Akihiro Sagawa : webservices: Add support for decoding decimal numeric character references.

[Alexandre Julliard]

Module: wine
Branch: master
Commit: dd4c126e956281269dca5a9025159dd1c802ed27
URL:    http://source.winehq.org/git/wine.git/?a=commit;h=dd4c126e956281269dca5a9025159dd1c802ed27

Author: Akihiro Sagawa <sagawa.aki at gmail.com>
Date:   Mon Oct 10 22:53:14 2016 +0900

webservices: Add support for decoding decimal numeric character references.

Signed-off-by: Akihiro Sagawa <sagawa.aki at gmail.com>
Signed-off-by: Hans Leidekker <hans at codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard at winehq.org>

---

 dlls/webservices/reader.c       | 50 +++++++++++++++++++++++++++++------------
 dlls/webservices/tests/reader.c |  7 ++++++
 2 files changed, 43 insertions(+), 14 deletions(-)

diff --git a/dlls/webservices/reader.c b/dlls/webservices/reader.c
index 655288a..de92603 100644
--- a/dlls/webservices/reader.c
+++ b/dlls/webservices/reader.c
@@ -1249,23 +1249,45 @@ static HRESULT decode_text( const unsigned char *str, ULONG len, unsigned char *
                 int len_utf8, cp = 0;
 
                 p++; len--;
-                if (!len || *p != 'x') return WS_E_INVALID_FORMAT;
-                p++; len--;
-
-                start = len;
-                while (len && isxdigit( *p )) { p++; len--; };
                 if (!len) return WS_E_INVALID_FORMAT;
-
-                p -= nb_digits = start - len;
-                if (!nb_digits || nb_digits > 5 || p[nb_digits] != ';') return WS_E_INVALID_FORMAT;
-                for (i = 0; i < nb_digits; i++)
+                else if (*p == 'x')
+                {
+                    p++; len--;
+
+                    start = len;
+                    while (len && isxdigit( *p )) { p++; len--; };
+                    if (!len) return WS_E_INVALID_FORMAT;
+
+                    p -= nb_digits = start - len;
+                    if (!nb_digits || nb_digits > 5 || p[nb_digits] != ';') return WS_E_INVALID_FORMAT;
+                    for (i = 0; i < nb_digits; i++)
+                    {
+                        cp *= 16;
+                        if (*p >= '0' && *p <= '9') cp += *p - '0';
+                        else if (*p >= 'a' && *p <= 'f') cp += *p - 'a' + 10;
+                        else cp += *p - 'A' + 10;
+                        p++;
+                    }
+                }
+                else if (isdigit( *p ))
                 {
-                    cp *= 16;
-                    if (*p >= '0' && *p <= '9') cp += *p - '0';
-                    else if (*p >= 'a' && *p <= 'f') cp += *p - 'a' + 10;
-                    else cp += *p - 'A' + 10;
-                    p++;
+                    while (len && *p == '0') { p++; len--; };
+                    if (!len) return WS_E_INVALID_FORMAT;
+
+                    start = len;
+                    while (len && isdigit( *p )) { p++; len--; };
+                    if (!len) return WS_E_INVALID_FORMAT;
+
+                    p -= nb_digits = start - len;
+                    if (!nb_digits || nb_digits > 7 || p[nb_digits] != ';') return WS_E_INVALID_FORMAT;
+                    for (i = 0; i < nb_digits; i++)
+                    {
+                        cp *= 10;
+                        cp += *p - '0';
+                        p++;
+                    }
                 }
+                else return WS_E_INVALID_FORMAT;
                 p++; len--;
                 if ((len_utf8 = codepoint_to_utf8( cp, q )) < 0) return WS_E_INVALID_FORMAT;
                 *ret_len += len_utf8;
diff --git a/dlls/webservices/tests/reader.c b/dlls/webservices/tests/reader.c
index 76de14a..f0fe729 100644
--- a/dlls/webservices/tests/reader.c
+++ b/dlls/webservices/tests/reader.c
@@ -3611,10 +3611,14 @@ static void test_entities(void)
     static const char str26[] = "<t>&#xffff;</t>";
     static const char str27[] = "<t><</t>";
     static const char str28[] = "<t>&#x0;</t>";
+    static const char str29[] = "<t>�</t>";
+    static const char str30[] = "<t>A</t>";
+    static const char str31[] = "<t>ｱ</t>";
     static const char res4[] = {0xea, 0xaa, 0xaa, 0x00};
     static const char res5[] = {0xf2, 0xaa, 0xaa, 0xaa, 0x00};
     static const char res21[] = {0xed, 0x9f, 0xbf, 0x00};
     static const char res24[] = {0xee, 0x80, 0x80, 0x00};
+    static const char res31[] = {0xef, 0xbd, 0xb1, 0x00};
     static const struct
     {
         const char *str;
@@ -3651,6 +3655,9 @@ static void test_entities(void)
         { str26, WS_E_INVALID_FORMAT },
         { str27, WS_E_INVALID_FORMAT },
         { str28, WS_E_INVALID_FORMAT },
+        { str29, WS_E_INVALID_FORMAT },
+        { str30, S_OK, "A" },
+        { str31, S_OK, res31 },
     };
     HRESULT hr;
     WS_XML_READER *reader;