Check in the pristine copy of ICU 51.2

icu51/source/common/ucharstrie.cpp

Index: icu51/source/common/ucharstrie.cpp
===================================================================
--- icu51/source/common/ucharstrie.cpp	(revision 0)
+++ icu51/source/common/ucharstrie.cpp	(revision 0)
@@ -0,0 +1,411 @@
+/*
+*******************************************************************************
+*   Copyright (C) 2010-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  ucharstrie.h
+*   encoding:   US-ASCII
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2010nov14
+*   created by: Markus W. Scherer
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/appendable.h"
+#include "unicode/ucharstrie.h"
+#include "unicode/uobject.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "uassert.h"
+
+U_NAMESPACE_BEGIN
+
+UCharsTrie::~UCharsTrie() {
+    uprv_free(ownedArray_);
+}
+
+UStringTrieResult
+UCharsTrie::current() const {
+    const UChar *pos=pos_;
+    if(pos==NULL) {
+        return USTRINGTRIE_NO_MATCH;
+    } else {
+        int32_t node;
+        return (remainingMatchLength_<0 && (node=*pos)>=kMinValueLead) ?
+                valueResult(node) : USTRINGTRIE_NO_VALUE;
+    }
+}
+
+UStringTrieResult
+UCharsTrie::firstForCodePoint(UChar32 cp) {
+    return cp<=0xffff ?
+        first(cp) :
+        (USTRINGTRIE_HAS_NEXT(first(U16_LEAD(cp))) ?
+            next(U16_TRAIL(cp)) :
+            USTRINGTRIE_NO_MATCH);
+}
+
+UStringTrieResult
+UCharsTrie::nextForCodePoint(UChar32 cp) {
+    return cp<=0xffff ?
+        next(cp) :
+        (USTRINGTRIE_HAS_NEXT(next(U16_LEAD(cp))) ?
+            next(U16_TRAIL(cp)) :
+            USTRINGTRIE_NO_MATCH);
+}
+
+UStringTrieResult
+UCharsTrie::branchNext(const UChar *pos, int32_t length, int32_t uchar) {
+    // Branch according to the current unit.
+    if(length==0) {
+        length=*pos++;
+    }
+    ++length;
+    // The length of the branch is the number of units to select from.
+    // The data structure encodes a binary search.
+    while(length>kMaxBranchLinearSubNodeLength) {
+        if(uchar<*pos++) {
+            length>>=1;
+            pos=jumpByDelta(pos);
+        } else {
+            length=length-(length>>1);
+            pos=skipDelta(pos);
+        }
+    }
+    // Drop down to linear search for the last few units.
+    // length>=2 because the loop body above sees length>kMaxBranchLinearSubNodeLength>=3
+    // and divides length by 2.
+    do {
+        if(uchar==*pos++) {
+            UStringTrieResult result;
+            int32_t node=*pos;
+            if(node&kValueIsFinal) {
+                // Leave the final value for getValue() to read.
+                result=USTRINGTRIE_FINAL_VALUE;
+            } else {
+                // Use the non-final value as the jump delta.
+                ++pos;
+                // int32_t delta=readValue(pos, node);
+                int32_t delta;
+                if(node<kMinTwoUnitValueLead) {
+                    delta=node;
+                } else if(node<kThreeUnitValueLead) {
+                    delta=((node-kMinTwoUnitValueLead)<<16)|*pos++;
+                } else {
+                    delta=(pos[0]<<16)|pos[1];
+                    pos+=2;
+                }
+                // end readValue()
+                pos+=delta;
+                node=*pos;
+                result= node>=kMinValueLead ? valueResult(node) : USTRINGTRIE_NO_VALUE;
+            }
+            pos_=pos;
+            return result;
+        }
+        --length;
+        pos=skipValue(pos);
+    } while(length>1);
+    if(uchar==*pos++) {
+        pos_=pos;
+        int32_t node=*pos;
+        return node>=kMinValueLead ? valueResult(node) : USTRINGTRIE_NO_VALUE;
+    } else {
+        stop();
+        return USTRINGTRIE_NO_MATCH;
+    }
+}
+
+UStringTrieResult
+UCharsTrie::nextImpl(const UChar *pos, int32_t uchar) {
+    int32_t node=*pos++;
+    for(;;) {
+        if(node<kMinLinearMatch) {
+            return branchNext(pos, node, uchar);
+        } else if(node<kMinValueLead) {
+            // Match the first of length+1 units.
+            int32_t length=node-kMinLinearMatch;  // Actual match length minus 1.
+            if(uchar==*pos++) {
+                remainingMatchLength_=--length;
+                pos_=pos;
+                return (length<0 && (node=*pos)>=kMinValueLead) ?
+                        valueResult(node) : USTRINGTRIE_NO_VALUE;
+            } else {
+                // No match.
+                break;
+            }
+        } else if(node&kValueIsFinal) {
+            // No further matching units.
+            break;
+        } else {
+            // Skip intermediate value.
+            pos=skipNodeValue(pos, node);
+            node&=kNodeTypeMask;
+        }
+    }
+    stop();
+    return USTRINGTRIE_NO_MATCH;
+}
+
+UStringTrieResult
+UCharsTrie::next(int32_t uchar) {
+    const UChar *pos=pos_;
+    if(pos==NULL) {
+        return USTRINGTRIE_NO_MATCH;
+    }
+    int32_t length=remainingMatchLength_;  // Actual remaining match length minus 1.
+    if(length>=0) {
+        // Remaining part of a linear-match node.
+        if(uchar==*pos++) {
+            remainingMatchLength_=--length;
+            pos_=pos;
+            int32_t node;
+            return (length<0 && (node=*pos)>=kMinValueLead) ?
+                    valueResult(node) : USTRINGTRIE_NO_VALUE;
+        } else {
+            stop();
+            return USTRINGTRIE_NO_MATCH;
+        }
+    }
+    return nextImpl(pos, uchar);
+}
+
+UStringTrieResult
+UCharsTrie::next(const UChar *s, int32_t sLength) {
+    if(sLength<0 ? *s==0 : sLength==0) {
+        // Empty input.
+        return current();
+    }
+    const UChar *pos=pos_;
+    if(pos==NULL) {
+        return USTRINGTRIE_NO_MATCH;
+    }
+    int32_t length=remainingMatchLength_;  // Actual remaining match length minus 1.
+    for(;;) {
+        // Fetch the next input unit, if there is one.
+        // Continue a linear-match node without rechecking sLength<0.
+        int32_t uchar;
+        if(sLength<0) {
+            for(;;) {
+                if((uchar=*s++)==0) {
+                    remainingMatchLength_=length;
+                    pos_=pos;
+                    int32_t node;
+                    return (length<0 && (node=*pos)>=kMinValueLead) ?
+                            valueResult(node) : USTRINGTRIE_NO_VALUE;
+                }
+                if(length<0) {
+                    remainingMatchLength_=length;
+                    break;
+                }
+                if(uchar!=*pos) {
+                    stop();
+                    return USTRINGTRIE_NO_MATCH;
+                }
+                ++pos;
+                --length;
+            }
+        } else {
+            for(;;) {
+                if(sLength==0) {
+                    remainingMatchLength_=length;
+                    pos_=pos;
+                    int32_t node;
+                    return (length<0 && (node=*pos)>=kMinValueLead) ?
+                            valueResult(node) : USTRINGTRIE_NO_VALUE;
+                }
+                uchar=*s++;
+                --sLength;
+                if(length<0) {
+                    remainingMatchLength_=length;
+                    break;
+                }
+                if(uchar!=*pos) {
+                    stop();
+                    return USTRINGTRIE_NO_MATCH;
+                }
+                ++pos;
+                --length;
+            }
+        }
+        int32_t node=*pos++;
+        for(;;) {
+            if(node<kMinLinearMatch) {
+                UStringTrieResult result=branchNext(pos, node, uchar);
+                if(result==USTRINGTRIE_NO_MATCH) {
+                    return USTRINGTRIE_NO_MATCH;
+                }
+                // Fetch the next input unit, if there is one.
+                if(sLength<0) {
+                    if((uchar=*s++)==0) {
+                        return result;
+                    }
+                } else {
+                    if(sLength==0) {
+                        return result;
+                    }
+                    uchar=*s++;
+                    --sLength;
+                }
+                if(result==USTRINGTRIE_FINAL_VALUE) {
+                    // No further matching units.
+                    stop();
+                    return USTRINGTRIE_NO_MATCH;
+                }
+                pos=pos_;  // branchNext() advanced pos and wrote it to pos_ .
+                node=*pos++;
+            } else if(node<kMinValueLead) {
+                // Match length+1 units.
+                length=node-kMinLinearMatch;  // Actual match length minus 1.
+                if(uchar!=*pos) {
+                    stop();
+                    return USTRINGTRIE_NO_MATCH;
+                }
+                ++pos;
+                --length;
+                break;
+            } else if(node&kValueIsFinal) {
+                // No further matching units.
+                stop();
+                return USTRINGTRIE_NO_MATCH;
+            } else {
+                // Skip intermediate value.
+                pos=skipNodeValue(pos, node);
+                node&=kNodeTypeMask;
+            }
+        }
+    }
+}
+
+const UChar *
+UCharsTrie::findUniqueValueFromBranch(const UChar *pos, int32_t length,
+                                      UBool haveUniqueValue, int32_t &uniqueValue) {
+    while(length>kMaxBranchLinearSubNodeLength) {
+        ++pos;  // ignore the comparison unit
+        if(NULL==findUniqueValueFromBranch(jumpByDelta(pos), length>>1, haveUniqueValue, uniqueValue)) {
+            return NULL;
+        }
+        length=length-(length>>1);
+        pos=skipDelta(pos);
+    }
+    do {
+        ++pos;  // ignore a comparison unit
+        // handle its value
+        int32_t node=*pos++;
+        UBool isFinal=(UBool)(node>>15);
+        node&=0x7fff;
+        int32_t value=readValue(pos, node);
+        pos=skipValue(pos, node);
+        if(isFinal) {
+            if(haveUniqueValue) {
+                if(value!=uniqueValue) {
+                    return NULL;
+                }
+            } else {
+                uniqueValue=value;
+                haveUniqueValue=TRUE;
+            }
+        } else {
+            if(!findUniqueValue(pos+value, haveUniqueValue, uniqueValue)) {
+                return NULL;
+            }
+            haveUniqueValue=TRUE;
+        }
+    } while(--length>1);
+    return pos+1;  // ignore the last comparison unit
+}
+
+UBool
+UCharsTrie::findUniqueValue(const UChar *pos, UBool haveUniqueValue, int32_t &uniqueValue) {
+    int32_t node=*pos++;
+    for(;;) {
+        if(node<kMinLinearMatch) {
+            if(node==0) {
+                node=*pos++;
+            }
+            pos=findUniqueValueFromBranch(pos, node+1, haveUniqueValue, uniqueValue);
+            if(pos==NULL) {
+                return FALSE;
+            }
+            haveUniqueValue=TRUE;
+            node=*pos++;
+        } else if(node<kMinValueLead) {
+            // linear-match node
+            pos+=node-kMinLinearMatch+1;  // Ignore the match units.
+            node=*pos++;
+        } else {
+            UBool isFinal=(UBool)(node>>15);
+            int32_t value;
+            if(isFinal) {
+                value=readValue(pos, node&0x7fff);
+            } else {
+                value=readNodeValue(pos, node);
+            }
+            if(haveUniqueValue) {
+                if(value!=uniqueValue) {
+                    return FALSE;
+                }
+            } else {
+                uniqueValue=value;
+                haveUniqueValue=TRUE;
+            }
+            if(isFinal) {
+                return TRUE;
+            }
+            pos=skipNodeValue(pos, node);
+            node&=kNodeTypeMask;
+        }
+    }
+}
+
+int32_t
+UCharsTrie::getNextUChars(Appendable &out) const {
+    const UChar *pos=pos_;
+    if(pos==NULL) {
+        return 0;
+    }
+    if(remainingMatchLength_>=0) {
+        out.appendCodeUnit(*pos);  // Next unit of a pending linear-match node.
+        return 1;
+    }
+    int32_t node=*pos++;
+    if(node>=kMinValueLead) {
+        if(node&kValueIsFinal) {
+            return 0;
+        } else {
+            pos=skipNodeValue(pos, node);
+            node&=kNodeTypeMask;
+        }
+    }
+    if(node<kMinLinearMatch) {
+        if(node==0) {
+            node=*pos++;
+        }
+        out.reserveAppendCapacity(++node);
+        getNextBranchUChars(pos, node, out);
+        return node;
+    } else {
+        // First unit of the linear-match node.
+        out.appendCodeUnit(*pos);
+        return 1;
+    }
+}
+
+void
+UCharsTrie::getNextBranchUChars(const UChar *pos, int32_t length, Appendable &out) {
+    while(length>kMaxBranchLinearSubNodeLength) {
+        ++pos;  // ignore the comparison unit
+        getNextBranchUChars(jumpByDelta(pos), length>>1, out);
+        length=length-(length>>1);
+        pos=skipDelta(pos);
+    }
+    do {
+        out.appendCodeUnit(*pos++);
+        pos=skipValue(pos);
+    } while(--length>1);
+    out.appendCodeUnit(*pos);
+}
+
+U_NAMESPACE_END
Property changes on: icu51/source/common/ucharstrie.cpp
___________________________________________________________________
Added: svn:eol-style
   + LF