Updated regex code with tentative support for partial matches

[SVN r8160]

demo/regress/parse.cpp

@@ -77,6 +77,7 @@ flag_info flag_data[] = {
                           { BOOST_RE_STR("match_any"), 9, match_any, 3 },
                           { BOOST_RE_STR("match_not_null"), 14, match_not_null, 3 },
                           { BOOST_RE_STR("match_continuous"), 16, match_continuous, 3 },
+                          { BOOST_RE_STR("match_partial"), 13, match_partial, 3 },
 
                           { BOOST_RE_STR("format_sed"), 10, format_sed, 3 },
                           { BOOST_RE_STR("format_perl"), 11, format_perl, 3 },

demo/regress/test1252.txt

@@ -30,3 +30,5 @@
 
 
 
+
+

demo/regress/tests.cpp

@@ -236,7 +236,7 @@ void cpp_tests(const reg_expression<C, T, A>& e, bool recurse = true)
          begin_error();
          cout << "Expression did not compile using regex++ API" << endl;
       }
-      else if(recurse)
+      else if((recurse) && ((flags[3] & match_partial) == 0))
          cpp_eh_tests(e);
    }
    else if(flags[4] & REG_GREP)
@@ -299,7 +299,7 @@ void cpp_tests(const reg_expression<C, T, A>& e, bool recurse = true)
                (m[-1].first - x) << "," << (m[-1].second - x) << ") expected (0" <<
                "," << matches[0] << ")" << endl;
          }
-         if((m[-2].first != m[0].second) || (m[-2].second != y))
+         if(((m[-2].first != m[0].second) || (m[-2].second != y)) && ((flags[3] & match_partial) == 0))
          {
             begin_error();
             cout << "regex++ API result mismatch in $' (match -2), found (" <<
@@ -327,6 +327,26 @@ void cpp_tests(const reg_expression<C, T, A>& e, bool recurse = true)
                begin_error();
                cout << "regex++ API result mismatch in regex_search(const std::string&, match_results&, const reg_expression&, int)" << endl;
             }
+            //
+            // partial match should give same result as full match
+            // provided a full match is expected:
+            //
+            if(matches[0] > 0)
+            {
+               if(regex_search(x, y, m, e, flags[3] | boost::match_partial))
+               {
+                  if(compare_result(sm, m) == false)
+                  {
+                     begin_error();
+                     cout << "regex++ API result mismatch in regex_search when enabling match_partial" << endl;
+                  }
+               }
+               else
+               {
+                  begin_error();
+                  cout << "regex++ API result: match not found when match_partial specified" << endl;
+               }
+            }
             if(s.find(char_t(0)) == std::basic_string<char_t>::npos)
             {
                match_results<const char_t*> ssm;
@@ -615,7 +635,7 @@ void run_tests()
 #if !defined(TEST_UNICODE)
    try
    {
-      if((flags[2] == regbase::normal) && (has_nulls(search_text.begin(), search_text.end()) == false))
+      if(((flags[3] & match_partial) == 0) && (flags[2] == regbase::normal) && (has_nulls(search_text.begin(), search_text.end()) == false))
       {
          RegEx e;
          e.SetExpression(expression.c_str(), flags[0] & REG_ICASE);

demo/regress/tests.txt

@@ -850,6 +850,12 @@ abc|\w+? abcd 0 3
 (?:a+(b+)) xaaabbba 1 7 4 7
 a+(?#b+)b+ xaaabbba 1 7
 
+;
+; try some partial matches:
+- match_partial match_default normal REG_EXTENDED REG_NO_POSIX_TEST
+(xyz)(.*)abc xyzaaab -1 -1 0 3 3 7
+(xyz)(.*)abc xyz -1 -1 0 3 3 3
+(xyz)(.*)abc xy -1 -1 -1 -1 -1 -1
 
 
 

include/boost/cregex.hpp

@@ -171,7 +171,9 @@ enum match_flags
    match_not_null = match_any << 1,                  // string can't be null
    match_continuous = match_not_null << 1,           // each grep match must continue from
                                                      // uninterupted from the previous one
-   match_stop = match_continuous << 1,               // stop after first match (grep)
+   match_partial = match_continuous << 1,            // find partial matches
+   
+   match_stop = match_partial << 1,                  // stop after first match (grep)
    match_max = match_stop
 };
 

include/boost/re_detail/regex_match.hpp

@@ -263,7 +263,7 @@ bool query_match_aux(iterator first,
                      iterator last, 
                      match_results<iterator, Allocator>& m, 
                      const reg_expression<charT, traits, Allocator2>& e, 
-                     unsigned flags, 
+                     unsigned flags,
                      _priv_match_data<iterator, Allocator>& pd,
                      iterator* restart)
 {
@@ -287,6 +287,7 @@ bool query_match_aux(iterator first,
 
    const re_syntax_base* ptr = access::first(e);
    bool match_found = false;
+   bool have_partial_match = false;
    bool need_push_match = (e.mark_count() > 1);
    int cur_acc = -1;    // no active accumulator
    pd.set_accumulator_size(access::repeat_count(e));
@@ -748,7 +749,7 @@ bool query_match_aux(iterator first,
    //
    // if we get to here then we've run out of characters to match against,
    // we could however still have non-character regex items left
-   if(ptr->can_be_null == 0)
+   if((ptr->can_be_null == 0) && ((flags & match_partial) == 0))
       goto failure;
    while(true)
    {
@@ -838,7 +839,7 @@ bool query_match_aux(iterator first,
 
          // see if we can skip the repeat:
          if(((unsigned int)accumulators[cur_acc] >= ((re_repeat*)ptr)->min)
-            && (ptr->can_be_null & mask_skip))
+            && ((ptr->can_be_null & mask_skip) || (flags & match_partial)))
          {
             // don't push failure info, there's no point:
             ptr = ((re_repeat*)ptr)->alt.p;
@@ -847,7 +848,7 @@ bool query_match_aux(iterator first,
 
          // otherwise see if we can take the repeat:
          if(((unsigned int)accumulators[cur_acc] < ((re_repeat*)ptr)->max)
-               && ((ptr->can_be_null & (mask_take | mask_skip)) == (mask_take | mask_skip)))
+               && (((ptr->can_be_null & (mask_take | mask_skip)) == (mask_take | mask_skip))) || (flags & match_partial))
          {
             // move to next item in list:
             ++accumulators[cur_acc];
@@ -870,6 +871,18 @@ bool query_match_aux(iterator first,
 
    failure:
 
+   //
+   // check for possible partial match:
+   //
+   if((flags & match_partial)
+      && !match_found          // no full match already
+      && (base != first)       // some charcters have been consumed
+      && (first == last))      // end of input has been reached
+   {
+      have_partial_match = true;
+      m.maybe_assign(temp_match);
+   }
+
    if(prev_record.empty() == false)
    {
       ptr = prev_record.peek();
@@ -931,7 +944,7 @@ bool query_match_aux(iterator first,
       }
    }
 
-   if(match_found)
+   if(match_found || have_partial_match)
       return true;
 
    // if we get to here then everything has failed

include/boost/regex.hpp

@@ -898,7 +898,8 @@ public:
 
    size_type BOOST_RE_CALL size()const
    {
-      return (*this)[0].matched ? ref->cmatches : 0;
+      //return (*this)[0].matched ? ref->cmatches : 0;
+      return ref->cmatches;
    }
 
    const sub_match<iterator>& BOOST_RE_CALL operator[](int n) const

src/c_regex_traits.cpp

@@ -559,7 +559,7 @@ bool BOOST_RE_CALL c_regex_traits<wchar_t>::lookup_collatename(std::basic_string
    scoped_array<char> buf(new char[len]);
    strnarrow(buf.get(), len, s.c_str());
    std::string t_out;
-   bool result = base_type::do_lookup_collate(t_out, buf.get());
+   bool result = do_lookup_collate(t_out, buf.get());
    if(t_out.size() == 0) result = false;
    if(result)
    {
@@ -990,7 +990,7 @@ jm_uintfast32_t BOOST_RE_CALL c_regex_traits<wchar_t>::lookup_classname(const wc
    unsigned int len = strnarrow((char*)NULL, 0, s.c_str());
    scoped_array<char> buf(new char[len]);
    strnarrow(buf.get(), len, s.c_str());
-   len =  base_type::do_lookup_class(buf.get());
+   len =  do_lookup_class(buf.get());
    return len;
 }
 

src/w32_regex_traits.cpp

@@ -761,7 +761,7 @@ jm_uintfast32_t BOOST_RE_CALL w32_regex_traits<wchar_t>::lookup_classname(const
    unsigned int len = strnarrow((char*)NULL, 0, s.c_str());
    scoped_array<char> buf(new char[len]);
    strnarrow(buf.get(), len, s.c_str());
-   len =  base_type::do_lookup_class(buf.get());
+   len =  do_lookup_class(buf.get());
    return len;
 }