regExp: Add support for case-insensitive patterns

From https://github.com/OpenFOAM/OpenFOAM-2.2.x/pull/1

applications/test/regex/Test-regex.C

@@ -2,7 +2,7 @@
   =========                 |
   \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox
    \\    /   O peration     |
-    \\  /    A nd           | Copyright (C) 2011 OpenFOAM Foundation
+    \\  /    A nd           | Copyright (C) 2011-2015 OpenFOAM Foundation
      \\/     M anipulation  |
 -------------------------------------------------------------------------------
 License
@@ -22,6 +22,7 @@ License
     along with OpenFOAM.  If not, see <http://www.gnu.org/licenses/>.
 
 Description
+    Tests for regular expressions
 
 \*---------------------------------------------------------------------------*/
 
@@ -39,14 +40,13 @@ using namespace Foam;
 
 int main(int argc, char *argv[])
 {
-
     List<Tuple2<string, string> > rawList(IFstream("testRegexps")());
-    Info<< "input list:" << rawList << endl;
+    Info<< "Test expressions:" << rawList << endl;
     IOobject::writeDivider(Info) << endl;
 
     List<string> groups;
 
-    // report matches:
+    // Report matches:
     forAll(rawList, elemI)
     {
         const string& pat = rawList[elemI].first();
@@ -60,50 +60,87 @@ int main(int argc, char *argv[])
             Info<< "true";
             if (re.ngroups())
             {
-                Info<< " groups:" << groups;
+                Info<< nl << "groups: " << groups;
             }
         }
         else
         {
-            Info<< "false";
             if (re.search(str))
             {
                 Info<< " partial match";
             }
+            else
+            {
+                Info<< "false";
+            }
         }
         Info<< endl;
     }
 
-    Info<<"test regExp(const char*) ..." << endl;
+    Info<< nl << "test regExp(const char*) ..." << endl;
     string me("Mark");
 
-    if (regExp("[Mm]ar[ck]").match(me))
+    // Handling of null strings
-    {
-        Info<< "matched: " << me << endl;
-    }
-    else
-    {
-        Info<< "no match" << endl;
-    }
-
-    if (regExp("").match(me))
-    {
-        Info<< "matched: " << me << endl;
-    }
-    else
-    {
-        Info<< "no match" << endl;
-    }
-
     if (regExp(NULL).match(me))
     {
-        Info<< "matched: " << me << endl;
+        Info<< "fail - matched: " << me << endl;
+    }
+    else
+    {
+        Info<< "pass - null pointer is no expression" << endl;
+    }
+
+    // Normal match
+    if (regExp("[Mm]ar[ck]").match(me))
+    {
+        Info<< "pass - matched: " << me << endl;
     }
     else
     {
         Info<< "no match" << endl;
     }
 
+    // Match ignore case
+    if (regExp("mar[ck]", true).match(me))
+    {
+        Info<< "pass - matched: " << me << endl;
+    }
+    else
+    {
+        Info<< "no match" << endl;
+    }
+
+    // Embedded prefix for match ignore case
+    if (regExp("(?i)mar[ck]").match(me))
+    {
+        Info<< "pass - matched: " << me << endl;
+    }
+    else
+    {
+        Info<< "no match" << endl;
+    }
+
+    // Handling of empty expression
+    if (regExp("").match(me))
+    {
+        Info<< "fail - matched: " << me << endl;
+    }
+    else
+    {
+        Info<< "pass - no match on empty expression" << endl;
+    }
+
+    // Embedded prefix - but expression is empty
+    if (regExp("(?i)").match(me))
+    {
+        Info<< "fail - matched: " << me << endl;
+    }
+    else
+    {
+        Info<< "pass - no match on empty expression" << endl;
+    }
+
+
     Info<< endl;
 
     return 0;

applications/test/regex/testRegexps

@@ -8,14 +8,30 @@
 
 // * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
 
-// pattern, string
+// Pattern, String
 (
-    ( "a.*" "abc" )
+    ( "a.*"  "abc" )        // true
-    ( "a.*" "bac" )
+    ( "a.*"  "bac" )        // false - partial match only
-    ( "a.*" "abcd" )
+    ( "a.*"  "abcd" )       // true
-    ( "a.*" "def" )
+    ( "a.*"  "def" )        // false
-    ( "d(.*)f" "def" )
+    ( ".*a.*"  "Abc" )      // false
-    ( " *([A-Za-z]+) *= *([^ /]+) *(//.*)?"   " keyword = value // settings" )
+    ( "(?i).*a.*"  "Abc" )  // true
+    ( "d(.*)f"  "def" )     // true
+    (
+        " *([A-Za-z]+) *= *([^ /]+) *(//.*)?"
+        "   keyword  = value // comment"
+    )   // true
+
+    (
+        "[[:digit:]]"
+        "contains 1 or more digits"
+    )   // false - partial match only
+
+    (
+        "[[:digit:]]+-[[:digit:]]+-[[:digit:]]+-[[:digit:]]+"
+        "1-905-123-2234"
+    )   // true
+
 )
 
 // * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

src/OSspecific/POSIX/regExp.C

@@ -2,7 +2,7 @@
   =========                 |
   \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox
    \\    /   O peration     |
-    \\  /    A nd           | Copyright (C) 2011-2012 OpenFOAM Foundation
+    \\  /    A nd           | Copyright (C) 2011-2015 OpenFOAM Foundation
      \\/     M anipulation  |
 -------------------------------------------------------------------------------
 License
@@ -23,13 +23,61 @@ License
 
 \*---------------------------------------------------------------------------*/
 
-#include <sys/types.h>
-
 #include "regExp.H"
-#include "label.H"
 #include "string.H"
 #include "List.H"
-#include "IOstreams.H"
+
+// * * * * * * * * * * * * * Private Member Functions  * * * * * * * * * * * //
+
+template<class StringType>
+bool Foam::regExp::matchGrouping
+(
+    const std::string& str,
+    List<StringType>& groups
+) const
+{
+    if (preg_ && str.size())
+    {
+        size_t nmatch = ngroups() + 1;
+        regmatch_t pmatch[nmatch];
+
+        // Also verify that the entire string was matched.
+        // pmatch[0] is the entire match
+        // pmatch[1..] are the (...) sub-groups
+        if
+        (
+            regexec(preg_, str.c_str(), nmatch, pmatch, 0) == 0
+         && (pmatch[0].rm_so == 0 && pmatch[0].rm_eo == label(str.size()))
+        )
+        {
+            groups.setSize(ngroups());
+            label groupI = 0;
+
+            for (size_t matchI = 1; matchI < nmatch; matchI++)
+            {
+                if (pmatch[matchI].rm_so != -1 && pmatch[matchI].rm_eo != -1)
+                {
+                    groups[groupI] = str.substr
+                    (
+                        pmatch[matchI].rm_so,
+                        pmatch[matchI].rm_eo - pmatch[matchI].rm_so
+                    );
+                }
+                else
+                {
+                    groups[groupI].clear();
+                }
+                groupI++;
+            }
+
+            return true;
+        }
+    }
+
+    groups.clear();
+    return false;
+}
+
 
 // * * * * * * * * * * * * * * * * Constructors  * * * * * * * * * * * * * * //
 
@@ -69,18 +117,34 @@ void Foam::regExp::set(const char* pattern, const bool ignoreCase) const
 {
     clear();
 
-    // avoid NULL pointer and zero-length patterns
+    // Avoid NULL pointer and zero-length patterns
     if (pattern && *pattern)
     {
-        preg_ = new regex_t;
-
         int cflags = REG_EXTENDED;
         if (ignoreCase)
         {
             cflags |= REG_ICASE;
         }
 
-        int err = regcomp(preg_, pattern, cflags);
+        const char* pat = pattern;
+
+        // Check for embedded prefix for ignore-case
+        // this is the only embedded prefix we support
+        // - a simple check is sufficient
+        if (!strncmp(pattern, "(?i)", 4))
+        {
+            cflags |= REG_ICASE;
+            pat += 4;
+
+            // avoid zero-length patterns
+            if (!*pat)
+            {
+                return;
+            }
+        }
+
+        preg_ = new regex_t;
+        int err = regcomp(preg_, pat, cflags);
 
         if (err != 0)
         {
@@ -89,7 +153,7 @@ void Foam::regExp::set(const char* pattern, const bool ignoreCase) const
 
             FatalErrorIn
             (
-                "regExp::set(const char*)"
+                "regExp::set(const char*, const bool ignoreCase)"
             )   << "Failed to compile regular expression '" << pattern << "'"
                 << nl << errbuf
                 << exit(FatalError);
@@ -143,7 +207,7 @@ bool Foam::regExp::match(const std::string& str) const
         size_t nmatch = 1;
         regmatch_t pmatch[1];
 
-        // also verify that the entire string was matched
+        // Also verify that the entire string was matched
         // pmatch[0] is the entire match
         if
         (
@@ -159,48 +223,23 @@ bool Foam::regExp::match(const std::string& str) const
 }
 
 
-bool Foam::regExp::match(const string& str, List<string>& groups) const
+bool Foam::regExp::match
+(
+    const std::string& str,
+    List<std::string>& groups
+) const
 {
-    if (preg_ && str.size())
+    return matchGrouping(str, groups);
-    {
+}
-        size_t nmatch = ngroups() + 1;
-        regmatch_t pmatch[nmatch];
 
-        // also verify that the entire string was matched
-        // pmatch[0] is the entire match
-        // pmatch[1..] are the (...) sub-groups
-        if
-        (
-            regexec(preg_, str.c_str(), nmatch, pmatch, 0) == 0
-         && (pmatch[0].rm_so == 0 && pmatch[0].rm_eo == label(str.size()))
-        )
-        {
-            groups.setSize(ngroups());
-            label groupI = 0;
 
-            for (size_t matchI = 1; matchI < nmatch; matchI++)
+bool Foam::regExp::match
-            {
+(
-                if (pmatch[matchI].rm_so != -1 && pmatch[matchI].rm_eo != -1)
+    const std::string& str,
-                {
+    List<Foam::string>& groups
-                    groups[groupI] = str.substr
+) const
-                    (
+{
-                        pmatch[matchI].rm_so,
+    return matchGrouping(str, groups);
-                        pmatch[matchI].rm_eo - pmatch[matchI].rm_so
-                    );
-                }
-                else
-                {
-                    groups[groupI].clear();
-                }
-                groupI++;
-            }
-
-            return true;
-        }
-    }
-
-    groups.clear();
-    return false;
 }
 
 

src/OSspecific/POSIX/regExp.H

@@ -2,7 +2,7 @@
   =========                 |
   \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox
    \\    /   O peration     |
-    \\  /    A nd           | Copyright (C) 2011 OpenFOAM Foundation
+    \\  /    A nd           | Copyright (C) 2011-2015 OpenFOAM Foundation
      \\/     M anipulation  |
 -------------------------------------------------------------------------------
 License
@@ -27,6 +27,9 @@ Class
 Description
     Wrapper around POSIX extended regular expressions.
 
+    The PCRE '(?i)' extension is provided to compile the regular expression
+    as being case-insensitive.
+
 SeeAlso
     The manpage regex(7) for more information about POSIX regular expressions.
     These differ somewhat from \c Perl and \c sed regular expressions.
@@ -51,6 +54,7 @@ namespace Foam
 class string;
 template<class T> class List;
 
+
 /*---------------------------------------------------------------------------*\
                            Class regExp Declaration
 \*---------------------------------------------------------------------------*/
@@ -71,6 +75,16 @@ class regExp
         //- Disallow default bitwise assignment
         void operator=(const regExp&);
 
+        //- Return true if it matches and sets the sub-groups matched.
+        //  Templated to support both std::string and Foam::string
+        template<class StringType>
+        bool matchGrouping
+        (
+            const std::string&,
+            List<StringType>& groups
+        ) const;
+
+
 public:
 
     // Static Member Functions
@@ -135,15 +149,14 @@ public:
 
         // Editing
 
-            //- Compile pattern into a regular expression, optionally ignoring
+            //- Compile pattern into a regular expression,
-            //  case
+            //  optionally ignoring case
             void set(const char*, const bool ignoreCase=false) const;
 
-            //- Compile pattern into a regular expression, optionally ignoring
+            //- Compile pattern into a regular expression,
-            //  case
+            //  optionally ignoring case
             void set(const std::string&, const bool ignoreCase=false) const;
 
-
             //- Release precompiled expression.
             //  Returns true if precompiled expression existed before clear
             bool clear() const;
@@ -161,7 +174,11 @@ public:
 
             //- Return true if it matches and sets the sub-groups matched
             //  The begin-of-line (^) and end-of-line ($) anchors are implicit
-            bool match(const string&, List<string>& groups) const;
+            bool match(const std::string&, List<std::string>& groups) const;
+
+            //- Return true if it matches and sets the sub-groups matched
+            //  The begin-of-line (^) and end-of-line ($) anchors are implicit
+            bool match(const std::string&, List<string>& groups) const;
 
             //- Return true if the regex was found within string
             bool search(const std::string& str) const