Add utility function to fixup Windows file paths

The idea is to iterate over a (canonicalized) path with forward
slashes by components, i.e. file and directory names. Each name
is then looked at by iterating over components, i.e. everything
between dots.

If a component is an illegal name, like COM1 or AUX, we add an
underscore. If it contains illegal characters, like : or \, we
re-map that character into unicode private use area.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

5 files changed, 226 insertions, 1 deletions

diff --git a/include/util/util.h b/include/util/util.h
index db6a712..0c632de 100644
--- a/include/util/util.h
+++ b/include/util/util.h
@@ -79,4 +79,6 @@ SQFS_INTERNAL int hex_decode(const char *in, size_t in_sz,
 SQFS_INTERNAL int base64_decode(const char *in, size_t in_len,
 				sqfs_u8 *out, size_t *out_len);
 
+SQFS_INTERNAL char *fix_win32_filename(const char *path);
+
 #endif /* SQFS_UTIL_H */
diff --git a/lib/util/Makemodule.am b/lib/util/Makemodule.am
index ec38b7a..c102b9b 100644
--- a/lib/util/Makemodule.am
+++ b/lib/util/Makemodule.am
@@ -16,6 +16,7 @@ libutil_a_SOURCES += lib/util/source_date_epoch.c
 libutil_a_SOURCES += lib/util/file_cmp.c
 libutil_a_SOURCES += lib/util/hex_decode.c
 libutil_a_SOURCES += lib/util/base64_decode.c
+libutil_a_SOURCES += lib/util/fix_win32_filename.c
 libutil_a_CFLAGS = $(AM_CFLAGS)
 libutil_a_CPPFLAGS = $(AM_CPPFLAGS)
 
diff --git a/lib/util/fix_win32_filename.c b/lib/util/fix_win32_filename.c
new file mode 100644
index 0000000..948de66
--- /dev/null
+++ b/lib/util/fix_win32_filename.c
@@ -0,0 +1,164 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * fix_win32_filename.c
+ *
+ * Copyright (C) 2024 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "util/util.h"
+
+#include <string.h>
+#include <stdlib.h>
+
+#ifdef _MSC_VER
+#define strncasecmp _strnicmp
+#define strcasecmp _stricmp
+#endif
+
+typedef struct {
+	size_t used;
+	size_t available;
+	char buffer[];
+} buffer_t;
+
+static buffer_t *buffer_append(buffer_t *buf, const char *data, size_t count)
+{
+	size_t bufspace, needed;
+
+	if (buf == NULL) {
+		buf = calloc(1, sizeof(*buf) + 128);
+		if (buf == NULL)
+			return NULL;
+
+		buf->used = 1;
+		buf->available = 128;
+		buf->buffer[0] = '\0';
+	}
+
+	bufspace = buf->available;
+	needed = buf->used + count;
+
+	while (bufspace < needed)
+		bufspace += 128;
+
+	if (bufspace != buf->available) {
+		void *new_buf = realloc(buf, sizeof(*buf) + bufspace);
+		if (new_buf == NULL) {
+			free(buf);
+			return NULL;
+		}
+		buf = new_buf;
+		buf->available = bufspace;
+	}
+
+	buf->used -= 1;
+	memcpy(buf->buffer + buf->used, data, count);
+	buf->used += count;
+	buf->buffer[buf->used++] = '\0';
+	return buf;
+}
+
+static const char *bad_names[] = {
+	"CON", "PRN", "AUX", "NUL",
+	"COM1", "COM2", "COM3", "COM4", "COM5", "COM6", "COM7", "COM8", "COM9",
+	"LPT1", "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9",
+};
+
+static buffer_t *handle_component(buffer_t *buf, const char *comp, size_t len)
+{
+	for (size_t i = 0; i < sizeof(bad_names) / sizeof(bad_names[0]); ++i) {
+		if (!strncasecmp(comp, bad_names[i], len)) {
+			buf = buffer_append(buf, comp, len);
+			if (buf != NULL)
+				buf = buffer_append(buf, "_", 1);
+			return buf;
+		}
+	}
+
+	while (len > 0) {
+		sqfs_u8 value, rep[3];
+		size_t i = 0;
+
+		for (i = 0; i < len; ++i) {
+			if (comp[i] < 0x20 || comp[i] == 0x7F)
+				break;
+			if (comp[i] == '<' || comp[i] == '>' || comp[i] == ':')
+				break;
+			if (comp[i] == '|' || comp[i] == '?' || comp[i] == '*')
+				break;
+			if (comp[i] == '\\' || comp[i] == '\"')
+				break;
+		}
+
+		if (i > 0) {
+			buf = buffer_append(buf, comp, i);
+			if (buf == NULL || i == len)
+				break;
+		}
+
+		value = comp[i++];
+		comp += i;
+		len -= i;
+
+		rep[0] = 0xEF;
+		rep[1] = 0x80 | ((value >> 6) & 0x3f);
+		rep[2] = 0x80 | ( value       & 0x3f);
+
+		buf = buffer_append(buf, (const char *)rep, 3);
+		if (buf == NULL)
+			break;
+	}
+
+	return buf;
+}
+
+static buffer_t *handle_name(buffer_t *buf, const char *name, size_t len)
+{
+	char *sep;
+
+	while ((sep = memchr(name, '.', len)) != NULL) {
+		buf = handle_component(buf, name, sep - name);
+		if (buf == NULL)
+			return NULL;
+
+		buf = buffer_append(buf, ".", 1);
+		if (buf == NULL)
+			return NULL;
+
+		len -= sep - name + 1;
+		name = sep + 1;
+	}
+
+	return handle_component(buf, name, len);
+}
+
+char *fix_win32_filename(const char *path)
+{
+	buffer_t *buf = NULL;
+	char *sep, *out;
+	size_t len;
+
+	while ((sep = strchr(path, '/')) != NULL) {
+		buf = handle_name(buf, path, sep - path);
+		if (buf == NULL)
+			return NULL;
+
+		buf = buffer_append(buf, "/", 1);
+		if (buf == NULL)
+			return NULL;
+
+		path = sep + 1;
+	}
+
+	buf = handle_name(buf, path, strlen(path));
+	if (buf == NULL)
+		return NULL;
+
+	len = buf->used;
+	memmove(buf, buf->buffer, len);
+
+	out = realloc(buf, len);
+	if (out == NULL)
+		out = (char *)buf;
+
+	return out;
+}
diff --git a/tests/libutil/Makemodule.am b/tests/libutil/Makemodule.am
index e039282..c783853 100644
--- a/tests/libutil/Makemodule.am
+++ b/tests/libutil/Makemodule.am
@@ -36,9 +36,12 @@ test_hex_decode_LDADD = libutil.a libcompat.a
 test_base64_decode_SOURCES = tests/libutil/base64_decode.c
 test_base64_decode_LDADD = libutil.a libcompat.a
 
+test_fix_win32_filename_SOURCES = tests/libutil/fix_win32_filename.c
+test_fix_win32_filename_LDADD = libutil.a libcompat.a
+
 LIBUTIL_TESTS = \
 	test_str_table test_rbtree test_xxhash test_threadpool test_ismemzero \
-	test_canonicalize_name test_filename_sane \
+	test_canonicalize_name test_filename_sane test_fix_win32_filename \
 	test_sdate_epoch test_hex_decode test_base64_decode
 
 check_PROGRAMS += $(LIBUTIL_TESTS)
diff --git a/tests/libutil/fix_win32_filename.c b/tests/libutil/fix_win32_filename.c
new file mode 100644
index 0000000..a4f71e8
--- /dev/null
+++ b/tests/libutil/fix_win32_filename.c
@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * fix_win32_filename.c
+ *
+ * Copyright (C) 2024 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "util/test.h"
+#include "util/util.h"
+
+static const struct {
+	const char *path;
+	const char *result;
+} test_data[] = {
+	{ "foo", "foo" },
+	{ "foo/bar", "foo/bar" },
+	{ "foo/bar.txt", "foo/bar.txt" },
+	{ "COM1", "COM1_" },
+	{ "COM1.txt", "COM1_.txt" },
+	{ "foo.aux", "foo.aux_" },
+	{ "foo/bar/test.LPT1/bla", "foo/bar/test.LPT1_/bla" },
+	{ "C:\\/foo/COM1.bla/bar",
+	  "C\xEF\x80\xBA\xEF\x81\x9c/foo/COM1_.bla/bar" },
+};
+
+int main(int argc, char **argv)
+{
+	(void)argc; (void)argv;
+
+	for (size_t i = 0; i < sizeof(test_data) / sizeof(test_data[0]); ++i) {
+		char *result = fix_win32_filename(test_data[i].path);
+		size_t out_len = strlen(test_data[i].result);
+
+		if (result == NULL) {
+			fprintf(stderr, "OOM for test case %u (%s)?\n",
+				(unsigned int)i, test_data[i].path);
+			return EXIT_FAILURE;
+		}
+
+		if (out_len != strlen(result) ||
+		    memcmp(result, test_data[i].result, out_len) != 0) {
+			fprintf(stderr,
+				"Mismatch for %s -> %s, got %s instead!\n",
+				test_data[i].path, test_data[i].result,
+				result);
+			free(result);
+			return EXIT_FAILURE;
+		}
+
+		free(result);
+	}
+
+	return EXIT_SUCCESS;
+}