scripts/validate_website_integrity.py

#!/usr/bin/env python3
"""Validate static website links, anchors, image assets, and JSON data.

This is a local integrity check for the GitHub Pages / Hugging Face static
website. It intentionally does not fetch external URLs; it verifies that the
published local surface is self-consistent.
"""

from __future__ import annotations

import argparse
import json
import re
from dataclasses import dataclass
from datetime import datetime, timezone
from html.parser import HTMLParser
from pathlib import Path
from urllib.parse import unquote, urlsplit

from PIL import Image


ROOT = Path(__file__).resolve().parents[1]
DEFAULT_DOCS = ROOT / "docs"
DEFAULT_OUTPUT = DEFAULT_DOCS / "data/website_integrity.json"
DEFAULT_SITE_BASE = "/ropedia-xperience-10m-task-suite/"
LOCAL_ATTRS = {"href", "src"}
SKIP_SCHEMES = {"http", "https", "mailto", "tel", "data", "javascript"}
IMAGE_SUFFIXES = {".jpg", ".jpeg", ".png", ".webp"}


@dataclass
class Reference:
    source: Path
    tag: str
    attr: str
    raw: str
    path_part: str
    fragment: str


class SiteParser(HTMLParser):
    def __init__(self, source: Path):
        super().__init__(convert_charrefs=True)
        self.source = source
        self.ids: list[str] = []
        self.references: list[Reference] = []
        self.images: list[str] = []

    def handle_starttag(self, tag: str, attrs):
        attr_map = dict(attrs)
        element_id = attr_map.get("id") or attr_map.get("name")
        if element_id:
            self.ids.append(element_id)

        for attr in LOCAL_ATTRS:
            raw = attr_map.get(attr)
            if not raw:
                continue
            parsed = urlsplit(raw)
            if parsed.scheme in SKIP_SCHEMES or parsed.netloc:
                continue
            path_part = unquote(parsed.path)
            fragment = unquote(parsed.fragment)
            if not path_part and not fragment:
                continue
            self.references.append(Reference(self.source, tag, attr, raw, path_part, fragment))
            if tag == "img" and attr == "src":
                self.images.append(path_part)


def parse_html(path: Path) -> SiteParser:
    parser = SiteParser(path)
    parser.feed(path.read_text(encoding="utf-8", errors="ignore"))
    return parser


def normalize_path_part(path_part: str, site_base: str) -> str:
    if path_part in {"", "/"}:
        return path_part
    normalized_base = "/" + site_base.strip("/") + "/"
    if path_part == normalized_base:
        return "/index.html"
    if path_part.startswith(normalized_base):
        return "/" + path_part[len(normalized_base):]
    return path_part


def resolve_reference(docs_root: Path, source: Path, path_part: str, site_base: str) -> Path:
    path_part = normalize_path_part(path_part, site_base)
    if not path_part:
        return source
    base = docs_root if path_part.startswith("/") else source.parent
    resolved = (base / path_part.lstrip("/")).resolve()
    if resolved.is_dir():
        return resolved / "index.html"
    return resolved


def relative(path: Path, root: Path) -> str:
    try:
        return path.relative_to(root).as_posix()
    except ValueError:
        return path.as_posix()


def duplicate_ids(ids: list[str]) -> list[dict]:
    counts: dict[str, int] = {}
    for item in ids:
        counts[item] = counts.get(item, 0) + 1
    return [{"id": item, "count": count} for item, count in sorted(counts.items()) if count > 1]


def image_record(path: Path, docs_root: Path) -> dict:
    suffix = path.suffix.lower()
    record = {
        "path": relative(path, docs_root),
        "exists": path.exists(),
        "bytes": path.stat().st_size if path.exists() else 0,
    }
    if not path.exists():
        return record
    if suffix in IMAGE_SUFFIXES:
        with Image.open(path) as image:
            record.update({
                "width": int(image.width),
                "height": int(image.height),
                "format": image.format,
            })
    elif suffix == ".svg":
        text = path.read_text(encoding="utf-8", errors="ignore")
        viewbox = re.search(r'viewBox=["\']([^"\']+)["\']', text)
        record["format"] = "SVG"
        record["has_viewbox"] = bool(viewbox)
    return record


def validate(docs_root: Path, site_base: str) -> dict:
    html_files = sorted(docs_root.glob("*.html"))
    parsers = {path: parse_html(path) for path in html_files}
    anchors_by_file = {path: set(parser.ids) for path, parser in parsers.items()}

    missing_targets = []
    missing_anchors = []
    local_references = []
    external_reference_count = 0
    image_paths: set[Path] = set()

    for path, parser in parsers.items():
        text = path.read_text(encoding="utf-8", errors="ignore")
        external_reference_count += len(re.findall(r'https?://', text))
        for ref in parser.references:
            target = resolve_reference(docs_root, path, ref.path_part, site_base)
            target_rel = relative(target, docs_root)
            source_rel = relative(ref.source, docs_root)
            local_references.append({
                "source": source_rel,
                "tag": ref.tag,
                "attr": ref.attr,
                "raw": ref.raw,
                "target": target_rel,
                "fragment": ref.fragment,
            })
            if not target.exists():
                missing_targets.append({
                    "source": source_rel,
                    "raw": ref.raw,
                    "target": target_rel,
                })
                continue
            if ref.tag == "img" and ref.attr == "src":
                image_paths.add(target)
            if ref.fragment:
                anchor_target = target if target.suffix.lower() == ".html" else path
                if ref.path_part and target.suffix.lower() != ".html":
                    continue
                anchors = anchors_by_file.get(anchor_target)
                if anchors is None and anchor_target.exists() and anchor_target.suffix.lower() == ".html":
                    anchors = set(parse_html(anchor_target).ids)
                    anchors_by_file[anchor_target] = anchors
                if anchors is not None and ref.fragment not in anchors:
                    missing_anchors.append({
                        "source": source_rel,
                        "raw": ref.raw,
                        "target": relative(anchor_target, docs_root),
                        "fragment": ref.fragment,
                    })

    json_records = []
    invalid_json = []
    for path in sorted((docs_root / "data").glob("*.json")):
        rel_path = relative(path, docs_root)
        try:
            payload = json.loads(path.read_text(encoding="utf-8"))
        except json.JSONDecodeError as exc:
            invalid_json.append({"path": rel_path, "error": str(exc)})
            continue
        json_records.append({
            "path": rel_path,
            "bytes": path.stat().st_size,
            "top_level_type": type(payload).__name__,
        })

    images = []
    invalid_images = []
    for path in sorted(image_paths):
        try:
            record = image_record(path, docs_root)
            images.append(record)
            if not record.get("exists") or record.get("bytes", 0) <= 0:
                invalid_images.append(record)
            if path.suffix.lower() in IMAGE_SUFFIXES and (record.get("width", 0) <= 0 or record.get("height", 0) <= 0):
                invalid_images.append(record)
            if path.suffix.lower() == ".svg" and not record.get("has_viewbox"):
                invalid_images.append(record)
        except Exception as exc:  # noqa: BLE001 - report image validation failures.
            invalid_images.append({"path": relative(path, docs_root), "error": str(exc)})

    duplicate_id_records = [
        {"path": relative(path, docs_root), "duplicates": duplicate_ids(parser.ids)}
        for path, parser in parsers.items()
    ]
    duplicate_id_records = [item for item in duplicate_id_records if item["duplicates"]]

    semantic_checks = []
    semantic_layout_failures = []
    index_path = docs_root / "index.html"
    index_text = index_path.read_text(encoding="utf-8", errors="ignore") if index_path.exists() else ""
    roadmap_payload = None
    roadmap_json_error = None
    roadmap_path = docs_root / "data/research_roadmap.json"
    try:
        roadmap_payload = json.loads(roadmap_path.read_text(encoding="utf-8"))
    except Exception as exc:  # noqa: BLE001 - report malformed/missing roadmap data.
        roadmap_json_error = str(exc)

    def section_pos(section_id: str) -> int:
        match = re.search(rf'<section\b[^>]*\bid="{re.escape(section_id)}"', index_text)
        return match.start() if match else -1

    suite_start = section_pos("suite")
    suite_end = section_pos("pipeline")
    suite_text = index_text[suite_start:suite_end] if suite_start >= 0 and suite_end > suite_start else ""
    overview_pos = section_pos("overview")
    protocol_pos = section_pos("protocol")
    evidence_pos = section_pos("evidence")
    dataset_start = section_pos("dataset-card")
    dataset_end = section_pos("suite")
    dataset_text = index_text[dataset_start:dataset_end] if dataset_start >= 0 and dataset_end > dataset_start else ""
    raw_sample_start = section_pos("raw-sample")
    raw_sample_end = section_pos("suite")
    raw_sample_text = index_text[raw_sample_start:raw_sample_end] if raw_sample_start >= 0 and raw_sample_end > raw_sample_start else ""
    roadmap_page = docs_root / "research_roadmap.html"
    roadmap_page_text = roadmap_page.read_text(encoding="utf-8", errors="ignore") if roadmap_page.exists() else ""
    semantic_rules = [
        (
            "project_tabs_have_six_groups",
            'data-tab-key=',
            None,
            "The long research page should expose six top-level tabs, including a dedicated Directions tab.",
        ),
        (
            "project_sections_are_assigned_to_tabs",
            'data-project-tab=',
            None,
            "Every major research section should be assigned to a tab group.",
        ),
        (
            "project_hash_router_preserves_deep_links",
            'activateTabForHash',
            None,
            "Deep links should open the correct tab instead of landing on hidden content.",
        ),
        (
            "homepage_hidden_attribute_not_overridden",
            "[hidden] { display: none !important; }",
            None,
            "Elements hidden by JavaScript should stay hidden even when component CSS sets display values.",
        ),
        (
            "project_tabs_use_accessible_roles",
            'role="tab"',
            None,
            "The tabbed research dashboard should expose tablist/tab semantics.",
        ),
        (
            "project_sections_are_labeled_tabpanels",
            'role="tabpanel"',
            None,
            "Every tabbed research section should expose a labeled panel role.",
        ),
        (
            "project_tabs_update_selected_state",
            'aria-selected',
            None,
            "Tab activation should update selected state for assistive technology.",
        ),
        (
            "project_tabs_support_keyboard_navigation",
            'moveProjectTabFocus',
            None,
            "Keyboard users should be able to switch project tabs with arrow, Home, and End keys.",
        ),
        (
            "project_overview_precedes_progress_ledger",
            '<section id="overview">',
            '<section id="evidence">',
            "The project overview should appear before the deeper progress ledger.",
        ),
        (
            "project_status_links_json",
            'data/project_status.json',
            None,
            "The website should expose the machine-readable project status.",
        ),
        (
            "roadmap_links_json",
            'data/research_roadmap.json',
            None,
            "The website should expose the machine-readable research roadmap.",
        ),
        (
            "interactive_roadmap_page_linked",
            'research_roadmap.html',
            None,
            "The project site should link to the dedicated interactive research roadmap page.",
        ),
        (
            "interactive_roadmap_links_json",
            'data/research_roadmap_interactive.json',
            None,
            "The project site should expose the machine-readable interactive roadmap contract.",
        ),
        (
            "interactive_roadmap_loads_generated_json",
            'data/research_roadmap_interactive.json',
            None,
            "The interactive roadmap page should load the generated roadmap JSON contract.",
        ),
        (
            "interactive_roadmap_tracks_four_directions",
            'Research tracks',
            None,
            "The interactive roadmap page should expose the four research-track control surface.",
        ),
        (
            "reader_resources_link_qwen_status",
            'results/omni_finetune/DATA_ACCESS_STATUS.md',
            None,
            "The website should expose the Qwen3-Omni scale-up status.",
        ),
        (
            "roadmap_html_matches_json_phases",
            "",
            None,
            "The roadmap section should show every stage defined in research_roadmap.json.",
        ),
        (
            "roadmap_status_chips_match_json",
            "",
            None,
            "The roadmap status chips should match the phase statuses in research_roadmap.json.",
        ),
        (
            "evaluation_protocol_between_overview_and_progress",
            '<section id="protocol">',
            '<section id="evidence">',
            "The evaluation protocol should appear before the deeper evidence ledger.",
        ),
        (
            "evaluation_protocol_links_json",
            'data/evaluation_protocol.json',
            None,
            "The website should expose the machine-readable evaluation protocol.",
        ),
        (
            "visual_figures_link_task_suite_image",
            'assets/task_suite_infographic.png',
            None,
            "The website should expose the main task-suite figure.",
        ),
        (
            "foundation_direction_cards_explain_one_sample_io",
            'class="foundation-io-panel"',
            None,
            "The three foundation direction cards should explain one-sample training inputs and outputs.",
        ),
        (
            "suite_task_map_precedes_radar_surface",
            '<div class="figure-pan" id="task-suite-map">',
            'class="chart radar-chart unified-radar-chart"',
            "The Suite anchor should show the task-suite map before the radar/results surface.",
        ),
        (
            "raw_sample_stream_ledger_contains_seven_modalities",
            "",
            None,
            "The raw sample browser should expose the seven source streams without a separate repeated atlas component.",
        ),
        (
            "raw_sample_browser_links_modality_metadata",
            'data/modality_atlas.json',
            None,
            "The raw sample browser should keep the machine-readable modality metadata link.",
        ),
        (
            "dataset_card_section_links_official_dataset",
            'https://huggingface.co/datasets/ropedia-ai/xperience-10m',
            None,
            "The dataset-card section should link the official gated dataset.",
        ),
        (
            "dataset_card_section_links_public_sample",
            'https://huggingface.co/datasets/ropedia-ai/xperience-10m-sample',
            None,
            "The dataset-card section should link the public sample dataset used here.",
        ),
        (
            "dataset_card_section_states_raw_data_boundary",
            'Raw MP4, HDF5, RRD files',
            None,
            "The dataset-card section should state that raw Xperience-10M files are not redistributed.",
        ),
        (
            "raw_sample_browser_section_present",
            '<section id="raw-sample"',
            None,
            "The website should expose a direct raw public sample browser under the Data tab.",
        ),
        (
            "raw_sample_browser_links_manifest",
            'data/raw_sample_files.json',
            None,
            "The raw sample browser should expose a machine-readable file manifest.",
        ),
        (
            "raw_sample_browser_has_video_player",
            'id="rawVideo"',
            None,
            "The raw sample browser should include a native video player for MP4 streams.",
        ),
        (
            "raw_sample_browser_has_audio_player",
            'id="rawAudio"',
            None,
            "The raw sample browser should include a native audio player for the embedded MP4 audio stream.",
        ),
        (
            "raw_sample_browser_lists_core_files",
            'visualization.rrd',
            None,
            "The raw sample browser should list the annotation, MP4, and optional RRD files.",
        ),
        (
            "raw_sample_browser_links_preview_assets",
            'assets/raw-sample-preview/fisheye_cam0_preview.mp4',
            None,
            "The raw sample browser should link browser-playable preview clips for MP4 streams.",
        ),
        (
            "raw_sample_browser_links_full_raw_source",
            'open full raw source',
            None,
            "The raw sample browser should keep direct links to the complete raw source files.",
        ),
        (
            "task_player_surface_present",
            'id="taskPlayer"',
            None,
            "The website should expose the interactive task walkthrough/player.",
        ),
        (
            "task_player_uses_walkthrough_json",
            'data/task_walkthroughs.json',
            None,
            "The task player and task cards should read the generated walkthrough JSON.",
        ),
        (
            "task_cards_use_human_research_names",
            'Action Recognition',
            None,
            "The public task surface should use readable research task names.",
        ),
    ]
    for name, marker, after_marker, reason in semantic_rules:
        if name == "project_tabs_have_six_groups":
            tab_count = index_text.count(marker)
            passed = tab_count == 6
            detail = {"tab_count": tab_count}
        elif name == "project_sections_are_assigned_to_tabs":
            section_count = index_text.count(marker)
            passed = section_count >= 19
            detail = {"section_count": section_count}
        elif name == "project_hash_router_preserves_deep_links":
            marker_count = index_text.count(marker)
            passed = marker_count >= 2 and "sectionTabMap" in index_text
            detail = {"marker_count": marker_count, "has_section_tab_map": "sectionTabMap" in index_text}
        elif name == "project_tabs_use_accessible_roles":
            tab_role_count = index_text.count(marker)
            project_tab_count = index_text.count("data-tab-key=")
            nested_tab_count = index_text.count("data-panel-target=")
            passed = (
                'role="tablist"' in index_text
                and project_tab_count == 6
                and nested_tab_count >= 4
                and tab_role_count >= project_tab_count + nested_tab_count
            )
            detail = {
                "tab_role_count": tab_role_count,
                "project_tab_count": project_tab_count,
                "nested_tab_count": nested_tab_count,
                "has_tablist": 'role="tablist"' in index_text,
            }
        elif name == "project_sections_are_labeled_tabpanels":
            panel_count = index_text.count(marker)
            passed = panel_count >= 19 and index_text.count('aria-labelledby="tab-') >= 19
            detail = {
                "panel_count": panel_count,
                "labeled_panel_count": index_text.count('aria-labelledby="tab-'),
            }
        elif name == "project_tabs_update_selected_state":
            selected_count = index_text.count(marker)
            passed = selected_count >= 6 and 'setAttribute("aria-selected"' in index_text
            detail = {
                "selected_count": selected_count,
                "updates_selected_state": 'setAttribute("aria-selected"' in index_text,
            }
        elif name == "project_tabs_support_keyboard_navigation":
            marker_count = index_text.count(marker)
            passed = marker_count >= 2 and "ArrowRight" in index_text and "Home" in index_text and "End" in index_text
            detail = {
                "marker_count": marker_count,
                "has_arrow_navigation": "ArrowRight" in index_text and "ArrowLeft" in index_text,
                "has_home_end_navigation": "Home" in index_text and "End" in index_text,
            }
        elif name == "roadmap_html_matches_json_phases":
            phase_names = [str(phase.get("name", "")) for phase in (roadmap_payload or {}).get("phases", [])]
            missing_names = [phase for phase in phase_names if phase and phase not in index_text]
            passed = bool(phase_names) and not missing_names and not roadmap_json_error
            detail = {
                "phase_count": len(phase_names),
                "missing_phase_names": missing_names,
                "roadmap_json_error": roadmap_json_error,
            }
        elif name == "roadmap_status_chips_match_json":
            statuses = [str(phase.get("status", "")).lower() for phase in (roadmap_payload or {}).get("phases", [])]
            missing_statuses = sorted({status for status in statuses if status and f'data-status="{status}"' not in index_text})
            passed = bool(statuses) and not missing_statuses and not roadmap_json_error
            detail = {
                "phase_count": len(statuses),
                "statuses": statuses,
                "missing_statuses": missing_statuses,
                "roadmap_json_error": roadmap_json_error,
            }
        elif name == "raw_sample_stream_ledger_contains_seven_modalities":
            modality_terms = [
                "Video",
                "Audio",
                "Depth",
                "Pose / SLAM",
                "Motion capture",
                "Inertial",
                "Language",
            ]
            present_terms = [term for term in modality_terms if term in raw_sample_text]
            passed = len(present_terms) == len(modality_terms)
            detail = {
                "modality_count": len(present_terms),
                "missing_modalities": [term for term in modality_terms if term not in present_terms],
            }
        elif name == "foundation_direction_cards_explain_one_sample_io":
            panel_count = index_text.count(marker)
            required_terms = [
                "Sample input",
                "Training output",
                "Existing hooks",
                "Spatial intelligence models",
                "Human-video world models",
                "Vision-language-action models",
            ]
            missing_terms = [term for term in required_terms if term not in index_text]
            passed = panel_count == 3 and not missing_terms
            detail = {"panel_count": panel_count, "missing_terms": missing_terms}
        elif name.startswith("dataset_card_section_"):
            marker_count = dataset_text.count(marker)
            passed = marker_count >= 1
            detail = {"marker_count": marker_count}
        elif name.startswith("raw_sample_browser_"):
            if name == "raw_sample_browser_section_present":
                marker_count = len(re.findall(r'<section\b[^>]*\bid=["\']raw-sample["\']', index_text))
            elif name == "raw_sample_browser_links_modality_metadata":
                marker_count = raw_sample_text.count(marker)
            else:
                marker_count = index_text.count(marker)
            passed = marker_count >= 1
            detail = {"marker_count": marker_count}
        elif name == "project_overview_precedes_progress_ledger":
            passed = overview_pos >= 0 and evidence_pos >= 0 and overview_pos < evidence_pos
            detail = {"overview_index": overview_pos, "evidence_index": evidence_pos}
        elif name == "evaluation_protocol_between_overview_and_progress":
            passed = overview_pos >= 0 and protocol_pos >= 0 and evidence_pos >= 0 and overview_pos < protocol_pos < evidence_pos
            detail = {"overview_index": overview_pos, "protocol_index": protocol_pos, "evidence_index": evidence_pos}
        elif name in {
            "project_status_links_json",
            "roadmap_links_json",
            "interactive_roadmap_page_linked",
            "interactive_roadmap_links_json",
            "homepage_hidden_attribute_not_overridden",
            "reader_resources_link_qwen_status",
            "visual_figures_link_task_suite_image",
            "task_player_surface_present",
            "task_player_uses_walkthrough_json",
            "task_cards_use_human_research_names",
        }:
            marker_count = index_text.count(marker)
            passed = marker_count >= 1
            detail = {"marker_count": marker_count}
        elif name in {
            "interactive_roadmap_loads_generated_json",
            "interactive_roadmap_tracks_four_directions",
        }:
            marker_count = roadmap_page_text.count(marker)
            passed = marker_count >= 1 and roadmap_page.exists()
            detail = {"marker_count": marker_count, "page_exists": roadmap_page.exists()}
        elif name == "evaluation_protocol_links_json":
            marker_count = index_text.count(marker)
            passed = marker_count >= 1
            detail = {"marker_count": marker_count}
        else:
            marker_pos = suite_text.find(marker)
            after_pos = suite_text.find(after_marker or "")
            passed = marker_pos >= 0 and after_pos >= 0 and marker_pos < after_pos
            detail = {"first_marker_index": marker_pos, "second_marker_index": after_pos}
        check = {"name": name, "status": "pass" if passed else "fail", "reason": reason, **detail}
        semantic_checks.append(check)
        if not passed:
            semantic_layout_failures.append(check)

    failures = {
        "missing_targets": missing_targets,
        "missing_anchors": missing_anchors,
        "duplicate_ids": duplicate_id_records,
        "invalid_json": invalid_json,
        "invalid_images": invalid_images,
        "semantic_layout": semantic_layout_failures,
    }
    failure_count = sum(len(items) for items in failures.values())

    return {
        "status": "pass" if failure_count == 0 else "fail",
        "generated_at_utc": datetime.now(timezone.utc).isoformat(timespec="seconds"),
        "docs_root": relative(docs_root, ROOT),
        "site_base": site_base,
        "summary": {
            "html_pages": len(html_files),
            "local_references": len(local_references),
            "external_reference_count": external_reference_count,
            "json_files": len(json_records),
            "image_assets_referenced": len(images),
            "failure_count": failure_count,
        },
        "failures": failures,
        "semantic_checks": semantic_checks,
        "html_pages": [
            {
                "path": relative(path, docs_root),
                "id_count": len(parser.ids),
                "reference_count": len(parser.references),
                "image_count": len(parser.images),
            }
            for path, parser in parsers.items()
        ],
        "json_files": json_records,
        "images": images,
    }


def main() -> int:
    parser = argparse.ArgumentParser()
    parser.add_argument("--docs-root", type=Path, default=DEFAULT_DOCS)
    parser.add_argument("--output", type=Path, default=DEFAULT_OUTPUT)
    parser.add_argument("--site-base", default=DEFAULT_SITE_BASE)
    args = parser.parse_args()

    report = validate(args.docs_root.resolve(), args.site_base)
    args.output.parent.mkdir(parents=True, exist_ok=True)
    args.output.write_text(json.dumps(report, indent=2) + "\n", encoding="utf-8")
    print(f"{report['status'].upper()}: wrote {args.output}")
    if report["status"] != "pass":
        for kind, items in report["failures"].items():
            for item in items[:20]:
                print(f"- {kind}: {item}")
        return 1
    return 0


if __name__ == "__main__":
    raise SystemExit(main())