Changeset 1098 for trunk

Timestamp:

Jun 13, 2010, 6:46:29 PM (11 years ago)

Author:

Peter Johansson

Message:

closes #450. extend filename matching tp allow [xyz] and ?. Also fixed a bug when having asterisk (*) in end of pattern string

Location:

trunk

Files:

• lib/Configuration.cc (modified) (1 diff)
• lib/utility.cc (modified) (3 diffs)
• lib/utility.h (modified) (1 diff)
• test/utility_test.cc (modified) (3 diffs)

trunk/lib/Configuration.cc

@@ -366 +366 @@
     std::string res;
     std::vector<std::string> vec;
-    if (!regexp(str.begin(), str.end(), dic.first.begin(),dic.first.end(),vec)){
+    if (!regexp(dic.first, str, vec)) {
       std::stringstream mess;
       mess << "invalid config file: "

trunk/lib/utility.cc

@@ -158 +158 @@
   bool fnmatch(const std::string& pattern, const std::string& str)
   {
-    int res = ::fnmatch(pattern.c_str(), str.c_str(), FNM_PERIOD);
+    int res = ::fnmatch(pattern.c_str(), str.c_str(), 0);
     if (res==0)
       return true;
@@ -223 +223 @@
 
 
-  bool regexp(std::string::const_iterator first1, 
-              std::string::const_iterator last1,
-              std::string::const_iterator first2,
-              std::string::const_iterator last2,
+  bool regexp(const std::string& pattern, const std::string& str,
               std::vector<std::string>& vec)
   {
-    if (vec.empty())
-      vec.push_back("");
-
-    // first two cases when ranges are empty
-    if (first1==last1 && first2==last2){
-      vec.pop_back();
-      return true;
-    }
+    bool regexp__(std::string::const_iterator first1, 
+                  std::string::const_iterator last1,
+                  std::string::const_iterator first2,
+                  std::string::const_iterator last2,
+                  std::vector<std::string>::iterator item);
+    
+    // find number of special chars
+    size_t count=0;
+    for (std::string::const_iterator i=pattern.begin(); i!=pattern.end(); ++i)
+      if (*i=='*' || *i=='?' || *i=='[')
+        ++count;
+    vec.resize(count);
+    return regexp__(pattern.begin(), pattern.end(), str.begin(), str.end(),
+                    vec.begin());
+  }
+
+  bool regexp__(std::string::const_iterator first1, 
+                std::string::const_iterator last1,
+                std::string::const_iterator first2,
+                std::string::const_iterator last2,
+                std::vector<std::string>::iterator item)
+  {
+    if (first1==last1) {
+      return first2==last2;
+    }
+    if (*first1 == '*') {
+      if (first2<last2) {
+        item->push_back(*first2);
+        if (regexp__(first1, last1, first2+1, last2, item))
+          return true;
+        item->resize(item->size()-1);
+      }
+      return regexp__(first1+1, last1, first2, last2, item+1);
+    }
+    if (*first1 == '?') {
+      if (first2==last2)
+        return false;
+      *item = *first2;
+      return regexp__(first1+1, last1, first2+1, last2, item+1);
+    }
+    if (*first1 == '[') {
+      if (first2==last2)
+        return false;
+      bool found = false;
+      while (*first1 != ']') {
+        if (*first1 == *first2) {
+          found = true;
+          *item = *first2;
+        }
+        ++first1;
+        assert(first1!=last1);
+      }
+      return regexp__(first1+1, last1, first2+1, last2, item+1);
+      /*
+      while (*first2 != ']') {
+        if (*first2==*first1) {
+          found = true;
+          vec.back() = *first1;
+          vec.push_back("");
+        }
+        ++first2;
+        assert(first2!=last2);
+      }
+      if (!found)
+        return false;
+      return regexp(first1+1, last1, first2+1, last2, vec);
+      */
+    }
+
+    if (first2==last2)
+      return false;
+    if (*first1 != *first2)
+      return false;
+    return regexp__(first1+1, last1, first2+1, last2, item);
+
+    /*
+
+
+
+    if (first1==last1)
+      return false;
+    if (*first1 != *first2)
+      return false;
+    return regexp(first1+1, last1, first2+1, last2, vec);
+    
+    /// hm
     if (first1==last1 || first2==last2)
       return false;
@@ -274 +349 @@
       return false;
     return regexp(++first1, last1, ++first2, last2, vec);
+    */
   }
 

trunk/lib/utility.h

@@ -165 +165 @@
 
   /**
-     \return true if first range match second range
-
-     Second range may contain wildcards (such as '*' and '?'), in
-     which case vector \a vec is filled with the corresponding strings
-     in first range. The algorithm is greedy, i.e., wildcard * is will
+     \return true if \a str matches \a pattern
+
+     \a pattern may contain wildcards '*', '?' and \a vec will contain
+     the matching string in \a str. If it's not a match, \a vec is
+     undefined. The algorithm is greedy, i.e., wildcards '*' will
      consume as many charcters as possible.
 
      \note \a vec is supposed to be empty
    */
-  bool regexp(std::string::const_iterator first1, 
-              std::string::const_iterator end1,
-              std::string::const_iterator first2,
-              std::string::const_iterator end2,
+  bool regexp(const std::string& pattern, const std::string& str,
               std::vector<std::string>& vec);
 

trunk/test/utility_test.cc

@@ -43 +43 @@
   ok &= test_fnmatch(true,"peter", "peter"); 
   ok &= test_fnmatch(false,"peter", "peterj"); 
+
+  ok &= test_fnmatch(true,"*", "peterj"); 
   ok &= test_fnmatch(true,"p*", "peterj"); 
   ok &= test_fnmatch(true, "p*", "peter"); 
   ok &= test_fnmatch(false, "p*j", "peter"); 
+  ok &= test_fnmatch(true, "p*j", "peterj"); 
   ok &= test_fnmatch(true, "*peter", "peter"); 
+
+  ok &= test_fnmatch(true, "p?ter", "peter"); 
+  ok &= test_fnmatch(false, "p?er", "peter"); 
+  ok &= test_fnmatch(false, "p?eter", "peter"); 
 
   ok &= test_fnmatch(true, "filename", "filename");
   ok &= test_fnmatch(true, "*name", "filename");
   ok &= test_fnmatch(true, "[fa]il?name", "filename");
+  ok &= test_fnmatch(true, "[fa]*il?name", "ffilename");
+
+  ok &= test_fnmatch(true, "[fa]*il?name", "fafafailename");
+  ok &= test_fnmatch(false, "[fa]?il?name", "ilename");
+  ok &= test_fnmatch(false, "?[fa]il?name", "ilename");
+  ok &= test_fnmatch(true, "[fa]il?name", "filename");
+  ok &= test_fnmatch(false, "[fa]?il?name", "fafafailename");
+
   ok &= test_fnmatch(true, "*name", "/path/to/filename");
   ok &= test_fnmatch(true, "*name", "file.name");
-  // posix dictates that leading persion can not be matched by wildcard
-  ok &= test_fnmatch(false, "*.txt", ".file.txt");
+  // posix dictates that leading period can not be matched by
+  // wildcard, but here we allow match
+  ok &= test_fnmatch(true, "*.txt", ".file.txt");
+
 
   std::vector<std::string> vec;
   ok &= test_regexp(true,"abcde", "abcde", vec); 
+  vec.push_back("c");
+  ok &= test_regexp(true,"ab?de", "abcde", vec); 
+  vec[0] = "bcd";
+  ok &= test_regexp(true,"a*e", "abcde", vec); 
   vec.push_back("");
-  ok &= test_regexp(true,"abcde", "abcd?e", vec); 
-  vec[0]="c";
-  ok &= test_regexp(true,"abcde", "ab?de", vec); 
-  vec[0] = "bcd";
-  ok &= test_regexp(true,"abcde", "a*e", vec); 
-  vec.push_back("");
-  ok &= test_regexp(true,"abcddf", "a*d*f", vec); 
+  ok &= test_regexp(true,"a*d*f", "abcddf", vec); 
   vec[0] = "bc";
   vec[1] = "ef";
-  ok &= test_regexp(true,"abcdefg", "a*d*g", vec); 
+  ok &= test_regexp(true,"a*d*g", "abcdefg", vec); 
   vec.push_back("");
-  ok &= test_regexp(true,"abcdefg", "a*d*?g", vec); 
   vec[1]="e";
   vec[2]="f";
-  ok &= test_regexp(true,"abcdefg", "a*d??g", vec); 
+  ok &= test_regexp(true,"a*d*?g", "abcdefg", vec); 
+  ok &= test_regexp(true,"a*d??g", "abcdefg", vec); 
+  vec.resize(2);
+  vec[0]="f";
+  vec[1]="e";
+  ok &= test_regexp(true, "[fa]il?name", "filename", vec);
 
   if (ok)
@@ -84 +103 @@
   using namespace theplu::svndigest;
   bool res = fnmatch(a.c_str(), b.c_str());
-  if (res == answ)
+  // check that fnmatch and regexp agree
+  std::vector<std::string> v;
+  bool res2 = regexp(a, b, v);
+  if (res == answ && res2==answ)
     return true;
-  std::cerr << "fnmatch(" << a << ", " << b << ") results "
-            << res
-            << ". Expects " << answ << std::endl;
+  if (res!=answ)
+    std::cerr << "fnmatch(" << a << ", " << b << ") results "
+              << res
+              << ". Expects " << answ << std::endl;
+  if (res2!=answ)
+    std::cerr << "regexp(" << b << ", " << a << ") results "
+              << res2
+              << ". Expects " << answ << std::endl;
   return false;
 }
@@ -107 +134 @@
   using namespace theplu::svndigest;
   std::vector<std::string> v;
-  bool res = regexp(a.begin(), a.end(), b.begin(), b.end(), v);
+  bool res = regexp(a, b, v);
   if (res!=ans || v!=vec) {
     std::cerr << "regexp(" << a << ", " << b << ") results "