From 3aaa4e2a534267dda2a22e97db3efcb8e4757536 Mon Sep 17 00:00:00 2001
From: Pierre Le Marre <dev@wismill.eu>
Date: Mon, 30 Oct 2023 15:51:34 +0100
Subject: [PATCH] rules: early detection of invalid encoding

---
 src/xkbcomp/rules.c               |  25 +++++++++++++--------
 test/data/rules/utf-16be_with_bom | Bin 0 -> 1146 bytes
 test/data/rules/utf-16le_with_bom | Bin 0 -> 1146 bytes
 test/data/rules/utf-32be          | Bin 0 -> 2288 bytes
 test/rules-file.c                 |  36 ++++++++++++++++++++++++++++++
 5 files changed, 52 insertions(+), 9 deletions(-)
 create mode 100644 test/data/rules/utf-16be_with_bom
 create mode 100644 test/data/rules/utf-16le_with_bom
 create mode 100644 test/data/rules/utf-32be

diff --git a/src/xkbcomp/rules.c b/src/xkbcomp/rules.c
index daa4f3d..918fe31 100644
--- a/src/xkbcomp/rules.c
+++ b/src/xkbcomp/rules.c
@@ -1084,29 +1084,36 @@ read_rules_file(struct xkb_context *ctx,
                 FILE *file,
                 const char *path)
 {
-    bool ret = false;
+    bool ret;
     char *string;
     size_t size;
     struct scanner scanner;
 
-    ret = map_file(file, &string, &size);
-    if (!ret) {
+    if (!map_file(file, &string, &size)) {
         log_err(ctx, XKB_LOG_MESSAGE_NO_ID,
                 "Couldn't read rules file \"%s\": %s\n",
                 path, strerror(errno));
-        goto out;
+        return false;
     }
 
     scanner_init(&scanner, matcher->ctx, string, size, path, NULL);
 
-    /* Skip UTF-8 encoded BOM (U+FEFF) */
-    /* See: https://www.unicode.org/faq/utf_bom.html#bom5 */
-    scanner_str(&scanner, "\xef\xbb\xbf", 3);
+    /* Basic detection of wrong character encoding.
+       The first character relevant to the grammar must be ASCII:
+       whitespace, !, / (for comment) */
+    if (!scanner_check_supported_char_encoding(&scanner)) {
+        scanner_err(&scanner,
+            "This could be a file encoding issue. "
+            "Supported encodings must be backward compatible with ASCII.");
+        scanner_err(&scanner,
+            "E.g. ISO/CEI 8859 and UTF-8 are supported "
+            "but UTF-16, UTF-32 and CP1026 are not.");
+        unmap_file(string, size);
+        return false;
+    }
 
     ret = matcher_match(matcher, &scanner, include_depth, string, size, path);
-
     unmap_file(string, size);
-out:
     return ret;
 }
 
diff --git a/test/data/rules/utf-16be_with_bom b/test/data/rules/utf-16be_with_bom
new file mode 100644
index 0000000000000000000000000000000000000000..ea44babc56209518ae6860a59719fc3a70f6ee8c
GIT binary patch
literal 1146
zcmb7^Pfx-?5XIk_Pf>3wF~*Ar6E7xk^3O%jq)HKDOARe5iJx8l4NGiySprSdc6N5&
zn>Vx5{`~5kE783k^r(TZuxm{<Q-w9*oh!p~>r8XKsMHwWc<r6@d;&StnJ#pxTR0MZ
z5^pStQf^~NiJyUfWqm`gJKo12`ydD0L!GFnBfQ0iKgSNQYdmkTt^QS1pe(AuH`;+U
zl-rMEPuC1#F1I$URXd@AC6&}<*=BLdb7EGgUlJ9veXzHn11UXdc8$FPQ|K97j@_;%
z&gUu#`I;JizCPN&$haW$NIt5!>3BR(Q@_u+U*Rq-4%xNaVfMqOq^7Iy)b4jQK3k36
z5Y<&{8-JKi`<e=rrkqi`v-^~8<GO0~8JyY@B{fr8GNs`JTbF!{MYw5<;|%f530Q?0
Ie-;n@2MrpgrT_o{

literal 0
HcmV?d00001

diff --git a/test/data/rules/utf-16le_with_bom b/test/data/rules/utf-16le_with_bom
new file mode 100644
index 0000000000000000000000000000000000000000..9faf37e9a5cefc91d7544f7874c00ee309b63b38
GIT binary patch
literal 1146
zcmb7^O;5r=5Qg6~zoOn$Vu%+HCSFY7<i|zNq)Jg@OARe5iN9Tah9$PUEP;lUZg=N-
z=bh=m@6SL3CA!nS9yHV?>sqgxsbV$do-1SJo-@t$tWp#9ChOff-;a<Zo#|W`x<Mn+
zC-J6|DCIJRl=vC+3(wc&y5)Wh*@ql*jdY@(j@T_W`Z??HbH&aZYO8-26_iC4?2UJ@
zM!EAid%9K#bGc_@t=cISEUBa>%eIPho)fde{gS9q?Sr!g9Z2axb86}pOra;ZoOQdJ
zxSp#d)N5+=_4;`KBIAP0WBF8X)A4worhZ>>zr$T#9ND$oVfMqOq^7Iy)b4jQK3k37
zkm~BS%^&8|eoX~VQ{GX#v-_NGb6vgq3eIhblbR_lnbL5AtxG;*5pEjeoDtu=0jn_M
IpT(#D0I|cSrT_o{

literal 0
HcmV?d00001

diff --git a/test/data/rules/utf-32be b/test/data/rules/utf-32be
new file mode 100644
index 0000000000000000000000000000000000000000..588e32e07ff105a4446bdf57ea97e51d97782c47
GIT binary patch
literal 2288
zcmcJRNlU~)5Jq$MSCls$5%J<d@FM8R<)UX9mr-%WUGT^I9;ToXv*G0M=uRiqU)9%@
z%v4piR{ao+2XF{SumhVA@B&wG4bKoT1Fvu40s_kWuHC~KJis})<{03!-gkZsYVN`+
ztiuNEfq4YHgE5(lYg2H{9K3e_-20Mwz&2RdJ_Ia*nk%5D8ji9I3$O?QcTnc9mg`+?
z)z928TK6aX_Chb!iW;?i?gWPDrdAn0M0Zl_$G|al^K56tPvI5x$+NPSKDm^h8|yxL
zq4)0RUeUoD_l<i3d%(Fp_E&7??ir|~cc-A9`fX=3B){@Zy7J4M%9ytN>|cB91FX@y
zty$08wDfu&erLXIIjR#iWnDu&Or0{mujZuC&pi5!soB-rU3ru~oBw~lC7!Xp`Usx8
z+D-lr(aKs&f4j8C_O(fWo_^-{*t+%9cSqf^Jnf@5wy$kDl-}B>dZHhGOTEtD>tNnQ
d_EMWp#+LmyL@zbI$isKQSoiXs@IOnV_BW|Tq^SS^

literal 0
HcmV?d00001

diff --git a/test/rules-file.c b/test/rules-file.c
index 302aa68..726ec89 100644
--- a/test/rules-file.c
+++ b/test/rules-file.c
@@ -106,6 +106,42 @@ main(int argc, char *argv[])
     };
     assert(test_rules(ctx, &test_utf_8_with_bom));
 
+    struct test_data test_utf_16le_with_bom = {
+        .rules = "utf-16le_with_bom",
+
+        .model = "my_model", .layout = "my_layout", .variant = "my_variant",
+        .options = "my_option",
+
+        .keycodes = "my_keycodes", .types = "my_types",
+        .compat = "my_compat|some:compat",
+        .symbols = "my_symbols+extra_variant",
+    };
+    assert(!test_rules(ctx, &test_utf_16le_with_bom));
+
+    struct test_data test_utf_16be_with_bom = {
+        .rules = "utf-16be_with_bom",
+
+        .model = "my_model", .layout = "my_layout", .variant = "my_variant",
+        .options = "my_option",
+
+        .keycodes = "my_keycodes", .types = "my_types",
+        .compat = "my_compat|some:compat",
+        .symbols = "my_symbols+extra_variant",
+    };
+    assert(!test_rules(ctx, &test_utf_16be_with_bom));
+
+    struct test_data test_utf_32be = {
+        .rules = "utf-32be",
+
+        .model = "my_model", .layout = "my_layout", .variant = "my_variant",
+        .options = "my_option",
+
+        .keycodes = "my_keycodes", .types = "my_types",
+        .compat = "my_compat|some:compat",
+        .symbols = "my_symbols+extra_variant",
+    };
+    assert(!test_rules(ctx, &test_utf_32be));
+
     struct test_data test1 = {
         .rules = "simple",