ICU-64260.0.1.tar.gz

[apple/icu.git] / icuSources / common / unames.cpp

diff --git a/icuSources/common/unames.cpp b/icuSources/common/unames.cpp
index a2035c48dfc9c6bf4a658ed4b2cdbee06bdb3776..d9f61cac1575a9acfe9f8be9e153c8a4b0db5438 100644 (file)
--- a/icuSources/common/unames.cpp
+++ b/icuSources/common/unames.cpp
@@ -1,3 +1,5 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html

 /*
 ******************************************************************************
 *
 /*
 ******************************************************************************
 *


@@ -6,7 +8,7 @@
 *
 ******************************************************************************
 *   file name:  unames.c
 *
 ******************************************************************************
 *   file name:  unames.c

-*   encoding:   US-ASCII
+*   encoding:   UTF-8

 *   tab size:   8 (not used)
 *   indentation:4
 *
 *   tab size:   8 (not used)
 *   indentation:4
 *


@@ -464,7 +466,7 @@ static uint16_t getExtName(uint32_t code, char *buffer, uint16_t bufferLength) {
         buffer[--i] = (v < 10 ? '0' + v : 'A' + v - 10);
     }
     buffer += ndigits;
         buffer[--i] = (v < 10 ? '0' + v : 'A' + v - 10);
     }
     buffer += ndigits;

-    length += ndigits;
+    length += static_cast<uint16_t>(ndigits);

     WRITE_CHAR(buffer, bufferLength, length, '>');
 
     return length;
     WRITE_CHAR(buffer, bufferLength, length, '>');
 
     return length;


@@ -1524,7 +1526,7 @@ u_charFromName(UCharNameChoice nameChoice,
     uint32_t i;
     UChar32 cp = 0;
     char c0;
     uint32_t i;
     UChar32 cp = 0;
     char c0;

-    UChar32 error = 0xffff;     /* Undefined, but use this for backwards compatibility. */
+    static constexpr UChar32 error = 0xffff;     /* Undefined, but use this for backwards compatibility. */

 
     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
         return error;
 
     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
         return error;


@@ -1558,39 +1560,45 @@ u_charFromName(UCharNameChoice nameChoice,
 
     /* try extended names first */
     if (lower[0] == '<') {
 
     /* try extended names first */
     if (lower[0] == '<') {

-        if (nameChoice == U_EXTENDED_CHAR_NAME) {
+        if (nameChoice == U_EXTENDED_CHAR_NAME && lower[--i] == '>') {

             // Parse a string like "<category-HHHH>" where HHHH is a hex code point.
             // Parse a string like "<category-HHHH>" where HHHH is a hex code point.

-            if (lower[--i] == '>' && i >= 3 && lower[--i] != '-') {
-                while (i >= 3 && lower[--i] != '-') {}
-
-                if (i >= 2 && lower[i] == '-') {
-                    uint32_t cIdx;
-
-                    lower[i] = 0;
-
-                    for (++i; lower[i] != '>'; ++i) {
-                        if (lower[i] >= '0' && lower[i] <= '9') {
-                            cp = (cp << 4) + lower[i] - '0';
-                        } else if (lower[i] >= 'a' && lower[i] <= 'f') {
-                            cp = (cp << 4) + lower[i] - 'a' + 10;
-                        } else {
-                            *pErrorCode = U_ILLEGAL_CHAR_FOUND;
-                            return error;
-                        }
-                    }
+            uint32_t limit = i;
+            while (i >= 3 && lower[--i] != '-') {}
+
+            // There should be 1 to 8 hex digits.
+            int32_t hexLength = limit - (i + 1);
+            if (i >= 2 && lower[i] == '-' && 1 <= hexLength && hexLength <= 8) {
+                uint32_t cIdx;
+
+                lower[i] = 0;

 
 

-                    /* Now validate the category name.
-                       We could use a binary search, or a trie, if
-                       we really wanted to. */
+                for (++i; i < limit; ++i) {
+                    if (lower[i] >= '0' && lower[i] <= '9') {
+                        cp = (cp << 4) + lower[i] - '0';
+                    } else if (lower[i] >= 'a' && lower[i] <= 'f') {
+                        cp = (cp << 4) + lower[i] - 'a' + 10;
+                    } else {
+                        *pErrorCode = U_ILLEGAL_CHAR_FOUND;
+                        return error;
+                    }
+                    // Prevent signed-integer overflow and out-of-range code points.
+                    if (cp > UCHAR_MAX_VALUE) {
+                        *pErrorCode = U_ILLEGAL_CHAR_FOUND;
+                        return error;
+                    }
+                }

 
 

-                    for (lower[i] = 0, cIdx = 0; cIdx < UPRV_LENGTHOF(charCatNames); ++cIdx) {
+                /* Now validate the category name.
+                   We could use a binary search, or a trie, if
+                   we really wanted to. */
+                uint8_t cat = getCharCat(cp);
+                for (lower[i] = 0, cIdx = 0; cIdx < UPRV_LENGTHOF(charCatNames); ++cIdx) {

 
 

-                        if (!uprv_strcmp(lower + 1, charCatNames[cIdx])) {
-                            if (getCharCat(cp) == cIdx) {
-                                return cp;
-                            }
-                            break;
+                    if (!uprv_strcmp(lower + 1, charCatNames[cIdx])) {
+                        if (cat == cIdx) {
+                            return cp;

                         }
                         }

+                        break;

                     }
                 }
             }
                     }
                 }
             }