1//===- StringMatcher.cpp - Generate a matcher for input strings -----------===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9// This file implements the StringMatcher class.
10//
11//===----------------------------------------------------------------------===//
12
13#include "llvm/TableGen/StringMatcher.h"
14#include "llvm/ADT/StringRef.h"
15#include "llvm/Support/ErrorHandling.h"
16#include "llvm/Support/raw_ostream.h"
17#include <cassert>
18#include <map>
19#include <string>
20#include <utility>
21#include <vector>
22
23using namespace llvm;
24
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.
28static unsigned
29FindFirstNonCommonLetter(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];
35
36    for (const StringMatcher::StringPair *Match : Matches)
37      if (Match->first[i] != Letter)
38        return i;
39  }
40
41  return Matches[0]->first.size();
42}
43
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.
47///
48/// \return - True if control can leave the emitted code fragment.
49bool StringMatcher::EmitStringMatcherForChar(
50    const std::vector<const StringPair *> &Matches, unsigned CharNo,
51    unsigned IndentCount, bool IgnoreDuplicates) const {
52  assert(!Matches.empty() && "Must have at least one string to match!");
53  std::string Indent(IndentCount * 2 + 4, ' ');
54
55  // If we have verified that the entire string matches, we're done: output the
56  // matching code.
57  if (CharNo == Matches[0]->first.size()) {
58    if (Matches.size() > 1 && !IgnoreDuplicates)
59      report_fatal_error("Had duplicate keys to match on");
60
61    // If the to-execute code has \n's in it, indent each subsequent line.
62    StringRef Code = Matches[0]->second;
63
64    std::pair<StringRef, StringRef> Split = Code.split('\n');
65    OS << Indent << Split.first << "\t // \"" << Matches[0]->first << "\"\n";
66
67    Code = Split.second;
68    while (!Code.empty()) {
69      Split = Code.split('\n');
70      OS << Indent << Split.first << "\n";
71      Code = Split.second;
72    }
73    return false;
74  }
75
76  // Bucket the matches by the character we are comparing.
77  std::map<char, std::vector<const StringPair*>> MatchesByLetter;
78
79  for (const StringPair *Match : Matches)
80    MatchesByLetter[Match->first[CharNo]].push_back(Match);
81
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;
87
88    // Emit code to break out if the prefix doesn't match.
89    if (NumChars == 1) {
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";
95    } else {
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";
102    }
103
104    return EmitStringMatcherForChar(Matches, FirstNonCommonLetter, IndentCount,
105                                    IgnoreDuplicates);
106  }
107
108  // Otherwise, we have multiple possible things, emit a switch on the
109  // character.
110  OS << Indent << "switch (" << StrVariableName << "[" << CharNo << "]) {\n";
111  OS << Indent << "default: break;\n";
112
113  for (const auto &LI : MatchesByLetter) {
114    // TODO: escape hard stuff (like \n) if we ever care about it.
115    OS << Indent << "case '" << LI.first << "':\t // " << LI.second.size()
116       << " string";
117    if (LI.second.size() != 1)
118      OS << 's';
119    OS << " to match.\n";
120    if (EmitStringMatcherForChar(LI.second, CharNo + 1, IndentCount + 1,
121                                 IgnoreDuplicates))
122      OS << Indent << "  break;\n";
123  }
124
125  OS << Indent << "}\n";
126  return true;
127}
128
129/// Emit - Top level entry point.
130///
131void StringMatcher::Emit(unsigned Indent, bool IgnoreDuplicates) const {
132  // If nothing to match, just fall through.
133  if (Matches.empty()) return;
134
135  // First level categorization: group strings by length.
136  std::map<unsigned, std::vector<const StringPair*>> MatchesByLength;
137
138  for (const StringPair &Match : Matches)
139    MatchesByLength[Match.first.size()].push_back(&Match);
140
141  // Output a switch statement on length and categorize the elements within each
142  // bin.
143  OS.indent(Indent*2+2) << "switch (" << StrVariableName << ".size()) {\n";
144  OS.indent(Indent*2+2) << "default: break;\n";
145
146  for (const auto &LI : MatchesByLength) {
147    OS.indent(Indent * 2 + 2)
148        << "case " << LI.first << ":\t // " << LI.second.size() << " string"
149        << (LI.second.size() == 1 ? "" : "s") << " to match.\n";
150    if (EmitStringMatcherForChar(LI.second, 0, Indent, IgnoreDuplicates))
151      OS.indent(Indent*2+4) << "break;\n";
152  }
153
154  OS.indent(Indent*2+2) << "}\n";
155}
156