xref: /aosp_15_r20/external/llvm/utils/TableGen/SequenceToOffsetTable.h (revision 9880d6810fe72a1726cb53787c6711e909410d58)
1*9880d681SAndroid Build Coastguard Worker //===-- SequenceToOffsetTable.h - Compress similar sequences ----*- C++ -*-===//
2*9880d681SAndroid Build Coastguard Worker //
3*9880d681SAndroid Build Coastguard Worker //                     The LLVM Compiler Infrastructure
4*9880d681SAndroid Build Coastguard Worker //
5*9880d681SAndroid Build Coastguard Worker // This file is distributed under the University of Illinois Open Source
6*9880d681SAndroid Build Coastguard Worker // License. See LICENSE.TXT for details.
7*9880d681SAndroid Build Coastguard Worker //
8*9880d681SAndroid Build Coastguard Worker //===----------------------------------------------------------------------===//
9*9880d681SAndroid Build Coastguard Worker //
10*9880d681SAndroid Build Coastguard Worker // SequenceToOffsetTable can be used to emit a number of null-terminated
11*9880d681SAndroid Build Coastguard Worker // sequences as one big array.  Use the same memory when a sequence is a suffix
12*9880d681SAndroid Build Coastguard Worker // of another.
13*9880d681SAndroid Build Coastguard Worker //
14*9880d681SAndroid Build Coastguard Worker //===----------------------------------------------------------------------===//
15*9880d681SAndroid Build Coastguard Worker 
16*9880d681SAndroid Build Coastguard Worker #ifndef LLVM_UTILS_TABLEGEN_SEQUENCETOOFFSETTABLE_H
17*9880d681SAndroid Build Coastguard Worker #define LLVM_UTILS_TABLEGEN_SEQUENCETOOFFSETTABLE_H
18*9880d681SAndroid Build Coastguard Worker 
19*9880d681SAndroid Build Coastguard Worker #include "llvm/Support/raw_ostream.h"
20*9880d681SAndroid Build Coastguard Worker #include <algorithm>
21*9880d681SAndroid Build Coastguard Worker #include <cassert>
22*9880d681SAndroid Build Coastguard Worker #include <cctype>
23*9880d681SAndroid Build Coastguard Worker #include <functional>
24*9880d681SAndroid Build Coastguard Worker #include <map>
25*9880d681SAndroid Build Coastguard Worker 
26*9880d681SAndroid Build Coastguard Worker namespace llvm {
27*9880d681SAndroid Build Coastguard Worker 
28*9880d681SAndroid Build Coastguard Worker /// SequenceToOffsetTable - Collect a number of terminated sequences of T.
29*9880d681SAndroid Build Coastguard Worker /// Compute the layout of a table that contains all the sequences, possibly by
30*9880d681SAndroid Build Coastguard Worker /// reusing entries.
31*9880d681SAndroid Build Coastguard Worker ///
32*9880d681SAndroid Build Coastguard Worker /// @tparam SeqT The sequence container. (vector or string).
33*9880d681SAndroid Build Coastguard Worker /// @tparam Less A stable comparator for SeqT elements.
34*9880d681SAndroid Build Coastguard Worker template<typename SeqT, typename Less = std::less<typename SeqT::value_type> >
35*9880d681SAndroid Build Coastguard Worker class SequenceToOffsetTable {
36*9880d681SAndroid Build Coastguard Worker   typedef typename SeqT::value_type ElemT;
37*9880d681SAndroid Build Coastguard Worker 
38*9880d681SAndroid Build Coastguard Worker   // Define a comparator for SeqT that sorts a suffix immediately before a
39*9880d681SAndroid Build Coastguard Worker   // sequence with that suffix.
40*9880d681SAndroid Build Coastguard Worker   struct SeqLess : public std::binary_function<SeqT, SeqT, bool> {
41*9880d681SAndroid Build Coastguard Worker     Less L;
operatorSeqLess42*9880d681SAndroid Build Coastguard Worker     bool operator()(const SeqT &A, const SeqT &B) const {
43*9880d681SAndroid Build Coastguard Worker       return std::lexicographical_compare(A.rbegin(), A.rend(),
44*9880d681SAndroid Build Coastguard Worker                                           B.rbegin(), B.rend(), L);
45*9880d681SAndroid Build Coastguard Worker     }
46*9880d681SAndroid Build Coastguard Worker   };
47*9880d681SAndroid Build Coastguard Worker 
48*9880d681SAndroid Build Coastguard Worker   // Keep sequences ordered according to SeqLess so suffixes are easy to find.
49*9880d681SAndroid Build Coastguard Worker   // Map each sequence to its offset in the table.
50*9880d681SAndroid Build Coastguard Worker   typedef std::map<SeqT, unsigned, SeqLess> SeqMap;
51*9880d681SAndroid Build Coastguard Worker 
52*9880d681SAndroid Build Coastguard Worker   // Sequences added so far, with suffixes removed.
53*9880d681SAndroid Build Coastguard Worker   SeqMap Seqs;
54*9880d681SAndroid Build Coastguard Worker 
55*9880d681SAndroid Build Coastguard Worker   // Entries in the final table, or 0 before layout was called.
56*9880d681SAndroid Build Coastguard Worker   unsigned Entries;
57*9880d681SAndroid Build Coastguard Worker 
58*9880d681SAndroid Build Coastguard Worker   // isSuffix - Returns true if A is a suffix of B.
isSuffix(const SeqT & A,const SeqT & B)59*9880d681SAndroid Build Coastguard Worker   static bool isSuffix(const SeqT &A, const SeqT &B) {
60*9880d681SAndroid Build Coastguard Worker     return A.size() <= B.size() && std::equal(A.rbegin(), A.rend(), B.rbegin());
61*9880d681SAndroid Build Coastguard Worker   }
62*9880d681SAndroid Build Coastguard Worker 
63*9880d681SAndroid Build Coastguard Worker public:
SequenceToOffsetTable()64*9880d681SAndroid Build Coastguard Worker   SequenceToOffsetTable() : Entries(0) {}
65*9880d681SAndroid Build Coastguard Worker 
66*9880d681SAndroid Build Coastguard Worker   /// add - Add a sequence to the table.
67*9880d681SAndroid Build Coastguard Worker   /// This must be called before layout().
add(const SeqT & Seq)68*9880d681SAndroid Build Coastguard Worker   void add(const SeqT &Seq) {
69*9880d681SAndroid Build Coastguard Worker     assert(Entries == 0 && "Cannot call add() after layout()");
70*9880d681SAndroid Build Coastguard Worker     typename SeqMap::iterator I = Seqs.lower_bound(Seq);
71*9880d681SAndroid Build Coastguard Worker 
72*9880d681SAndroid Build Coastguard Worker     // If SeqMap contains a sequence that has Seq as a suffix, I will be
73*9880d681SAndroid Build Coastguard Worker     // pointing to it.
74*9880d681SAndroid Build Coastguard Worker     if (I != Seqs.end() && isSuffix(Seq, I->first))
75*9880d681SAndroid Build Coastguard Worker       return;
76*9880d681SAndroid Build Coastguard Worker 
77*9880d681SAndroid Build Coastguard Worker     I = Seqs.insert(I, std::make_pair(Seq, 0u));
78*9880d681SAndroid Build Coastguard Worker 
79*9880d681SAndroid Build Coastguard Worker     // The entry before I may be a suffix of Seq that can now be erased.
80*9880d681SAndroid Build Coastguard Worker     if (I != Seqs.begin() && isSuffix((--I)->first, Seq))
81*9880d681SAndroid Build Coastguard Worker       Seqs.erase(I);
82*9880d681SAndroid Build Coastguard Worker   }
83*9880d681SAndroid Build Coastguard Worker 
empty()84*9880d681SAndroid Build Coastguard Worker   bool empty() const { return Seqs.empty(); }
85*9880d681SAndroid Build Coastguard Worker 
size()86*9880d681SAndroid Build Coastguard Worker   unsigned size() const {
87*9880d681SAndroid Build Coastguard Worker     assert(Entries && "Call layout() before size()");
88*9880d681SAndroid Build Coastguard Worker     return Entries;
89*9880d681SAndroid Build Coastguard Worker   }
90*9880d681SAndroid Build Coastguard Worker 
91*9880d681SAndroid Build Coastguard Worker   /// layout - Computes the final table layout.
layout()92*9880d681SAndroid Build Coastguard Worker   void layout() {
93*9880d681SAndroid Build Coastguard Worker     assert(Entries == 0 && "Can only call layout() once");
94*9880d681SAndroid Build Coastguard Worker     // Lay out the table in Seqs iteration order.
95*9880d681SAndroid Build Coastguard Worker     for (typename SeqMap::iterator I = Seqs.begin(), E = Seqs.end(); I != E;
96*9880d681SAndroid Build Coastguard Worker          ++I) {
97*9880d681SAndroid Build Coastguard Worker       I->second = Entries;
98*9880d681SAndroid Build Coastguard Worker       // Include space for a terminator.
99*9880d681SAndroid Build Coastguard Worker       Entries += I->first.size() + 1;
100*9880d681SAndroid Build Coastguard Worker     }
101*9880d681SAndroid Build Coastguard Worker   }
102*9880d681SAndroid Build Coastguard Worker 
103*9880d681SAndroid Build Coastguard Worker   /// get - Returns the offset of Seq in the final table.
get(const SeqT & Seq)104*9880d681SAndroid Build Coastguard Worker   unsigned get(const SeqT &Seq) const {
105*9880d681SAndroid Build Coastguard Worker     assert(Entries && "Call layout() before get()");
106*9880d681SAndroid Build Coastguard Worker     typename SeqMap::const_iterator I = Seqs.lower_bound(Seq);
107*9880d681SAndroid Build Coastguard Worker     assert(I != Seqs.end() && isSuffix(Seq, I->first) &&
108*9880d681SAndroid Build Coastguard Worker            "get() called with sequence that wasn't added first");
109*9880d681SAndroid Build Coastguard Worker     return I->second + (I->first.size() - Seq.size());
110*9880d681SAndroid Build Coastguard Worker   }
111*9880d681SAndroid Build Coastguard Worker 
112*9880d681SAndroid Build Coastguard Worker   /// emit - Print out the table as the body of an array initializer.
113*9880d681SAndroid Build Coastguard Worker   /// Use the Print function to print elements.
114*9880d681SAndroid Build Coastguard Worker   void emit(raw_ostream &OS,
115*9880d681SAndroid Build Coastguard Worker             void (*Print)(raw_ostream&, ElemT),
116*9880d681SAndroid Build Coastguard Worker             const char *Term = "0") const {
117*9880d681SAndroid Build Coastguard Worker     assert(Entries && "Call layout() before emit()");
118*9880d681SAndroid Build Coastguard Worker     for (typename SeqMap::const_iterator I = Seqs.begin(), E = Seqs.end();
119*9880d681SAndroid Build Coastguard Worker          I != E; ++I) {
120*9880d681SAndroid Build Coastguard Worker       OS << "  /* " << I->second << " */ ";
121*9880d681SAndroid Build Coastguard Worker       for (typename SeqT::const_iterator SI = I->first.begin(),
122*9880d681SAndroid Build Coastguard Worker              SE = I->first.end(); SI != SE; ++SI) {
123*9880d681SAndroid Build Coastguard Worker         Print(OS, *SI);
124*9880d681SAndroid Build Coastguard Worker         OS << ", ";
125*9880d681SAndroid Build Coastguard Worker       }
126*9880d681SAndroid Build Coastguard Worker       OS << Term << ",\n";
127*9880d681SAndroid Build Coastguard Worker     }
128*9880d681SAndroid Build Coastguard Worker   }
129*9880d681SAndroid Build Coastguard Worker };
130*9880d681SAndroid Build Coastguard Worker 
131*9880d681SAndroid Build Coastguard Worker // Helper function for SequenceToOffsetTable<string>.
printChar(raw_ostream & OS,char C)132*9880d681SAndroid Build Coastguard Worker static inline void printChar(raw_ostream &OS, char C) {
133*9880d681SAndroid Build Coastguard Worker   unsigned char UC(C);
134*9880d681SAndroid Build Coastguard Worker   if (isalnum(UC) || ispunct(UC)) {
135*9880d681SAndroid Build Coastguard Worker     OS << '\'';
136*9880d681SAndroid Build Coastguard Worker     if (C == '\\' || C == '\'')
137*9880d681SAndroid Build Coastguard Worker       OS << '\\';
138*9880d681SAndroid Build Coastguard Worker     OS << C << '\'';
139*9880d681SAndroid Build Coastguard Worker   } else {
140*9880d681SAndroid Build Coastguard Worker     OS << unsigned(UC);
141*9880d681SAndroid Build Coastguard Worker   }
142*9880d681SAndroid Build Coastguard Worker }
143*9880d681SAndroid Build Coastguard Worker 
144*9880d681SAndroid Build Coastguard Worker } // end namespace llvm
145*9880d681SAndroid Build Coastguard Worker 
146*9880d681SAndroid Build Coastguard Worker #endif
147