use gix::ObjectId;
use gix::bstr::ByteSlice as _;
use gix::objs::tree::EntryKind;
use line_numbers::{LineNumber, SingleLineSpan};
use rustc_hash::FxHashSet;
use tracing::info;

/// Blobs larger than this are treated as binary.
const LARGE_FILE_THRESHOLD_BYTES: u64 = 1024 * 1024;

#[derive(Debug)]
pub struct FileContent {
    pub(crate) path: String,
    pub(crate) oid: String,
    pub(crate) size: usize,
    pub(crate) is_binary: bool,
    pub(crate) is_submodule: bool,
    /// Raw file bytes, inlined when the `oid` is not fetchable by the client (interdiff's
    /// synthetic rebased tree). `None` for ordinary diffs and for binary/oversized blobs.
    pub(crate) content: Option<Vec<u8>>,
}

impl From<FileContent> for crate::protocol::v1::FileContent {
    fn from(f: FileContent) -> Self {
        Self {
            path: f.path,
            oid: f.oid,
            size: f.size as u64,
            is_binary: f.is_binary,
            is_submodule: f.is_submodule,
            content: f.content,
        }
    }
}

#[derive(Debug, Clone)]
pub struct Hunk {
    pub(crate) novel_lhs: FxHashSet<LineNumber>,
    pub(crate) novel_rhs: FxHashSet<LineNumber>,
    pub(crate) lines: Vec<(Option<LineNumber>, Option<LineNumber>)>,
}

impl From<Hunk> for crate::protocol::v1::Hunk {
    fn from(h: Hunk) -> Self {
        // Sort the novel line sets for deterministic output.
        let sorted = |set: FxHashSet<LineNumber>| -> Vec<u32> {
            let mut v: Vec<u32> = set.into_iter().map(|n| n.0).collect();
            v.sort_unstable();
            v
        };
        Self {
            novel_lhs: sorted(h.novel_lhs),
            novel_rhs: sorted(h.novel_rhs),
            lines: h
                .lines
                .into_iter()
                .map(|(lhs, rhs)| crate::protocol::v1::LinePair {
                    lhs: lhs.map(|n| n.0),
                    rhs: rhs.map(|n| n.0),
                })
                .collect(),
        }
    }
}

/// A matched token (an atom, a delimiter, or a comment word).
#[derive(PartialEq, Eq, Debug, Clone)]
pub enum MatchKind {
    // TBD
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub struct MatchedPos {
    pub(crate) kind: MatchKind,
    pub(crate) pos: SingleLineSpan,
}

#[derive(Debug)]
pub struct Diff {
    pub(crate) lhs_src: FileContent,
    pub(crate) rhs_src: FileContent,
    pub(crate) hunks: Vec<Hunk>,

    #[allow(unused)]
    pub(crate) lhs_positions: Vec<MatchedPos>,
    #[allow(unused)]
    pub(crate) rhs_positions: Vec<MatchedPos>,

    /// If the two files do not have exactly the same bytes, the
    /// number of bytes in each file.
    pub(crate) has_byte_changes: Option<(usize, usize)>,
    pub(crate) has_syntactic_changes: bool,
}

impl From<Diff> for crate::protocol::v1::FileDiff {
    fn from(d: Diff) -> Self {
        Self {
            lhs_src: Some(d.lhs_src.into()),
            rhs_src: Some(d.rhs_src.into()),
            hunks: d.hunks.into_iter().map(Into::into).collect(),
            has_byte_changes: d.has_byte_changes.map(|(lhs, rhs)| {
                crate::protocol::v1::ByteChanges {
                    lhs: lhs as u64,
                    rhs: rhs as u64,
                }
            }),
            has_syntactic_changes: d.has_syntactic_changes,
        }
    }
}

/// Diff two trees, yielding one [`Diff`] per changed file (line-level [`Hunk`]s included).
///
/// Returns a lazy iterator of `anyhow::Result<Diff>`.
pub fn diff<'repo>(
    repo: &'repo gix::Repository,
    old_tree: &gix::Tree<'_>,
    new_tree: &gix::Tree<'_>,
    embed_content: bool,
) -> anyhow::Result<DiffIter<'repo>> {
    let changes = repo.diff_tree_to_tree(Some(old_tree), Some(new_tree), None)?;
    let mut cache = repo.diff_resource_cache_for_tree_diff()?;
    cache.filter.options.large_file_threshold_bytes = LARGE_FILE_THRESHOLD_BYTES;
    Ok(DiffIter {
        repo,
        changes: changes.into_iter(),
        cache,
        embed_content,
    })
}

/// Lazy iterator over per-file [`Diff`]s.
pub struct DiffIter<'repo> {
    repo: &'repo gix::Repository,
    changes: std::vec::IntoIter<gix::object::tree::diff::ChangeDetached>,
    cache: gix::diff::blob::Platform,
    embed_content: bool,
}

impl Iterator for DiffIter<'_> {
    type Item = anyhow::Result<Diff>;

    fn next(&mut self) -> Option<Self::Item> {
        loop {
            let change = self.changes.next()?;
            let (old, new) = change_sides(&change);
            // Trees are recursed into by gix and never diffed directly; skip defensively.
            if old.is_kind(EntryKind::Tree) || new.is_kind(EntryKind::Tree) {
                continue;
            }
            // Submodules (gitlinks) can't go through the blob pipeline — their commit lives in
            // another repo. Emit the commit oids directly for the caller to compare.
            if old.is_kind(EntryKind::Commit) || new.is_kind(EntryKind::Commit) {
                return Some(Ok(submodule_diff(self.repo, &change, old, new)));
            }
            let result = build_diff(self.repo, &change, &mut self.cache, self.embed_content);
            self.cache.clear_resource_cache_keep_allocation();
            return Some(result);
        }
    }
}

enum Side {
    Absent,
    Present { oid: ObjectId, kind: EntryKind },
}

impl Side {
    fn is_kind(&self, wanted: EntryKind) -> bool {
        matches!(self, Side::Present { kind, .. } if *kind == wanted)
    }
}

/// Decompose a change into its `(old, new)` sides regardless of variant.
fn change_sides(change: &gix::object::tree::diff::ChangeDetached) -> (Side, Side) {
    use gix::object::tree::diff::ChangeDetached as C;
    let present = |oid, mode: gix::object::tree::EntryMode| Side::Present {
        oid,
        kind: mode.kind(),
    };
    match change {
        C::Addition { id, entry_mode, .. } => (Side::Absent, present(*id, *entry_mode)),
        C::Deletion { id, entry_mode, .. } => (present(*id, *entry_mode), Side::Absent),
        C::Modification {
            previous_id,
            previous_entry_mode,
            id,
            entry_mode,
            ..
        } => (
            present(*previous_id, *previous_entry_mode),
            present(*id, *entry_mode),
        ),
        C::Rewrite {
            source_id,
            source_entry_mode,
            id,
            entry_mode,
            ..
        } => (
            present(*source_id, *source_entry_mode),
            present(*id, *entry_mode),
        ),
    }
}

/// Build a [`Diff`] for a submodule (gitlink) change: no hunks, both sides carry their commit
/// oid with `is_submodule` set. The caller compares the oids as submodule pointers.
fn submodule_diff(
    repo: &gix::Repository,
    change: &gix::object::tree::diff::ChangeDetached,
    old: Side,
    new: Side,
) -> Diff {
    let path = change.location().to_string();
    let side = |s: Side| FileContent {
        path: path.clone(),
        oid: match s {
            Side::Absent => ObjectId::null(repo.object_hash()).to_string(),
            Side::Present { oid, .. } => oid.to_string(),
        },
        size: 0,
        is_binary: false,
        is_submodule: true,
        content: None,
    };
    Diff {
        lhs_src: side(old),
        rhs_src: side(new),
        hunks: Vec::new(),
        lhs_positions: Vec::new(),
        rhs_positions: Vec::new(),
        has_byte_changes: None,
        has_syntactic_changes: false,
    }
}

/// Build the [`Diff`] for a single changed file.
fn build_diff(
    repo: &gix::Repository,
    change: &gix::object::tree::diff::ChangeDetached,
    cache: &mut gix::diff::blob::Platform,
    embed_content: bool,
) -> anyhow::Result<Diff> {
    use gix::diff::blob::platform::prepare_diff::Operation;
    use gix::diff::blob::platform::resource::Data;
    use gix::diff::blob::{Diff as BlobDiff, InternedInput};
    use gix::prelude::TreeDiffChangeExt as _;

    let change = change.attach(repo, repo);
    let blob = change.diff(cache)?;
    let out = blob.resource_cache.prepare_diff()?;

    // Per-side file metadata, derived from the prepared resources.
    let side =
        |res: &gix::diff::blob::platform::Resource<'_>, embed_content: bool| -> FileContent {
            let (size, is_binary, content) = match res.data {
                // Only textual buffers carry inlinable bytes; binary/missing sides fall back to oid.
                Data::Buffer { buf, .. } => (buf.len(), false, embed_content.then(|| buf.to_vec())),
                Data::Binary { size } => (size as usize, true, None),
                Data::Missing => (0, false, None),
            };
            FileContent {
                path: res.rela_path.to_string(),
                oid: res.id.to_string(),
                size,
                is_binary,
                is_submodule: false,
                content,
            }
        };
    let mut lhs_src = side(&out.old, embed_content);
    let mut rhs_src = side(&out.new, false);

    let has_byte_changes = if lhs_src.size == rhs_src.size && out.old.id == out.new.id {
        None
    } else {
        Some((lhs_src.size, rhs_src.size))
    };

    let hunks = match out.operation {
        Operation::InternalDiff { algorithm } => {
            let input = InternedInput::new(out.old.intern_source(), out.new.intern_source());
            let mut d = BlobDiff::compute(algorithm, &input);
            d.postprocess_lines(&input);
            d.hunks()
                .map(|h| {
                    let before: Vec<LineNumber> = h.before.clone().map(LineNumber).collect();
                    let after: Vec<LineNumber> = h.after.clone().map(LineNumber).collect();
                    let n = before.len().max(after.len());
                    let lines = (0..n)
                        .map(|i| (before.get(i).copied(), after.get(i).copied()))
                        .collect();
                    Hunk {
                        novel_lhs: before.iter().copied().collect(),
                        novel_rhs: after.iter().copied().collect(),
                        lines,
                    }
                })
                .collect()
        }
        Operation::SourceOrDestinationIsBinary => {
            lhs_src.is_binary = true;
            rhs_src.is_binary = true;
            Vec::new()
        }
        Operation::ExternalCommand { .. } => unreachable!("we disabled that"),
    };

    Ok(Diff {
        lhs_src,
        rhs_src,
        hunks,
        lhs_positions: Vec::new(),
        rhs_positions: Vec::new(),
        has_byte_changes,
        has_syntactic_changes: false,
    })
}

/// Compute the two trees whose diff is the interdiff between an old patch
/// version (`from_base..from_head`) and a new one (`to_base..to_head`).
///
/// Rebase `from_base..from_head` onto `to_base` and return rebased tree so
/// caller can diff between `rebased_tree` and `to_head.tree`
pub fn prepare_interdiff<'repo>(
    repo: &'repo gix::Repository,
    (from_base_id, from_head_id): (ObjectId, ObjectId),
    to_base_id: ObjectId,
) -> anyhow::Result<gix::Tree<'repo>> {
    let from_head = repo.find_commit(from_head_id)?;

    let from_base = repo.find_commit(from_base_id)?;
    let to_base = repo.find_commit(to_base_id)?;

    // Endpoint trees for the 3-way merge.
    let from_base_tree = from_base.tree_id()?.detach();
    let from_head_tree = from_head.tree_id()?.detach();
    let to_base_tree = to_base.tree_id()?.detach();

    let options = repo.tree_merge_options()?;
    let labels = gix::merge::blob::builtin_driver::text::Labels {
        ancestor: Some("from.base".as_bytes().as_bstr()),
        current: Some("to.base".as_bytes().as_bstr()),
        other: Some("from.head".as_bytes().as_bstr()),
    };
    let mut outcome = repo.merge_trees(
        from_base_tree,
        to_base_tree,
        from_head_tree,
        labels,
        options,
    )?;

    if !outcome.conflicts.is_empty() {
        info!(
            conflicts = outcome.conflicts.len(),
            "interdiff rebase produced conflicts; tree contains conflict markers"
        );
    }

    let rebased_id = outcome.tree.write()?.detach();
    let rebased_tree = repo.find_tree(rebased_id)?;
    Ok(rebased_tree)
}