intl/icu/source/common/rbbisetb.h@b8a032363ba2 (annotated)
intl/icu/source/common/rbbisetb.h
Thu, 22 Jan 2015 13:21:57 +0100
- author
- Michael Schloh von Bennewitz <michael@schloh.com>
- date
- Thu, 22 Jan 2015 13:21:57 +0100
- branch
- TOR_BUG_9701
- changeset 15
- b8a032363ba2
- permissions
- -rw-r--r--
Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6
| michael@0 | 1 | // |
| michael@0 | 2 | // rbbisetb.h |
| michael@0 | 3 | /* |
| michael@0 | 4 | ********************************************************************** |
| michael@0 | 5 | * Copyright (c) 2001-2005, International Business Machines |
| michael@0 | 6 | * Corporation and others. All Rights Reserved. |
| michael@0 | 7 | ********************************************************************** |
| michael@0 | 8 | */ |
| michael@0 | 9 | |
| michael@0 | 10 | #ifndef RBBISETB_H |
| michael@0 | 11 | #define RBBISETB_H |
| michael@0 | 12 | |
| michael@0 | 13 | #include "unicode/utypes.h" |
| michael@0 | 14 | #include "unicode/uobject.h" |
| michael@0 | 15 | #include "rbbirb.h" |
| michael@0 | 16 | #include "uvector.h" |
| michael@0 | 17 | |
| michael@0 | 18 | struct UNewTrie; |
| michael@0 | 19 | |
| michael@0 | 20 | U_NAMESPACE_BEGIN |
| michael@0 | 21 | |
| michael@0 | 22 | // |
| michael@0 | 23 | // RBBISetBuilder Derives the character categories used by the runtime RBBI engine |
| michael@0 | 24 | // from the Unicode Sets appearing in the source RBBI rules, and |
| michael@0 | 25 | // creates the TRIE table used to map from Unicode to the |
| michael@0 | 26 | // character categories. |
| michael@0 | 27 | // |
| michael@0 | 28 | |
| michael@0 | 29 | |
| michael@0 | 30 | // |
| michael@0 | 31 | // RangeDescriptor |
| michael@0 | 32 | // |
| michael@0 | 33 | // Each of the non-overlapping character ranges gets one of these descriptors. |
| michael@0 | 34 | // All of them are strung together in a linked list, which is kept in order |
| michael@0 | 35 | // (by character) |
| michael@0 | 36 | // |
| michael@0 | 37 | class RangeDescriptor : public UMemory { |
| michael@0 | 38 | public: |
| michael@0 | 39 | UChar32 fStartChar; // Start of range, unicode 32 bit value. |
| michael@0 | 40 | UChar32 fEndChar; // End of range, unicode 32 bit value. |
| michael@0 | 41 | int32_t fNum; // runtime-mapped input value for this range. |
| michael@0 | 42 | UVector *fIncludesSets; // vector of the the original |
| michael@0 | 43 | // Unicode sets that include this range. |
| michael@0 | 44 | // (Contains ptrs to uset nodes) |
| michael@0 | 45 | RangeDescriptor *fNext; // Next RangeDescriptor in the linked list. |
| michael@0 | 46 | |
| michael@0 | 47 | RangeDescriptor(UErrorCode &status); |
| michael@0 | 48 | RangeDescriptor(const RangeDescriptor &other, UErrorCode &status); |
| michael@0 | 49 | ~RangeDescriptor(); |
| michael@0 | 50 | void split(UChar32 where, UErrorCode &status); // Spit this range in two at "where", with |
| michael@0 | 51 | // where appearing in the second (higher) part. |
| michael@0 | 52 | void setDictionaryFlag(); // Check whether this range appears as part of |
| michael@0 | 53 | // the Unicode set named "dictionary" |
| michael@0 | 54 | |
| michael@0 | 55 | private: |
| michael@0 | 56 | RangeDescriptor(const RangeDescriptor &other); // forbid copying of this class |
| michael@0 | 57 | RangeDescriptor &operator=(const RangeDescriptor &other); // forbid copying of this class |
| michael@0 | 58 | }; |
| michael@0 | 59 | |
| michael@0 | 60 | |
| michael@0 | 61 | // |
| michael@0 | 62 | // RBBISetBuilder Handles processing of Unicode Sets from RBBI rules. |
| michael@0 | 63 | // |
| michael@0 | 64 | // Starting with the rules parse tree from the scanner, |
| michael@0 | 65 | // |
| michael@0 | 66 | // - Enumerate the set of UnicodeSets that are referenced |
| michael@0 | 67 | // by the RBBI rules. |
| michael@0 | 68 | // - compute a derived set of non-overlapping UnicodeSets |
| michael@0 | 69 | // that will correspond to columns in the state table for |
| michael@0 | 70 | // the RBBI execution engine. |
| michael@0 | 71 | // - construct the trie table that maps input characters |
| michael@0 | 72 | // to set numbers in the non-overlapping set of sets. |
| michael@0 | 73 | // |
| michael@0 | 74 | |
| michael@0 | 75 | |
| michael@0 | 76 | class RBBISetBuilder : public UMemory { |
| michael@0 | 77 | public: |
| michael@0 | 78 | RBBISetBuilder(RBBIRuleBuilder *rb); |
| michael@0 | 79 | ~RBBISetBuilder(); |
| michael@0 | 80 | |
| michael@0 | 81 | void build(); |
| michael@0 | 82 | void addValToSets(UVector *sets, uint32_t val); |
| michael@0 | 83 | void addValToSet (RBBINode *usetNode, uint32_t val); |
| michael@0 | 84 | int32_t getNumCharCategories() const; // CharCategories are the same as input symbol set to the |
| michael@0 | 85 | // runtime state machine, which are the same as |
| michael@0 | 86 | // columns in the DFA state table |
| michael@0 | 87 | int32_t getTrieSize() /*const*/; // Size in bytes of the serialized Trie. |
| michael@0 | 88 | void serializeTrie(uint8_t *where); // write out the serialized Trie. |
| michael@0 | 89 | UChar32 getFirstChar(int32_t val) const; |
| michael@0 | 90 | UBool sawBOF() const; // Indicate whether any references to the {bof} pseudo |
| michael@0 | 91 | // character were encountered. |
| michael@0 | 92 | #ifdef RBBI_DEBUG |
| michael@0 | 93 | void printSets(); |
| michael@0 | 94 | void printRanges(); |
| michael@0 | 95 | void printRangeGroups(); |
| michael@0 | 96 | #else |
| michael@0 | 97 | #define printSets() |
| michael@0 | 98 | #define printRanges() |
| michael@0 | 99 | #define printRangeGroups() |
| michael@0 | 100 | #endif |
| michael@0 | 101 | |
| michael@0 | 102 | private: |
| michael@0 | 103 | void numberSets(); |
| michael@0 | 104 | |
| michael@0 | 105 | RBBIRuleBuilder *fRB; // The RBBI Rule Compiler that owns us. |
| michael@0 | 106 | UErrorCode *fStatus; |
| michael@0 | 107 | |
| michael@0 | 108 | RangeDescriptor *fRangeList; // Head of the linked list of RangeDescriptors |
| michael@0 | 109 | |
| michael@0 | 110 | UNewTrie *fTrie; // The mapping TRIE that is the end result of processing |
| michael@0 | 111 | uint32_t fTrieSize; // the Unicode Sets. |
| michael@0 | 112 | |
| michael@0 | 113 | // Groups correspond to character categories - |
| michael@0 | 114 | // groups of ranges that are in the same original UnicodeSets. |
| michael@0 | 115 | // fGroupCount is the index of the last used group. |
| michael@0 | 116 | // fGroupCount+1 is also the number of columns in the RBBI state table being compiled. |
| michael@0 | 117 | // State table column 0 is not used. Column 1 is for end-of-input. |
| michael@0 | 118 | // column 2 is for group 0. Funny counting. |
| michael@0 | 119 | int32_t fGroupCount; |
| michael@0 | 120 | |
| michael@0 | 121 | UBool fSawBOF; |
| michael@0 | 122 | |
| michael@0 | 123 | RBBISetBuilder(const RBBISetBuilder &other); // forbid copying of this class |
| michael@0 | 124 | RBBISetBuilder &operator=(const RBBISetBuilder &other); // forbid copying of this class |
| michael@0 | 125 | }; |
| michael@0 | 126 | |
| michael@0 | 127 | |
| michael@0 | 128 | |
| michael@0 | 129 | U_NAMESPACE_END |
| michael@0 | 130 | #endif |
