ref_comp!r}"
                )

    return len(errors) == 0, errors