commit e7752e7e1996c6b387a19a3b989976ab3fd76f75
parent eb4fced7dfc80539b1c04a1cbf95210754850328
Author: Drew DeVault <sir@cmpwn.com>
Date: Sat, 29 Jan 2022 11:22:37 +0100
mime: load system database
This also corrects a use-after-free error with the storage tables by
switching to slices of pointers rather than slices of values.
Signed-off-by: Drew DeVault <sir@cmpwn.com>
Diffstat:
6 files changed, 110 insertions(+), 62 deletions(-)
diff --git a/mime/database.ha b/mime/database.ha
@@ -10,11 +10,11 @@ export type mimetype = struct {
// List of media types with statically allocated fields (though the list itself
// is dynamically allocated).
-let static_db: []mimetype = [];
+let static_db: []*mimetype = [];
// List of media types with heap-allocated fields, used when loading mime types
// from the system database.
-let heap_db: []mimetype = [];
+let heap_db: []*mimetype = [];
def MIME_BUCKETS: size = 256;
@@ -25,42 +25,45 @@ let exttable: [MIME_BUCKETS][]*mimetype = [[]...];
// Registers a Media Type and its extensions in the internal MIME database. This
// function is designed to be used by @init functions for modules which
// implement new Media Types.
-export fn register(mime: mimetype...) void = {
+export fn register(mime: *mimetype...) void = {
let i = len(static_db);
append(static_db, mime...);
for (i < len(static_db); i += 1) {
- const item = &static_db[i];
- const hash = fnv::string(item.mime);
- let bucket = &mimetable[hash % len(mimetable)];
- append(bucket, item);
+ hashtable_insert(static_db[i]);
+ };
+};
+
+fn hashtable_insert(item: *mimetype) void = {
+ const hash = fnv::string(item.mime);
+ let bucket = &mimetable[hash % len(mimetable)];
+ append(bucket, item);
- for (let i = 0z; i < len(item.exts); i += 1) {
- const hash = fnv::string(item.exts[i]);
- let bucket = &exttable[hash % len(exttable)];
- append(bucket, item);
- };
+ for (let i = 0z; i < len(item.exts); i += 1) {
+ const hash = fnv::string(item.exts[i]);
+ let bucket = &exttable[hash % len(exttable)];
+ append(bucket, item);
};
};
-const builtins: [_]mimetype = [
- mimetype {
- mime = "text/plain",
- exts = ["txt"],
- },
- mimetype {
- mime = "text/x-hare",
- exts = ["ha"],
- },
-];
+const text_plain: mimetype = mimetype {
+ mime = "text/plain",
+ exts = ["txt"],
+};
+
+const text_hare: mimetype = mimetype {
+ mime = "text/x-hare",
+ exts = ["ha"],
+};
@init fn init() void = {
- register(builtins...);
+ register(&text_plain, &text_hare);
};
@fini fn fini() void = {
for (let i = 0z; i < len(heap_db); i += 1) {
free(heap_db[i].mime);
strings::freeall(heap_db[i].exts);
+ free(heap_db[i]);
};
free(heap_db);
free(static_db);
diff --git a/mime/lookup.ha b/mime/lookup.ha
@@ -1,5 +1,6 @@
use hash::fnv;
use strings;
+use io; // XXX: TEMP
// Looks up a Media Type based on the mime type string, returning null if
// unknown.
diff --git a/mime/system.ha b/mime/system.ha
@@ -0,0 +1,76 @@
+use bufio;
+use errors;
+use encoding::utf8;
+use fs;
+use io;
+use os;
+use strings;
+
+// Path to the system MIME database.
+export def SYSTEM_DB: str = "/etc/mime.types";
+
+@init fn init() void = {
+ // Done in a separate function so we can discard errors here
+ load_systemdb(): void;
+};
+
+fn load_systemdb() (void | fs::error | io::error) = {
+ const file = os::open(SYSTEM_DB)?;
+ defer io::close(file);
+
+ let buf: [os::BUFSIZ]u8 = [0...];
+ const file = bufio::buffered(file, buf, []);
+ defer io::close(&file);
+
+ for (true) {
+ const line = match (bufio::scanline(&file)?) {
+ case let bytes: []u8 =>
+ yield match (strings::try_fromutf8(bytes)) {
+ case utf8::invalid =>
+ io::errorln("Warning: /etc/mime.types contains invalid UTF-8");
+ return;
+ case let s: str =>
+ yield s;
+ };
+ case io::EOF =>
+ break;
+ };
+ defer free(line);
+
+ const line = strings::trim(line);
+ if (strings::hasprefix(line, "#") || len(line) == 0) {
+ continue;
+ };
+
+ const items = strings::cut(line, "\t");
+ const mime = strings::trim(items.0),
+ exts = strings::trim(items.1);
+ if (len(exts) == 0) {
+ continue;
+ };
+
+ const tok = strings::tokenize(exts, " ");
+ let entry = alloc(mimetype {
+ mime = strings::dup(mime),
+ exts = [],
+ });
+ for (true) {
+ const ext = match (strings::next_token(&tok)) {
+ case let tok: str =>
+ yield strings::trim(tok);
+ case void =>
+ break;
+ };
+ append(entry.exts, strings::dup(ext));
+ };
+ register_heap(entry);
+ };
+};
+
+fn register_heap(mime: *mimetype...) void = {
+ let i = len(heap_db);
+ append(heap_db, mime...);
+ for (i < len(heap_db); i += 1) {
+ hashtable_insert(heap_db[i]);
+ };
+};
diff --git a/os/+linux/fs.ha b/os/+linux/fs.ha
@@ -3,6 +3,7 @@ use fs;
use path;
use rt;
use strings;
+use io;
@init fn init() void = {
static let cwd_fs = os_filesystem { ... };
diff --git a/scripts/gen-stdlib b/scripts/gen-stdlib
@@ -794,11 +794,15 @@ math() {
}
mime() {
+ # This module is not built by default because gen-stdlib does not do a good
+ # job of resolving @init dependency ordering issues
gen_srcs mime \
database.ha \
lookup.ha \
- parse.ha
- gen_ssa mime ascii errors string
+ parse.ha \
+ system.ha
+ gen_ssa mime ascii errors string hash::fnv encoding::utf8 bufio \
+ errors fs io os
}
net() {
@@ -1216,7 +1220,6 @@ linux::io_uring linux
linux::vdso linux
math
math::random
-mime
net linux freebsd
net::dial
net::dns
diff --git a/stdlib.mk b/stdlib.mk
@@ -444,12 +444,6 @@ stdlib_deps_any+=$(stdlib_math_random_any)
stdlib_math_random_linux=$(stdlib_math_random_any)
stdlib_math_random_freebsd=$(stdlib_math_random_any)
-# gen_lib mime (any)
-stdlib_mime_any=$(HARECACHE)/mime/mime-any.o
-stdlib_deps_any+=$(stdlib_mime_any)
-stdlib_mime_linux=$(stdlib_mime_any)
-stdlib_mime_freebsd=$(stdlib_mime_any)
-
# gen_lib net (linux)
stdlib_net_linux=$(HARECACHE)/net/net-linux.o
stdlib_deps_linux+=$(stdlib_net_linux)
@@ -1318,18 +1312,6 @@ $(HARECACHE)/math/random/math_random-any.ssa: $(stdlib_math_random_any_srcs) $(s
@HARECACHE=$(HARECACHE) $(HAREC) $(HAREFLAGS) -o $@ -Nmath::random \
-t$(HARECACHE)/math/random/math_random.td $(stdlib_math_random_any_srcs)
-# mime (+any)
-stdlib_mime_any_srcs= \
- $(STDLIB)/mime/database.ha \
- $(STDLIB)/mime/lookup.ha \
- $(STDLIB)/mime/parse.ha
-
-$(HARECACHE)/mime/mime-any.ssa: $(stdlib_mime_any_srcs) $(stdlib_rt) $(stdlib_ascii_$(PLATFORM)) $(stdlib_errors_$(PLATFORM)) $(stdlib_string_$(PLATFORM))
- @printf 'HAREC \t$@\n'
- @mkdir -p $(HARECACHE)/mime
- @HARECACHE=$(HARECACHE) $(HAREC) $(HAREFLAGS) -o $@ -Nmime \
- -t$(HARECACHE)/mime/mime.td $(stdlib_mime_any_srcs)
-
# net (+linux)
stdlib_net_linux_srcs= \
$(STDLIB)/net/+linux.ha \
@@ -2262,12 +2244,6 @@ testlib_deps_any+=$(testlib_math_random_any)
testlib_math_random_linux=$(testlib_math_random_any)
testlib_math_random_freebsd=$(testlib_math_random_any)
-# gen_lib mime (any)
-testlib_mime_any=$(TESTCACHE)/mime/mime-any.o
-testlib_deps_any+=$(testlib_mime_any)
-testlib_mime_linux=$(testlib_mime_any)
-testlib_mime_freebsd=$(testlib_mime_any)
-
# gen_lib net (linux)
testlib_net_linux=$(TESTCACHE)/net/net-linux.o
testlib_deps_linux+=$(testlib_net_linux)
@@ -3172,18 +3148,6 @@ $(TESTCACHE)/math/random/math_random-any.ssa: $(testlib_math_random_any_srcs) $(
@HARECACHE=$(TESTCACHE) $(HAREC) $(TESTHAREFLAGS) -o $@ -Nmath::random \
-t$(TESTCACHE)/math/random/math_random.td $(testlib_math_random_any_srcs)
-# mime (+any)
-testlib_mime_any_srcs= \
- $(STDLIB)/mime/database.ha \
- $(STDLIB)/mime/lookup.ha \
- $(STDLIB)/mime/parse.ha
-
-$(TESTCACHE)/mime/mime-any.ssa: $(testlib_mime_any_srcs) $(testlib_rt) $(testlib_ascii_$(PLATFORM)) $(testlib_errors_$(PLATFORM)) $(testlib_string_$(PLATFORM))
- @printf 'HAREC \t$@\n'
- @mkdir -p $(TESTCACHE)/mime
- @HARECACHE=$(TESTCACHE) $(HAREC) $(TESTHAREFLAGS) -o $@ -Nmime \
- -t$(TESTCACHE)/mime/mime.td $(testlib_mime_any_srcs)
-
# net (+linux)
testlib_net_linux_srcs= \
$(STDLIB)/net/+linux.ha \
