From e02f4091b868e1adff6155196fc063876a4f3789 Mon Sep 17 00:00:00 2001
From: Quinn <quinnthepigeon@proton.me>
Date: Mon, 25 Aug 2025 14:29:45 +0200
Subject: [PATCH] start writing code for `*.mcX` file parsing, starting with
 indexing logic.

---
 src/dat/mcx.c | 24 ++++++++++++++++++++++++
 src/dat/mcx.h | 15 +++++++++++++++
 2 files changed, 39 insertions(+)
 create mode 100644 src/dat/mcx.c

diff --git a/src/dat/mcx.c b/src/dat/mcx.c
new file mode 100644
index 0000000..44f4f81
--- /dev/null
+++ b/src/dat/mcx.c
@@ -0,0 +1,24 @@
+#include "mcx.h"
+
+#include <endian.h>
+#include <stdint.h>
+#include <stdlib.h>
+
+#include "../util/compat/endian.h"
+#include "../util/intdef.h"
+
+/* an `*.mcX` contains a `0x2000` byte long table, the first `0x1000` containing
+ * `0x400` entries of chunk data.
+ * This chunk data is big-endian, where bytes `0xFFFFFF00` represent the `0x1000` sector offset.
+ * From the start, and bytes `0x000000FF` represent the length in `0x1000` sectors. */
+void mcx_index(const u8 *restrict buf, struct mcx_chunk *restrict chunks) {
+	const u32 *ptr = (u32 *)buf;
+	for (uint i = 0; i < 0x400; i++) {
+		u32 dat = be32toh(ptr[i]);
+		chunks[i] = (struct mcx_chunk){
+			.idx = (dat >> 8) * 0x1000,
+			.len = (dat & 0xFF) * 0x1000,
+			.time = be32toh(ptr[i + 0x400]),
+		};
+	}
+}
diff --git a/src/dat/mcx.h b/src/dat/mcx.h
index 2a97db4..97dd584 100644
--- a/src/dat/mcx.h
+++ b/src/dat/mcx.h
@@ -2,6 +2,21 @@
 // Licensed under the MIT Licence. See LICENSE for details
 #pragma once
 
+#include <stdlib.h>
+
+#include "../util/atrb.h"
+#include "../util/intdef.h"
+
+/* contains chunk metadata */
+struct mcx_chunk {
+	size_t idx; // byte offset for start of chunk data
+	u32 len;    // byte length of chunk (+ padding)
+	u32 time;      // modification time in epoch seconds
+};
+
+/* indexes the chunks in an `*.mcX` file, writing `0x400` of entries to `chunks` */
+void mcx_index(const u8 *restrict buf, struct mcx_chunk *restrict chunks) NONNULL((1, 2));
+
 /* the MCR (Minecraft region) and MCA (Minecraft anvil) files are similar
  * MCA is the newer variant, where it includes:
  *  - a world height of 256, rather than 128.