1*bf9a7730SZachary Turner //===-- ConstString.cpp -----------------------------------------*- C++ -*-===//
2*bf9a7730SZachary Turner //
3*bf9a7730SZachary Turner //                     The LLVM Compiler Infrastructure
4*bf9a7730SZachary Turner //
5*bf9a7730SZachary Turner // This file is distributed under the University of Illinois Open Source
6*bf9a7730SZachary Turner // License. See LICENSE.TXT for details.
7*bf9a7730SZachary Turner //
8*bf9a7730SZachary Turner //===----------------------------------------------------------------------===//
9*bf9a7730SZachary Turner 
10*bf9a7730SZachary Turner #include "lldb/Utility/ConstString.h"
11*bf9a7730SZachary Turner 
12*bf9a7730SZachary Turner // C Includes
13*bf9a7730SZachary Turner // C++ Includes
14*bf9a7730SZachary Turner #include <array>
15*bf9a7730SZachary Turner #include <mutex>
16*bf9a7730SZachary Turner 
17*bf9a7730SZachary Turner // Other libraries and framework includes
18*bf9a7730SZachary Turner #include "llvm/ADT/StringExtras.h"
19*bf9a7730SZachary Turner #include "llvm/ADT/StringMap.h"
20*bf9a7730SZachary Turner #include "llvm/Support/RWMutex.h"
21*bf9a7730SZachary Turner 
22*bf9a7730SZachary Turner // Project includes
23*bf9a7730SZachary Turner #include "lldb/Utility/Stream.h"
24*bf9a7730SZachary Turner 
25*bf9a7730SZachary Turner using namespace lldb_private;
26*bf9a7730SZachary Turner 
27*bf9a7730SZachary Turner class Pool {
28*bf9a7730SZachary Turner public:
29*bf9a7730SZachary Turner   typedef const char *StringPoolValueType;
30*bf9a7730SZachary Turner   typedef llvm::StringMap<StringPoolValueType, llvm::BumpPtrAllocator>
31*bf9a7730SZachary Turner       StringPool;
32*bf9a7730SZachary Turner   typedef llvm::StringMapEntry<StringPoolValueType> StringPoolEntryType;
33*bf9a7730SZachary Turner 
34*bf9a7730SZachary Turner   static StringPoolEntryType &
35*bf9a7730SZachary Turner   GetStringMapEntryFromKeyData(const char *keyData) {
36*bf9a7730SZachary Turner     char *ptr = const_cast<char *>(keyData) - sizeof(StringPoolEntryType);
37*bf9a7730SZachary Turner     return *reinterpret_cast<StringPoolEntryType *>(ptr);
38*bf9a7730SZachary Turner   }
39*bf9a7730SZachary Turner 
40*bf9a7730SZachary Turner   size_t GetConstCStringLength(const char *ccstr) const {
41*bf9a7730SZachary Turner     if (ccstr != nullptr) {
42*bf9a7730SZachary Turner       const uint8_t h = hash(llvm::StringRef(ccstr));
43*bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
44*bf9a7730SZachary Turner       const StringPoolEntryType &entry = GetStringMapEntryFromKeyData(ccstr);
45*bf9a7730SZachary Turner       return entry.getKey().size();
46*bf9a7730SZachary Turner     }
47*bf9a7730SZachary Turner     return 0;
48*bf9a7730SZachary Turner   }
49*bf9a7730SZachary Turner 
50*bf9a7730SZachary Turner   StringPoolValueType GetMangledCounterpart(const char *ccstr) const {
51*bf9a7730SZachary Turner     if (ccstr != nullptr) {
52*bf9a7730SZachary Turner       const uint8_t h = hash(llvm::StringRef(ccstr));
53*bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
54*bf9a7730SZachary Turner       return GetStringMapEntryFromKeyData(ccstr).getValue();
55*bf9a7730SZachary Turner     }
56*bf9a7730SZachary Turner     return nullptr;
57*bf9a7730SZachary Turner   }
58*bf9a7730SZachary Turner 
59*bf9a7730SZachary Turner   bool SetMangledCounterparts(const char *key_ccstr, const char *value_ccstr) {
60*bf9a7730SZachary Turner     if (key_ccstr != nullptr && value_ccstr != nullptr) {
61*bf9a7730SZachary Turner       {
62*bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(key_ccstr));
63*bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
64*bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(key_ccstr).setValue(value_ccstr);
65*bf9a7730SZachary Turner       }
66*bf9a7730SZachary Turner       {
67*bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(value_ccstr));
68*bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
69*bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(value_ccstr).setValue(key_ccstr);
70*bf9a7730SZachary Turner       }
71*bf9a7730SZachary Turner       return true;
72*bf9a7730SZachary Turner     }
73*bf9a7730SZachary Turner     return false;
74*bf9a7730SZachary Turner   }
75*bf9a7730SZachary Turner 
76*bf9a7730SZachary Turner   const char *GetConstCString(const char *cstr) {
77*bf9a7730SZachary Turner     if (cstr != nullptr)
78*bf9a7730SZachary Turner       return GetConstCStringWithLength(cstr, strlen(cstr));
79*bf9a7730SZachary Turner     return nullptr;
80*bf9a7730SZachary Turner   }
81*bf9a7730SZachary Turner 
82*bf9a7730SZachary Turner   const char *GetConstCStringWithLength(const char *cstr, size_t cstr_len) {
83*bf9a7730SZachary Turner     if (cstr != nullptr)
84*bf9a7730SZachary Turner       return GetConstCStringWithStringRef(llvm::StringRef(cstr, cstr_len));
85*bf9a7730SZachary Turner     return nullptr;
86*bf9a7730SZachary Turner   }
87*bf9a7730SZachary Turner 
88*bf9a7730SZachary Turner   const char *GetConstCStringWithStringRef(const llvm::StringRef &string_ref) {
89*bf9a7730SZachary Turner     if (string_ref.data()) {
90*bf9a7730SZachary Turner       const uint8_t h = hash(string_ref);
91*bf9a7730SZachary Turner 
92*bf9a7730SZachary Turner       {
93*bf9a7730SZachary Turner         llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
94*bf9a7730SZachary Turner         auto it = m_string_pools[h].m_string_map.find(string_ref);
95*bf9a7730SZachary Turner         if (it != m_string_pools[h].m_string_map.end())
96*bf9a7730SZachary Turner           return it->getKeyData();
97*bf9a7730SZachary Turner       }
98*bf9a7730SZachary Turner 
99*bf9a7730SZachary Turner       llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
100*bf9a7730SZachary Turner       StringPoolEntryType &entry =
101*bf9a7730SZachary Turner           *m_string_pools[h]
102*bf9a7730SZachary Turner                .m_string_map.insert(std::make_pair(string_ref, nullptr))
103*bf9a7730SZachary Turner                .first;
104*bf9a7730SZachary Turner       return entry.getKeyData();
105*bf9a7730SZachary Turner     }
106*bf9a7730SZachary Turner     return nullptr;
107*bf9a7730SZachary Turner   }
108*bf9a7730SZachary Turner 
109*bf9a7730SZachary Turner   const char *
110*bf9a7730SZachary Turner   GetConstCStringAndSetMangledCounterPart(const char *demangled_cstr,
111*bf9a7730SZachary Turner                                           const char *mangled_ccstr) {
112*bf9a7730SZachary Turner     if (demangled_cstr != nullptr) {
113*bf9a7730SZachary Turner       const char *demangled_ccstr = nullptr;
114*bf9a7730SZachary Turner 
115*bf9a7730SZachary Turner       {
116*bf9a7730SZachary Turner         llvm::StringRef string_ref(demangled_cstr);
117*bf9a7730SZachary Turner         const uint8_t h = hash(string_ref);
118*bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
119*bf9a7730SZachary Turner 
120*bf9a7730SZachary Turner         // Make string pool entry with the mangled counterpart already set
121*bf9a7730SZachary Turner         StringPoolEntryType &entry =
122*bf9a7730SZachary Turner             *m_string_pools[h]
123*bf9a7730SZachary Turner                  .m_string_map.insert(std::make_pair(string_ref, mangled_ccstr))
124*bf9a7730SZachary Turner                  .first;
125*bf9a7730SZachary Turner 
126*bf9a7730SZachary Turner         // Extract the const version of the demangled_cstr
127*bf9a7730SZachary Turner         demangled_ccstr = entry.getKeyData();
128*bf9a7730SZachary Turner       }
129*bf9a7730SZachary Turner 
130*bf9a7730SZachary Turner       {
131*bf9a7730SZachary Turner         // Now assign the demangled const string as the counterpart of the
132*bf9a7730SZachary Turner         // mangled const string...
133*bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(mangled_ccstr));
134*bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
135*bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(mangled_ccstr).setValue(demangled_ccstr);
136*bf9a7730SZachary Turner       }
137*bf9a7730SZachary Turner 
138*bf9a7730SZachary Turner       // Return the constant demangled C string
139*bf9a7730SZachary Turner       return demangled_ccstr;
140*bf9a7730SZachary Turner     }
141*bf9a7730SZachary Turner     return nullptr;
142*bf9a7730SZachary Turner   }
143*bf9a7730SZachary Turner 
144*bf9a7730SZachary Turner   const char *GetConstTrimmedCStringWithLength(const char *cstr,
145*bf9a7730SZachary Turner                                                size_t cstr_len) {
146*bf9a7730SZachary Turner     if (cstr != nullptr) {
147*bf9a7730SZachary Turner       const size_t trimmed_len = std::min<size_t>(strlen(cstr), cstr_len);
148*bf9a7730SZachary Turner       return GetConstCStringWithLength(cstr, trimmed_len);
149*bf9a7730SZachary Turner     }
150*bf9a7730SZachary Turner     return nullptr;
151*bf9a7730SZachary Turner   }
152*bf9a7730SZachary Turner 
153*bf9a7730SZachary Turner   //------------------------------------------------------------------
154*bf9a7730SZachary Turner   // Return the size in bytes that this object and any items in its
155*bf9a7730SZachary Turner   // collection of uniqued strings + data count values takes in
156*bf9a7730SZachary Turner   // memory.
157*bf9a7730SZachary Turner   //------------------------------------------------------------------
158*bf9a7730SZachary Turner   size_t MemorySize() const {
159*bf9a7730SZachary Turner     size_t mem_size = sizeof(Pool);
160*bf9a7730SZachary Turner     for (const auto &pool : m_string_pools) {
161*bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(pool.m_mutex);
162*bf9a7730SZachary Turner       for (const auto &entry : pool.m_string_map)
163*bf9a7730SZachary Turner         mem_size += sizeof(StringPoolEntryType) + entry.getKey().size();
164*bf9a7730SZachary Turner     }
165*bf9a7730SZachary Turner     return mem_size;
166*bf9a7730SZachary Turner   }
167*bf9a7730SZachary Turner 
168*bf9a7730SZachary Turner protected:
169*bf9a7730SZachary Turner   uint8_t hash(const llvm::StringRef &s) const {
170*bf9a7730SZachary Turner     uint32_t h = llvm::HashString(s);
171*bf9a7730SZachary Turner     return ((h >> 24) ^ (h >> 16) ^ (h >> 8) ^ h) & 0xff;
172*bf9a7730SZachary Turner   }
173*bf9a7730SZachary Turner 
174*bf9a7730SZachary Turner   struct PoolEntry {
175*bf9a7730SZachary Turner     mutable llvm::sys::SmartRWMutex<false> m_mutex;
176*bf9a7730SZachary Turner     StringPool m_string_map;
177*bf9a7730SZachary Turner   };
178*bf9a7730SZachary Turner 
179*bf9a7730SZachary Turner   std::array<PoolEntry, 256> m_string_pools;
180*bf9a7730SZachary Turner };
181*bf9a7730SZachary Turner 
182*bf9a7730SZachary Turner //----------------------------------------------------------------------
183*bf9a7730SZachary Turner // Frameworks and dylibs aren't supposed to have global C++
184*bf9a7730SZachary Turner // initializers so we hide the string pool in a static function so
185*bf9a7730SZachary Turner // that it will get initialized on the first call to this static
186*bf9a7730SZachary Turner // function.
187*bf9a7730SZachary Turner //
188*bf9a7730SZachary Turner // Note, for now we make the string pool a pointer to the pool, because
189*bf9a7730SZachary Turner // we can't guarantee that some objects won't get destroyed after the
190*bf9a7730SZachary Turner // global destructor chain is run, and trying to make sure no destructors
191*bf9a7730SZachary Turner // touch ConstStrings is difficult.  So we leak the pool instead.
192*bf9a7730SZachary Turner //----------------------------------------------------------------------
193*bf9a7730SZachary Turner static Pool &StringPool() {
194*bf9a7730SZachary Turner   static std::once_flag g_pool_initialization_flag;
195*bf9a7730SZachary Turner   static Pool *g_string_pool = nullptr;
196*bf9a7730SZachary Turner 
197*bf9a7730SZachary Turner   std::call_once(g_pool_initialization_flag,
198*bf9a7730SZachary Turner                  []() { g_string_pool = new Pool(); });
199*bf9a7730SZachary Turner 
200*bf9a7730SZachary Turner   return *g_string_pool;
201*bf9a7730SZachary Turner }
202*bf9a7730SZachary Turner 
203*bf9a7730SZachary Turner ConstString::ConstString(const char *cstr)
204*bf9a7730SZachary Turner     : m_string(StringPool().GetConstCString(cstr)) {}
205*bf9a7730SZachary Turner 
206*bf9a7730SZachary Turner ConstString::ConstString(const char *cstr, size_t cstr_len)
207*bf9a7730SZachary Turner     : m_string(StringPool().GetConstCStringWithLength(cstr, cstr_len)) {}
208*bf9a7730SZachary Turner 
209*bf9a7730SZachary Turner ConstString::ConstString(const llvm::StringRef &s)
210*bf9a7730SZachary Turner     : m_string(StringPool().GetConstCStringWithLength(s.data(), s.size())) {}
211*bf9a7730SZachary Turner 
212*bf9a7730SZachary Turner bool ConstString::operator<(const ConstString &rhs) const {
213*bf9a7730SZachary Turner   if (m_string == rhs.m_string)
214*bf9a7730SZachary Turner     return false;
215*bf9a7730SZachary Turner 
216*bf9a7730SZachary Turner   llvm::StringRef lhs_string_ref(m_string,
217*bf9a7730SZachary Turner                                  StringPool().GetConstCStringLength(m_string));
218*bf9a7730SZachary Turner   llvm::StringRef rhs_string_ref(
219*bf9a7730SZachary Turner       rhs.m_string, StringPool().GetConstCStringLength(rhs.m_string));
220*bf9a7730SZachary Turner 
221*bf9a7730SZachary Turner   // If both have valid C strings, then return the comparison
222*bf9a7730SZachary Turner   if (lhs_string_ref.data() && rhs_string_ref.data())
223*bf9a7730SZachary Turner     return lhs_string_ref < rhs_string_ref;
224*bf9a7730SZachary Turner 
225*bf9a7730SZachary Turner   // Else one of them was nullptr, so if LHS is nullptr then it is less than
226*bf9a7730SZachary Turner   return lhs_string_ref.data() == nullptr;
227*bf9a7730SZachary Turner }
228*bf9a7730SZachary Turner 
229*bf9a7730SZachary Turner Stream &lldb_private::operator<<(Stream &s, const ConstString &str) {
230*bf9a7730SZachary Turner   const char *cstr = str.GetCString();
231*bf9a7730SZachary Turner   if (cstr != nullptr)
232*bf9a7730SZachary Turner     s << cstr;
233*bf9a7730SZachary Turner 
234*bf9a7730SZachary Turner   return s;
235*bf9a7730SZachary Turner }
236*bf9a7730SZachary Turner 
237*bf9a7730SZachary Turner size_t ConstString::GetLength() const {
238*bf9a7730SZachary Turner   return StringPool().GetConstCStringLength(m_string);
239*bf9a7730SZachary Turner }
240*bf9a7730SZachary Turner 
241*bf9a7730SZachary Turner bool ConstString::Equals(const ConstString &lhs, const ConstString &rhs,
242*bf9a7730SZachary Turner                          const bool case_sensitive) {
243*bf9a7730SZachary Turner   if (lhs.m_string == rhs.m_string)
244*bf9a7730SZachary Turner     return true;
245*bf9a7730SZachary Turner 
246*bf9a7730SZachary Turner   // Since the pointers weren't equal, and identical ConstStrings always have
247*bf9a7730SZachary Turner   // identical pointers,
248*bf9a7730SZachary Turner   // the result must be false for case sensitive equality test.
249*bf9a7730SZachary Turner   if (case_sensitive)
250*bf9a7730SZachary Turner     return false;
251*bf9a7730SZachary Turner 
252*bf9a7730SZachary Turner   // perform case insensitive equality test
253*bf9a7730SZachary Turner   llvm::StringRef lhs_string_ref(
254*bf9a7730SZachary Turner       lhs.m_string, StringPool().GetConstCStringLength(lhs.m_string));
255*bf9a7730SZachary Turner   llvm::StringRef rhs_string_ref(
256*bf9a7730SZachary Turner       rhs.m_string, StringPool().GetConstCStringLength(rhs.m_string));
257*bf9a7730SZachary Turner   return lhs_string_ref.equals_lower(rhs_string_ref);
258*bf9a7730SZachary Turner }
259*bf9a7730SZachary Turner 
260*bf9a7730SZachary Turner int ConstString::Compare(const ConstString &lhs, const ConstString &rhs,
261*bf9a7730SZachary Turner                          const bool case_sensitive) {
262*bf9a7730SZachary Turner   // If the iterators are the same, this is the same string
263*bf9a7730SZachary Turner   const char *lhs_cstr = lhs.m_string;
264*bf9a7730SZachary Turner   const char *rhs_cstr = rhs.m_string;
265*bf9a7730SZachary Turner   if (lhs_cstr == rhs_cstr)
266*bf9a7730SZachary Turner     return 0;
267*bf9a7730SZachary Turner   if (lhs_cstr && rhs_cstr) {
268*bf9a7730SZachary Turner     llvm::StringRef lhs_string_ref(
269*bf9a7730SZachary Turner         lhs_cstr, StringPool().GetConstCStringLength(lhs_cstr));
270*bf9a7730SZachary Turner     llvm::StringRef rhs_string_ref(
271*bf9a7730SZachary Turner         rhs_cstr, StringPool().GetConstCStringLength(rhs_cstr));
272*bf9a7730SZachary Turner 
273*bf9a7730SZachary Turner     if (case_sensitive) {
274*bf9a7730SZachary Turner       return lhs_string_ref.compare(rhs_string_ref);
275*bf9a7730SZachary Turner     } else {
276*bf9a7730SZachary Turner       return lhs_string_ref.compare_lower(rhs_string_ref);
277*bf9a7730SZachary Turner     }
278*bf9a7730SZachary Turner   }
279*bf9a7730SZachary Turner 
280*bf9a7730SZachary Turner   if (lhs_cstr)
281*bf9a7730SZachary Turner     return +1; // LHS isn't nullptr but RHS is
282*bf9a7730SZachary Turner   else
283*bf9a7730SZachary Turner     return -1; // LHS is nullptr but RHS isn't
284*bf9a7730SZachary Turner }
285*bf9a7730SZachary Turner 
286*bf9a7730SZachary Turner void ConstString::Dump(Stream *s, const char *fail_value) const {
287*bf9a7730SZachary Turner   if (s != nullptr) {
288*bf9a7730SZachary Turner     const char *cstr = AsCString(fail_value);
289*bf9a7730SZachary Turner     if (cstr != nullptr)
290*bf9a7730SZachary Turner       s->PutCString(cstr);
291*bf9a7730SZachary Turner   }
292*bf9a7730SZachary Turner }
293*bf9a7730SZachary Turner 
294*bf9a7730SZachary Turner void ConstString::DumpDebug(Stream *s) const {
295*bf9a7730SZachary Turner   const char *cstr = GetCString();
296*bf9a7730SZachary Turner   size_t cstr_len = GetLength();
297*bf9a7730SZachary Turner   // Only print the parens if we have a non-nullptr string
298*bf9a7730SZachary Turner   const char *parens = cstr ? "\"" : "";
299*bf9a7730SZachary Turner   s->Printf("%*p: ConstString, string = %s%s%s, length = %" PRIu64,
300*bf9a7730SZachary Turner             static_cast<int>(sizeof(void *) * 2),
301*bf9a7730SZachary Turner             static_cast<const void *>(this), parens, cstr, parens,
302*bf9a7730SZachary Turner             static_cast<uint64_t>(cstr_len));
303*bf9a7730SZachary Turner }
304*bf9a7730SZachary Turner 
305*bf9a7730SZachary Turner void ConstString::SetCString(const char *cstr) {
306*bf9a7730SZachary Turner   m_string = StringPool().GetConstCString(cstr);
307*bf9a7730SZachary Turner }
308*bf9a7730SZachary Turner 
309*bf9a7730SZachary Turner void ConstString::SetString(const llvm::StringRef &s) {
310*bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringWithLength(s.data(), s.size());
311*bf9a7730SZachary Turner }
312*bf9a7730SZachary Turner 
313*bf9a7730SZachary Turner void ConstString::SetCStringWithMangledCounterpart(const char *demangled,
314*bf9a7730SZachary Turner                                                    const ConstString &mangled) {
315*bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringAndSetMangledCounterPart(
316*bf9a7730SZachary Turner       demangled, mangled.m_string);
317*bf9a7730SZachary Turner }
318*bf9a7730SZachary Turner 
319*bf9a7730SZachary Turner bool ConstString::GetMangledCounterpart(ConstString &counterpart) const {
320*bf9a7730SZachary Turner   counterpart.m_string = StringPool().GetMangledCounterpart(m_string);
321*bf9a7730SZachary Turner   return (bool)counterpart;
322*bf9a7730SZachary Turner }
323*bf9a7730SZachary Turner 
324*bf9a7730SZachary Turner void ConstString::SetCStringWithLength(const char *cstr, size_t cstr_len) {
325*bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringWithLength(cstr, cstr_len);
326*bf9a7730SZachary Turner }
327*bf9a7730SZachary Turner 
328*bf9a7730SZachary Turner void ConstString::SetTrimmedCStringWithLength(const char *cstr,
329*bf9a7730SZachary Turner                                               size_t cstr_len) {
330*bf9a7730SZachary Turner   m_string = StringPool().GetConstTrimmedCStringWithLength(cstr, cstr_len);
331*bf9a7730SZachary Turner }
332*bf9a7730SZachary Turner 
333*bf9a7730SZachary Turner size_t ConstString::StaticMemorySize() {
334*bf9a7730SZachary Turner   // Get the size of the static string pool
335*bf9a7730SZachary Turner   return StringPool().MemorySize();
336*bf9a7730SZachary Turner }
337