//! Moving the working copy to another state (spec §12.3, §7.7).

use super::*;

#[derive(Debug, Default, Clone, PartialEq, Eq)]

pub struct CheckoutReport {

    pub written: usize,

    pub removed: usize,

    pub conflicted: bool,

    /// Nested repositories that could not be moved, and other remarks.

    pub notes: Vec<String>,

    /// How many changes of the state left behind the new one lacks.

    pub lacks: usize,

}

/// Text files rendered at a time by a checkout: bounds its memory.

const CHECKOUT_BATCH: usize = 2048;

impl Repo {

    /// Move the working copy to the state whose heads are `heads` (spec §12.3):

    /// the working directory is rewritten to that state and a fresh

    /// working-copy change is started on it. The bookmark does not move.

    pub fn checkout(

        &self,

        heads: &[ChangeHash],

        discard: bool,

        command: &str,

    ) -> Result<CheckoutReport> {

        self.checkout_keeping(heads, discard, command, &BTreeSet::new())

    }

    /// [`Repo::checkout`], leaving the nested repositories at

    /// `keep_subrepos` as they are (their pins are carried work).

    pub(crate) fn checkout_keeping(

        &self,

        heads: &[ChangeHash],

        discard: bool,

        command: &str,

        keep_subrepos: &BTreeSet<String>,

    ) -> Result<CheckoutReport> {

        let _wc = self.lock_working_copy()?;

        let _all = crate::timing::phase("checkout");

        // A working copy clean by the manifest needs no snapshot; otherwise a

        // status snapshot records what changed, and rewrites the manifest

        // with the graph key of every file it proves unchanged. Either way

        // the manifest then says, for each file on disk, which graph it

        // renders, so the base state itself is not needed (oxygen#35).

        let changed_paths: Vec<String> = if self.clean_by_manifest()? {

            Vec::new()

        } else {

            let snap = self.snapshot_status(command)?;

            self.refuse_dirty(&snap, discard)?;

            snap.statuses.iter().map(|s| s.path.clone()).collect()

        };

        let (_, wc_now) = self.working_copy()?;

        let old_manifest = self.read_manifest()?.map(|(_, _, m)| m).unwrap_or_default();

        let changed: HashSet<&str> = changed_paths.iter().map(String::as_str).collect();

        // The target's history; its file graphs are built a batch at a time

        // for the files that must be written, and dropped after. It becomes

        // the working copy's base, which the state index holds.

        let mut target = self.load_state_indexed(heads)?;

        let heads = target.maximal(heads);

        let tv = crate::timing::phase("checkout: target view");

        let tview = target.tree().view();

        drop(tv);

        let mut report = CheckoutReport::default();

        // Paths this file system cannot hold side by side (spec §7.7).

        report.notes.extend(

            self.platform

                .problems(tview.entries.keys().map(String::as_str)),

        );

        report.conflicted = tview.is_conflicted();

        report.lacks = self.count_lacking(&wc_now.deps, &target)?;

        // Remove what the base or the working copy has and the target does

        // not. A nested repository is never deleted (spec §7.8).

        let mut dirs_to_prune: BTreeSet<PathBuf> = BTreeSet::new();

        let gone: BTreeSet<&str> = old_manifest

            .entries

            .keys()

            .map(String::as_str)

            .chain(changed.iter().copied())

            .filter(|p| !tview.entries.contains_key(*p))

            .collect();

        for path in gone {

            let dir = self.root.join(path);

            if dir.join(OXY_DIR).exists() || dir.join(".git").exists() {

                report

                    .notes

                    .push(format!("{path}: nested repository left in place"));

                continue;

            }

            let full = self.safe_path(path)?;

            if fs::symlink_metadata(&full).is_err() {

                continue;

            }

            fs::remove_file(&full)?;

            report.removed += 1;

            if let Some(parent) = full.parent() {

                dirs_to_prune.insert(parent.to_path_buf());

            }

        }

        // The manifest after the checkout, built from what is known.

        let plan = crate::timing::phase("checkout: plan");

        let mut entries: BTreeMap<String, FsEntry> = BTreeMap::new();

        let hex = |b: &[u8]| *blake3::hash(b).as_bytes();

        // Text files to render, attribute files first (the others need them).

        let mut texts: Vec<(String, NodeId, Mode)> = Vec::new();

        for (path, e) in tview.entries.iter() {

            if e.kind == NodeKind::Dir || !self.in_scope(path) {

                continue;

            }

            // The manifest entry of a file the working copy has not changed,

            // with the right executable bit.

            let exec_ok =

                |m: &prompt::ManifestEntry| !self.platform.exec_bits || m.exec == e.mode.exec;

            let entry = old_manifest

                .entries

                .get(path)

                .filter(|m| !changed.contains(path.as_str()) && exec_ok(m));

            if e.kind == NodeKind::File && !e.mode.opaque {

                // The manifest proves the file on disk equal to the rendering

                // of a graph whose key is the target's: nothing to read,

                // render or write.

                let key = target.graph_key(&e.node);

                let proven = |m: &&prompt::ManifestEntry| {

                    !is_attributes(path) && key.is_some() && m.graph == key

                };

                match entry.filter(proven) {

                    Some(m) => {

                        entries.insert(

                            path.clone(),

                            FsEntry {

                                content_hash: Some(m.hash),

                                graph_key: m.graph,

                                unchanged: true,

                                stat: Some((m.size, m.mtime_ns)),

                                ..FsEntry::new(NodeKind::File, e.mode.exec, Vec::new())

                            },

                        );

                    }

                    None => texts.push((path.clone(), e.node, e.mode)),

                }

                continue;

            }

            // Opaque files, symlinks and subrepos: their blobs.

            let blobs = target.live_blobs(&e.node);

            if blobs.len() > 1 {

                report.conflicted = true;

            }

            let Some(blob) = blobs.first().copied() else {

                continue;

            };

            if e.kind == NodeKind::Subrepo {

                if keep_subrepos.contains(path) {

                    continue;

                }

                let pin = crate::subrepo::Pin::decode(&self.read_blob(&blob)?)?;

                let sync = self.sync_subrepo(path, &pin, false)?;

                if matches!(

                    sync,

                    SubrepoSync::Updated | SubrepoSync::KeptBookmark { .. }

                ) {

                    report.written += 1;

                }

                report.notes.extend(sync.note(path));

                continue;

            }

            if let Some(m) = entry

                .filter(|m| e.kind == NodeKind::File && blobs.len() == 1 && m.blob == Some(blob))

            {

                entries.insert(

                    path.clone(),

                    FsEntry {

                        content_hash: Some(m.hash),

                        blob: (e.kind == NodeKind::File).then_some(blob),

                        stat: Some((m.size, m.mtime_ns)),

                        ..FsEntry::new(e.kind, e.mode.exec, Vec::new())

                    },

                );

                continue;

            }

            let bytes = self.read_blob(&blob)?;

            let full = self.root.join(path);

            if e.kind == NodeKind::Symlink {

                let same = fs::read_link(&full)

                    .is_ok_and(|t| t.to_string_lossy().as_bytes() == bytes.as_slice());

                if !same {

                    self.write_symlink(path, &bytes)?;

                    report.written += 1;

                }

                entries.insert(

                    path.clone(),

                    FsEntry {

                        content_hash: Some(hex(&bytes)),

                        ..FsEntry::new(NodeKind::Symlink, false, Vec::new())

                    },

                );

            } else {

                if !self.same_on_disk(&full, &bytes, e.mode.exec) {

                    self.write_file(path, &bytes, e.mode.exec)?;

                    report.written += 1;

                }

                entries.insert(

                    path.clone(),

                    FsEntry {

                        content_hash: Some(hex(&bytes)),

                        blob: (blobs.len() == 1).then_some(blob),

                        ..FsEntry::new(NodeKind::File, e.mode.exec, Vec::new())

                    },

                );

            }

            for (i, b) in blobs.iter().enumerate().skip(1) {

                self.write_file(

                    &format!("{path}.oxy-conflict-{}", i + 1),

                    &self.read_blob(b)?,

                    e.mode.exec,

                )?;

            }

        }

        drop(plan);

        let _render = crate::timing::phase("checkout: render and write");

        // Render and write text files a batch at a time; attribute files

        // come first, since the others' line endings and locks follow them.

        texts.sort_by_key(|(p, _, _)| !is_attributes(p));

        let keys: HashMap<NodeId, [u8; 32]> = texts

            .iter()

            .filter_map(|(_, n, _)| Some((*n, target.graph_key(n)?)))

            .collect();

        let mut attrs = crate::attributes::Attributes::default();

        let attr_count = texts.iter().filter(|(p, _, _)| is_attributes(p)).count();

        let mut done = 0usize;

        let mut batches: Vec<&[(String, NodeId, Mode)]> = Vec::new();

        if attr_count > 0 {

            batches.push(&texts[..attr_count]);

        }

        batches.extend(texts[attr_count..].chunks(CHECKOUT_BATCH));

        for batch in batches {

            use rayon::prelude::*;

            let nodes: Vec<NodeId> = batch.iter().map(|(_, n, _)| *n).collect();

            // A file added once and never edited is its text: no graph.

            let single: Vec<Option<crate::objects::TextRef>> =

                nodes.iter().map(|n| target.single_text(n)).collect();

            let graphs: Vec<NodeId> = nodes

                .iter()

                .zip(&single)

                .filter(|(_, s)| s.is_none())

                .map(|(n, _)| *n)

                .collect();

            let mut texts = {

                let _t = crate::timing::phase("checkout: read single texts");

                let wanted: Vec<crate::objects::TextRef> =

                    single.iter().flatten().copied().collect();

                self.single_texts(&wanted)?.into_iter()

            };

            {

                let _t = crate::timing::phase("checkout: build graphs");

                self.load_graphs_of(&mut target, &graphs)?;

            }

            let rendered: Vec<(Vec<u8>, bool)> = {

                let _t = crate::timing::phase("checkout: render");

                let mut out: Vec<Option<(Vec<u8>, bool)>> = single

                    .iter()

                    .map(|s| s.as_ref().and_then(|_| texts.next().map(|t| (t, false))))

                    .collect();

                let from_graphs: Vec<(Vec<u8>, bool)> = graphs

                    .par_iter()

                    .map(|n| {

                        let (bytes, m) = target.render_file(n);

                        (bytes, m.is_conflicted())

                    })

                    .collect();

                let mut from_graphs = from_graphs.into_iter();

                for o in out.iter_mut().filter(|o| o.is_none()) {

                    *o = from_graphs.next();

                }

                out.into_iter()

                    .map(|o| o.expect("every file rendered"))

                    .collect()

            };

            #[cfg(test)]

            {

                let singles: Vec<NodeId> = nodes

                    .iter()

                    .zip(&single)

                    .filter(|(_, s)| s.is_some())

                    .map(|(n, _)| *n)

                    .collect();

                self.load_graphs_of(&mut target, &singles)?;

                for (n, (bytes, _)) in nodes.iter().zip(&rendered) {

                    assert_eq!(

                        *bytes,

                        target.render_file(n).0,

                        "a single text is its file's rendering"

                    );

                }

            }

            target.drop_graphs(&nodes);

            let _t = crate::timing::phase("checkout: write");

            if done == 0 && attr_count > 0 {

                attrs = crate::attributes::Attributes::from_files(

                    batch

                        .iter()

                        .zip(&rendered)

                        .map(|((p, _, _), (b, _))| (p.as_str(), b.as_slice())),

                );

            }

            done += batch.len();

            let on_disk: Vec<Vec<u8>> = batch

                .iter()

                .zip(rendered.iter())

                .map(|((path, _, _), (bytes, _))| {

                    if attrs.lookup(path).eol == Some(crate::attributes::Eol::Crlf) {

                        crate::attributes::to_crlf(bytes)

                    } else {

                        bytes.clone()

                    }

                })

                .collect();

            // Each directory checked (no symlink on the way) and made once,

            // then the files compared and written on every core.

            let mut dirs: BTreeSet<&str> = BTreeSet::new();

            for (path, _, _) in batch {

                if let Some((dir, _)) = path.rsplit_once('/') {

                    dirs.insert(dir);

                }

            }

            for dir in dirs {

                let full = self.safe_path(dir)?;

                if !full.is_dir() {

                    if fs::symlink_metadata(&full).is_ok() {

                        remove_any(&full)?;

                    }

                    fs::create_dir_all(&full)?;

                }

            }

            for (path, _, _) in batch {

                check_path_name(path)?;

            }

            let (root, exec_bits) = (&self.root, self.platform.exec_bits);

            // Compared on every core; new files written on every core too,

            // but files that exist rewritten one at a time: on ext4,

            // rewriting in parallel was six times slower (each truncation

            // starts writing back the old blocks).

            let plans: Vec<Disk> = batch

                .par_iter()

                .zip(on_disk.par_iter())

                .map(|((path, _, mode), bytes)| {

                    compare_on_disk(&root.join(path), bytes, mode.exec, exec_bits)

                })

                .collect();

            let fresh: Vec<usize> = (0..batch.len())

                .filter(|&i| plans[i] == Disk::Absent)

                .collect();

            fresh

                .par_iter()

                .map(|&i| write_file_at(&root.join(&batch[i].0), &on_disk[i], batch[i].2.exec))

                .collect::<std::io::Result<()>>()?;

            for (i, plan) in plans.iter().enumerate() {

                if *plan == Disk::Differs {

                    write_file_at(&root.join(&batch[i].0), &on_disk[i], batch[i].2.exec)?;

                }

            }

            report.written += plans.iter().filter(|p| **p != Disk::Same).count();

            for (((path, node, mode), (_, conflicted)), on_disk) in

                batch.iter().zip(rendered).zip(on_disk)

            {

                let a = attrs.lookup(path);

                report.conflicted |= conflicted;

                entries.insert(

                    path.clone(),

                    FsEntry {

                        content_hash: Some(hex(&on_disk)),

                        graph_key: (!conflicted && a.eol.is_none())

                            .then(|| keys.get(node).copied())

                            .flatten(),

                        ..FsEntry::new(NodeKind::File, mode.exec, Vec::new())

                    },

                );

            }

        }

        // Lockable files are read-only until locked (spec §18), whether

        // written now or left as they were.

        for (path, e) in &entries {

            if e.kind == NodeKind::File && attrs.lookup(path).lockable && !self.is_locked(path) {

                let _ = set_readonly(&self.root.join(path), true);

            }

        }

        // Prune directories left empty, deepest first.

        for d in dirs_to_prune.iter().rev() {

            let mut d = d.clone();

            while d != self.root

                && fs::read_dir(&d)

                    .map(|mut i| i.next().is_none())

                    .unwrap_or(false)

            {

                fs::remove_dir(&d)?;

                let Some(p) = d.parent() else { break };

                d = p.to_path_buf();

            }

        }

        drop(_render);

        let recording = crate::timing::phase("checkout: record");

        let (op_ref, op) = self.head_op()?;

        let wc = self.new_working_copy(heads.clone(), "");

        let wc_hash = self.store.put_change(&wc)?;

        let mut view = op.view.clone();

        view.workspaces.insert(self.workspace, wc_hash);

        self.record_op(&[op_ref], command, view)?;

        drop(recording);

        let _t = crate::timing::phase("checkout: manifest");

        // The directories the new tree has, with any the old manifest kept

        // that still exist (holding only ignored files, say), statted now,

        // after the writes: the next check then need not list the tree. A

        // directory written just now is within the racy window, so for a

        // moment the check lists anyway.

        if !old_manifest.dirs.is_empty() {