possibility to evaluate and extract sub expressions

mrw/regexp.cpp

@@ -9,28 +9,46 @@
     @license LGPL, see file <a href="license.html">COPYING</a>
 
     $Log$
+    Revision 1.2  2004/12/16 13:09:31  marc
+    possibility to evaluate and extract sub expressions
+
     Revision 1.1  2004/12/14 20:20:30  marc
     initial version
 
 
 */
 
+#include <mrw/string.hpp>
 #include <mrw/regexp.hpp>
 #include <mrw/exception.hpp>
 
 namespace mrw {
 
-  RegExp::RegExp(const std::string& pattern, int flags)
-    throw(std::exception, std::bad_exception) {
-    if (regcomp(&_regex, pattern.c_str(), flags|nosub))
-      throw(mrw::invalid_argument(pattern));
+  RegExp::RegExp(const std::string& pattern, bool hassub, int flags)
+    throw(std::exception):
+    _hassub(hassub) {
+      if (flags&nosub) throw mrw::invalid_argument("nosub");
+      if (regcomp(&_regex, pattern.c_str(), (_hassub?flags:(flags|nosub))))
+        throw mrw::invalid_argument(pattern);
   }
 
   RegExp::~RegExp() throw() {
     regfree(&_regex);
   }
 
-  bool RegExp::operator()(const std::string& text) const throw() {
-    return !regexec(const_cast<regex_t*>(&_regex), text.c_str(), 0, 0, 0);
+  bool RegExp::operator()(const std::string& text) throw(std::bad_exception) {
+    if (_hassub)
+      return !regexec(&_regex, (_text=text).c_str(), MAX_SUB, _sub, 0);
+    else
+      return !regexec(&_regex, text.c_str(), 0, 0, 0);
   }
+
+  std::string RegExp::operator[](unsigned int n) const throw(std::exception) {
+    if (!_hassub)
+      throw mrw::invalid_argument("initialized with no sub expressions");
+    if (n>=MAX_SUB || _sub[n].rm_so<0 || _sub[n].rm_eo<0)
+      throw mrw::invalid_argument(mrw::string(n));
+    return _text.substr(_sub[n].rm_so, _sub[n].rm_eo-_sub[n].rm_so);
+  }
+
 }

mrw/regexp.hpp

@@ -9,6 +9,9 @@
     @license LGPL, see file <a href="license.html">COPYING</a>
 
     $Log$
+    Revision 1.2  2004/12/16 13:09:31  marc
+    possibility to evaluate and extract sub expressions
+
     Revision 1.1  2004/12/14 20:20:30  marc
     initial version
 
@@ -57,6 +60,9 @@ namespace mrw {
   
   public:
 
+    /// The maximum number of sub expressions that are evaluated.
+    static const unsigned int MAX_SUB = 99;
+
     /** @brief flags that influence regular expressions
 
         Flag @c newline treats a newline in the text to be compared as
@@ -82,16 +88,20 @@ namespace mrw {
         The regular expression is compiled on instanciation and can
         then be matced several times on different texts.
 
-        @param pattern the regular expression pattern, thee the @c man
+        @param pattern the regular expression pattern, see the @c man
                        page for POSIX regular expressions (on linux: @c
                        info&nbsp;7&nbsp;regex)
-        @param flags   special flags, they default to extended|nosub and
+        @param hassub  pass @c true if you want to evaluate sub expressions
+        @param flags   special flags, they default to extended and
                        should consist of the Flag values combined with @c |
+                       flag @c nosub must not be used, because it is set
+                       automatically if necessary
 
-        @throw std::invalid_argument if pattern compilation fails
+        @throw mrw::invalid_argument if pattern compilation fails or @c nosub
+                                     was set in @c flags
     */
-    RegExp(const std::string& pattern, int flags = extended)
-      throw(std::exception, std::bad_exception);
+    RegExp(const std::string& pattern, bool hassub=false, int flags=extended)
+      throw(std::exception);
 
     /** @brief cleans up expression from memory */
     ~RegExp() throw();
@@ -102,11 +112,28 @@ namespace mrw {
         @return
           - true if @c text matches
           - false otherwise */
-    bool operator()(const std::string& text) const throw();
+    bool operator()(const std::string& text) throw(std::bad_exception);
+
+    /** @brief get the n-th sub expression of the last matched text
+
+        If the RegExp was instanciated with @c Regexp(pattern, true),
+        so that sub expressions are evaluated, then you can get the
+        n-th matched sub expression.
+
+        @param n the number of sub expression to get, get the n-th sub
+                 expression
+
+        @throw mrw::invalid_argument if this subexpression is not
+                                     available
+    */
+    std::string operator[](unsigned int n) const throw(std::exception);
   
   private:
   
     regex_t _regex;
+    bool _hassub;
+    regmatch_t _sub[MAX_SUB];
+    std::string _text;
   
   };
 }

mrw/regexp_test.cpp

@@ -9,6 +9,9 @@
     @license LGPL, see file <a href="license.html">COPYING</a>
 
     $Log$
+    Revision 1.2  2004/12/16 13:09:31  marc
+    possibility to evaluate and extract sub expressions
+
     Revision 1.1  2004/12/14 20:20:30  marc
     initial version
 
@@ -16,6 +19,7 @@
 */
 
 #include <mrw/regexp.hpp>
+#include <mrw/exception.hpp>
 
 #include <cppunit/TestFixture.h>
 #include <cppunit/ui/text/TestRunner.h>
@@ -31,9 +35,23 @@ public:
     CPPUNIT_ASSERT(!findHalloWelt(""));
     CPPUNIT_ASSERT(!findHalloWelt(" Hallo Welt "));
     CPPUNIT_ASSERT(findHalloWelt("HalloWelt"));
+    mrw::RegExp extractTest("^Guten (.*) (Herr|Frau) (.*)$", true);
+    CPPUNIT_ASSERT(extractTest("Guten Tag Frau Zuercher"));
+    CPPUNIT_ASSERT(extractTest[1]=="Tag" &&
+                   extractTest[2]=="Frau" &&
+                   extractTest[3]=="Zuercher");
+  }
+  void ExceptionTest() {
+    mrw::RegExp extractTest("^Guten (.*) (Herr|Frau) (.*)$", true);
+    CPPUNIT_ASSERT(extractTest("Guten Tag Herr Schweizer"));
+    CPPUNIT_ASSERT(extractTest[1]=="Tag" &&
+                   extractTest[2]=="Herr" &&
+                   extractTest[3]=="Schweizer");
+    std::string s = extractTest[4];
   }
   CPPUNIT_TEST_SUITE(RegExpTest);
   CPPUNIT_TEST(CheckRegExp);
+  CPPUNIT_TEST_EXCEPTION(ExceptionTest, mrw::invalid_argument);
   CPPUNIT_TEST_SUITE_END();
 };
 CPPUNIT_TEST_SUITE_REGISTRATION(RegExpTest);