1//===- YAMLBench - Benchmark the YAMLParser implementation ----------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This program executes the YAMLParser on differntly sized YAML texts and
11// outputs the run time.
12//
13//===----------------------------------------------------------------------===//
14
15
16#include "llvm/ADT/SmallString.h"
17#include "llvm/Support/Casting.h"
18#include "llvm/Support/CommandLine.h"
19#include "llvm/Support/MemoryBuffer.h"
20#include "llvm/Support/raw_ostream.h"
21#include "llvm/Support/SourceMgr.h"
22#include "llvm/Support/system_error.h"
23#include "llvm/Support/Timer.h"
24#include "llvm/Support/YAMLParser.h"
25
26using namespace llvm;
27
28static cl::opt<bool>
29  DumpTokens( "tokens"
30            , cl::desc("Print the tokenization of the file.")
31            , cl::init(false)
32            );
33
34static cl::opt<bool>
35  DumpCanonical( "canonical"
36               , cl::desc("Print the canonical YAML for this file.")
37               , cl::init(false)
38               );
39
40static cl::opt<std::string>
41 Input(cl::Positional, cl::desc("<input>"));
42
43static cl::opt<bool>
44  Verify( "verify"
45        , cl::desc(
46            "Run a quick verification useful for regression testing")
47        , cl::init(false)
48        );
49
50static cl::opt<unsigned>
51  MemoryLimitMB("memory-limit", cl::desc(
52                  "Do not use more megabytes of memory"),
53                cl::init(1000));
54
55struct indent {
56  unsigned distance;
57  indent(unsigned d) : distance(d) {}
58};
59
60static raw_ostream &operator <<(raw_ostream &os, const indent &in) {
61  for (unsigned i = 0; i < in.distance; ++i)
62    os << "  ";
63  return os;
64}
65
66static void dumpNode( yaml::Node *n
67                    , unsigned Indent = 0
68                    , bool SuppressFirstIndent = false) {
69  if (!n)
70    return;
71  if (!SuppressFirstIndent)
72    outs() << indent(Indent);
73  StringRef Anchor = n->getAnchor();
74  if (!Anchor.empty())
75    outs() << "&" << Anchor << " ";
76  if (yaml::ScalarNode *sn = dyn_cast<yaml::ScalarNode>(n)) {
77    SmallString<32> Storage;
78    StringRef Val = sn->getValue(Storage);
79    outs() << "!!str \"" << yaml::escape(Val) << "\"";
80  } else if (yaml::SequenceNode *sn = dyn_cast<yaml::SequenceNode>(n)) {
81    outs() << "!!seq [\n";
82    ++Indent;
83    for (yaml::SequenceNode::iterator i = sn->begin(), e = sn->end();
84                                      i != e; ++i) {
85      dumpNode(i, Indent);
86      outs() << ",\n";
87    }
88    --Indent;
89    outs() << indent(Indent) << "]";
90  } else if (yaml::MappingNode *mn = dyn_cast<yaml::MappingNode>(n)) {
91    outs() << "!!map {\n";
92    ++Indent;
93    for (yaml::MappingNode::iterator i = mn->begin(), e = mn->end();
94                                     i != e; ++i) {
95      outs() << indent(Indent) << "? ";
96      dumpNode(i->getKey(), Indent, true);
97      outs() << "\n";
98      outs() << indent(Indent) << ": ";
99      dumpNode(i->getValue(), Indent, true);
100      outs() << ",\n";
101    }
102    --Indent;
103    outs() << indent(Indent) << "}";
104  } else if (yaml::AliasNode *an = dyn_cast<yaml::AliasNode>(n)){
105    outs() << "*" << an->getName();
106  } else if (dyn_cast<yaml::NullNode>(n)) {
107    outs() << "!!null null";
108  }
109}
110
111static void dumpStream(yaml::Stream &stream) {
112  for (yaml::document_iterator di = stream.begin(), de = stream.end(); di != de;
113       ++di) {
114    outs() << "%YAML 1.2\n"
115           << "---\n";
116    yaml::Node *n = di->getRoot();
117    if (n)
118      dumpNode(n);
119    else
120      break;
121    outs() << "\n...\n";
122  }
123}
124
125static void benchmark( llvm::TimerGroup &Group
126                     , llvm::StringRef Name
127                     , llvm::StringRef JSONText) {
128  llvm::Timer BaseLine((Name + ": Loop").str(), Group);
129  BaseLine.startTimer();
130  char C = 0;
131  for (llvm::StringRef::iterator I = JSONText.begin(),
132                                 E = JSONText.end();
133       I != E; ++I) { C += *I; }
134  BaseLine.stopTimer();
135  volatile char DontOptimizeOut = C; (void)DontOptimizeOut;
136
137  llvm::Timer Tokenizing((Name + ": Tokenizing").str(), Group);
138  Tokenizing.startTimer();
139  {
140    yaml::scanTokens(JSONText);
141  }
142  Tokenizing.stopTimer();
143
144  llvm::Timer Parsing((Name + ": Parsing").str(), Group);
145  Parsing.startTimer();
146  {
147    llvm::SourceMgr SM;
148    llvm::yaml::Stream stream(JSONText, SM);
149    stream.skip();
150  }
151  Parsing.stopTimer();
152}
153
154static std::string createJSONText(size_t MemoryMB, unsigned ValueSize) {
155  std::string JSONText;
156  llvm::raw_string_ostream Stream(JSONText);
157  Stream << "[\n";
158  size_t MemoryBytes = MemoryMB * 1024 * 1024;
159  while (JSONText.size() < MemoryBytes) {
160    Stream << " {\n"
161           << "  \"key1\": \"" << std::string(ValueSize, '*') << "\",\n"
162           << "  \"key2\": \"" << std::string(ValueSize, '*') << "\",\n"
163           << "  \"key3\": \"" << std::string(ValueSize, '*') << "\"\n"
164           << " }";
165    Stream.flush();
166    if (JSONText.size() < MemoryBytes) Stream << ",";
167    Stream << "\n";
168  }
169  Stream << "]\n";
170  Stream.flush();
171  return JSONText;
172}
173
174int main(int argc, char **argv) {
175  llvm::cl::ParseCommandLineOptions(argc, argv);
176  if (Input.getNumOccurrences()) {
177    OwningPtr<MemoryBuffer> Buf;
178    if (MemoryBuffer::getFileOrSTDIN(Input, Buf))
179      return 1;
180
181    llvm::SourceMgr sm;
182    if (DumpTokens) {
183      yaml::dumpTokens(Buf->getBuffer(), outs());
184    }
185
186    if (DumpCanonical) {
187      yaml::Stream stream(Buf->getBuffer(), sm);
188      dumpStream(stream);
189    }
190  }
191
192  if (Verify) {
193    llvm::TimerGroup Group("YAML parser benchmark");
194    benchmark(Group, "Fast", createJSONText(10, 500));
195  } else if (!DumpCanonical && !DumpTokens) {
196    llvm::TimerGroup Group("YAML parser benchmark");
197    benchmark(Group, "Small Values", createJSONText(MemoryLimitMB, 5));
198    benchmark(Group, "Medium Values", createJSONText(MemoryLimitMB, 500));
199    benchmark(Group, "Large Values", createJSONText(MemoryLimitMB, 50000));
200  }
201
202  return 0;
203}
204