1 //===- StringMatcher.cpp - Generate a matcher for input strings -----------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file implements the StringMatcher class.
12 //===----------------------------------------------------------------------===//
14 #include "llvm/ADT/StringRef.h"
15 #include "llvm/Support/raw_ostream.h"
16 #include "llvm/TableGen/StringMatcher.h"
25 /// FindFirstNonCommonLetter - Find the first character in the keys of the
26 /// string pairs that is not shared across the whole set of strings. All
27 /// strings are assumed to have the same length.
29 FindFirstNonCommonLetter(const std::vector<const
30 StringMatcher::StringPair*> &Matches) {
31 assert(!Matches.empty());
32 for (unsigned i = 0, e = Matches[0]->first.size(); i != e; ++i) {
33 // Check to see if letter i is the same across the set.
34 char Letter = Matches[0]->first[i];
36 for (unsigned str = 0, e = Matches.size(); str != e; ++str)
37 if (Matches[str]->first[i] != Letter)
41 return Matches[0]->first.size();
44 /// EmitStringMatcherForChar - Given a set of strings that are known to be the
45 /// same length and whose characters leading up to CharNo are the same, emit
46 /// code to verify that CharNo and later are the same.
48 /// \return - True if control can leave the emitted code fragment.
50 EmitStringMatcherForChar(const std::vector<const StringPair*> &Matches,
51 unsigned CharNo, unsigned IndentCount) const {
52 assert(!Matches.empty() && "Must have at least one string to match!");
53 std::string Indent(IndentCount*2+4, ' ');
55 // If we have verified that the entire string matches, we're done: output the
57 if (CharNo == Matches[0]->first.size()) {
58 assert(Matches.size() == 1 && "Had duplicate keys to match on");
60 // If the to-execute code has \n's in it, indent each subsequent line.
61 StringRef Code = Matches[0]->second;
63 std::pair<StringRef, StringRef> Split = Code.split('\n');
64 OS << Indent << Split.first << "\t // \"" << Matches[0]->first << "\"\n";
67 while (!Code.empty()) {
68 Split = Code.split('\n');
69 OS << Indent << Split.first << "\n";
75 // Bucket the matches by the character we are comparing.
76 std::map<char, std::vector<const StringPair*>> MatchesByLetter;
78 for (unsigned i = 0, e = Matches.size(); i != e; ++i)
79 MatchesByLetter[Matches[i]->first[CharNo]].push_back(Matches[i]);
82 // If we have exactly one bucket to match, see how many characters are common
83 // across the whole set and match all of them at once.
84 if (MatchesByLetter.size() == 1) {
85 unsigned FirstNonCommonLetter = FindFirstNonCommonLetter(Matches);
86 unsigned NumChars = FirstNonCommonLetter-CharNo;
88 // Emit code to break out if the prefix doesn't match.
90 // Do the comparison with if (Str[1] != 'f')
91 // FIXME: Need to escape general characters.
92 OS << Indent << "if (" << StrVariableName << "[" << CharNo << "] != '"
93 << Matches[0]->first[CharNo] << "')\n";
94 OS << Indent << " break;\n";
96 // Do the comparison with if memcmp(Str.data()+1, "foo", 3).
97 // FIXME: Need to escape general strings.
98 OS << Indent << "if (memcmp(" << StrVariableName << ".data()+" << CharNo
99 << ", \"" << Matches[0]->first.substr(CharNo, NumChars) << "\", "
100 << NumChars << ") != 0)\n";
101 OS << Indent << " break;\n";
104 return EmitStringMatcherForChar(Matches, FirstNonCommonLetter, IndentCount);
107 // Otherwise, we have multiple possible things, emit a switch on the
109 OS << Indent << "switch (" << StrVariableName << "[" << CharNo << "]) {\n";
110 OS << Indent << "default: break;\n";
112 for (std::map<char, std::vector<const StringPair*>>::iterator LI =
113 MatchesByLetter.begin(), E = MatchesByLetter.end(); LI != E; ++LI) {
114 // TODO: escape hard stuff (like \n) if we ever care about it.
115 OS << Indent << "case '" << LI->first << "':\t // "
116 << LI->second.size() << " string";
117 if (LI->second.size() != 1) OS << 's';
118 OS << " to match.\n";
119 if (EmitStringMatcherForChar(LI->second, CharNo+1, IndentCount+1))
120 OS << Indent << " break;\n";
123 OS << Indent << "}\n";
127 /// Emit - Top level entry point.
129 void StringMatcher::Emit(unsigned Indent) const {
130 // If nothing to match, just fall through.
131 if (Matches.empty()) return;
133 // First level categorization: group strings by length.
134 std::map<unsigned, std::vector<const StringPair*>> MatchesByLength;
136 for (unsigned i = 0, e = Matches.size(); i != e; ++i)
137 MatchesByLength[Matches[i].first.size()].push_back(&Matches[i]);
139 // Output a switch statement on length and categorize the elements within each
141 OS.indent(Indent*2+2) << "switch (" << StrVariableName << ".size()) {\n";
142 OS.indent(Indent*2+2) << "default: break;\n";
144 for (std::map<unsigned, std::vector<const StringPair*>>::iterator LI =
145 MatchesByLength.begin(), E = MatchesByLength.end(); LI != E; ++LI) {
146 OS.indent(Indent*2+2) << "case " << LI->first << ":\t // "
148 << " string" << (LI->second.size() == 1 ? "" : "s") << " to match.\n";
149 if (EmitStringMatcherForChar(LI->second, 0, Indent))
150 OS.indent(Indent*2+4) << "break;\n";
153 OS.indent(Indent*2+2) << "}\n";