module snapshot

import os
import time
import src.storage

fn temp_root(name string) !string {
	dir := os.join_path(os.temp_dir(), 'koh3_snap_${os.getpid()}_${name}')
	os.rmdir_all(dir) or {}
	os.mkdir_all(dir, os.MkdirParams{})!
	return dir
}

fn put(root string, rel string, text string) ! {
	full := os.join_path(root, rel)
	os.mkdir_all(os.dir(full), os.MkdirParams{})!
	os.write_file(full, text)!
}

fn test_ignore_matches_the_documented_shapes() {
	ig := parse('
# a comment
*.log
build/
/anchored.txt
out/**
docs/*.md
')
	assert ig.matches('error.log', false)
	assert ig.matches('deep/nested/error.log', false)
	assert !ig.matches('error.txt', false)
	// A trailing slash names a directory: the directory itself and everything
	// below it, but not a file that shares the name.
	assert ig.matches('build', true)
	assert ig.matches('build/thing.o', false)
	assert ig.matches('src/build/nested.o', false)
	assert !ig.matches('build', false)
	assert ig.matches('anchored.txt', false)
	assert !ig.matches('sub/anchored.txt', false)
	assert ig.matches('out/anything/at/all', false)
	assert ig.matches('docs/readme.md', false)
	assert !ig.matches('docs/readme.txt', false)
	assert !ig.matches('src/main.v', false)
}

// A stripped path is banned as itself: its `*` and `?` must not become
// wildcards, in `.kohignore` or in the walk's ban.
fn test_a_stripped_path_with_wildcard_characters_bans_only_itself() {
	root := temp_root('literal_ban')!
	defer {
		os.rmdir_all(root) or {}
	}
	assert add_ignore(root, 'keys/a*b?.env')!
	assert os.read_file(os.join_path(root, ignore_file))! == '/keys/a\\*b\\?.env\n'
	for ig in [load(root), Ignore{}.banning(['keys/a*b?.env'])] {
		assert ig.matches('keys/a*b?.env', false)
		assert !ig.matches('keys/axxbz.env', false)
		assert !ig.matches('keys/a*b?.env.bak', false)
	}
	assert parse('\\\\x').matches('\\x', false)
}

fn test_hard_ignores_are_not_overridable() {
	for path in ['.koh', '.koh-wal', '.koh-shm', '.koh-v2', '.koh-v2/snapshots/x.json', '.git',
		'.git/config', 'LOG.md'] {
		assert hard_ignored(path), '${path} should be hard ignored'
	}
	assert !hard_ignored('.kohignore')
	assert !hard_ignored('src/log.md')
	// A repo inside the repo keeps its own history: restoring an old copy of
	// its .koh or .git would roll that history back.
	for path in ['inner/.koh', 'inner/.koh-wal', 'inner/.koh-shm', 'a/b/.git', 'vendored/.git/HEAD',
		'old/.koh-v2/snapshots/x.json', 'koh2repo/.koh/objects/ab/cd'] {
		assert hard_ignored(path), '${path} should be hard ignored'
	}
	assert !hard_ignored('inner/.kohignore')
	assert !hard_ignored('docs/.github/workflow.yml')
	assert !hard_ignored('notes/.koh.md')
	// A pattern list cannot un-ignore them, and `.kohignore` itself is content.
	ig := parse('')
	assert ig.matches('.koh', false)
	assert !ig.matches('.kohignore', false)
}

fn test_walk_skips_ignored_files_directories_and_symlinks() {
	mut store := storage.create(temp_root('walk')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'keep.txt', 'keep\n')!
	put(root, 'nested/deep/file.txt', 'deep\n')!
	put(root, 'skip.log', 'noise\n')!
	put(root, 'build/output.o', 'binary\n')!
	put(root, '.kohignore', '*.log\nbuild/\n')!
	os.symlink('keep.txt', os.join_path(root, 'link.txt')) or {}

	entries := walk(store, root, load(root))!
	mut paths := []string{}
	for entry in entries {
		paths << entry.path
	}
	assert paths == ['.kohignore', 'keep.txt', 'nested/deep/file.txt']
}

fn test_walk_records_the_executable_bit() {
	mut store := storage.create(temp_root('exec')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'run.sh', '#!/bin/sh\necho hi\n')!
	put(root, 'plain.txt', 'hello\n')!
	os.chmod(os.join_path(root, 'run.sh'), 0o755)!

	mut executable := false
	mut plain_executable := true
	for entry in walk(store, root, Ignore{})! {
		if entry.path == 'run.sh' {
			executable = entry.executable
		}
		if entry.path == 'plain.txt' {
			plain_executable = entry.executable
		}
	}
	assert executable
	assert !plain_executable
}

fn test_legal_path_is_the_shared_save_path_rule() {
	for path in ['src/main.v', 'café/猫.txt', '.kohignore'] {
		assert legal_path(path), path
	}
	for path in ['', '/etc/passwd', 'a//b', './a', 'a/../b', 'a\\b', 'Icon\r', 'a\u0080b',
		[u8(0xff)].bytestr()] {
		assert !legal_path(path), path
	}
}

// Only bytes outside a valid UTF-8 sequence become \xNN; Kepr's escapePath prints the same.
fn test_escape_path_escapes_only_the_invalid_bytes() {
	assert escape_path('é' + [u8(0xff)].bytestr() + 'é') == 'é\\xffé'
	assert escape_path([u8(0xed), 0xa0, 0x80].bytestr()) == '\\xed\\xa0\\x80'
	assert escape_path('a\u0085\\b\r') == 'a\\u0085\\\\b\\x0d'
}

fn test_walk_skips_and_reports_unsafe_names() {
	root := temp_root('unsafe_names')!
	defer {
		os.rmdir_all(root) or {}
	}
	store := storage.create(root)!
	defer {
		store.close()
	}
	put(root, 'safe.txt', 'kept\n')!
	os.write_file('${root}/bad\\name.txt', 'skipped\n')!
	os.write_file('${root}/Icon\r', 'skipped\n')!
	mut hashing := Hashing{}
	walked := walk_report(store, root, Ignore{}, mut hashing)!
	assert walked.files.map(it.path) == ['safe.txt']
	// Only a file that had to be read counts; unsafe names are never read.
	assert hashing.count == 1
	assert walked.unsafe_paths == ['Icon\\x0d', 'bad\\\\name.txt']
}

fn test_write_tree_refuses_every_path_outside_the_root_before_writing() {
	base := temp_root('restore_containment')!
	defer {
		os.rmdir_all(base) or {}
	}
	root := os.join_path(base, 'repo')
	os.mkdir(root)!
	store := storage.create(root)!
	defer {
		store.close()
	}
	for path in ['../outside.txt', '/etc/koh-test', 'a/../../outside.txt', 'bad\\name', 'control\rname'] {
		if _ := write_tree(store, root, Ignore{}, [FileEntry{
			path: path
		}], []) {
			assert false, '${path} must be refused'
		}
	}
	assert !os.exists(os.join_path(base, 'outside.txt'))
}

fn test_prune_empty_dirs_never_touches_a_sibling_with_the_roots_prefix() {
	base := temp_root('prune_prefix')!
	defer {
		os.rmdir_all(base) or {}
	}
	root := os.join_path(base, 'repo')
	sibling := os.join_path(base, 'repo2')
	os.mkdir(root)!
	os.mkdir(sibling)!
	prune_empty_dirs(root, sibling)
	assert os.is_dir(sibling)
}

// The point of nanosecond timestamps: a file rewritten inside the same second
// with the same size must not reuse the cached hash, or a save would record
// content that is no longer on disk.
fn test_cache_does_not_reuse_a_same_size_same_second_edit() {
	mut store := storage.create(temp_root('cache')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'file.txt', 'aaaa\n')!
	first := walk(store, root, Ignore{})!
	assert first[0].hash == storage.hash_hex('aaaa\n'.bytes())

	put(root, 'file.txt', 'bbbb\n')!
	second := walk(store, root, Ignore{})!
	assert second[0].hash == storage.hash_hex('bbbb\n'.bytes())
	assert second[0].hash != first[0].hash
}

// Linux stamps files from a clock that ticks every few milliseconds, so a
// same-size rewrite in the tick the file was hashed in leaves size, mtime and
// ctime unchanged. A hash taken that close to the last change is never cached;
// the next walk hashes again. Tools such as `cp -p` keep an old mtime while
// rewriting, so the newer of the two times decides.
fn test_cache_skips_a_file_changed_moments_ago() {
	second := i64(1_000_000_000)
	now := 100 * second
	old := now - 10 * second
	assert cacheable(Stat{ mtime_ns: old, ctime_ns: old }, now)
	assert !cacheable(Stat{ mtime_ns: now - second, ctime_ns: now - second }, now)
	assert !cacheable(Stat{ mtime_ns: old, ctime_ns: now - second }, now)

	mut store := storage.create(temp_root('racy')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'fresh.txt', 'new\n')!
	put(root, 'copied.txt', 'kept mtime\n')!
	hour_ago := time.now().unix() - 3600
	os.utime(os.join_path(root, 'copied.txt'), hour_ago, hour_ago)!
	walk(store, root, Ignore{})!
	assert store.load_cache()!.len == 0
}

// `status` must work while another agent's save holds the write lock: the
// cache only saves time, so a busy repo hashes without caching, at once.
fn test_a_walk_does_not_wait_for_another_writer() {
	mut store := storage.create(temp_root('busy_walk')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'a.txt', 'a\n')!
	// Only a file whose last change is older than the racy window is cached,
	// and ctime cannot be set back, so the test waits it out.
	time.sleep(racy_window_ns * time.nanosecond + 100 * time.millisecond)
	writer := storage.open(root)!
	defer {
		writer.close()
	}
	writer.begin()!
	started := time.now()
	files := walk(store, root, Ignore{})!
	assert files.len == 1
	assert time.since(started) < time.second
	writer.rollback()
}

fn test_restore_writes_removes_and_leaves_ignored_files_alone() {
	mut store := storage.create(temp_root('restore')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'keep.txt', 'recorded\n')!
	put(root, 'gone/inside.txt', 'also recorded\n')!
	put(root, 'noise.log', 'ignored, must survive\n')!
	put(root, '.kohignore', '*.log\n')!
	target := walk(store, root, load(root))!
	store_objects(store, root, target, unsafe { nil })!

	// Drift the working tree: clobber one file, delete another, and add a file
	// the recorded tree has never seen.
	os.write_file(os.join_path(root, 'keep.txt'), 'clobbered\n')!
	os.rm(full_path(root, 'gone/inside.txt'))!
	put(root, 'extra.txt', 'never saved\n')!

	// The removal list names an ignored file (which survives), a file the
	// target still has (rewritten, not removed), and one genuinely extra file.
	report := restore_files(store, root, load(root), target,
		['extra.txt', '.kohignore', 'noise.log'])!

	// The clobbered and the deleted file are written; the unchanged
	// .kohignore is left as it is.
	assert report.written == 2
	assert report.removed == 1
	assert os.read_file(full_path(root, 'keep.txt'))! == 'recorded\n'
	assert os.read_file(full_path(root, 'gone/inside.txt'))! == 'also recorded\n'
	assert !os.exists(full_path(root, 'extra.txt'))
	assert os.read_file(full_path(root, 'noise.log'))! == 'ignored, must survive\n'
	assert os.read_file(full_path(root, '.kohignore'))! == '*.log\n'
}

fn test_restore_keeps_the_executable_bit() {
	mut store := storage.create(temp_root('restore_exec')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'run.sh', '#!/bin/sh\necho hi\n')!
	os.chmod(os.join_path(root, 'run.sh'), 0o755)!
	target := walk(store, root, Ignore{})!
	store_objects(store, root, target, unsafe { nil })!
	os.chmod(os.join_path(root, 'run.sh'), 0o644)!

	restore_files(store, root, Ignore{}, target, [])!
	mode := (lstat(os.join_path(root, 'run.sh')) or { panic('missing') }).mode
	assert is_executable(mode)
}

fn test_diff_reports_kinds_counts_and_binary_files() {
	mut store := storage.create(temp_root('diff')!)!
	root := store.root
	defer {
		store.close()
		os.rmdir_all(root) or {}
	}
	put(root, 'text.txt', 'one\ntwo\n')!
	put(root, 'same.txt', 'unchanged\n')!
	put(root, 'gone.txt', 'bye\n')!
	before := walk(store, root, Ignore{})!
	store_objects(store, root, before, unsafe { nil })!

	os.write_file(os.join_path(root, 'text.txt'), 'one\nthree\nfour\n')!
	os.rm(full_path(root, 'gone.txt'))!
	os.write_file(os.join_path(root, 'added.txt'), 'new\n')!
	os.write_bytes(os.join_path(root, 'blob.bin'), [u8(0), 0x01, 0x02])!
	after := walk(store, root, Ignore{})!

	// A save's side of a diff is bytes from the object store; only the side
	// that is the working tree reads from disk.
	deltas := diff_trees(store, Tree{
		files: before
	}, Tree{
		root:  root
		files: after
	})!
	mut by_path := map[string]Delta{}
	for delta in deltas {
		by_path[delta.path] = delta
	}
	assert deltas.len == 4
	assert by_path['text.txt'].kind == 'modified'
	assert by_path['text.txt'].added == 2
	assert by_path['text.txt'].removed == 1
	assert by_path['added.txt'].kind == 'added'
	assert by_path['gone.txt'].kind == 'removed'
	assert by_path['blob.bin'].binary
	assert by_path['blob.bin'].patch == binary_notice
	assert 'same.txt' !in by_path
}

fn full_path(root string, rel string) string {
	return os.join_path(root, rel)
}