Merge pull request #936 from m-holger/hex

Refactor QUtil::hex_decode

4 files changed, 39 insertions, 65 deletions

diff --git a/include/qpdf/QUtil.hh b/include/qpdf/QUtil.hh
index b42fe195..4d46f630 100644
--- a/include/qpdf/QUtil.hh
+++ b/include/qpdf/QUtil.hh
@@ -223,6 +223,11 @@ namespace QUtil
     QPDF_DLL
     std::string hex_decode(std::string const&);
 
+    // Decode a single hex digit into a char in the range 0 <= char < 16. Return
+    // a char >= 16 if digit is not a valid hex digit.
+    QPDF_DLL
+    inline constexpr char hex_decode_char(char digit) noexcept;
+
     // Set stdin, stdout to binary mode
     QPDF_DLL
     void binary_stdout();
@@ -550,8 +555,7 @@ namespace QUtil
 inline bool
 QUtil::is_hex_digit(char ch)
 {
-    return ('0' <= ch && ch <= '9') || ('a' <= ch && ch <= 'f') ||
-        ('A' <= ch && ch <= 'F');
+    return hex_decode_char(ch) < '\20';
 }
 
 inline bool
@@ -603,4 +607,13 @@ QUtil::hex_encode_char(char c)
         '#', hexchars[static_cast<unsigned char>(c) >> 4], hexchars[c & 0x0f]};
 }
 
+inline constexpr char
+QUtil::hex_decode_char(char digit) noexcept
+{
+    return digit <= '9' && digit >= '0'
+        ? char(digit - '0')
+        : (digit >= 'a' ? char(digit - 'a' + 10)
+                        : (digit >= 'A' ? char(digit - 'A' + 10) : '\20'));
+}
+
 #endif // QUTIL_HH
diff --git a/libqpdf/JSON.cc b/libqpdf/JSON.cc
index cb60eabc..fbf06f88 100644
--- a/libqpdf/JSON.cc
+++ b/libqpdf/JSON.cc
@@ -1121,12 +1121,8 @@ JSONParser::getToken()
 
             case ls_u4:
                 using ui = unsigned int;
-                if ('0' <= *p && *p <= '9') {
-                    u_value = 16 * u_value + (ui(*p) - ui('0'));
-                } else if ('a' <= *p && *p <= 'f') {
-                    u_value = 16 * u_value + (10 + ui(*p) - ui('a'));
-                } else if ('A' <= *p && *p <= 'F') {
-                    u_value = 16 * u_value + (10 + ui(*p) - ui('A'));
+                if (ui val = ui(QUtil::hex_decode_char(*p)); val < 16) {
+                    u_value = 16 * u_value + val;
                 } else {
                     tokenError();
                 }
diff --git a/libqpdf/QPDFTokenizer.cc b/libqpdf/QPDFTokenizer.cc
index 990d5b65..d8d457ab 100644
--- a/libqpdf/QPDFTokenizer.cc
+++ b/libqpdf/QPDFTokenizer.cc
@@ -449,18 +449,9 @@ QPDFTokenizer::inNameHex1(char ch)
 {
     this->hex_char = ch;
 
-    if ('0' <= ch && ch <= '9') {
-        this->char_code = 16 * (int(ch) - int('0'));
+    if (char hval = QUtil::hex_decode_char(ch); hval < '\20') {
+        this->char_code = int(hval) << 4;
         this->state = st_name_hex2;
-
-    } else if ('A' <= ch && ch <= 'F') {
-        this->char_code = 16 * (10 + int(ch) - int('A'));
-        this->state = st_name_hex2;
-
-    } else if ('a' <= ch && ch <= 'f') {
-        this->char_code = 16 * (10 + int(ch) - int('a'));
-        this->state = st_name_hex2;
-
     } else {
         QTC::TC("qpdf", "QPDFTokenizer bad name 1");
         this->error_message = "name with stray # will not work with PDF >= 1.2";
@@ -475,15 +466,8 @@ QPDFTokenizer::inNameHex1(char ch)
 void
 QPDFTokenizer::inNameHex2(char ch)
 {
-    if ('0' <= ch && ch <= '9') {
-        this->char_code += int(ch) - int('0');
-
-    } else if ('A' <= ch && ch <= 'F') {
-        this->char_code += 10 + int(ch) - int('A');
-
-    } else if ('a' <= ch && ch <= 'f') {
-        this->char_code += 10 + int(ch) - int('a');
-
+    if (char hval = QUtil::hex_decode_char(ch); hval < '\20') {
+        this->char_code |= int(hval);
     } else {
         QTC::TC("qpdf", "QPDFTokenizer bad name 2");
         this->error_message = "name with stray # will not work with PDF >= 1.2";
@@ -675,16 +659,8 @@ QPDFTokenizer::inLiteral(char ch)
 void
 QPDFTokenizer::inHexstring(char ch)
 {
-    if ('0' <= ch && ch <= '9') {
-        this->char_code = 16 * (int(ch) - int('0'));
-        this->state = st_in_hexstring_2nd;
-
-    } else if ('A' <= ch && ch <= 'F') {
-        this->char_code = 16 * (10 + int(ch) - int('A'));
-        this->state = st_in_hexstring_2nd;
-
-    } else if ('a' <= ch && ch <= 'f') {
-        this->char_code = 16 * (10 + int(ch) - int('a'));
+    if (char hval = QUtil::hex_decode_char(ch); hval < '\20') {
+        this->char_code = int(hval) << 4;
         this->state = st_in_hexstring_2nd;
 
     } else if (ch == '>') {
@@ -706,16 +682,8 @@ QPDFTokenizer::inHexstring(char ch)
 void
 QPDFTokenizer::inHexstring2nd(char ch)
 {
-    if ('0' <= ch && ch <= '9') {
-        this->val += char(this->char_code + int(ch) - int('0'));
-        this->state = st_in_hexstring;
-
-    } else if ('A' <= ch && ch <= 'F') {
-        this->val += char(this->char_code + 10 + int(ch) - int('A'));
-        this->state = st_in_hexstring;
-
-    } else if ('a' <= ch && ch <= 'f') {
-        this->val += char(this->char_code + 10 + int(ch) - int('a'));
+    if (char hval = QUtil::hex_decode_char(ch); hval < '\20') {
+        this->val += char(this->char_code) | hval;
         this->state = st_in_hexstring;
 
     } else if (ch == '>') {
diff --git a/libqpdf/QUtil.cc b/libqpdf/QUtil.cc
index bae067b6..03301d9d 100644
--- a/libqpdf/QUtil.cc
+++ b/libqpdf/QUtil.cc
@@ -783,28 +783,25 @@ std::string
 QUtil::hex_decode(std::string const& input)
 {
     std::string result;
-    size_t pos = 0;
+    // We know result.size() <= 0.5 * input.size() + 1. However, reserving
+    // string space for this upper bound has a negative impact.
+    bool first = true;
+    char decoded;
     for (auto ch: input) {
-        bool skip = false;
-        if ((ch >= 'A') && (ch <= 'F')) {
-            ch = QIntC::to_char(ch - 'A' + 10);
-        } else if ((ch >= 'a') && (ch <= 'f')) {
-            ch = QIntC::to_char(ch - 'a' + 10);
-        } else if ((ch >= '0') && (ch <= '9')) {
-            ch = QIntC::to_char(ch - '0');
-        } else {
-            skip = true;
-        }
-        if (!skip) {
-            if (pos == 0) {
-                result.push_back(static_cast<char>(ch << 4));
-                pos = 1;
+        ch = hex_decode_char(ch);
+        if (ch < '\20') {
+            if (first) {
+                decoded = static_cast<char>(ch << 4);
+                first = false;
             } else {
-                result[result.length() - 1] |= ch;
-                pos = 0;
+                result.push_back(decoded | ch);
+                first = true;
             }
         }
     }
+    if (!first) {
+        result.push_back(decoded);
+    }
     return result;
 }