//! Real public-domain / CC0 media for the example seed: the manifest that names
//! each asset, and the fetch step that turns it into bytes.
//!
//! The seed's other phases are pure data in `creators.rs`. Media cannot be,
//! because a real asset is a file somebody has to choose, license-check, and
//! host. So the choosing lives in `media-manifest.toml` (curated by hand, checked
//! into the repo, embedded in the binary) and everything downstream of it is
//! mechanical: fetch by URL, verify the digest, upload through the storage layer,
//! record the attribution.
//!
//! # Curation state is per asset, not per manifest
//!
//! An entry with no `url` is *declared but not yet curated*. It resolves to
//! nothing and [`super::media`] falls back to the generated placeholder for that
//! one slot. So the manifest doubles as the curation checklist: every id the seed
//! can use is listed, and the ones still carrying grey boxes are exactly the ones
//! with an empty `url`. Partial curation is a first-class state, and the box
//! stays seedable throughout.
//!
//! # Failure is loud, and it happens before any write
//!
//! A curated asset that will not fetch, or that fetches to the wrong bytes, fails
//! the whole seed. It does not silently degrade to the placeholder: that is the
//! failure mode `sando/deploy/mnw-testnot-smoke.sh` exists to catch, and a demo
//! box that quietly reverts to grey squares is worse than one that refuses to
//! reseed. Resolution runs to completion before the seed touches the database, so
//! a failure leaves the previous catalog standing.

use std::collections::HashMap;
use std::path::PathBuf;
use std::time::Duration;

use serde::Deserialize;

/// The curated manifest, embedded so a deployed binary carries it. Override with
/// [`MANIFEST_PATH_ENV`] when iterating locally.
const EMBEDDED_MANIFEST: &str = include_str!("media-manifest.toml");

/// Path to a manifest file to read instead of the embedded copy.
pub const MANIFEST_PATH_ENV: &str = "SEED_MEDIA_MANIFEST";

/// Directory holding fetched assets between runs. Defaults to
/// `{temp_dir}/mnw-seed-media`.
pub const CACHE_DIR_ENV: &str = "SEED_MEDIA_CACHE";

/// How long a single asset fetch may take.
const FETCH_TIMEOUT: Duration = Duration::from_mins(1);

/// Refuse an asset larger than this. Demo media, not a distribution channel; a
/// multi-gigabyte URL in the manifest is a mistake, not a big file.
const MAX_ASSET_BYTES: u64 = 64 * 1024 * 1024;

/// One curated file: where it comes from, what it is, and who to credit.
#[derive(Debug, Clone, Deserialize)]
pub struct Asset {
    /// Stable id referenced from `creators.rs` (`ItemSpec::media`,
    /// `ItemSpec::cover`, `ProjectSpec::cover`).
    pub id: String,
    /// Direct download URL. Absent = declared but not yet curated; the slot keeps
    /// its generated placeholder.
    #[serde(default)]
    pub url: Option<String>,
    /// Lowercase hex SHA-256 of the fetched bytes. Absent on a curated asset is
    /// allowed but warned about: the run logs the digest it saw so it can be
    /// pinned. Present and mismatched is a hard failure.
    #[serde(default)]
    pub sha256: Option<String>,
    /// Content type to upload under (`audio/wav`, `image/jpeg`, ...).
    pub media_type: String,
    /// Filename to store the object as, and to show on the download.
    pub filename: String,
    /// SPDX-ish licence string. Public domain / CC0 only; see the manifest header.
    pub license: String,
    /// Human title of the work, for the credits page.
    pub title: String,
    /// Creator to credit. `None` for anonymous or corporate-anonymous works.
    #[serde(default)]
    pub author: Option<String>,
    /// The page a visitor can reach to verify the licence claim. Not the direct
    /// download URL, the landing page.
    pub source: String,
}

impl Asset {
    /// Whether this asset has been curated (has somewhere to fetch from).
    fn is_curated(&self) -> bool {
        self.url.as_deref().is_some_and(|u| !u.trim().is_empty())
    }
}

/// Shape of the TOML file: a flat array of assets.
#[derive(Debug, Deserialize)]
struct ManifestFile {
    #[serde(default)]
    asset: Vec<Asset>,
}

/// Why the manifest could not be loaded, or an asset could not be resolved.
#[derive(Debug, thiserror::Error)]
pub enum ManifestError {
    /// The override path was set but unreadable.
    #[error("cannot read media manifest at {path}: {source}")]
    Read {
        path: PathBuf,
        #[source]
        source: std::io::Error,
    },
    /// The manifest is not valid TOML, or does not match the schema.
    #[error("media manifest is not valid: {0}")]
    Parse(#[from] toml::de::Error),
    /// Two entries claim the same id, so a reference is ambiguous.
    #[error("media manifest declares id {0:?} more than once")]
    DuplicateId(String),
    /// One or more curated assets failed to fetch or verify. Collected rather
    /// than returned one at a time: a curation pass wants the whole list.
    #[error("{} media asset(s) failed to resolve:\n{}", .0.len(), .0.join("\n"))]
    Unresolved(Vec<String>),
}

/// The parsed manifest, indexed by asset id.
#[derive(Debug, Default)]
pub struct Manifest {
    assets: Vec<Asset>,
}

impl Manifest {
    /// Load the manifest: the file named by [`MANIFEST_PATH_ENV`] if set,
    /// otherwise the copy embedded at build time.
    pub fn load() -> Result<Self, ManifestError> {
        match std::env::var(MANIFEST_PATH_ENV) {
            Ok(path) if !path.trim().is_empty() => {
                let path = PathBuf::from(path);
                let text =
                    std::fs::read_to_string(&path).map_err(|source| ManifestError::Read {
                        path: path.clone(),
                        source,
                    })?;
                tracing::info!(path = %path.display(), "example seed: using media manifest override");
                Self::parse(&text)
            }
            _ => Self::parse(EMBEDDED_MANIFEST),
        }
    }

    /// Parse and validate manifest text. Separated from [`Self::load`] so the
    /// embedded manifest can be checked by a unit test with no filesystem.
    pub fn parse(text: &str) -> Result<Self, ManifestError> {
        let file: ManifestFile = toml::from_str(text)?;
        let mut seen = std::collections::HashSet::with_capacity(file.asset.len());
        for asset in &file.asset {
            if !seen.insert(asset.id.as_str()) {
                return Err(ManifestError::DuplicateId(asset.id.clone()));
            }
        }
        Ok(Self { assets: file.asset })
    }

    /// Every declared id, curated or not. Used by the roster-coverage test.
    pub fn ids(&self) -> impl Iterator<Item = &str> {
        self.assets.iter().map(|a| a.id.as_str())
    }

    /// Fetch every curated asset, verify its digest, and return the resolved set.
    ///
    /// Uncurated entries are skipped (their slots keep the placeholder). Any
    /// curated asset that fails is collected; the call returns
    /// [`ManifestError::Unresolved`] listing all of them rather than stopping at
    /// the first.
    pub async fn resolve(&self) -> Result<ResolvedAssets, ManifestError> {
        let curated: Vec<&Asset> = self.assets.iter().filter(|a| a.is_curated()).collect();
        let total = self.assets.len();
        if curated.is_empty() {
            tracing::warn!(
                declared = total,
                "example seed: no media curated yet; every slot keeps its generated placeholder"
            );
            return Ok(ResolvedAssets::default());
        }
        tracing::info!(
            curated = curated.len(),
            declared = total,
            "example seed: resolving curated media"
        );

        let cache = cache_dir();
        if let Err(e) = std::fs::create_dir_all(&cache) {
            tracing::warn!(dir = %cache.display(), error = ?e, "example seed: media cache unusable; fetching every asset fresh");
        }

        crate::crypto::install_default_crypto_provider();
        let client = reqwest::Client::builder()
            .timeout(FETCH_TIMEOUT)
            // Wikimedia's User-Agent policy refuses generic agents outright, and
            // some of the manifest is hosted there. Name the project and give a
            // contact, which is what the policy asks for.
            .user_agent("makenotwork-example-seed/1.0 (+https://makenot.work; info@makenot.work)")
            .build()
            .map_err(|e| ManifestError::Unresolved(vec![format!("http client: {e}")]))?;

        let mut resolved = HashMap::with_capacity(curated.len());
        let mut failures = Vec::new();
        for asset in curated {
            match fetch_asset(&client, &cache, asset).await {
                Ok(bytes) => {
                    resolved.insert(asset.id.clone(), (asset.clone(), bytes));
                }
                Err(reason) => failures.push(format!("  {}: {reason}", asset.id)),
            }
        }

        if !failures.is_empty() {
            return Err(ManifestError::Unresolved(failures));
        }
        Ok(ResolvedAssets { assets: resolved })
    }
}

/// Curated media, fetched and verified, ready to upload.
#[derive(Debug, Default)]
pub struct ResolvedAssets {
    assets: HashMap<String, (Asset, Vec<u8>)>,
}

impl ResolvedAssets {
    /// The asset behind an id, if it was curated and resolved.
    pub fn get(&self, id: &str) -> Option<(&Asset, &[u8])> {
        self.assets.get(id).map(|(a, b)| (a, b.as_slice()))
    }

    /// The asset behind an optional reference, so call sites can pass
    /// `spec.cover` straight through.
    pub fn lookup(&self, id: Option<&str>) -> Option<(&Asset, &[u8])> {
        self.get(id?)
    }

    /// How many assets resolved.
    pub fn len(&self) -> usize {
        self.assets.len()
    }

    /// Whether nothing resolved (every slot is on its placeholder).
    pub fn is_empty(&self) -> bool {
        self.assets.is_empty()
    }

    /// A markdown credits list: one line per resolved asset, title linked to the
    /// source page, with author and licence. Sorted by title so a reseed with the
    /// same manifest produces the same page.
    ///
    /// Empty string when nothing resolved, so the caller can skip publishing.
    pub fn attribution_markdown(&self) -> String {
        if self.assets.is_empty() {
            return String::new();
        }
        let mut lines: Vec<String> = self
            .assets
            .values()
            .map(|(a, _)| {
                let author = a
                    .author
                    .as_deref()
                    .map_or_else(String::new, |author| format!(" by {author}"));
                format!("- [{}]({}){} — {}", a.title, a.source, author, a.license)
            })
            .collect();
        lines.sort();
        lines.dedup();
        lines.join("\n")
    }
}

/// Where fetched assets are cached between runs.
fn cache_dir() -> PathBuf {
    match std::env::var(CACHE_DIR_ENV) {
        Ok(dir) if !dir.trim().is_empty() => PathBuf::from(dir),
        _ => std::env::temp_dir().join("mnw-seed-media"),
    }
}

/// Fetch one asset, preferring the cache, and verify its digest.
///
/// The cache is keyed by id, and a cached file is only trusted when the manifest
/// pins a digest and the file matches it. An unpinned asset is re-fetched every
/// run: without a digest there is nothing to tell a stale cache entry from a
/// current one.
async fn fetch_asset(
    client: &reqwest::Client,
    cache: &std::path::Path,
    asset: &Asset,
) -> Result<Vec<u8>, String> {
    let cached = cache.join(&asset.id);
    if let (Some(want), Ok(bytes)) = (asset.sha256.as_deref(), std::fs::read(&cached))
        && digest_hex(&bytes).eq_ignore_ascii_case(want.trim())
    {
        tracing::debug!(id = %asset.id, "example seed: media cache hit");
        return Ok(bytes);
    }

    let url = asset.url.as_deref().unwrap_or_default().trim();
    let response = client
        .get(url)
        .send()
        .await
        .map_err(|e| format!("fetching {url}: {e}"))?;
    if !response.status().is_success() {
        return Err(format!("fetching {url}: HTTP {}", response.status()));
    }
    // Refuse an oversized body before buffering it, when the server declares one.
    if let Some(len) = response.content_length()
        && len > MAX_ASSET_BYTES
    {
        return Err(format!(
            "fetching {url}: {len} bytes exceeds the {MAX_ASSET_BYTES}-byte asset ceiling"
        ));
    }
    let bytes = response
        .bytes()
        .await
        .map_err(|e| format!("reading {url}: {e}"))?
        .to_vec();
    if bytes.len() as u64 > MAX_ASSET_BYTES {
        return Err(format!(
            "fetching {url}: {} bytes exceeds the {MAX_ASSET_BYTES}-byte asset ceiling",
            bytes.len()
        ));
    }
    if bytes.is_empty() {
        return Err(format!("fetching {url}: empty body"));
    }

    let got = digest_hex(&bytes);
    match asset.sha256.as_deref().map(str::trim) {
        Some(want) if !want.is_empty() => {
            if !got.eq_ignore_ascii_case(want) {
                return Err(format!(
                    "digest mismatch for {url}: manifest pins {want}, fetched {got}. \
                     The asset changed at the source; re-check the licence before repinning."
                ));
            }
        }
        _ => tracing::warn!(
            id = %asset.id,
            sha256 = %got,
            "example seed: asset is unpinned; add sha256 = \"{got}\" to media-manifest.toml"
        ),
    }

    if let Err(e) = std::fs::write(&cached, &bytes) {
        tracing::debug!(id = %asset.id, error = ?e, "example seed: could not cache asset");
    }
    tracing::info!(id = %asset.id, bytes = bytes.len(), "example seed: fetched media asset");
    Ok(bytes)
}

/// Lowercase hex SHA-256.
fn digest_hex(bytes: &[u8]) -> String {
    use sha2::{Digest, Sha256};
    hex::encode(Sha256::digest(bytes))
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn embedded_manifest_parses() {
        Manifest::parse(EMBEDDED_MANIFEST).expect("the embedded manifest must always be valid");
    }

    #[test]
    fn duplicate_ids_are_refused() {
        let text = r#"
[[asset]]
id = "a"
media_type = "image/png"
filename = "a.png"
license = "CC0-1.0"
title = "A"
source = "https://example.test/a"

[[asset]]
id = "a"
media_type = "image/png"
filename = "b.png"
license = "CC0-1.0"
title = "B"
source = "https://example.test/b"
"#;
        assert!(matches!(
            Manifest::parse(text).unwrap_err(),
            ManifestError::DuplicateId(id) if id == "a"
        ));
    }

    #[test]
    fn an_asset_without_a_url_is_uncurated() {
        let text = r#"
[[asset]]
id = "a"
url = "  "
media_type = "image/png"
filename = "a.png"
license = "CC0-1.0"
title = "A"
source = "https://example.test/a"
"#;
        let manifest = Manifest::parse(text).unwrap();
        assert!(!manifest.assets[0].is_curated());
    }

    #[tokio::test]
    async fn an_uncurated_manifest_never_reaches_for_the_network() {
        // Synthetic rather than the embedded manifest, which is now fully
        // curated: this is a claim about the code path, and it should keep
        // holding whatever the shipped manifest looks like.
        let text = r#"
[[asset]]
id = "a"
media_type = "image/png"
filename = "a.png"
license = "CC0-1.0"
title = "A"
source = "https://example.test/a"
"#;
        let resolved = Manifest::parse(text).unwrap().resolve().await.unwrap();
        assert!(resolved.is_empty());
    }

    #[test]
    fn every_shipped_asset_is_curated_and_pinned() {
        // An unpinned asset still works, but it re-fetches every run and cannot
        // detect the file changing at the source. The shipped manifest was
        // curated in one pass with digests taken from the bytes that arrived,
        // so anything unpinned here is an entry someone added without running
        // it.
        let manifest = Manifest::load().expect("manifest loads");
        for asset in &manifest.assets {
            assert!(asset.is_curated(), "{} has no url", asset.id);
            let digest = asset.sha256.as_deref().unwrap_or_default().trim();
            assert_eq!(digest.len(), 64, "{} is not pinned", asset.id);
            assert!(
                digest.chars().all(|c| c.is_ascii_hexdigit()),
                "{} has a malformed digest",
                asset.id
            );
        }
    }

    /// Fetch every shipped asset and check it against its pin.
    ///
    /// Ignored by default: it is ~90 MB over the network and depends on two
    /// museums staying up, neither of which belongs in a normal test run. Run it
    /// after editing the manifest, which is the moment it earns its cost:
    ///
    ///   cargo test --lib seed::manifest -- --ignored --nocapture
    #[tokio::test]
    #[ignore = "network: fetches every asset in the manifest"]
    async fn every_shipped_asset_actually_resolves() {
        let manifest = Manifest::load().expect("manifest loads");
        let declared = manifest.assets.len();
        let resolved = manifest
            .resolve()
            .await
            .expect("every asset should resolve");
        assert_eq!(
            resolved.len(),
            declared,
            "resolved {} of {declared} assets",
            resolved.len()
        );
        assert!(!resolved.attribution_markdown().is_empty());
    }

    #[test]
    fn every_shipped_asset_is_public_domain_and_verifiable() {
        // Rule 1 and rule 2 of the manifest header, enforced rather than
        // trusted. These files sit on a public box under a licence claim this
        // repo makes; a CC-BY asset slipping in is a licensing problem, and a
        // `source` that is not a reachable page makes the claim uncheckable.
        let manifest = Manifest::load().expect("manifest loads");
        for asset in &manifest.assets {
            let licence = asset.license.to_ascii_lowercase();
            assert!(
                licence.contains("cc0") || licence.contains("public domain"),
                "{} is licensed {:?}, which is not public domain or CC0",
                asset.id,
                asset.license
            );
            assert!(
                asset.source.starts_with("https://"),
                "{} has no https source page",
                asset.id
            );
            assert!(
                asset
                    .url
                    .as_deref()
                    .is_some_and(|u| u.starts_with("https://")),
                "{} is not fetched over https",
                asset.id
            );
        }
    }

    #[test]
    fn attribution_lists_resolved_assets_only() {
        let asset = Asset {
            id: "x".into(),
            url: Some("https://example.test/x.jpg".into()),
            sha256: None,
            media_type: "image/jpeg".into(),
            filename: "x.jpg".into(),
            license: "CC0-1.0".into(),
            title: "A Study".into(),
            author: Some("A. Person".into()),
            source: "https://example.test/x".into(),
        };
        let mut assets = HashMap::new();
        assets.insert("x".to_string(), (asset, vec![1, 2, 3]));
        let resolved = ResolvedAssets { assets };
        assert_eq!(
            resolved.attribution_markdown(),
            "- [A Study](https://example.test/x) by A. Person — CC0-1.0"
        );
        assert!(ResolvedAssets::default().attribution_markdown().is_empty());
    }

    #[test]
    fn digest_is_lowercase_hex_sha256() {
        // Known vector: SHA-256 of the empty string.
        assert_eq!(
            digest_hex(b""),
            "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
        );
    }
}