extractor: Add cum extractor groundwork

Signed-off-by: Chloe M. <chloe@mensia.org>

.gitignore

@@ -1,2 +1,3 @@
 /hole
+/cumxct
 *.o

Makefile

@@ -7,8 +7,12 @@
 #
 
 .PHONY: all
-all: core
+all: core extractor
 
 .PHONY: core
 core:
 	cd core/; $(MAKE)
+
+.PHONY: extractor
+extractor:
+	cd extractor/; $(MAKE)

extractor/Makefile

@@ -0,0 +1,23 @@
+#
+# Copyright (c) 2026, Chloe M.
+# Provided under the BSD-3 clause
+#
+# Description: Cum extractor build script
+# Author: 	   Chloe M.
+#
+
+include ../mk/global.mk
+.SILENT:
+
+CFILES = $(shell find . -name "*.c")
+OFILES = $(CFILES:.c=.o)
+
+.PHONY: all
+all: ../cumxct
+
+.PHONY: hole
+../cumxct: $(OFILES)
+	$(CC) $^ -o $@
+
+%.o: %.c
+	$(CC) -c $(CFLAGS) $< -o $@

extractor/cum_xct.c

@@ -0,0 +1,140 @@
+/*
+ * Copyright (c) 2026, Chloe M.
+ * Provided under the BSD-3 clause.
+ *
+ * Description:  CUMHOLE extraction module
+ * Author:       Chloe M.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <string.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <stddef.h>
+
+#define REF_MAGIC "CUMHOLE"
+#define REF_MAGIC_LEN 8
+#define REF_PATH_LEN 256
+
+/*
+ * The offset table is made up of references, this
+ * structure represents a reference.
+ *
+ * @magic:      Reference magic number
+ * @path:       Path this reference refers to
+ * @size:       Size of data being referenced
+ * @data_off:   Offset to data being referenced
+ */
+struct hole_ref {
+    char magic[REF_MAGIC_LEN];
+    char path[REF_PATH_LEN];
+    size_t size;
+    size_t data_off;
+} __attribute__((packed));
+
+/* Globals */
+static char *input_path = NULL;
+
+static void
+help(void)
+{
+    printf("usage: ./cumxct [flags]\n");
+    printf("[-h]    Display this help menu\n");
+    printf("[-i]    Input CUMHOLE bundle\n");
+}
+
+static void
+do_parse(struct hole_ref *ref)
+{
+    if (ref == NULL) {
+        return;
+    }
+
+    for (;;) {
+        if (memcmp(ref->magic, REF_MAGIC, REF_MAGIC_LEN) != 0) {
+            break;
+        }
+
+        printf("found '%s' with size %ld\n", ref->path, ref->size);
+        ++ref;
+    }
+}
+
+static void
+input_parse(void)
+{
+    int fd;
+    char *data;
+    size_t archive_size;
+    ssize_t nret;
+
+    fd = open(input_path, O_RDONLY);
+    if (fd < 0) {
+        printf("fatal: could not open input file\n");
+        perror("open");
+        return;
+    }
+
+    /* Obtain the size of the buffer */
+    archive_size = lseek(fd, 0, SEEK_END);
+    lseek(fd, 0, SEEK_SET);
+
+    /* Allocate the actual buffer */
+    data = malloc(archive_size);
+    if (data == NULL) {
+        printf("fatal: could not allocate archive buffer\n");
+        close(fd);
+        return;
+    }
+
+    /* Populate it with bytes */
+    nret = read(fd, data, archive_size);
+    if (nret <= 0) {
+        printf("fatal: could not read archive\n");
+        perror("read");
+        return;
+    }
+
+    do_parse((struct hole_ref *)data);
+    close(fd);
+    free(data);
+}
+
+int
+main(int argc, char **argv)
+{
+    int opt;
+
+    if (argc < 2) {
+        printf("fatal: too few arguments\n");
+        help();
+        return -1;
+    }
+
+    while ((opt = getopt(argc, argv, "hi:")) != -1) {
+        switch (opt) {
+        case 'h':
+            help();
+            return -1;
+        case 'i':
+            input_path = strdup(optarg);
+            if (input_path == NULL) {
+                printf("fatal: out of memory\n");
+                return -1;
+            }
+
+            break;
+        }
+    }
+
+    if (input_path == NULL) {
+        printf("fatal: expected input bundle\n");
+        help();
+        return -1;
+    }
+
+    input_parse();
+    free(input_path);
+    return 0;
+}