Validate numeric input using regex and simplify code.

bcoconni · Feb 9, 2025 · 7c647c6 · 7c647c6
1 parent 210c71f
commit 7c647c6
Show file tree

Hide file tree

Showing 2 changed files with 61 additions and 14 deletions.
diff --git a/src/input_output/string_utilities.cpp b/src/input_output/string_utilities.cpp
@@ -41,6 +41,7 @@ INCLUDES
 #include <iostream>
 #include <sstream>
 #include <stdio.h>
+#include <regex>
 #ifdef __APPLE__
 #include <xlocale.h>
 #else
@@ -84,26 +85,26 @@ struct CNumericLocale
  */
 double atof_locale_c(const string& input)
 {
-  string v = input;
-  trim(v);
+  static const std::regex number_format(R"(^\s*[+-]?(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?\s*$)");
+  const char* first = input.c_str();
 
-  if (v.empty()) {
-    InvalidNumber e("Expecting a numeric attribute value, but got an empty string");
+  while (*first) {
+    if (!isspace(*first)) break;
+    first++;
+  }
+
+  if (!*first) {
+    InvalidNumber e("Expecting a numeric attribute value, but only got spaces");
     cerr << e.what() << endl;
     throw e;
   }
 
-  if (v.find_first_not_of("+-.0123456789Ee") != std::string::npos) {
+  if (!std::regex_match(input, number_format)) {
     InvalidNumber e("Expecting a numeric attribute value, but got: " + input);
     cerr << e.what() << endl;
     throw e;
   }
 
-  const char* first = v.c_str();
-
-  //Ignoring the leading '+' sign
-  if (*first == '+') ++first;
-
   CNumericLocale numeric_c;
   errno = 0;          // Reset the error code
   double value = strtod_l(first, nullptr, numeric_c.Locale);

diff --git a/tests/unit_tests/StringUtilitiesTest.h b/tests/unit_tests/StringUtilitiesTest.h
@@ -89,14 +89,60 @@ class StringUtilitiesTest : public CxxTest::TestSuite
   }
 
   void testAtofLocaleC() {
-    TS_ASSERT_EQUALS(atof_locale_c("+1  "), 1.0);
+    // Test legal numbers
+    TS_ASSERT_EQUALS(atof_locale_c("0.0"), 0.0);
+    TS_ASSERT_EQUALS(atof_locale_c("+1"), 1.0);
+    TS_ASSERT_EQUALS(atof_locale_c("1"), 1.0);
+    TS_ASSERT_EQUALS(atof_locale_c("-1"), -1.0);
     TS_ASSERT_EQUALS(atof_locale_c(" 123.4"), 123.4);
+    TS_ASSERT_EQUALS(atof_locale_c("123.4 "), 123.4);
+    TS_ASSERT_EQUALS(atof_locale_c(" 123.4 "), 123.4);
+    TS_ASSERT_EQUALS(atof_locale_c(".25"), 0.25);
+    TS_ASSERT_EQUALS(atof_locale_c("1.e1"), 10.);
+    TS_ASSERT_EQUALS(atof_locale_c("1e1"), 10.);
+    TS_ASSERT_EQUALS(atof_locale_c(".1e1"), 1.);
+    TS_ASSERT_EQUALS(atof_locale_c("+.1e1"), 1.);
+    TS_ASSERT_EQUALS(atof_locale_c("-.1e1"), -1.);
+    TS_ASSERT_EQUALS(atof_locale_c("31.4e1"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("+3.14e+2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("+3.14e2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("+3.14e-2"), 0.0314);
+    TS_ASSERT_EQUALS(atof_locale_c("3.14e+2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("3.14e2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("3.14e-2"), 0.0314);
+    TS_ASSERT_EQUALS(atof_locale_c("-3.14e+2"), -314);
+    TS_ASSERT_EQUALS(atof_locale_c("-3.14e2"), -314);
     TS_ASSERT_EQUALS(atof_locale_c("-3.14e-2"), -0.0314);
+    TS_ASSERT_EQUALS(atof_locale_c("+3.14E+2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("+3.14E2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("+3.14E-2"), 0.0314);
+    TS_ASSERT_EQUALS(atof_locale_c("3.14E+2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("3.14E2"), 314);
+    TS_ASSERT_EQUALS(atof_locale_c("3.14E-2"), 0.0314);
+    TS_ASSERT_EQUALS(atof_locale_c("-3.14E+2"), -314);
+    TS_ASSERT_EQUALS(atof_locale_c("-3.14E2"), -314);
+    TS_ASSERT_EQUALS(atof_locale_c("-3.14E-2"), -0.0314);
+    // Test rounded down numbers
     TS_ASSERT_EQUALS(atof_locale_c("1E-999"), 0.0);
     TS_ASSERT_EQUALS(atof_locale_c("-1E-999"), 0.0);
-    TS_ASSERT_EQUALS(atof_locale_c("0.0"), 0.0);
-    TS_ASSERT_THROWS(atof_locale_c("1E+999"), BaseException&);
-    TS_ASSERT_THROWS(atof_locale_c("-1E+999"), BaseException&);
+    // Test invalid numbers
+    TS_ASSERT_THROWS(atof_locale_c("1E+999"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("-1E+999"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("invalid"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("1.0.0"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("1E-"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("E-2"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("."), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c(".E"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c(".E2"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c(".E-2"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("1.2E"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("1.2E+"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("1.2E1.0"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("--1"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c("++1"), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c(""), InvalidNumber&);
+    TS_ASSERT_THROWS(atof_locale_c(" "), InvalidNumber&);
   }
 
 private: