wip: simplify run_init().

1 files changed, 160 insertions, 56 deletions

diff --git a/vcx b/vcx
index 7a98e1f..559d449 100644
--- a/vcx
+++ b/vcx
@@ -58,26 +58,7 @@ if ($cmd eq 'init') {
 sub run_init {
 	make_path(OBJ_DIR, REV_DIR, TMP_DIR);
 	touch_file(INDEX);
-
 	my $rev_id = to_hex_id(0);
-	my $rev_file = File::Spec->catfile(REV_DIR, $rev_id);
-
-	my $empty_tree = sha1_hex(""); 
-	touch_file(File::Spec->catfile(OBJ_DIR, $empty_tree));
-
-	my $ts  = time();
-	my $msg = "Revision zero.\n";
-	my $msg_len = length($msg);
-	
-	open my $fh, '>:raw', $rev_file or die $!;
-	print $fh "tree:$empty_tree\n";
-	print $fh "parent:none\n";
-	print $fh "date:$ts\n";
-	print $fh "len:$msg_len\n";
-	print $fh "\n"; # The "Header/body" separator
-	print $fh $msg;
-	close $fh;
-
 	write_file(HEAD, "$rev_id\n");
 	print "Initialized repository.\n";
 }
@@ -163,6 +144,118 @@ sub run_add {
 }
 
 sub run_commit {
+	my ($msg) = @_;
+	
+	# Get message and parent
+	$msg //= launch_editor(); 
+	die "Commit aborted: empty message\n" unless $msg && $msg =~ /\S/;
+	
+	my $parent_id = read_head();
+	
+	my $it = stream_index();
+	my %patches;
+	
+	# New index (temporary)
+	my ($idx_fh, $idx_path) = tempfile(DIR => TMP_DIR, UNLINK => 0);
+	binmode $idx_fh, ":raw";
+
+	# Tree logic
+	my $sha_tree = Digest::SHA->new(1);
+	my @tree_lines;
+	my $tree_size = 0;
+	my $use_disk_tree = 0;
+	my ($tree_fh, $tree_path);
+
+	while (my $entry = $it->()) {
+		my $final_hash_for_tree = $entry->{base_hash};
+
+		if ($entry->{staged_hash} ne $entry->{base_hash}) {
+			my $patch = calculate_delta($entry);
+			my $compressed = compress_data($patch);
+
+			# THE PIVOT DECISION (50% Rule)
+			if (length($compressed) < ($entry->{size} * 0.5)) {
+				# Efficiency Win: Store Patch
+				$patches{"$entry->{path}.patch"} = $compressed;
+				# Tree continues to point to OLD base (maintaining delta chain)
+				$final_hash_for_tree = $entry->{base_hash};
+			} else {
+				# Pivot: Store full object as a new Keyframe
+				my $obj_path = File::Spec->catfile(OBJ_DIR, $entry->{staged_hash});
+				if (!-e $obj_path) {
+					copy($entry->{path}, $obj_path) or die "Pivot failed: $!";
+				}
+				# Tree points to NEW hash (resetting delta chain)
+				$final_hash_for_tree = $entry->{staged_hash};
+			}
+		}
+
+		# --- A. Update Tree State ---
+		my $t_line = "$final_hash_for_tree\t$entry->{path}\n";
+		$sha_tree->add($t_line);
+		$tree_size += length($t_line);
+
+		if (!$use_disk_tree && $tree_size > MEM_LIMIT) {
+			($tree_fh, $tree_path) = tempfile(DIR => TMP_DIR, UNLINK => 0);
+			binmode $tree_fh, ":raw";
+			print $tree_fh @tree_lines;
+			@tree_lines = ();
+			$use_disk_tree = 1;
+		}
+		$use_disk_tree ? print $tree_fh $t_line : push @tree_lines, $t_line;
+
+		# --- B. Update Index State ---
+		# Current staged_hash becomes the next base_hash
+		# We also refresh metadata from the actual file
+		my @stats = stat($entry->{path});
+		print $idx_fh join("\t", 
+			$entry->{staged_hash}, # New Staged
+			$entry->{staged_hash}, # New Base (Index always tracks "Current")
+			$stats[9],			 # Fresh mtime
+			$stats[7],			 # Fresh size
+			$entry->{path}
+		) . "\n";
+	}
+
+	# 4. Finalize Tree Object
+	my $tree_hash = $sha_tree->hexdigest;
+	my $final_tree_obj = File::Spec->catfile(OBJ_DIR, $tree_hash);
+	
+	if (!-e $final_tree_obj) {
+		if ($use_disk_tree) {
+			close $tree_fh;
+			rename($tree_path, $final_tree_obj) or die $!;
+		} else {
+			write_file($final_tree_obj, join("", @tree_lines));
+		}
+	} else {
+		unlink($tree_path) if $use_disk_tree;
+	}
+
+	# 5. Assemble Unified Revision File
+	my $next_id = to_hex_id(from_hex_id($parent_id) + 1);
+	my $rev_file = File::Spec->catfile(REV_DIR, $next_id);
+	
+	my $tar_payload = create_tarball(\%patches);
+	my $ts = time();
+	my $msg_final = "$msg\n";
+	my $msg_len = length($msg_final);
+
+	open my $rfh, '>:raw', $rev_file or die $!;
+	print $rfh "tree:$tree_hash\n";
+	print $rfh "parent:$parent_id\n";
+	print $rfh "date:$ts\n";
+	print $rfh "len:$msg_len\n\n";
+	print $rfh $msg_final;
+	print $rfh $tar_payload;
+	close $rfh;
+
+	# 6. Final Atomicity: Commit Index and HEAD
+	close $idx_fh;
+	rename($idx_path, INDEX) or die "Index update failed: $!";
+	write_file(HEAD, "$next_id\n");
+
+	print "Revision $next_id committed.\n";
 }
 
 sub run_log {
@@ -209,30 +302,6 @@ sub run_log {
 	select($old_fh);
 }
 
-sub stage_file {
-	my ($src, $obj, $tmp) = @_;
-	make_path(dirname($tmp));
-	
-	# We want the link inside tmp/ to point to "obj/HASH"
-	# relative to project root.
-	my $rel_target = $obj; 
-	
-	unlink($tmp) if -e $tmp || -l $tmp;
-	symlink($rel_target, $tmp) or die "Failed to symlink $tmp: $!";
-	print "[Staged File] $src\n";
-}
-
-sub stage_link {
-	my ($src, $tmp) = @_;
-	my $target = readlink($src); 
-	
-	make_path(dirname($tmp));
-	unlink($tmp) if -e $tmp || -l $tmp;
-	# For workspace symlinks, we clone the target
-	symlink($target, $tmp) or die "Failed to symlink $tmp: $!";
-	print "[Staged link] $src -> $target\n";
-}
-
 sub make_patch {
 	my ($src, $obj_path, $patches) = @_;
 
@@ -314,19 +383,6 @@ sub apply_bin_patch {
 	close $fh;
 }
 
-sub save_patches {
-	my ($patches) = @_;
-
-	return unless keys %$patches;
-
-	my $tar = Archive::Tar->new;
-	while (my ($obj_name, $patch) = each %$patches) {
-		$tar->add_data("$obj_name.patch", $patch);
-	}
-
-	$tar->write(TMP_DIFF, 1); # gzip
-}
-
 # Convert decimal to a padded 7-char hex string
 sub to_hex_id { sprintf("%07x", $_[0]) }
 
@@ -500,3 +556,51 @@ sub matches_paths {
     return 0;
 }
 
+sub snapshot_tree {
+	my $it = stream_index();
+	
+	my @buf;
+	my $use_disk = 0;
+	my $total_size = 0;
+	my $chunk_size = 1024 * 64; 
+	my $sha = Digest::SHA->new(1);
+
+	my ($tmp_fh, $tmp_path);
+
+	while (my $entry = $it->()) {
+		my $line = "$entry->{s_hash}\t$entry->{path}\n";
+		$sha->add($line);
+		$total_size += length($line);
+
+		if (!$use_disk && $total_size > MEM_LIMIT) {
+			($tmp_fh, $tmp_path) = tempfile();
+			$tmp_fh->setvbuf(undef, POSIX::_IOFBF(), $chunk_size); 
+			binmode $tmp_fh, ":raw";
+			print $tmp_fh @buf;
+			@buf = ();
+			$use_disk = 1;
+		}
+
+		if ($use_disk) {
+			print $tmp_fh $line;
+		} else {
+			push @buf, $line;
+		}
+	}
+
+	my $tree_hash = $sha->hexdigest;
+	my $obj_path = File::Spec->catfile(OBJ_DIR, $tree_hash);
+	if (!-e $obj_path) {
+		if ($use_disk) {
+			close $tmp_fh;
+			rename($tmp_path, $obj_path) or die "Rename failed: $!";
+		} else {
+			write_file($obj_path, join("", @buffer));
+		}
+	} else {
+		unlink($tmp_path) if $use_disk;
+	}
+
+	return $tree_hash;
+}
+