commit edfc9d54897c7455eb32e1bada8639d8a3929f8b
parent 2682bea8663f8fb818209ef82d122af8e3db3312
Author: Martin Ashby <martin@ashbysoft.com>
Date: Sun, 18 Aug 2024 22:02:59 +0100
Extract key-value-log-message parsing code from structure which uses it
(git commit)
It's clearly a little less efficient, maybe we could do some comptime
magic to make it runtime efficient.
Diffstat:
| M | build.zig | | | 1 | + |
| M | src/root.zig | | | 180 | ++++++++++++++++++++++++++++++++++++++++++------------------------------------- |
2 files changed, 96 insertions(+), 85 deletions(-)
diff --git a/build.zig b/build.zig
@@ -38,6 +38,7 @@ pub fn build(b: *std.Build) void {
});
const run_lib_unit_tests = b.addRunArtifact(lib_unit_tests);
+ b.installArtifact(lib_unit_tests);
const exe_unit_tests = b.addTest(.{
.root_source_file = b.path("src/main.zig"),
diff --git a/src/root.zig b/src/root.zig
@@ -304,65 +304,115 @@ const ObjectKind = enum {
};
const Commit = struct {
- const max_parents = 4;
+ _kvlm: Kvlm,
tree: []const u8,
- parents: std.ArrayList([]const u8),
+ parents: std.ArrayListUnmanaged([]const u8),
author: []const u8,
committer: ?[]const u8,
gpgsig: ?[]const u8,
message: []const u8,
pub fn parse(a: std.mem.Allocator, z_reader: anytype) !Commit {
+ var kvlm = try Kvlm.parse(a, z_reader);
+ errdefer kvlm.deinit(a);
+ return .{
+ ._kvlm = kvlm,
+ .tree = if (kvlm.headers.get("tree")) |tree| tree.items[0] else return error.InvalidCommit,
+ .parents = kvlm.headers.get("parent") orelse .{},
+ .author = if (kvlm.headers.get("author")) |tree| tree.items[0] else return error.InvalidCommit,
+ .committer = if (kvlm.headers.get("committer")) |tree| tree.items[0] else null,
+ .gpgsig = if (kvlm.headers.get("gpgsig")) |tree| tree.items[0] else null,
+ .message = kvlm.message,
+ };
+ }
+
+ pub fn deinit(self: *Commit, a: std.mem.Allocator) void {
+ self._kvlm.deinit(a);
+ }
+};
+
+test "parse commit" {
+ const commit_str =
+ \\tree 29ff16c9c14e2652b22f8b78bb08a5a07930c147
+ \\parent 206941306e8a8af65b66eaaaea388a7ae24d49a0
+ \\author Thibault Polge <thibault@thb.lt> 1527025023 +0200
+ \\committer Thibault Polge <thibault@thb.lt> 1527025044 +0200
+ \\gpgsig -----BEGIN PGP SIGNATURE-----
+ \\
+ \\ iQIzBAABCAAdFiEExwXquOM8bWb4Q2zVGxM2FxoLkGQFAlsEjZQACgkQGxM2FxoL
+ \\ kGQdcBAAqPP+ln4nGDd2gETXjvOpOxLzIMEw4A9gU6CzWzm+oB8mEIKyaH0UFIPh
+ \\ rNUZ1j7/ZGFNeBDtT55LPdPIQw4KKlcf6kC8MPWP3qSu3xHqx12C5zyai2duFZUU
+ \\ wqOt9iCFCscFQYqKs3xsHI+ncQb+PGjVZA8+jPw7nrPIkeSXQV2aZb1E68wa2YIL
+ \\ 3eYgTUKz34cB6tAq9YwHnZpyPx8UJCZGkshpJmgtZ3mCbtQaO17LoihnqPn4UOMr
+ \\ V75R/7FjSuPLS8NaZF4wfi52btXMSxO/u7GuoJkzJscP3p4qtwe6Rl9dc1XC8P7k
+ \\ NIbGZ5Yg5cEPcfmhgXFOhQZkD0yxcJqBUcoFpnp2vu5XJl2E5I/quIyVxUXi6O6c
+ \\ /obspcvace4wy8uO0bdVhc4nJ+Rla4InVSJaUaBeiHTW8kReSFYyMmDCzLjGIu1q
+ \\ doU61OM3Zv1ptsLu3gUE6GU27iWYj2RWN3e3HE4Sbd89IFwLXNdSuM0ifDLZk7AQ
+ \\ WBhRhipCCgZhkj9g2NEk7jRVslti1NdN5zoQLaJNqSwO1MtxTmJ15Ksk3QP6kfLB
+ \\ Q52UWybBzpaP9HEd4XnR+HuQ4k2K0ns2KgNImsNvIyFwbpMUyUWLMPimaV1DWUXo
+ \\ 5SBjDB/V/W2JBFR+XKHFJeFwYhj7DD/ocsGr4ZMx/lgc8rjIBkI=
+ \\ =lgTX
+ \\ -----END PGP SIGNATURE-----
+ \\
+ \\Create first draft
+ ;
+ var fbs = std.io.fixedBufferStream(commit_str);
+ const rdr = fbs.reader();
+ var commit = try Commit.parse(std.testing.allocator, rdr);
+ defer commit.deinit(std.testing.allocator);
+}
+
+const Kvlm = struct {
+ headers: std.StringArrayHashMapUnmanaged(std.ArrayListUnmanaged([]const u8)),
+ message: []const u8,
+
+ pub fn parse(a: std.mem.Allocator, z_reader: anytype) !Kvlm {
var cr = std.io.countingReader(z_reader);
var reader = cr.reader();
var al = std.ArrayList(u8).init(a);
defer al.deinit();
- var tree: ?[]const u8 = null;
- var parents = std.ArrayList([]const u8).init(a);
- var author: ?[]const u8 = null;
- var committer: ?[]const u8 = null;
- var gpgsig: ?[]const u8 = null;
- var message: ?[]const u8 = null;
-
+ var headers = std.StringArrayHashMapUnmanaged(std.ArrayListUnmanaged([]const u8)){};
errdefer {
- if (tree) |t| a.free(t);
- for (parents.items) |parent| a.free(parent);
- parents.deinit();
- if (author) |t| a.free(t);
- if (committer) |t| a.free(t);
- if (gpgsig) |t| a.free(t);
- if (message) |t| a.free(t);
+ var it = headers.iterator();
+ while (it.next()) |i| {
+ a.free(i.key_ptr.*);
+ for (i.value_ptr.*.items) |ii| {
+ a.free(ii);
+ }
+ i.value_ptr.*.deinit(a);
+ }
+ headers.deinit(a);
}
- headers: while (true) {
+ headers_loop: while (true) {
const b = try reader.readByte();
switch (b) {
'\n' => {
const b2 = try reader.readByte();
if (b2 == ' ') {
try al.append(b);
- continue :headers;
+ continue :headers_loop;
}
var spl = std.mem.splitScalar(u8, al.items, ' ');
- const key = spl.first();
- const val = spl.rest();
- if (std.mem.eql(u8, key, "tree")) {
- tree = try a.dupe(u8, val);
- } else if (std.mem.eql(u8, key, "parent")) {
- try parents.append(try a.dupe(u8, val));
- } else if (std.mem.eql(u8, key, "author")) {
- author = try a.dupe(u8, val);
- } else if (std.mem.eql(u8, key, "committer")) {
- committer = try a.dupe(u8, val);
- } else if (std.mem.eql(u8, key, "gpgsig")) {
- gpgsig = try a.dupe(u8, val);
+ const key = try a.dupe(u8, spl.first());
+ errdefer a.free(key);
+ const val = try a.dupe(u8, spl.rest());
+ errdefer a.free(val);
+ var gpr = try headers.getOrPut(a, key);
+ if (gpr.found_existing) {
+ a.free(key);
+ try gpr.value_ptr.*.append(a, val);
+ } else {
+ gpr.value_ptr.* = .{};
+ try gpr.value_ptr.append(a, val);
}
+
al.clearRetainingCapacity();
if (b2 == '\n') {
- break :headers;
+ break :headers_loop;
} else {
try al.append(b2);
}
@@ -375,65 +425,25 @@ const Commit = struct {
// And the message is everything else
try reader.readAllArrayList(&al, 1_000_000);
- message = try al.toOwnedSlice();
+ const message = try al.toOwnedSlice();
+ errdefer a.free(message);
return .{
- .tree = tree orelse {
- std.log.err("no tree", .{});
- return error.InvalidCommit;
- },
- .parents = parents,
- .author = author orelse {
- std.log.err("no author", .{});
- return error.InvalidCommit;
- },
- .committer = committer,
- .gpgsig = gpgsig,
- .message = message orelse {
- std.log.err("no message", .{});
- return error.InvalidCommit;
- },
+ .headers = headers,
+ .message = message,
};
}
- pub fn deinit(self: *Commit, a: std.mem.Allocator) void {
- a.free(self.tree);
- for (self.parents.items) |parent| a.free(parent);
- self.parents.deinit();
- a.free(self.author);
- if (self.committer) |committer| a.free(committer);
- if (self.gpgsig) |gpgsig| a.free(gpgsig);
+ pub fn deinit(self: *Kvlm, a: std.mem.Allocator) void {
+ var it = self.headers.iterator();
+ while (it.next()) |i| {
+ a.free(i.key_ptr.*);
+ for (i.value_ptr.*.items) |ii| {
+ a.free(ii);
+ }
+ i.value_ptr.*.deinit(a);
+ }
+ self.headers.deinit(a);
a.free(self.message);
}
};
-
-test "parse commit" {
- const commit_str =
- \\tree 29ff16c9c14e2652b22f8b78bb08a5a07930c147
- \\parent 206941306e8a8af65b66eaaaea388a7ae24d49a0
- \\author Thibault Polge <thibault@thb.lt> 1527025023 +0200
- \\committer Thibault Polge <thibault@thb.lt> 1527025044 +0200
- \\gpgsig -----BEGIN PGP SIGNATURE-----
- \\
- \\ iQIzBAABCAAdFiEExwXquOM8bWb4Q2zVGxM2FxoLkGQFAlsEjZQACgkQGxM2FxoL
- \\ kGQdcBAAqPP+ln4nGDd2gETXjvOpOxLzIMEw4A9gU6CzWzm+oB8mEIKyaH0UFIPh
- \\ rNUZ1j7/ZGFNeBDtT55LPdPIQw4KKlcf6kC8MPWP3qSu3xHqx12C5zyai2duFZUU
- \\ wqOt9iCFCscFQYqKs3xsHI+ncQb+PGjVZA8+jPw7nrPIkeSXQV2aZb1E68wa2YIL
- \\ 3eYgTUKz34cB6tAq9YwHnZpyPx8UJCZGkshpJmgtZ3mCbtQaO17LoihnqPn4UOMr
- \\ V75R/7FjSuPLS8NaZF4wfi52btXMSxO/u7GuoJkzJscP3p4qtwe6Rl9dc1XC8P7k
- \\ NIbGZ5Yg5cEPcfmhgXFOhQZkD0yxcJqBUcoFpnp2vu5XJl2E5I/quIyVxUXi6O6c
- \\ /obspcvace4wy8uO0bdVhc4nJ+Rla4InVSJaUaBeiHTW8kReSFYyMmDCzLjGIu1q
- \\ doU61OM3Zv1ptsLu3gUE6GU27iWYj2RWN3e3HE4Sbd89IFwLXNdSuM0ifDLZk7AQ
- \\ WBhRhipCCgZhkj9g2NEk7jRVslti1NdN5zoQLaJNqSwO1MtxTmJ15Ksk3QP6kfLB
- \\ Q52UWybBzpaP9HEd4XnR+HuQ4k2K0ns2KgNImsNvIyFwbpMUyUWLMPimaV1DWUXo
- \\ 5SBjDB/V/W2JBFR+XKHFJeFwYhj7DD/ocsGr4ZMx/lgc8rjIBkI=
- \\ =lgTX
- \\ -----END PGP SIGNATURE-----
- \\
- \\Create first draft
- ;
- var fbs = std.io.fixedBufferStream(commit_str);
- const rdr = fbs.reader();
- var commit = try Commit.parse(std.testing.allocator, rdr);
- defer commit.deinit(std.testing.allocator);
-}
