//! Trees: sorted directory listings. A flattened tree ([`Files`]) maps
//! `a/b/c.txt` paths to file entries.

use std::collections::{BTreeMap, HashMap};

use crate::object::{Kind, Object};
use crate::{Id, Result, Source, corrupt};

#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
#[repr(u8)]
pub enum EntryKind {
    File = 1,
    Exec = 2,
    Link = 3,
    Tree = 4,
}

impl EntryKind {
    pub fn from_u8(b: u8) -> Option<EntryKind> {
        match b {
            1 => Some(EntryKind::File),
            2 => Some(EntryKind::Exec),
            3 => Some(EntryKind::Link),
            4 => Some(EntryKind::Tree),
            _ => None,
        }
    }

    pub fn name(self) -> &'static str {
        match self {
            EntryKind::File => "file",
            EntryKind::Exec => "exec",
            EntryKind::Link => "link",
            EntryKind::Tree => "tree",
        }
    }
}

/// A file in a flattened tree. `id` names a blob or, for big files, a chunk list.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct Entry {
    pub kind: EntryKind,
    pub id: Id,
}

pub type Files = BTreeMap<String, Entry>;

#[derive(Debug, Clone, PartialEq, Eq)]
pub struct TreeEntry {
    pub name: String,
    pub kind: EntryKind,
    pub id: Id,
}

pub fn encode(entries: &[TreeEntry]) -> Vec<u8> {
    let mut sorted: Vec<&TreeEntry> = entries.iter().collect();
    sorted.sort_by(|a, b| a.name.as_bytes().cmp(b.name.as_bytes()));
    let mut out = Vec::with_capacity(entries.len() * 48);
    for e in sorted {
        out.push(e.kind as u8);
        out.extend_from_slice(&(e.name.len() as u16).to_le_bytes());
        out.extend_from_slice(e.name.as_bytes());
        out.extend_from_slice(&e.id.0);
    }
    out
}

pub fn decode(data: &[u8]) -> Result<Vec<TreeEntry>> {
    let mut out = Vec::new();
    let mut at = 0;
    while at < data.len() {
        let kind = EntryKind::from_u8(data[at]).ok_or_else(|| corrupt("bad tree entry kind"))?;
        let len = u16::from_le_bytes(data.get(at + 1..at + 3).ok_or_else(|| corrupt("short tree"))?.try_into().unwrap()) as usize;
        let name_end = at + 3 + len;
        let name = std::str::from_utf8(data.get(at + 3..name_end).ok_or_else(|| corrupt("short tree"))?)
            .map_err(|_| corrupt("tree name isn't UTF-8"))?
            .to_string();
        if name.is_empty() || name.contains('/') || name == "." || name == ".." {
            return Err(corrupt("bad name in tree"));
        }
        let id = Id::from_slice(data.get(name_end..name_end + 32).ok_or_else(|| corrupt("short tree"))?).unwrap();
        out.push(TreeEntry { name, kind, id });
        at = name_end + 32;
    }
    Ok(out)
}

pub fn empty() -> Object {
    Object::new(Kind::Tree, Vec::new())
}

pub fn empty_id() -> Id {
    empty().id()
}

pub fn read(src: &dyn Source, id: &Id) -> Result<Vec<TreeEntry>> {
    let object = src.must(id)?;
    if object.kind != Kind::Tree {
        return Err(corrupt(format!("{} isn't a tree", id.short())));
    }
    decode(&object.data)
}

/// Every file under `tree`.
pub fn flatten(src: &dyn Source, tree: &Id) -> Result<Files> {
    let mut out = Files::new();
    walk(src, tree, "", &mut out)?;
    Ok(out)
}

fn walk(src: &dyn Source, tree: &Id, prefix: &str, out: &mut Files) -> Result<()> {
    for e in read(src, tree)? {
        let path = if prefix.is_empty() { e.name.clone() } else { format!("{prefix}/{}", e.name) };
        if e.kind == EntryKind::Tree {
            walk(src, &e.id, &path, out)?;
        } else {
            out.insert(path, Entry { kind: e.kind, id: e.id });
        }
    }
    Ok(())
}

/// Every object ID a tree reaches (the trees themselves, file blobs and
/// chunk lists, not the chunks inside chunk lists).
pub fn reachable(src: &dyn Source, tree: &Id, out: &mut Vec<Id>) -> Result<()> {
    out.push(*tree);
    for e in read(src, tree)? {
        if e.kind == EntryKind::Tree {
            reachable(src, &e.id, out)?;
        } else {
            out.push(e.id);
        }
    }
    Ok(())
}

/// Trees reached from `new` but not from `old`, walking only subtrees that
/// differ, plus the file IDs those subtrees hold that `old` doesn't have at
/// the same path.
pub fn new_since(src: &dyn Source, old: Option<&Id>, new: &Id, out: &mut Vec<Id>) -> Result<()> {
    if Some(new) == old {
        return Ok(());
    }
    out.push(*new);
    let old_entries: HashMap<String, TreeEntry> = match old {
        Some(o) => read(src, o)?.into_iter().map(|e| (e.name.clone(), e)).collect(),
        None => HashMap::new(),
    };
    for e in read(src, new)? {
        let before = old_entries.get(&e.name);
        if before.is_some_and(|b| b.id == e.id) {
            continue;
        }
        if e.kind == EntryKind::Tree {
            let old_sub = before.filter(|b| b.kind == EntryKind::Tree).map(|b| b.id);
            new_since(src, old_sub.as_ref(), &e.id, out)?;
        } else {
            out.push(e.id);
        }
    }
    Ok(())
}

/// Builds tree objects for `files`. Returns the root ID and every tree
/// object (the host stores the ones it doesn't have).
pub fn build(files: &Files) -> (Id, Vec<Object>) {
    let mut dirs: HashMap<&str, Vec<TreeEntry>> = HashMap::new();
    let mut subdirs: HashMap<&str, Vec<&str>> = HashMap::new();
    dirs.entry("").or_default();
    for (path, entry) in files {
        let (dir, name) = split(path);
        dirs.entry(dir).or_default().push(TreeEntry { name: name.to_string(), kind: entry.kind, id: entry.id });
        let mut d = dir;
        while !d.is_empty() {
            let (parent, _) = split(d);
            let known = subdirs.entry(parent).or_default();
            if known.contains(&d) {
                break;
            }
            known.push(d);
            dirs.entry(d).or_default();
            d = parent;
        }
    }
    let mut order: Vec<&str> = dirs.keys().copied().collect();
    order.sort_by_key(|d| std::cmp::Reverse(if d.is_empty() { 0 } else { d.matches('/').count() + 1 }));
    let mut written: HashMap<&str, Id> = HashMap::with_capacity(order.len());
    let mut objects = Vec::with_capacity(order.len());
    for dir in order {
        let mut entries = dirs.remove(dir).unwrap_or_default();
        for child in subdirs.get(dir).map(Vec::as_slice).unwrap_or(&[]) {
            entries.push(TreeEntry { name: split(child).1.to_string(), kind: EntryKind::Tree, id: written[child] });
        }
        let object = Object::new(Kind::Tree, encode(&entries));
        let id = object.id();
        objects.push(object);
        written.insert(dir, id);
    }
    (written[""], objects)
}

/// "a/b/c" → ("a/b", "c"); "c" → ("", "c").
pub fn split(path: &str) -> (&str, &str) {
    match path.rfind('/') {
        Some(i) => (&path[..i], &path[i + 1..]),
        None => ("", path),
    }
}

/// Whether `path` is `prefix` or inside it.
pub fn under(path: &str, prefix: &str) -> bool {
    prefix.is_empty() || path == prefix || (path.starts_with(prefix) && path.as_bytes().get(prefix.len()) == Some(&b'/'))
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Change {
    Added(Entry),
    Removed(Entry),
    Modified(Entry, Entry),
}

/// What changed from `a` to `b`, by path.
pub fn diff(a: &Files, b: &Files) -> BTreeMap<String, Change> {
    let mut out = BTreeMap::new();
    for (path, old) in a {
        match b.get(path) {
            None => {
                out.insert(path.clone(), Change::Removed(*old));
            }
            Some(new) if new != old => {
                out.insert(path.clone(), Change::Modified(*old, *new));
            }
            _ => {}
        }
    }
    for (path, new) in b {
        if !a.contains_key(path) {
            out.insert(path.clone(), Change::Added(*new));
        }
    }
    out
}

/// A file to write into a tree: its path, kind and content.
pub struct Put {
    pub path: String,
    pub kind: EntryKind,
    pub content: Vec<u8>,
}

/// Whether `path` can name a file: relative, with no empty, `.` or `..`
/// parts, and no NUL.
pub fn valid_path(path: &str) -> bool {
    !path.is_empty() && !path.contains('\0') && path.split('/').all(|p| !p.is_empty() && p != "." && p != ".." && p.len() <= u16::MAX as usize)
}

/// `tree` with files written and deleted. Writing a path replaces whatever
/// was there (a folder of that name included); deleting a folder deletes
/// what's in it. Returns the new root and every object it needs that's new
/// (file contents, big files' chunks and trees).
pub fn edit(src: &dyn Source, tree: &Id, put: &[Put], delete: &[String]) -> Result<std::result::Result<(Id, Vec<Object>), String>> {
    if let Some(bad) = put.iter().map(|p| p.path.as_str()).chain(delete.iter().map(String::as_str)).find(|p| !valid_path(p)) {
        return Ok(Err(bad.to_string()));
    }
    let mut files = flatten(src, tree)?;
    for d in delete {
        files.retain(|p, _| !under(p, d));
    }
    let mut objects = Vec::new();
    for p in put {
        // A file and a folder can't share a path.
        files.retain(|q, _| !under(q, &p.path) && !under(&p.path, q));
        let parts = crate::chunk::file_objects(&p.content);
        files.insert(p.path.clone(), Entry { kind: p.kind, id: parts[0].id() });
        objects.extend(parts);
    }
    let (root, trees) = build(&files);
    objects.extend(trees);
    Ok(Ok((root, objects)))
}

/// What changed from tree `a` to tree `b`, by path, reading only the
/// folders that differ (the same answer as `diff` on both flattened).
pub fn diff_trees(src: &dyn Source, a: &Id, b: &Id) -> Result<BTreeMap<String, Change>> {
    let mut out = BTreeMap::new();
    diff_into(src, Some(a), Some(b), "", &mut out)?;
    Ok(out)
}

fn diff_into(src: &dyn Source, a: Option<&Id>, b: Option<&Id>, prefix: &str, out: &mut BTreeMap<String, Change>) -> Result<()> {
    if a == b {
        return Ok(());
    }
    let entries = |t: Option<&Id>| -> Result<BTreeMap<String, TreeEntry>> {
        Ok(match t {
            Some(t) => read(src, t)?.into_iter().map(|e| (e.name.clone(), e)).collect(),
            None => BTreeMap::new(),
        })
    };
    let (ea, eb) = (entries(a)?, entries(b)?);
    let path = |name: &str| if prefix.is_empty() { name.to_string() } else { format!("{prefix}/{name}") };
    let entry = |e: &TreeEntry| Entry { kind: e.kind, id: e.id };
    for (name, x) in &ea {
        let p = path(name);
        match eb.get(name) {
            Some(y) if x.kind == y.kind && x.id == y.id => {}
            Some(y) => match (x.kind == EntryKind::Tree, y.kind == EntryKind::Tree) {
                (true, true) => diff_into(src, Some(&x.id), Some(&y.id), &p, out)?,
                (true, false) => {
                    diff_into(src, Some(&x.id), None, &p, out)?;
                    out.insert(p, Change::Added(entry(y)));
                }
                (false, true) => {
                    out.insert(p.clone(), Change::Removed(entry(x)));
                    diff_into(src, None, Some(&y.id), &p, out)?;
                }
                (false, false) => {
                    out.insert(p, Change::Modified(entry(x), entry(y)));
                }
            },
            None if x.kind == EntryKind::Tree => diff_into(src, Some(&x.id), None, &p, out)?,
            None => {
                out.insert(p, Change::Removed(entry(x)));
            }
        }
    }
    for (name, y) in &eb {
        if ea.contains_key(name) {
            continue;
        }
        let p = path(name);
        if y.kind == EntryKind::Tree {
            diff_into(src, None, Some(&y.id), &p, out)?;
        } else {
            out.insert(p, Change::Added(entry(y)));
        }
    }
    Ok(())
}

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

    #[test]
    fn diff_trees_matches_flat_diff() {
        let mut src = MemSource::default();
        let mut put = |list: &[(&str, &str, EntryKind)]| -> Id {
            let files: Files = list.iter().map(|(p, c, k)| (p.to_string(), Entry { kind: *k, id: src.put(Object::blob(c.as_bytes().to_vec())) })).collect();
            let (root, trees) = build(&files);
            for t in trees {
                src.put(t);
            }
            root
        };
        let f = EntryKind::File;
        let a = put(&[("a.txt", "1", f), ("dir/x", "x", f), ("dir/sub/y", "y", f), ("same/z", "z", f), ("gone/q", "q", f), ("file_then_dir", "f", f), ("run", "r", f)]);
        let b = put(&[("a.txt", "2", f), ("dir/x", "x", f), ("dir/sub/y2", "y", f), ("same/z", "z", f), ("new/n", "n", f), ("file_then_dir/inner", "i", f), ("run", "r", EntryKind::Exec)]);
        let flat = diff(&flatten(&src, &a).unwrap(), &flatten(&src, &b).unwrap());
        assert_eq!(diff_trees(&src, &a, &b).unwrap(), flat);
        assert!(diff_trees(&src, &a, &a).unwrap().is_empty());
    }

    #[test]
    fn edit_writes_and_deletes() {
        let mut src = MemSource::default();
        let files: Files = [("a.txt", "1"), ("dir/x", "x"), ("dir/y", "y"), ("f", "f")]
            .iter()
            .map(|(p, c)| (p.to_string(), Entry { kind: EntryKind::File, id: src.put(Object::blob(c.as_bytes().to_vec())) }))
            .collect();
        let (root, trees) = build(&files);
        for t in trees {
            src.put(t);
        }
        let put = vec![
            Put { path: "a.txt".into(), kind: EntryKind::File, content: b"2".to_vec() },
            Put { path: "new/deep/n.sh".into(), kind: EntryKind::Exec, content: b"#!/bin/sh".to_vec() },
            Put { path: "f/inside".into(), kind: EntryKind::File, content: b"i".to_vec() },
        ];
        let (out, objects) = edit(&src, &root, &put, &["dir".to_string()]).unwrap().unwrap();
        for o in objects {
            src.put(o);
        }
        let got = flatten(&src, &out).unwrap();
        assert_eq!(got.keys().cloned().collect::<Vec<_>>(), vec!["a.txt", "f/inside", "new/deep/n.sh"]);
        assert_eq!(got["new/deep/n.sh"].kind, EntryKind::Exec);
        assert_eq!(edit(&src, &root, &[], &["../x".to_string()]).unwrap().unwrap_err(), "../x");
    }

    fn files(src: &mut MemSource, list: &[(&str, &str)]) -> Files {
        list.iter()
            .map(|(p, c)| (p.to_string(), Entry { kind: EntryKind::File, id: src.put(Object::blob(c.as_bytes().to_vec())) }))
            .collect()
    }

    #[test]
    fn build_and_flatten() {
        let mut src = MemSource::default();
        let f = files(&mut src, &[("a.txt", "a"), ("src/b.rs", "b"), ("src/x/c.rs", "c"), ("src.rs", "d")]);
        let (root, trees) = build(&f);
        for t in trees {
            src.put(t);
        }
        assert_eq!(flatten(&src, &root).unwrap(), f);
        let (empty, _) = build(&Files::new());
        assert_eq!(empty, empty_id());
    }

    #[test]
    fn new_objects_only() {
        let mut src = MemSource::default();
        let a = files(&mut src, &[("a.txt", "a"), ("src/b.rs", "b"), ("docs/x.md", "x")]);
        let (ra, ta) = build(&a);
        for t in ta {
            src.put(t);
        }
        let mut b = a.clone();
        b.insert("src/b.rs".into(), Entry { kind: EntryKind::File, id: src.put(Object::blob(b"b2".to_vec())) });
        let (rb, tb) = build(&b);
        for t in tb {
            src.put(t);
        }
        let mut out = Vec::new();
        new_since(&src, Some(&ra), &rb, &mut out).unwrap();
        // The root, src/ and the new blob; docs/ and a.txt are shared.
        assert_eq!(out.len(), 3);
    }

    #[test]
    fn under_paths() {
        assert!(under("src/a.rs", "src"));
        assert!(under("src", "src"));
        assert!(!under("srcx/a.rs", "src"));
        assert!(under("anything", ""));
    }
}