use std::collections::{HashMap, HashSet};
use std::marker::PhantomData;
use std::path::{PathBuf, absolute};
use std::str::FromStr;
use std::sync::Arc;
use std::{fmt, io};

use anyhow::Context;
use anyhow::{Error, anyhow};
use clap::Parser;
use clap::Subcommand;
use futures::{StreamExt, TryStreamExt, stream};
use rusqlite::fallible_iterator::FallibleIterator as _;
use rusqlite::{OptionalExtension, named_params, params};
use scraper::{Html, Selector};
use serde::de::{self, Visitor};
use serde::{Deserialize, Serialize};
use snix_castore::B3Digest;
use snix_castore::blobservice::BlobService;
use snix_castore::directoryservice::DirectoryService;
use snix_castore::{blobservice, directoryservice, import::fs::ingest_path};
use std::sync::Mutex;
use tokio::io::{AsyncReadExt, BufReader};
use tokio::sync::Semaphore;
use tokio::sync::mpsc::{Sender, channel};
use tokio_stream::wrappers::ReceiverStream;
use url::Url;

#[derive(Clone, Debug)]
enum Ingestable {
    Url(Url),
    Path(PathBuf),
}

#[derive(Debug, Clone)]
enum SampledWhen {
    Now,
    Before,
}

#[derive(Debug, Clone)]
struct SizedBlob {
    hash: B3Digest,
    n_bytes: u64,
}

#[derive(Debug, Clone)]
#[allow(dead_code)]
struct Sampled {
    sample_id: u32,
    uri: String,
    blob: Option<SizedBlob>,
    http_status: Option<u16>,
    epoch: u32,
    when: SampledWhen,
}

#[derive(Clone)]
enum FetchListingMessage {
    Sampled(Url, Sampled),
    Recurse(Url, usize, Sender<FetchListingMessage>),
}

impl std::fmt::Display for Ingestable {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            Ingestable::Url(url) => write!(f, "{}", url),
            Ingestable::Path(path_buf) => match path_buf.to_str() {
                Some(s) => write!(f, "{}", s),
                None => {
                    panic!("PathBuf::to_str failed")
                }
            },
        }
    }
}

fn parse_url_or_path(s: &str) -> Result<Ingestable, Error> {
    if s.is_empty() {
        Err(anyhow!("Empty path (url)"))
    } else if s.starts_with("./") || s.starts_with("/") {
        Ok(Ingestable::Path(PathBuf::from(s)))
    } else {
        let url = Url::parse(s)?;
        if url.scheme() == "file" {
            match url.to_file_path() {
                Ok(s) => Ok(Ingestable::Path(s)),
                Err(()) => Err(anyhow!(
                    "parse_url_or_path: couldn't convert Url ({}) to Path",
                    url
                )),
            }
        } else {
            Ok(Ingestable::Url(url))
        }
    }
}

fn data_path() -> PathBuf {
    let xdg_data_dir = std::env::var("XDG_DATA_DIR")
        .and_then(|s| Ok(PathBuf::from(s)))
        .or_else(|_| -> Result<PathBuf, Error> {
            match std::env::home_dir() {
                Some(p) => Ok(p.join(".local/share")),
                None => Err(anyhow!("...")), // FIXME
            }
        });
    match xdg_data_dir {
        Ok(p) => p.join("sidx"),
        Err(_) => PathBuf::from(".sidx"),
    }
}
fn default_castore_path() -> PathBuf {
    data_path().join("castore")
}
fn default_db_path() -> PathBuf {
    data_path().join("sidx.db")
}

#[derive(Subcommand)]
enum Command {
    Ingest {
        #[clap(value_parser = parse_url_or_path, num_args = 1)]
        inputs: Vec<Ingestable>,
    },
    FetchListing {
        #[clap(value_parser, long, default_value_t = 5)]
        max_depth: usize,
        #[clap(value_parser, long, default_value_t = 1024 * 1024)]
        html_max_bytes: u64,
        #[clap(value_parser, num_args = 1)]
        inputs: Vec<Url>,
    },
    ParseUrl {
        #[clap(value_parser, num_args = 1)]
        url: Vec<Url>,
    },
    DemoCudaManifest,
    FormatCudaManifest,
    ProcessCudaManifests {
        #[clap(short, long, action)]
        include_finished: bool,
    },
}

#[derive(Parser)]
struct Cli {
    #[clap(short, long, action)]
    refetch: bool,

    #[clap(short, long, value_parser, default_value_t = 2)]
    max_parallel: usize,

    #[clap(short, long, value_parser, default_value_os_t = default_db_path())]
    db_path: PathBuf,

    #[clap(short, long, value_parser, default_value_os_t = default_castore_path())]
    castore_path: PathBuf,

    #[command(subcommand)]
    command: Option<Command>,
}

struct SidxContext<BS, DS>
where
    BS: blobservice::BlobService + Clone + Send + 'static,
    DS: directoryservice::DirectoryService + Clone + Send + 'static,
{
    refetch: bool,
    max_parallel: usize,
    http: reqwest::Client,
    http_semaphore: Arc<Semaphore>,
    con: Arc<Mutex<rusqlite::Connection>>,
    blob_service: BS,
    dir_service: DS,
}

async fn open_context(
    refetch: bool,
    max_parallel: usize,
    db_path: PathBuf,
    castore_path: PathBuf,
) -> SidxContext<Arc<dyn BlobService>, Arc<dyn DirectoryService>> {
    if let Some(p) = db_path.parent() {
        let _ = std::fs::create_dir_all(p);
    }

    let con = rusqlite::Connection::open(&db_path).expect("Failed to construct Database object");
    con.pragma_update(None, "jorunal_mode", "wal").unwrap();
    con.pragma_update(None, "synchronous", "normal").unwrap();
    con.pragma_update(None, "temp_store", "memory").unwrap();
    con.pragma_update(None, "foreign_keys", "on").unwrap();
    con.execute_batch(include_str!("q/sidx-init.sql"))
        .expect("Failed to execute sidx-init.sql");
    let castore_path = absolute(castore_path).expect("Failed to canonicalize castore_path");
    let blob_service = blobservice::from_addr(&std::format!(
        "objectstore+file://{}",
        castore_path
            .join("blob")
            .to_str()
            .expect("Path::to_str unexpectedly broken")
    ))
    .await
    .expect("Couldn't initialize .castore/blob");
    let dir_service = directoryservice::from_addr(&std::format!(
        "objectstore+file://{}",
        castore_path
            .join("directories")
            .to_str()
            .expect("Path::to_str unexpectedly broken")
    ))
    .await
    .expect("Couldn't initialize .castore/directories");

    SidxContext::<Arc<dyn BlobService>, Arc<dyn DirectoryService>> {
        refetch,
        max_parallel,
        http: reqwest::Client::new(),
        http_semaphore: Arc::new(Semaphore::new(max_parallel)),
        con: Arc::new(Mutex::new(con)),
        blob_service,
        dir_service,
    }
}

impl<BS, DS> Drop for SidxContext<BS, DS>
where
    BS: BlobService + Clone,
    DS: DirectoryService + Clone,
{
    fn drop(&mut self) {
        let con = self
            .con
            .lock()
            .expect("Acquiring mutex for sqlite to run #pragma optimize before exit");
        con.pragma_update(None, "analysis_limit", 500).unwrap();
        con.pragma_query(None, "optimize", |_| Ok(())).unwrap();
    }
}

impl<BS: BlobService + Clone, DS: DirectoryService + Clone> SidxContext<BS, DS> {
    async fn latest_sample(&self, uri: &str) -> Result<Option<Sampled>, Error> {
        let lock = self.con.lock().unwrap();
        let mut find_sample = lock
            .prepare_cached(include_str!("q/latest-download.sql"))
            .expect("Failed to prepare latest-download.sql");
        find_sample
            .query_row(params![uri], |r| {
                <(u32, String, u64, Option<u16>, u32)>::try_from(r)
            })
            .optional()
            .context("db_latest_download.sql")
            .and_then(|maybe_tuple| match maybe_tuple {
                Some((sample_id, hash, n_bytes, http_code, epoch)) => Ok(Some(Sampled {
                    sample_id,
                    uri: uri.to_string(),
                    blob: Some(SizedBlob {
                        hash: B3Digest::from_str(&hash)?,
                        n_bytes,
                    }),
                    http_status: http_code,
                    epoch,
                    when: SampledWhen::Before,
                })),
                None => Ok(None),
            })
    }
    async fn db_add_sample(
        &self,
        uri: &str,
        hash: &Option<String>,
        http_code: &Option<u16>,
        content_type: &Option<String>,
    ) -> Result<(u32, u32), Error> {
        let lock = self.con.lock().expect("Locking mutex for db_add_sample");
        let mut add_sample = lock
            .prepare_cached(include_str!("q/add-sample.sql"))
            .context("Failed to prepare add-sample.sql")?;
        Ok(add_sample.query_row(
            named_params! {
                ":uri": uri,
                ":hash": hash,
                ":http_code": http_code,
                ":content_type": content_type
            },
            |row| <(u32, u32)>::try_from(row),
        )?)
    }
    async fn db_add_blob(&self, hash: &str, n_bytes: u64) -> Result<usize, Error> {
        let lock = self.con.lock().expect("db_add_blob: couldn't lock mutex?");
        let mut add_blob = lock
            .prepare_cached(include_str!("q/upsert-blob.sql"))
            .context("Failed to prepare upsert-blob.sql")?;
        Ok(add_blob.execute(params![hash, n_bytes,])?)
    }
    async fn db_add_uri(&self, uri: &str) -> Result<usize, Error> {
        let lock = self.con.lock().unwrap();
        let mut add_uri = lock
            .prepare_cached(include_str!("q/upsert-uri.sql"))
            .context("Failed to prepare upsert-uri.sql")?;

        Ok(add_uri.execute(params![uri])?)
    }
    async fn record_ingested_node(
        &self,
        uri: &str,
        blob: &Option<SizedBlob>,
        http_code: Option<u16>,
        content_type: Option<String>,
    ) -> Result<Sampled, Error> {
        let digest64 = if let Some(SizedBlob { hash, n_bytes }) = blob {
            let digest64 = format!("{}", hash);
            self.db_add_blob(&digest64, n_bytes.clone()).await?;
            Some(digest64)
        } else {
            None
        };
        self.db_add_uri(&uri).await?;
        let (sample_id, epoch) = self
            .db_add_sample(&uri, &digest64, &http_code, &content_type)
            .await?;
        Ok(Sampled {
            sample_id,
            uri: uri.to_string(),
            blob: blob.clone(),
            http_status: http_code,
            epoch,
            when: SampledWhen::Now,
        })
    }
    async fn download(&self, uri: &Url) -> Result<Sampled, Error> {
        let _permit = self.http_semaphore.acquire().await.unwrap();
        eprintln!("Downloading {:?}", uri.to_string());
        let uri_s = uri.to_string();
        let res = self
            .http
            .get(uri.clone())
            .send()
            .await
            .context(format!("Request::send failed early for {:?}", uri))?;

        let status = res.status();
        let status_code = status.as_u16();
        let content_type = res
            .headers()
            .get(reqwest::header::CONTENT_TYPE)
            .and_then(|x| x.to_str().ok())
            .map(|x| x.to_string());

        if status.is_success() {
            let mut r = tokio_util::io::StreamReader::new(
                res.bytes_stream().map_err(std::io::Error::other),
            );
            let mut w = self.blob_service.open_write().await;
            let n_bytes = match tokio::io::copy(&mut r, &mut w).await {
                Ok(n) => n,
                Err(e) => {
                    return Err(anyhow!(
                        "tokio::io::copy failed for uri={} with {}",
                        uri_s,
                        e
                    ));
                }
            };
            let digest = w.close().await?;

            self.record_ingested_node(
                &uri_s,
                &Some(SizedBlob {
                    hash: digest,
                    n_bytes,
                }),
                Some(status_code),
                content_type,
            )
            .await
        } else {
            self.record_ingested_node(&uri_s, &None, Some(status_code), content_type)
                .await
        }
    }
    async fn ensure_blob(&self, hash: &B3Digest) -> Result<(), Error> {
        if self
            .blob_service
            .has(hash)
            .await
            .context("ensure_has() accessing blob_service")?
        {
            Ok(())
        } else {
            let b64 = hash.to_string();
            let uris = {
                let con = self.con.lock().unwrap();
                let mut find_uris = con
                    .prepare_cached(include_str!("q/uris-of-hash.sql"))
                    .context("Preparing statement: q/uris-of-hash.sql")
                    .unwrap();
                find_uris
                    .query(named_params! {":hash": b64, ":limit": 100})?
                    .map(|b| b.get(0))
                    .collect::<Vec<String>>()?
            };
            if uris.is_empty() {
                return Err(anyhow!("No uris recorded for {}", b64));
            };
            for uri in uris {
                let url = match Url::parse(&uri) {
                    Ok(url) => url,
                    Err(_) => continue,
                };
                match self
                    .download(&url)
                    .await
                    .context("Redownloading missing blob for ensure_hash")
                {
                    Ok(Sampled {
                        sample_id: _,
                        uri: _,
                        blob,
                        http_status: _,
                        epoch: _,
                        when: _,
                    }) => {
                        if blob.map_or(false, |sb| sb.hash == *hash) {
                            return Ok(());
                        } else {
                            continue;
                        }
                    }
                    Err(_) => {
                        continue;
                    }
                }
            }
            Err(anyhow!(
                "All uris for {} are out of date (result in errors or different hashes)",
                b64
            ))
        }
    }
    async fn ensure_sampled_uri(&self, uri: &Url) -> Result<Sampled, Error> {
        /* TODO: flatten */
        if self.refetch {
            self.download(&uri).await
        } else {
            /* TODO: Add negative TTL */
            match self.latest_sample(&uri.to_string()).await? {
                Some(ingested) => match ingested.blob.clone() {
                    Some(SizedBlob { hash, n_bytes: _ }) => {
                        if self.blob_service.has(&hash).await? {
                            Ok(ingested)
                        } else {
                            self.download(&uri).await
                        }
                    }
                    None => self.download(&uri).await,
                },
                None => self.download(&uri).await,
            }
        }
    }
    async fn ingest(&self, inputs: &Vec<Ingestable>) -> Vec<Result<Option<Sampled>, Error>> {
        let samples = stream::iter(inputs.iter().map(|uri| {
            let blob_service = &self.blob_service;
            let dir_service = &self.dir_service;

            async move {
                let uri_s = uri.to_string();
                let latest_download = self.latest_sample(&uri_s).await?;
                if latest_download.is_some() {
                    return Ok(latest_download);
                }
                match uri {
                    Ingestable::Path(path) => {
                        match ingest_path::<_, _, _, &[u8]>(&blob_service, &dir_service, path, None)
                            .await?
                        {
                            snix_castore::Node::Directory { digest, size } => self
                                .record_ingested_node(
                                    &uri_s,
                                    &Some(SizedBlob {
                                        hash: digest,
                                        n_bytes: size,
                                    }),
                                    None,
                                    None,
                                )
                                .await
                                .map(Some),

                            snix_castore::Node::File {
                                digest,
                                size,
                                executable: _,
                            } => self
                                .record_ingested_node(
                                    &uri_s,
                                    &Some(SizedBlob {
                                        hash: digest,
                                        n_bytes: size,
                                    }),
                                    None,
                                    None,
                                )
                                .await
                                .map(Some),
                            snix_castore::Node::Symlink { target: _ } => {
                                Err(anyhow!("TODO: Figure out what to do with symlink roots"))
                            }
                        }
                    }
                    Ingestable::Url(url) => self.ensure_sampled_uri(url).await.map(Some),
                }
            }
        }))
        .buffer_unordered(self.max_parallel)
        .collect::<Vec<Result<Option<Sampled>, _>>>()
        .await;

        samples
    }

    fn extract_hrefs(content: &str) -> Result<Vec<String>, Error> {
        let sel = Selector::parse("a").map_err(|e| anyhow!(e.to_string()))?;
        let html = Html::parse_document(&content);

        Ok(html
            .select(&sel)
            .flat_map(|elt| elt.value().attr("href"))
            .map(|s| s.to_string())
            .collect::<Vec<_>>())
    }

    async fn fetch_from_listing_impl(
        self: Arc<Self>,
        url: Url,
        max_depth: usize,
        html_max_bytes: u64,
        tx: Sender<FetchListingMessage>,
    ) -> Result<(), Error> {
        let maybe_root = self.ensure_sampled_uri(&url).await;
        if let Err(ref e) = maybe_root {
            eprintln!("Couldn't download {}: {:?}", url, e);
        };
        let root = maybe_root?;
        tx.send(FetchListingMessage::Sampled(url.clone(), root.clone()))
            .await
            .context("Stopped accepting tasks before processing an Ingested notification")?;
        if max_depth <= 0 {
            return Ok(());
        }

        match root.blob {
            None => Err(anyhow!(
                "Couldn't download {}. Status code: {:?}",
                url,
                root.http_status
            )),
            Some(SizedBlob { hash, n_bytes }) => {
                if n_bytes > html_max_bytes {
                    return Ok(());
                }
                match self.blob_service.open_read(&hash).await? {
                    Some(mut reader) => {
                        let content = {
                            let mut br = BufReader::new(&mut *reader);
                            let mut content = String::new();
                            br.read_to_string(&mut content).await?;
                            content
                        };
                        let hrefs = Self::extract_hrefs(&content).unwrap_or(vec![]);
                        /* max_depth > 0 here */
                        for href in hrefs.clone() {
                            let next_url = url.join(&href).context("Constructing next_url")?;
                            tx.send(FetchListingMessage::Recurse(
                                next_url.clone(),
                                max_depth - 1,
                                tx.clone(),
                            ))
                            .await
                            .context("Stopped accepting tasks before finishing all hrefs")?;
                        }
                        {
                            let lock = self.con.lock().expect("Couldn't acquire Mutex?");
                            for href in hrefs {
                                let mut stmt =
                                    lock.prepare_cached(include_str!("q/add-str.sql"))?;
                                stmt.execute(params!["href"])?;

                                let next_url = url.join(&href).context("Constructing next_url")?;
                                let mut stmt =
                                    lock.prepare_cached(include_str!("q/add-uri-ref.sql"))?;
                                let digest64 = hash.to_string();
                                stmt.execute(named_params! {":source": digest64, ":target": next_url.to_string(), ":why": "href"})?;
                            }
                        };
                        Ok(())
                    }
                    None => Err(anyhow!("Couldn't read the ingested blob")),
                }
            }
        }
    }

    async fn fetch_from_listing(
        self: Arc<Self>,
        url: Url,
        max_depth: usize,
        html_max_bytes: u64,
    ) -> ReceiverStream<Sampled> {
        let mq_size = 10;

        /* TODO: move task queue to e.g. sqlite */
        let (tx, mut rx) = channel(mq_size);

        let (out_tx, out_rx) = channel(mq_size);

        tokio::spawn({
            async move {
                let mut seen: HashSet<String> = HashSet::new();
                {
                    let tx_moved = tx;
                    tx_moved
                        .send(FetchListingMessage::Recurse(
                            url,
                            max_depth,
                            tx_moved.clone(),
                        ))
                        .await
                        .expect("fetch_from_listing failed populating the queue");
                };
                while let Some(m) = rx.recv().await {
                    match m {
                        FetchListingMessage::Sampled(_url, ingested) => {
                            out_tx
                                .send(ingested)
                                .await
                                .expect("ReceiverStream failed to accept an Ingestable");
                        }
                        FetchListingMessage::Recurse(url, max_depth, tx) => {
                            if max_depth > 0 && !seen.contains(&url.to_string()) {
                                seen.insert(url.to_string());
                                tokio::spawn({
                                    let s = self.clone();
                                    let url = url.clone();
                                    async move {
                                        s.fetch_from_listing_impl(
                                            url,
                                            max_depth,
                                            html_max_bytes,
                                            tx,
                                        )
                                        .await
                                    }
                                });
                            }
                        }
                    }
                }
            }
        });
        ReceiverStream::new(out_rx)
    }
}

fn string_or_int<'de, T, D>(deserializer: D) -> Result<T, D::Error>
where
    T: Deserialize<'de> + TryFrom<u64> + FromStr<Err = std::num::ParseIntError>,
    D: serde::Deserializer<'de>,
{
    struct StringOrInt<T>(PhantomData<fn() -> T>);

    impl<'de, T> Visitor<'de> for StringOrInt<T>
    where
        T: Deserialize<'de> + TryFrom<u64> + FromStr<Err = std::num::ParseIntError>,
    {
        type Value = T;

        fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
            formatter.write_str("string or int")
        }

        fn visit_u64<E>(self, value: u64) -> Result<T, E>
        where
            E: de::Error,
        {
            T::try_from(value).map_err(|_e| de::Error::custom("ignored error"))
        }

        fn visit_str<E>(self, value: &str) -> Result<T, E>
        where
            E: de::Error,
        {
            FromStr::from_str(value).map_err(de::Error::custom)
        }
    }

    deserializer.deserialize_any(StringOrInt(PhantomData))
}

#[derive(Serialize, Deserialize, Debug, Clone)]
struct CudaArtifact {
    relative_path: String,
    sha256: String,
    md5: Option<String>,

    // Tha manifests export size as string instead of number
    #[serde(deserialize_with = "string_or_int")]
    size: i64,
}

#[derive(Serialize, Deserialize, Debug, Clone)]
#[serde(untagged)]
enum CudaArtifactsByTag {
    Single(CudaArtifact),
    Many {
        #[serde(flatten)]
        by_tag: HashMap<String, CudaArtifact>,
    },
}
impl IntoIterator for CudaArtifactsByTag {
    type Item = (Option<String>, CudaArtifact);
    type IntoIter = std::vec::IntoIter<Self::Item>;

    fn into_iter(self) -> std::vec::IntoIter<Self::Item> {
        match self {
            CudaArtifactsByTag::Single(art) => vec![(None, art)].into_iter(),
            CudaArtifactsByTag::Many { by_tag: by_compat } => by_compat
                .iter()
                .map(|(k, x)| (Some(k.clone()), x.clone()))
                .collect::<Vec<Self::Item>>()
                .into_iter(),
        }
    }
}
#[derive(Serialize, Deserialize, Debug, Clone)]
#[serde(untagged)]
enum CudaArtifactsByPlatform {
    Binary {
        #[serde(flatten)]
        by_platform: HashMap<String, CudaArtifactsByTag>,
    },
    Source {
        source: CudaArtifact,
    },
}

impl IntoIterator for CudaArtifactsByPlatform {
    type Item = (String, Option<String>, CudaArtifact);

    /* TODO: Figure out which is the trait that doesn't involve copying */
    type IntoIter = std::vec::IntoIter<(String, Option<String>, CudaArtifact)>;

    fn into_iter(self) -> Self::IntoIter {
        match self {
            CudaArtifactsByPlatform::Binary { by_platform } => by_platform
                .iter()
                .flat_map(|(platform, by_tag)| {
                    by_tag
                        .clone()
                        .into_iter()
                        .map(|(tag, artifact)| (platform.clone(), tag.clone(), artifact))
                })
                .collect::<Vec<Self::Item>>()
                .into_iter(),
            CudaArtifactsByPlatform::Source { source } => {
                (vec![("source".to_string(), None, source)]).into_iter()
            }
        }
    }
}

#[derive(Serialize, Deserialize, Debug)]
struct CudaJsonPackage {
    name: Option<String>,
    license: String,
    license_path: Option<String>,
    version: String,

    cuda_variant: Option<Vec<String>>,

    #[serde(flatten)]
    artifacts: CudaArtifactsByPlatform,
}

#[derive(Serialize, Deserialize, Debug)]
struct CudaJsonManifest {
    release_date: Option<String>,
    release_label: Option<String>,
    release_product: Option<String>,

    #[serde(flatten)]
    by_pname: HashMap<String, CudaJsonPackage>,
}

#[tokio::main]
async fn main() {
    let args = Cli::parse();

    let _cwd = std::env::current_dir().expect("Couldn't get CWD");
    let _host_name = std::env::var("HOSTNAME").map_or(None, Some);

    let ctx = Arc::new(
        open_context(
            args.refetch,
            args.max_parallel,
            args.db_path,
            args.castore_path,
        )
        .await,
    );

    match args.command {
        Some(Command::Ingest { inputs }) => {
            let samples = ctx.ingest(&inputs).await;
            for s in samples {
                match s {
                    Err(e) => {
                        eprintln!("Failed to fetch: {}", e);
                    }
                    Ok(None) => {}
                    Ok(Some(ingested)) => {
                        eprintln!("{:?}", ingested)
                    }
                }
            }
        }
        Some(Command::FetchListing {
            max_depth,
            html_max_bytes,
            inputs,
        }) => {
            let ingested: Vec<Sampled> = stream::iter(inputs)
                .then(async |i| {
                    let i = i.clone();
                    ctx.clone()
                        .fetch_from_listing(i, max_depth, html_max_bytes)
                        .await
                })
                .flatten_unordered(args.max_parallel)
                .collect()
                .await;
            for i in ingested {
                eprintln!("{:?}", i);
            }
        }
        Some(Command::ParseUrl { url: urls }) => {
            for url in urls {
                println!("{:?}", url);
            }
        }
        Some(Command::FormatCudaManifest) => {
            println!(
                "{}",
                serde_json::to_string(
                    &serde_json::from_reader::<_, CudaJsonManifest>(io::stdin()).unwrap()
                )
                .unwrap()
            );
        }
        Some(Command::DemoCudaManifest) => {
            println!(
                "{}",
                serde_json::to_string(&CudaJsonManifest {
                    release_date: Some("1984-01-01".to_string()),
                    release_label: Some("8.9.x".to_string()),
                    release_product: Some("cudnn".to_string()),
                    by_pname: HashMap::from([
                        (
                        "cudnn".to_string(),
                        CudaJsonPackage {
                            name: Some("cuDNN Library".to_string()),
                            license: "cudnn".to_string(),
                            license_path: Some("bar/foo".to_string()),
                            version: "8.9.7.6".to_string(),
                            cuda_variant: Some(vec!["11".to_string(), "12".to_string()]),
                            artifacts: CudaArtifactsByPlatform::Binary {
                                by_platform: HashMap::from([
                                    ("x86_64-linux".to_string(),
                                        CudaArtifactsByTag::Many {
                                            by_tag:
                                                HashMap::from([
                                                    ("cuda11"
                                                     .to_string(),
                                                     CudaArtifact{
                                                         relative_path:
                                                         "kek".to_string(),
                                                         sha256: "2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824".to_string(),
                                                         md5: Some("5d41402abc4b2a76b9719d911017c592".to_string()),
                                                         size: 5 })])})

                                ]),
                            }
                        }),
                        (
                        "cuda_samples".to_string(),
                        CudaJsonPackage {
                            name: Some("NVIDIA cuDNN samples".to_string()),
                            license: "cudnn".to_string(),
                            license_path: Some("foo/bar".to_string()),
                            version: "8.9.7.6".to_string(),
                            cuda_variant: None,
                            artifacts: CudaArtifactsByPlatform::Source {
                                source: CudaArtifact {
                                    relative_path: "/biba/boba/fifa".to_string(),
                                    sha256: "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855".to_string(),
                                    md5: Some("d41d8cd98f00b204e9800998ecf8427e".to_string()),
                                    size: 0,
                                }
                            }
                        }
                    ),])
                })
                .unwrap()
            );
        }
        Some(Command::ProcessCudaManifests { include_finished }) => {
            let manifests: Vec<(String, String, Option<u64>)> = {
                let con = ctx.con.lock().unwrap();
                con.execute_batch(include_str!("q/cuda-init.sql"))
                    .context("q/cuda-init.sql")
                    .unwrap();
                let mut find_manifests = con
                    .prepare_cached(include_str!("q/find-cuda-manifests.sql"))
                    .context("q/find-cuda-manifests.sql")
                    .unwrap();
                find_manifests
                    .query(named_params! {":include_finished": include_finished})
                    .context("q/find-cuda-manifests.sql")
                    .unwrap()
                    .map(|row| <(String, String, Option<u64>)>::try_from(row))
                    .collect()
                    .expect("Casting result of q/find-cuda-manifests.sql")
            };
            for m in &manifests {
                let b64 = m.1.clone();
                let b3 = match B3Digest::from_str(&b64) {
                    Ok(b3) => b3,
                    Err(e) => {
                        eprintln!("Invalid hash recorded for {:?}: {}", m, e);
                        continue;
                    }
                };
                if let Err(e) = ctx.ensure_blob(&b3).await {
                    eprintln!("Couldn't provision the blob for {:?}: {}", m, e);
                    continue;
                };
                let json = {
                    let mut reader = match ctx.blob_service.open_read(&b3).await {
                        Ok(Some(reader)) => reader,
                        Ok(None) => {
                            eprintln!("Blob doesn't exist after ensure_blob: {:?}", m);
                            continue;
                        }
                        Err(e) => {
                            eprintln!("Couldn't query the blob for {:?}: {}", m, e);
                            continue;
                        }
                    };
                    let mut json = String::new();
                    match reader.read_to_string(&mut json).await {
                        Ok(_) => (),
                        Err(e) => {
                            eprintln!("Couldn't read blob {:?}: {:?}", m, e);
                            continue;
                        }
                    };
                    json
                };
                let parsed: CudaJsonManifest = match serde_json::from_str(&json) {
                    Ok(m) => m,
                    Err(e) => {
                        eprintln!("Couldn't parse JSON for {:?}: {:?}", m, e);
                        continue;
                    }
                };
                {
                    let mut lock = ctx.con.lock().unwrap();
                    let tx = lock.transaction().unwrap();
                    {
                        let mut add_str = tx
                            .prepare_cached(include_str!("q/add-str.sql"))
                            .context("q/add-str.sql")
                            .unwrap();
                        let mut add_hash = tx
                            .prepare_cached(include_str!("q/upsert-blob.sql"))
                            .context("q/upsert-blob.sql")
                            .unwrap();
                        let mut add_manifest = tx
                            .prepare_cached(include_str!("q/add-cuda-manifest.sql"))
                            .context("q/add-cuda-manifest.sql")
                            .unwrap();
                        let mut add_comp = tx
                            .prepare_cached(include_str!("q/add-cuda-artifact.sql"))
                            .context("q/add-cuda-artifact.sql")
                            .unwrap();

                        add_hash.execute(params![b64, None::<usize>]).unwrap();
                        for s in vec![
                            &parsed.release_date,
                            &parsed.release_label,
                            &parsed.release_product,
                        ] {
                            add_str.execute((s,)).unwrap();
                        }
                        add_manifest
                            .execute(named_params! {
                                ":hash": b64,
                                ":release_date": parsed.release_date,
                                ":release_label": parsed.release_label,
                                ":release_product": parsed.release_product,
                            })
                            .context("Executing q/add-cuda-manifest.sql")
                            .unwrap();

                        for (pname, pkg) in parsed.by_pname {
                            for (platform, maybe_tag, comp) in pkg.artifacts.into_iter() {
                                let ps = named_params! {
                                    ":manifest": b64,
                                    ":name": pkg.name,
                                    ":pname": pname,
                                    ":license_name": pkg.license,
                                    ":license_path": pkg.license_path,
                                    ":version": pkg.version,
                                    ":sha256": comp.sha256,
                                    ":md5": comp.md5,
                                    ":platform": platform,
                                    ":relative_path": comp.relative_path,
                                    ":n_bytes": comp.size,
                                    ":compat_tag": maybe_tag
                                };
                                for h in &vec![Some(&comp.sha256), comp.md5.as_ref()] {
                                    add_hash.execute(params![h, None::<usize>]).unwrap();
                                }
                                for s in &vec![
                                    Some(&pname),
                                    pkg.name.as_ref(),
                                    Some(&pkg.license),
                                    pkg.license_path.as_ref(),
                                    Some(&pkg.version),
                                    Some(&platform.to_string()),
                                    Some(&comp.relative_path),
                                    maybe_tag.as_ref(),
                                ] {
                                    add_str.execute(params![s]).unwrap();
                                }
                                add_comp
                                    .execute(ps)
                                    .context("Executing q/add-cuda-artifact.sql")
                                    .unwrap();
                            }
                        }
                    }
                    tx.commit()
                        .expect("Couldn't commit transaction adding manifest or its component");
                }
            }
        }
        None => {}
    }
}