Implemented CharacterSet-based SBuf::find_first_of and find_first_not_of

diff --git a/src/Makefile.am b/src/Makefile.am
index 191c6de5a2e1900902f28d746b56d2cafe1dfe60..6cd5df7b99bd2b7f8c1a5e559d15e023264b8e61 100644 (file)
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -15,6 +15,7 @@ DNSSOURCE = \
        DnsLookupDetails.cc
 
 SBUF_SOURCE= \
+       base/CharacterSet.h \
        base/InstanceId.h \
        MemBlob.h \
        MemBlob.cc \

diff --git a/src/SBuf.cc b/src/SBuf.cc
index 26480b8dcdcedb001e6912422c7ae66d37683fd6..18053bd734628f16ff33cad83d84db88308c8383 100644 (file)
--- a/src/SBuf.cc
+++ b/src/SBuf.cc
@@ -688,12 +688,8 @@ SBuf::rfind(char c, SBuf::size_type endPos) const
 }
 
 SBuf::size_type
-SBuf::find_first_of(const SBuf &set, size_type startPos) const
+SBuf::find_first_of(const CharacterSet &set, size_type startPos) const
 {
-    // if set is 1 char big, use the char search. Stats updated there
-    if (set.length() == 1)
-        return find(set[0], startPos);
-
     ++stats.find;
 
     if (startPos == npos)
@@ -702,18 +698,39 @@ SBuf::find_first_of(const SBuf &set, size_type startPos) const
     if (startPos >= length())
         return npos;
 
-    if (set.length() == 0)
+    debugs(24, 7, "first of characterset " << set.name << " in id " << id);
+    char *cur = buf()+startPos;
+    const char *end = bufEnd();
+    while (cur < end) {
+        if (set[*cur])
+            return cur-buf();
+        ++cur;
+    }
+    debugs(24, 7, "not found");
+    return npos;
+}
+
+SBuf::size_type
+SBuf::find_first_not_of(const CharacterSet &set, size_type startPos) const
+{
+    ++stats.find;
+
+    if (startPos == npos)
         return npos;
 
-    debugs(24, 7, "any of '" << set << "' " << " in id " << id);
-    char *cur = buf()+startPos, *end = bufEnd();
+    if (startPos >= length())
+        return npos;
+
+    debugs(24, 7, "first not of characterset " << set.name << " in id " << id);
+    char *cur = buf()+startPos;
+    const char *end = bufEnd();
     while (cur < end) {
-        if (memchr(set.buf(), *cur, set.length()))
-            return (cur-buf());
+        if (!set[*cur])
+            return cur-buf();
         ++cur;
     }
     debugs(24, 7, "not found");
-    return npos;
+    return length();
 }
 
 /*

diff --git a/src/SBuf.h b/src/SBuf.h
index d374b11a0057b1dbe9d311a1c3ba9552cf85083c..3fcf9f1869154a29c4b5a63cf02a9019cd4990e9 100644 (file)
--- a/src/SBuf.h
+++ b/src/SBuf.h
@@ -29,6 +29,7 @@
 #ifndef SQUID_SBUF_H
 #define SQUID_SBUF_H
 
+#include "base/CharacterSet.h"
 #include "base/InstanceId.h"
 #include "MemBlob.h"
 #include "SBufExceptions.h"
@@ -513,7 +514,15 @@ public:
      * \param startPos if specified, ignore any occurrences before that position
      *   if npos, then npos is always returned
      */
-    size_type find_first_of(const SBuf &set, size_type startPos = 0) const;
+    size_type find_first_of(const CharacterSet &set, size_type startPos = 0) const;
+
+    /** Find first occurrence character NOT in character set
+     *
+     * \return length() if all characters in the SBuf are from set
+     * \param startPos if specified, ignore any occurrences before that position
+     *   if npos, then npos is always returned
+     */
+    size_type find_first_not_of(const CharacterSet &set, size_type startPos = 0) const;
 
     /** sscanf-alike
      *

diff --git a/src/base/CharacterSet.h b/src/base/CharacterSet.h
new file mode 100644 (file)
index 0000000..3c11d69
--- /dev/null
+++ b/src/base/CharacterSet.h
@@ -0,0 +1,44 @@
+#ifndef _SQUID_SRC_PARSER_CHARACTERSET_H
+#define _SQUID_SRC_PARSER_CHARACTERSET_H
+
+#include <vector>
+
+class CharacterSet
+{
+public:
+    CharacterSet(const char *label, const char * const c) : name(label), chars_(std::vector<bool>(256,false)) {
+        size_t clen = strlen(c);
+        for (size_t i = 0; i < clen; ++i)
+            chars_[static_cast<uint8_t>(c[i])] = true;
+    }
+
+    /// whether a given character exists in the set
+    bool operator[](unsigned char c) const {return chars_[static_cast<uint8_t>(c)];}
+
+    /// add a given char to the character set. c must be >= 0.
+    CharacterSet & add(const unsigned char c) {chars_[static_cast<uint8_t>(c)] = true; return *this; }
+
+    /// add all characters from the given CharacterSet to this one
+    const CharacterSet &operator +=(const CharacterSet &src) {
+        //precondition: src.chars_.size() == chars_.size()
+        std::vector<bool>::const_iterator s = src.chars_.begin();
+        const std::vector<bool>::const_iterator e = src.chars_.end();
+        std::vector<bool>::iterator d = chars_.begin();
+        while (s != e) {
+            if (*s)
+                *d = true;
+            ++s;
+            ++d;
+        }
+        return *this;
+    }
+
+    /// name of this character set
+    const char * name;
+
+private:
+    /// characters defined in this set
+    std::vector<bool> chars_; //std::vector<bool> is optimized
+};
+
+#endif /* _SQUID_SRC_PARSER_CHARACTERSET_H */

diff --git a/src/base/Makefile.am b/src/base/Makefile.am
index 1b84589d87d419fad2124ea79ae1f23f35e0c88b..b99beaebc9579533f4fe10fd625f4474461337f7 100644 (file)
--- a/src/base/Makefile.am
+++ b/src/base/Makefile.am
@@ -12,6 +12,7 @@ libbase_la_SOURCES = \
        AsyncJobCalls.h \
        AsyncCallQueue.cc \
        AsyncCallQueue.h \
+       CharacterSet.h \
        TidyPointer.h \
        CbcPointer.h \
        InstanceId.h \

diff --git a/src/icmp/Makefile.am b/src/icmp/Makefile.am
index 4be96995217a4a511002ff9d256ca09bf6dfe325..4cde92efeeb128aea51b9857cfd9cc9a3a1eac54 100644 (file)
--- a/src/icmp/Makefile.am
+++ b/src/icmp/Makefile.am
@@ -23,6 +23,7 @@ libexec_PROGRAMS = $(PINGER)
 noinst_LTLIBRARIES = libicmp-core.la libicmp.la
 
 SBUF_SOURCE= \
+       $(top_srcdir)/src/base/CharacterSet.h \
        $(top_srcdir)/src/SBuf.h \
        $(top_srcdir)/src/SBuf.cc \
        $(top_srcdir)/src/MemBlob.h \

diff --git a/src/tests/SBufFindTest.cc b/src/tests/SBufFindTest.cc
index 6bfd1d29156ec34b9fba63a7eb5e4bc934d04076..247b5a51644ba4d04e2f60fcd5db1d4ed0a6ee98 100644 (file)
--- a/src/tests/SBufFindTest.cc
+++ b/src/tests/SBufFindTest.cc
@@ -105,7 +105,7 @@ SBufFindTest::testFindFirstOf()
 {
     theFindString = theStringHay.find_first_of(theStringNeedle, thePos);
     theBareNeedlePos = theStringHay.find_first_of(theStringNeedle);
-    theFindSBuf = theSBufHay.find_first_of(theSBufNeedle, thePos);
+    theFindSBuf = theSBufHay.find_first_of(CharacterSet("cs",theSBufNeedle.c_str()), thePos);
     checkResults("find_first_of");
 }
 

diff --git a/src/tests/testSBuf.cc b/src/tests/testSBuf.cc
index b424444eb213cbe45dd3c329f08c2dedf06c691b..bc9b56efa287d2adc9c7f3bc9236ac3b4f878113 100644 (file)
--- a/src/tests/testSBuf.cc
+++ b/src/tests/testSBuf.cc
@@ -759,22 +759,46 @@ testSBuf::testFindFirstOf()
     SBuf::size_type idx;
 
     // not found
-    idx=haystack.find_first_of(SBuf("ADHRWYP"));
+    idx=haystack.find_first_of(CharacterSet("t1","ADHRWYP"));
     CPPUNIT_ASSERT_EQUAL(SBuf::npos,idx);
 
     // found at beginning
-    idx=haystack.find_first_of(SBuf("THANDF"));
+    idx=haystack.find_first_of(CharacterSet("t2","THANDF"));
     CPPUNIT_ASSERT_EQUAL(0U,idx);
 
     //found at end of haystack
-    idx=haystack.find_first_of(SBuf("QWERYVg"));
+    idx=haystack.find_first_of(CharacterSet("t3","QWERYVg"));
     CPPUNIT_ASSERT_EQUAL(haystack.length()-1,idx);
 
     //found in the middle of haystack
-    idx=haystack.find_first_of(SBuf("QWERqYV"));
+    idx=haystack.find_first_of(CharacterSet("t4","QWERqYV"));
     CPPUNIT_ASSERT_EQUAL(4U,idx);
 }
 
+void
+testSBuf::testFindFirstNotOf()
+{
+    SBuf haystack(literal);
+    SBuf::size_type idx;
+
+    // all chars from the set
+    idx=haystack.find_first_not_of(CharacterSet("t1",literal.c_str()));
+    CPPUNIT_ASSERT_EQUAL(haystack.length(),idx);
+
+    // found at beginning
+    idx=haystack.find_first_not_of(CharacterSet("t2","a"));
+    CPPUNIT_ASSERT_EQUAL(0U,idx);
+
+    //found at end of haystack
+    idx=haystack.find_first_not_of(CharacterSet("t3",literal.substr(0,literal.length()-1).c_str()));
+    CPPUNIT_ASSERT_EQUAL(haystack.length()-1,idx);
+
+    //found in the middle of haystack
+    idx=haystack.find_first_not_of(CharacterSet("t4","The"));
+    CPPUNIT_ASSERT_EQUAL(3U,idx);
+}
+
+
 void
 testSBuf::testAutoFind()
 {

diff --git a/src/tests/testSBuf.h b/src/tests/testSBuf.h
index 373eeef1185de06c27cb47bdd5f7a37697d00830..f4301e5cf79b5d5694acf7be1fbe392184fad3d0 100644 (file)
--- a/src/tests/testSBuf.h
+++ b/src/tests/testSBuf.h
@@ -35,6 +35,7 @@ class testSBuf : public CPPUNIT_NS::TestFixture
     CPPUNIT_TEST( testRFindChar );
     CPPUNIT_TEST( testRFindSBuf );
     CPPUNIT_TEST( testFindFirstOf );
+    CPPUNIT_TEST( testFindFirstNotOf );
     CPPUNIT_TEST( testPrintf );
     CPPUNIT_TEST( testScanf );
     CPPUNIT_TEST( testCopy );
@@ -79,6 +80,7 @@ protected:
     void testStartsWith();
     void testSBufStream();
     void testFindFirstOf();
+    void testFindFirstNotOf();
     void testAutoFind();
     void testStdStringOps();
 };