Refactoring ReHashList (using sorted ReSeqArray)

diff --git a/base/ReHashList.cpp b/base/ReHashList.cpp
index 1b32acc1dcfad45200c34e4d81db3a3beb126a08..2d60e735229e3b87b51c39d711333b45f320d79f 100644 (file)
--- a/base/ReHashList.cpp
+++ b/base/ReHashList.cpp
@@ -45,7 +45,7 @@ ReHashList::ReHashList(bool ignoreCase, int keyTagSize, int contentLengthSize,
        m_keys.setIgnoreCase(ignoreCase);
        m_keys.setSorted(true);
        m_keys.setSizes(keyTagSize, keyLengthSize);
-       m_keys.setSizes(0, contentLengthSize);
+       m_values.setSizes(0, contentLengthSize);
 }
 /** Destructor
  */
@@ -142,10 +142,6 @@ bool ReHashList::next(ReArrayPosition& position, ReByteBuffer* key,
  */
 void ReHashList::put(const Byte* key, size_t keyLength,
                const Byte* value, size_t valueLength){
-       if (keyLength == (size_t) -1)
-               keyLength = strlen(key);
-       if (valueLength == (size_t) -1)
-               valueLength = strlen(value);
        ReSeqArray::Index index;
        ReSeqArray::Tag tag;
        if (m_keys.binarySearch(key, keyLength, index, &tag)){
@@ -153,8 +149,7 @@ void ReHashList::put(const Byte* key, size_t keyLength,
                m_values.set((ReSeqArray::Index) tag, value, valueLength, 0);
        } else {
                // insert a new value.
-               tag = (ReSeqArray::Tag) m_values.count();
-               m_values.add(-1, value, valueLength);
+               tag = (ReSeqArray::Tag) m_values.add(-1, value, valueLength);
                m_keys.add(index, key, keyLength, tag);
        }
 }

diff --git a/base/ReSeqArray.cpp b/base/ReSeqArray.cpp
index e266dcd823392bcc4794967554a06de19d804b39..cca2b36a878899bff76fdb06bce123d39fe2384f 100644 (file)
--- a/base/ReSeqArray.cpp
+++ b/base/ReSeqArray.cpp
@@ -132,7 +132,7 @@ ReSeqArray::Index ReSeqArray::add(Index index, const Byte* source,
                m_list.insert(rc * m_entrySize, (Byte*) &seq, m_entrySize);
        } else {
                if (index >= count()){
-                       rc = m_content.length();
+                       rc = count();
                        m_list.append((Byte*) &seq, m_entrySize);
                }else{
                        m_list.insert(index * m_entrySize, (Byte*) &seq, m_entrySize);
@@ -161,7 +161,7 @@ bool ReSeqArray::binarySearch(const Byte* toFind, int length, Index& index,
        bool rc = false;
    int lbound = 0;
        int theCount = count();
-    int ubound = theCount;
+    int ubound = theCount - 1;
        int compareRc = 0;
     // binary search over the sorted vector:
     while(lbound <= ubound){
@@ -183,6 +183,8 @@ bool ReSeqArray::binarySearch(const Byte* toFind, int length, Index& index,
         else {
                        rc = true;
             index = half;
+                       if (tag != NULL)
+                               getLengthAndTag(seq, *tag);
             break;
         }
     }
@@ -561,11 +563,12 @@ void ReSeqArray::setSorted(bool onNotOff){
  * @param sizeOfLength 0: no length stored (constant length, stored in
  *                                                     m_commonSize).
  *                                                     Otherwise: length in byte: 1, 2, 3, 4 or 5
- * @param constantSize  0 or the size of the sequence if the length is not
+ * @param constantSize         INDIVIDUAL_SIZE: the length will be stored in each element
+ *                                                     0 or the size of the sequence if the length is not
  *                                                     indiviually stored. <br>
  *                                                     If > 0 <code>sizeOfLength</code>must be 0!
  */
-void ReSeqArray::setSizes(int sizeOfTag, int sizeOfLength, int constantSize){
+void ReSeqArray::setSizes(int sizeOfTag, int sizeOfLength, size_t constantSize){
        switch(m_sizeOfLength = sizeOfLength){
        case 0:
        case 1:
@@ -596,10 +599,10 @@ void ReSeqArray::setSizes(int sizeOfTag, int sizeOfLength, int constantSize){
        }
        m_offsetOfLength = sizeof(Index);
        m_offsetOfTag = sizeof(Index) + m_sizeOfLength;
-       if (sizeOfLength > 0 && constantSize > 0){
+       if (sizeOfLength > 0 && constantSize != INDIVIDUAL_SIZE){
                globalLogger()->sayF(LOG_ERROR | CAT_LIB, LC_SET_SIZES_1,
                                        i18n("collision of sizeOfLength $1 and constantSize $2"))
-                                       .arg(sizeOfLength).arg(constantSize).end();
+                                       .arg(sizeOfLength).arg((int) constantSize).end();
                constantSize = 0;
        }
        m_commonSize = constantSize;

diff --git a/base/ReSeqArray.hpp b/base/ReSeqArray.hpp
index 3f06bb2782549a0bc46c1bc8c38e3ec618b6d34d..5cb59fc0b212a4f95cfd47ff886b03bbd450ad89 100644 (file)
--- a/base/ReSeqArray.hpp
+++ b/base/ReSeqArray.hpp
@@ -35,9 +35,10 @@ public:
        ReSeqArray(const ReSeqArray& source);
        ReSeqArray& operator = (const ReSeqArray& source);
 public:
-       Index add(Index index, const Byte* source, size_t sourceLength, Tag tag = 0);
+       Index add(Index index, const Byte* source, size_t sourceLength,
+                       Tag tag = 0);
        bool binarySearch(const Byte* toFind, int length, Index& index,
-                               Tag* tag = NULL) const;
+                       Tag* tag = NULL) const;
        void clear();
        int compare(Index index1, Index index2);
        /** @brief Returns the count of defined entries in the list.
@@ -64,7 +65,7 @@ public:
        }
        void remove(Index index);
        void set(Index index, const Byte* source, size_t sourceLength, Tag tag);
-       void setSizes(int sizeOfTag, int sizeOfLength, int constantLength = 0);
+       void setSizes(int sizeOfTag, int sizeOfLength, size_t constantLength = INDIVIDUAL_SIZE);
        void setSorted(bool onNotOff);
        void setIgnoreCase(bool onNotOff);
        void sort();
@@ -76,7 +77,6 @@ protected:
                return m_content.buffer();
        }
        friend class ReHashList;
-       friend class ReStringList;
        /** @brief Returns the info of an entry of the list.
         * @param index         The index of the wanted entry.
         * @return The pointer of the entry.

diff --git a/cunit/cuReSeqArray.cpp b/cunit/cuReSeqArray.cpp
index bdfca227e7ac38fcdbab087bf845ad67b0c3da27..1a787d02233d374c7b0ab7a11be21517ba66f6ab 100644 (file)
--- a/cunit/cuReSeqArray.cpp
+++ b/cunit/cuReSeqArray.cpp
@@ -71,6 +71,8 @@ private:
                                list.add(ix, value.str(), -1, (ReSeqArray::Tag) ix*10);
                        }
                        for (int ix = 0; ix <= count; ix++){
+                               if (count == 2 && ix == 2)
+                                       ix = 2;
                                value.setLength(2);
                                value.buffer()[0] = 'a' + ix;
                                value.buffer()[1] = 'a';
@@ -97,11 +99,11 @@ private:
                list.add(-1, "abc", -1, 300);
                list.add(-1, "abc", -1, 350);
                list.add(-1, "cde", -1, 400);
-               log(false, "unsorted:");
-               list.dump(stdout);
+               //log(false, "unsorted:");
+               //list.dump(stdout);
                list.setSorted(true);
-               log(false, "sorted:");
-               list.dump(stdout);
+               //log(false, "sorted:");
+               //list.dump(stdout);
                checkElement(list, 0U, "AbC", 350);
                checkElement(list, 1U, "AbC", 300);
                checkElement(list, 2U, "AbCd", 100);
@@ -124,7 +126,7 @@ private:
                ReByteBuffer value, expectedValue;
                list.add(-1, "bcd", -1, 200);
                list.add(0, "abc", -1, 100);
-               list.dump(stdout);
+               // list.dump(stdout);
                checkEqu(0U, list.find("AbC", -1));
                checkEqu(1U, list.find("BCD", -1));
        }
@@ -170,7 +172,7 @@ private:
                        if (expectedTag != tag)
                                checkEqu(expectedTag, tag);
                }
-               list.dump(stdout);
+               //list.dump(stdout);
                expectedValue.setLength(0);
                for (size_t ix = 0; ix < maxIx; ix++){
                        expectedTag = -1;

diff --git a/cunit/testall.cpp b/cunit/testall.cpp
index d181b888b9a202a084b232a87215aa582cf469ab..2e8906a41dbe125a4197e7a0c5d3073ea51a6b91 100644 (file)
--- a/cunit/testall.cpp
+++ b/cunit/testall.cpp
@@ -12,9 +12,8 @@
 
 void testBase(){
        extern void testReHashList(void);
-       //testReHashList();
-       extern void testReSeqArray();
-       testReSeqArray();
+       testReHashList();
+
        extern void testReTestUnit();
        // testReTestUnit();
        extern void testReByteBuffer();