Use existing number parsing code

unicode-org · Sep 17, 2024 · 176a232 · 176a232
1 parent 1a8fdb5
commit 176a232
Showing 1 changed file with 22 additions and 142 deletions.
diff --git a/icu4c/source/i18n/messageformat2_function_registry.cpp b/icu4c/source/i18n/messageformat2_function_registry.cpp
@@ -14,6 +14,7 @@
 #include "unicode/messageformat2_function_registry.h"
 #include "unicode/smpdtfmt.h"
 #include "charstr.h"
+#include "double-conversion.h"
 #include "messageformat2_allocation.h"
 #include "messageformat2_function_registry_internal.h"
 #include "messageformat2_macros.h"
@@ -423,74 +424,6 @@ static FormattedPlaceholder notANumber(const FormattedPlaceholder& input) {
     return FormattedPlaceholder(input, FormattedValue(UnicodeString("NaN")));
 }
 
-// Returns true if `c` is in the interval [`first`, `last`]
-static bool inRange(UChar32 c, UChar32 first, UChar32 last) {
-    U_ASSERT(first < last);
-    return c >= first && c <= last;
-}
-
-static bool isDigit(UChar32 c) { return inRange(c, 0x0030, 0x0039); }
-
-static int32_t parseDigit(UChar32 c) {
-    switch(c) {
-    case u'0': {
-        return 0;
-    }
-    case u'1': {
-        return 1;
-    }
-    case u'2': {
-        return 2;
-    }
-    case u'3': {
-        return 3;
-    }
-    case u'4': {
-        return 4;
-    }
-    case u'5': {
-        return 5;
-    }
-    case u'6': {
-        return 6;
-    }
-    case u'7': {
-        return 7;
-    }
-    case u'8': {
-        return 8;
-    }
-    case u'9': {
-        return 9;
-    }
-    default: {
-        // Should be unreachable
-        return 0;
-    }
-    }
-}
-
-static int32_t parseDigits(const UnicodeString& s, int32_t& i) {
-    int32_t result = 0;
-    while (isDigit(s[i])) {
-        result *= 10;
-        result += parseDigit(s[i]);
-        i++;
-    }
-    return result;
-}
-
-static double parseDecimalPart(const UnicodeString& s, int32_t& i) {
-    int32_t firstDigit = i;
-    double result = (double) parseDigits(s, i);
-    int32_t denominator = 1;
-    for (int32_t j = i; j > firstDigit; j--) {
-        denominator *= 10;
-    }
-    result /= denominator;
-    return result;
-}
-
 static double parseNumberLiteral(const FormattedPlaceholder& input, UErrorCode& errorCode) {
     if (U_FAILURE(errorCode)) {
         return {};
@@ -504,84 +437,31 @@ static double parseNumberLiteral(const FormattedPlaceholder& input, UErrorCode&
         return {};
     }
 
-    int32_t index = 0;
-
-#define ERROR() errorCode = U_MF_OPERAND_MISMATCH_ERROR; return 0;
-#define IN_BOUNDS(inputStr, index) (index < inputStr.length())
-#define CHECK_BOUNDS(inputStr, index) if (!IN_BOUNDS(inputStr, index)) { ERROR(); }
-
-    CHECK_BOUNDS(inputStr, index);
-
-    // Parse the sign if present
-    double sign = 1;
-    if (inputStr[index] == HYPHEN) {
-        sign = -1;
-        index++;
-    }
-
-    CHECK_BOUNDS(inputStr, index);
-
-    // Parse the integer part
-    if (!isDigit(inputStr[index])) {
-        // Non-numeric first character after sign -- not valid
-        ERROR();
-    }
-
-    // First, check for leading zero with no decimal point
-    if (inputStr[index] == u'0') {
-        bool isZero = inputStr.length() == index + 1;
-        bool hasDecimalPart = inputStr.length() > index && inputStr[index + 1] == PERIOD;
-        if (!(isZero || hasDecimalPart)) {
-            ERROR();
-        }
-    }
-
-    double result = parseDigits(inputStr, index);
+    // Hack: Check for cases that are forbidden by the MF2 grammar
+    // but allowed by StringToDouble
+    int32_t len = inputStr.length();
 
-    if (IN_BOUNDS(inputStr, index) && inputStr[index] == PERIOD) {
-        index++;
-        CHECK_BOUNDS(inputStr, index);
-        if (isDigit(inputStr[index])) {
-            result += parseDecimalPart(inputStr, index);
-        } else {
-            // '.' not followed by a digit is an error
-            ERROR();
-        }
-    }
-
-    result *= sign;
-
-    if (IN_BOUNDS(inputStr, index) &&
-        (inputStr[index] == UPPERCASE_E || inputStr[index] == LOWERCASE_E)) {
-        double exponent;
-        bool positive = true;
-        index++;
-        CHECK_BOUNDS(inputStr, index);
-        // Parse sign if present
-        if (inputStr[index] == PLUS) {
-            index++;
-        } else if (inputStr[index] == HYPHEN) {
-            positive = false;
-            index++;
-        }
-        // Parse exponent digits
-        CHECK_BOUNDS(inputStr, index);
-        if (!isDigit(inputStr[index])) {
-            ERROR();
-        }
-        exponent = parseDigits(inputStr, index);
-        if (positive) {
-            result *= (exponent * 10);
-        } else {
-            result *= (exponent * -10);
-        }
+    if (len > 0 && ((inputStr[0] == '+')
+                    || (inputStr[0] == '0' && len > 1 && inputStr[1] != '.')
+                    || (inputStr[len - 1] == '.')
+                    || (inputStr[0] == '.'))) {
+        errorCode = U_MF_OPERAND_MISMATCH_ERROR;
+        return 0;
     }
 
-    // Make sure entire input is consumed
-    if (index != inputStr.length()) {
-        ERROR();
+    // Otherwise, convert to double using double_conversion::StringToDoubleConverter
+    using namespace double_conversion;
+    int processedCharactersCount = 0;
+    StringToDoubleConverter converter(0, 0, 0, "", "");
+    LocalArray<char> target(new char(len + 1));
+    int32_t bufferLen = inputStr.extract(0, len, target.getAlias());
+    U_ASSERT(bufferLen == len);
+    double result = converter.StringToDouble(target.getAlias(),
+                                             len,
+                                             &processedCharactersCount);
+    if (processedCharactersCount != len) {
+        errorCode = U_MF_OPERAND_MISMATCH_ERROR;
     }
-
     return result;
 }