1bf9a7730SZachary Turner //===-- ConstString.cpp -----------------------------------------*- C++ -*-===//
2bf9a7730SZachary Turner //
3bf9a7730SZachary Turner //                     The LLVM Compiler Infrastructure
4bf9a7730SZachary Turner //
5bf9a7730SZachary Turner // This file is distributed under the University of Illinois Open Source
6bf9a7730SZachary Turner // License. See LICENSE.TXT for details.
7bf9a7730SZachary Turner //
8bf9a7730SZachary Turner //===----------------------------------------------------------------------===//
9bf9a7730SZachary Turner 
10bf9a7730SZachary Turner #include "lldb/Utility/ConstString.h"
11bf9a7730SZachary Turner 
124479ac15SZachary Turner #include "lldb/Utility/Stream.h"
13bf9a7730SZachary Turner 
14bf9a7730SZachary Turner #include "llvm/ADT/StringMap.h"
154479ac15SZachary Turner #include "llvm/ADT/iterator.h"            // for iterator_facade_base
164479ac15SZachary Turner #include "llvm/Support/Allocator.h"       // for BumpPtrAllocator
17*560ce2c7SJonas Devlieghere #include "llvm/Support/DJB.h"             // for djbHash
184479ac15SZachary Turner #include "llvm/Support/FormatProviders.h" // for format_provider
19bf9a7730SZachary Turner #include "llvm/Support/RWMutex.h"
20c5f28e2aSKamil Rytarowski #include "llvm/Support/Threading.h"
214479ac15SZachary Turner 
224479ac15SZachary Turner #include <algorithm> // for min
234479ac15SZachary Turner #include <array>
244479ac15SZachary Turner #include <utility> // for make_pair, pair
254479ac15SZachary Turner 
264479ac15SZachary Turner #include <inttypes.h> // for PRIu64
274479ac15SZachary Turner #include <stdint.h>   // for uint8_t, uint32_t, uint64_t
284479ac15SZachary Turner #include <string.h>   // for size_t, strlen
29bf9a7730SZachary Turner 
30bf9a7730SZachary Turner using namespace lldb_private;
31bf9a7730SZachary Turner 
32bf9a7730SZachary Turner class Pool {
33bf9a7730SZachary Turner public:
34bf9a7730SZachary Turner   typedef const char *StringPoolValueType;
35bf9a7730SZachary Turner   typedef llvm::StringMap<StringPoolValueType, llvm::BumpPtrAllocator>
36bf9a7730SZachary Turner       StringPool;
37bf9a7730SZachary Turner   typedef llvm::StringMapEntry<StringPoolValueType> StringPoolEntryType;
38bf9a7730SZachary Turner 
39bf9a7730SZachary Turner   static StringPoolEntryType &
40bf9a7730SZachary Turner   GetStringMapEntryFromKeyData(const char *keyData) {
418070bf0aSPavel Labath     return StringPoolEntryType::GetStringMapEntryFromKeyData(keyData);
42bf9a7730SZachary Turner   }
43bf9a7730SZachary Turner 
448070bf0aSPavel Labath   static size_t GetConstCStringLength(const char *ccstr) {
45bf9a7730SZachary Turner     if (ccstr != nullptr) {
468070bf0aSPavel Labath       // Since the entry is read only, and we derive the entry entirely from the
478070bf0aSPavel Labath       // pointer, we don't need the lock.
48bf9a7730SZachary Turner       const StringPoolEntryType &entry = GetStringMapEntryFromKeyData(ccstr);
49bf9a7730SZachary Turner       return entry.getKey().size();
50bf9a7730SZachary Turner     }
51bf9a7730SZachary Turner     return 0;
52bf9a7730SZachary Turner   }
53bf9a7730SZachary Turner 
54bf9a7730SZachary Turner   StringPoolValueType GetMangledCounterpart(const char *ccstr) const {
55bf9a7730SZachary Turner     if (ccstr != nullptr) {
56bf9a7730SZachary Turner       const uint8_t h = hash(llvm::StringRef(ccstr));
57bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
58bf9a7730SZachary Turner       return GetStringMapEntryFromKeyData(ccstr).getValue();
59bf9a7730SZachary Turner     }
60bf9a7730SZachary Turner     return nullptr;
61bf9a7730SZachary Turner   }
62bf9a7730SZachary Turner 
63bf9a7730SZachary Turner   bool SetMangledCounterparts(const char *key_ccstr, const char *value_ccstr) {
64bf9a7730SZachary Turner     if (key_ccstr != nullptr && value_ccstr != nullptr) {
65bf9a7730SZachary Turner       {
66bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(key_ccstr));
67bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
68bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(key_ccstr).setValue(value_ccstr);
69bf9a7730SZachary Turner       }
70bf9a7730SZachary Turner       {
71bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(value_ccstr));
72bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
73bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(value_ccstr).setValue(key_ccstr);
74bf9a7730SZachary Turner       }
75bf9a7730SZachary Turner       return true;
76bf9a7730SZachary Turner     }
77bf9a7730SZachary Turner     return false;
78bf9a7730SZachary Turner   }
79bf9a7730SZachary Turner 
80bf9a7730SZachary Turner   const char *GetConstCString(const char *cstr) {
81bf9a7730SZachary Turner     if (cstr != nullptr)
82bf9a7730SZachary Turner       return GetConstCStringWithLength(cstr, strlen(cstr));
83bf9a7730SZachary Turner     return nullptr;
84bf9a7730SZachary Turner   }
85bf9a7730SZachary Turner 
86bf9a7730SZachary Turner   const char *GetConstCStringWithLength(const char *cstr, size_t cstr_len) {
87bf9a7730SZachary Turner     if (cstr != nullptr)
88bf9a7730SZachary Turner       return GetConstCStringWithStringRef(llvm::StringRef(cstr, cstr_len));
89bf9a7730SZachary Turner     return nullptr;
90bf9a7730SZachary Turner   }
91bf9a7730SZachary Turner 
92bf9a7730SZachary Turner   const char *GetConstCStringWithStringRef(const llvm::StringRef &string_ref) {
93bf9a7730SZachary Turner     if (string_ref.data()) {
94bf9a7730SZachary Turner       const uint8_t h = hash(string_ref);
95bf9a7730SZachary Turner 
96bf9a7730SZachary Turner       {
97bf9a7730SZachary Turner         llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
98bf9a7730SZachary Turner         auto it = m_string_pools[h].m_string_map.find(string_ref);
99bf9a7730SZachary Turner         if (it != m_string_pools[h].m_string_map.end())
100bf9a7730SZachary Turner           return it->getKeyData();
101bf9a7730SZachary Turner       }
102bf9a7730SZachary Turner 
103bf9a7730SZachary Turner       llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
104bf9a7730SZachary Turner       StringPoolEntryType &entry =
105bf9a7730SZachary Turner           *m_string_pools[h]
106bf9a7730SZachary Turner                .m_string_map.insert(std::make_pair(string_ref, nullptr))
107bf9a7730SZachary Turner                .first;
108bf9a7730SZachary Turner       return entry.getKeyData();
109bf9a7730SZachary Turner     }
110bf9a7730SZachary Turner     return nullptr;
111bf9a7730SZachary Turner   }
112bf9a7730SZachary Turner 
113bf9a7730SZachary Turner   const char *
114bf9a7730SZachary Turner   GetConstCStringAndSetMangledCounterPart(const char *demangled_cstr,
115bf9a7730SZachary Turner                                           const char *mangled_ccstr) {
116bf9a7730SZachary Turner     if (demangled_cstr != nullptr) {
117bf9a7730SZachary Turner       const char *demangled_ccstr = nullptr;
118bf9a7730SZachary Turner 
119bf9a7730SZachary Turner       {
120bf9a7730SZachary Turner         llvm::StringRef string_ref(demangled_cstr);
121bf9a7730SZachary Turner         const uint8_t h = hash(string_ref);
122bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
123bf9a7730SZachary Turner 
124bf9a7730SZachary Turner         // Make string pool entry with the mangled counterpart already set
125bf9a7730SZachary Turner         StringPoolEntryType &entry =
126bf9a7730SZachary Turner             *m_string_pools[h]
127bf9a7730SZachary Turner                  .m_string_map.insert(std::make_pair(string_ref, mangled_ccstr))
128bf9a7730SZachary Turner                  .first;
129bf9a7730SZachary Turner 
130bf9a7730SZachary Turner         // Extract the const version of the demangled_cstr
131bf9a7730SZachary Turner         demangled_ccstr = entry.getKeyData();
132bf9a7730SZachary Turner       }
133bf9a7730SZachary Turner 
134bf9a7730SZachary Turner       {
135bf9a7730SZachary Turner         // Now assign the demangled const string as the counterpart of the
136bf9a7730SZachary Turner         // mangled const string...
137bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(mangled_ccstr));
138bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
139bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(mangled_ccstr).setValue(demangled_ccstr);
140bf9a7730SZachary Turner       }
141bf9a7730SZachary Turner 
142bf9a7730SZachary Turner       // Return the constant demangled C string
143bf9a7730SZachary Turner       return demangled_ccstr;
144bf9a7730SZachary Turner     }
145bf9a7730SZachary Turner     return nullptr;
146bf9a7730SZachary Turner   }
147bf9a7730SZachary Turner 
148bf9a7730SZachary Turner   const char *GetConstTrimmedCStringWithLength(const char *cstr,
149bf9a7730SZachary Turner                                                size_t cstr_len) {
150bf9a7730SZachary Turner     if (cstr != nullptr) {
151bf9a7730SZachary Turner       const size_t trimmed_len = std::min<size_t>(strlen(cstr), cstr_len);
152bf9a7730SZachary Turner       return GetConstCStringWithLength(cstr, trimmed_len);
153bf9a7730SZachary Turner     }
154bf9a7730SZachary Turner     return nullptr;
155bf9a7730SZachary Turner   }
156bf9a7730SZachary Turner 
157bf9a7730SZachary Turner   //------------------------------------------------------------------
158bf9a7730SZachary Turner   // Return the size in bytes that this object and any items in its
159bf9a7730SZachary Turner   // collection of uniqued strings + data count values takes in
160bf9a7730SZachary Turner   // memory.
161bf9a7730SZachary Turner   //------------------------------------------------------------------
162bf9a7730SZachary Turner   size_t MemorySize() const {
163bf9a7730SZachary Turner     size_t mem_size = sizeof(Pool);
164bf9a7730SZachary Turner     for (const auto &pool : m_string_pools) {
165bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(pool.m_mutex);
166bf9a7730SZachary Turner       for (const auto &entry : pool.m_string_map)
167bf9a7730SZachary Turner         mem_size += sizeof(StringPoolEntryType) + entry.getKey().size();
168bf9a7730SZachary Turner     }
169bf9a7730SZachary Turner     return mem_size;
170bf9a7730SZachary Turner   }
171bf9a7730SZachary Turner 
172bf9a7730SZachary Turner protected:
173bf9a7730SZachary Turner   uint8_t hash(const llvm::StringRef &s) const {
174*560ce2c7SJonas Devlieghere     uint32_t h = llvm::djbHash(s);
175bf9a7730SZachary Turner     return ((h >> 24) ^ (h >> 16) ^ (h >> 8) ^ h) & 0xff;
176bf9a7730SZachary Turner   }
177bf9a7730SZachary Turner 
178bf9a7730SZachary Turner   struct PoolEntry {
179bf9a7730SZachary Turner     mutable llvm::sys::SmartRWMutex<false> m_mutex;
180bf9a7730SZachary Turner     StringPool m_string_map;
181bf9a7730SZachary Turner   };
182bf9a7730SZachary Turner 
183bf9a7730SZachary Turner   std::array<PoolEntry, 256> m_string_pools;
184bf9a7730SZachary Turner };
185bf9a7730SZachary Turner 
186bf9a7730SZachary Turner //----------------------------------------------------------------------
187bf9a7730SZachary Turner // Frameworks and dylibs aren't supposed to have global C++
188bf9a7730SZachary Turner // initializers so we hide the string pool in a static function so
189bf9a7730SZachary Turner // that it will get initialized on the first call to this static
190bf9a7730SZachary Turner // function.
191bf9a7730SZachary Turner //
192bf9a7730SZachary Turner // Note, for now we make the string pool a pointer to the pool, because
193bf9a7730SZachary Turner // we can't guarantee that some objects won't get destroyed after the
194bf9a7730SZachary Turner // global destructor chain is run, and trying to make sure no destructors
195bf9a7730SZachary Turner // touch ConstStrings is difficult.  So we leak the pool instead.
196bf9a7730SZachary Turner //----------------------------------------------------------------------
197bf9a7730SZachary Turner static Pool &StringPool() {
198c5f28e2aSKamil Rytarowski   static llvm::once_flag g_pool_initialization_flag;
199bf9a7730SZachary Turner   static Pool *g_string_pool = nullptr;
200bf9a7730SZachary Turner 
201c5f28e2aSKamil Rytarowski   llvm::call_once(g_pool_initialization_flag,
202bf9a7730SZachary Turner                  []() { g_string_pool = new Pool(); });
203bf9a7730SZachary Turner 
204bf9a7730SZachary Turner   return *g_string_pool;
205bf9a7730SZachary Turner }
206bf9a7730SZachary Turner 
207bf9a7730SZachary Turner ConstString::ConstString(const char *cstr)
208bf9a7730SZachary Turner     : m_string(StringPool().GetConstCString(cstr)) {}
209bf9a7730SZachary Turner 
210bf9a7730SZachary Turner ConstString::ConstString(const char *cstr, size_t cstr_len)
211bf9a7730SZachary Turner     : m_string(StringPool().GetConstCStringWithLength(cstr, cstr_len)) {}
212bf9a7730SZachary Turner 
213bf9a7730SZachary Turner ConstString::ConstString(const llvm::StringRef &s)
214bf9a7730SZachary Turner     : m_string(StringPool().GetConstCStringWithLength(s.data(), s.size())) {}
215bf9a7730SZachary Turner 
216bf9a7730SZachary Turner bool ConstString::operator<(const ConstString &rhs) const {
217bf9a7730SZachary Turner   if (m_string == rhs.m_string)
218bf9a7730SZachary Turner     return false;
219bf9a7730SZachary Turner 
2208070bf0aSPavel Labath   llvm::StringRef lhs_string_ref(GetStringRef());
2218070bf0aSPavel Labath   llvm::StringRef rhs_string_ref(rhs.GetStringRef());
222bf9a7730SZachary Turner 
223bf9a7730SZachary Turner   // If both have valid C strings, then return the comparison
224bf9a7730SZachary Turner   if (lhs_string_ref.data() && rhs_string_ref.data())
225bf9a7730SZachary Turner     return lhs_string_ref < rhs_string_ref;
226bf9a7730SZachary Turner 
227bf9a7730SZachary Turner   // Else one of them was nullptr, so if LHS is nullptr then it is less than
228bf9a7730SZachary Turner   return lhs_string_ref.data() == nullptr;
229bf9a7730SZachary Turner }
230bf9a7730SZachary Turner 
231bf9a7730SZachary Turner Stream &lldb_private::operator<<(Stream &s, const ConstString &str) {
232bf9a7730SZachary Turner   const char *cstr = str.GetCString();
233bf9a7730SZachary Turner   if (cstr != nullptr)
234bf9a7730SZachary Turner     s << cstr;
235bf9a7730SZachary Turner 
236bf9a7730SZachary Turner   return s;
237bf9a7730SZachary Turner }
238bf9a7730SZachary Turner 
239bf9a7730SZachary Turner size_t ConstString::GetLength() const {
2408070bf0aSPavel Labath   return Pool::GetConstCStringLength(m_string);
241bf9a7730SZachary Turner }
242bf9a7730SZachary Turner 
243bf9a7730SZachary Turner bool ConstString::Equals(const ConstString &lhs, const ConstString &rhs,
244bf9a7730SZachary Turner                          const bool case_sensitive) {
245bf9a7730SZachary Turner   if (lhs.m_string == rhs.m_string)
246bf9a7730SZachary Turner     return true;
247bf9a7730SZachary Turner 
248bf9a7730SZachary Turner   // Since the pointers weren't equal, and identical ConstStrings always have
249bf9a7730SZachary Turner   // identical pointers,
250bf9a7730SZachary Turner   // the result must be false for case sensitive equality test.
251bf9a7730SZachary Turner   if (case_sensitive)
252bf9a7730SZachary Turner     return false;
253bf9a7730SZachary Turner 
254bf9a7730SZachary Turner   // perform case insensitive equality test
2558070bf0aSPavel Labath   llvm::StringRef lhs_string_ref(lhs.GetStringRef());
2568070bf0aSPavel Labath   llvm::StringRef rhs_string_ref(rhs.GetStringRef());
257bf9a7730SZachary Turner   return lhs_string_ref.equals_lower(rhs_string_ref);
258bf9a7730SZachary Turner }
259bf9a7730SZachary Turner 
260bf9a7730SZachary Turner int ConstString::Compare(const ConstString &lhs, const ConstString &rhs,
261bf9a7730SZachary Turner                          const bool case_sensitive) {
262bf9a7730SZachary Turner   // If the iterators are the same, this is the same string
263bf9a7730SZachary Turner   const char *lhs_cstr = lhs.m_string;
264bf9a7730SZachary Turner   const char *rhs_cstr = rhs.m_string;
265bf9a7730SZachary Turner   if (lhs_cstr == rhs_cstr)
266bf9a7730SZachary Turner     return 0;
267bf9a7730SZachary Turner   if (lhs_cstr && rhs_cstr) {
2688070bf0aSPavel Labath     llvm::StringRef lhs_string_ref(lhs.GetStringRef());
2698070bf0aSPavel Labath     llvm::StringRef rhs_string_ref(rhs.GetStringRef());
270bf9a7730SZachary Turner 
271bf9a7730SZachary Turner     if (case_sensitive) {
272bf9a7730SZachary Turner       return lhs_string_ref.compare(rhs_string_ref);
273bf9a7730SZachary Turner     } else {
274bf9a7730SZachary Turner       return lhs_string_ref.compare_lower(rhs_string_ref);
275bf9a7730SZachary Turner     }
276bf9a7730SZachary Turner   }
277bf9a7730SZachary Turner 
278bf9a7730SZachary Turner   if (lhs_cstr)
279bf9a7730SZachary Turner     return +1; // LHS isn't nullptr but RHS is
280bf9a7730SZachary Turner   else
281bf9a7730SZachary Turner     return -1; // LHS is nullptr but RHS isn't
282bf9a7730SZachary Turner }
283bf9a7730SZachary Turner 
284bf9a7730SZachary Turner void ConstString::Dump(Stream *s, const char *fail_value) const {
285bf9a7730SZachary Turner   if (s != nullptr) {
286bf9a7730SZachary Turner     const char *cstr = AsCString(fail_value);
287bf9a7730SZachary Turner     if (cstr != nullptr)
288bf9a7730SZachary Turner       s->PutCString(cstr);
289bf9a7730SZachary Turner   }
290bf9a7730SZachary Turner }
291bf9a7730SZachary Turner 
292bf9a7730SZachary Turner void ConstString::DumpDebug(Stream *s) const {
293bf9a7730SZachary Turner   const char *cstr = GetCString();
294bf9a7730SZachary Turner   size_t cstr_len = GetLength();
295bf9a7730SZachary Turner   // Only print the parens if we have a non-nullptr string
296bf9a7730SZachary Turner   const char *parens = cstr ? "\"" : "";
297bf9a7730SZachary Turner   s->Printf("%*p: ConstString, string = %s%s%s, length = %" PRIu64,
298bf9a7730SZachary Turner             static_cast<int>(sizeof(void *) * 2),
299bf9a7730SZachary Turner             static_cast<const void *>(this), parens, cstr, parens,
300bf9a7730SZachary Turner             static_cast<uint64_t>(cstr_len));
301bf9a7730SZachary Turner }
302bf9a7730SZachary Turner 
303bf9a7730SZachary Turner void ConstString::SetCString(const char *cstr) {
304bf9a7730SZachary Turner   m_string = StringPool().GetConstCString(cstr);
305bf9a7730SZachary Turner }
306bf9a7730SZachary Turner 
307bf9a7730SZachary Turner void ConstString::SetString(const llvm::StringRef &s) {
308bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringWithLength(s.data(), s.size());
309bf9a7730SZachary Turner }
310bf9a7730SZachary Turner 
311bf9a7730SZachary Turner void ConstString::SetCStringWithMangledCounterpart(const char *demangled,
312bf9a7730SZachary Turner                                                    const ConstString &mangled) {
313bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringAndSetMangledCounterPart(
314bf9a7730SZachary Turner       demangled, mangled.m_string);
315bf9a7730SZachary Turner }
316bf9a7730SZachary Turner 
317bf9a7730SZachary Turner bool ConstString::GetMangledCounterpart(ConstString &counterpart) const {
318bf9a7730SZachary Turner   counterpart.m_string = StringPool().GetMangledCounterpart(m_string);
319bf9a7730SZachary Turner   return (bool)counterpart;
320bf9a7730SZachary Turner }
321bf9a7730SZachary Turner 
322bf9a7730SZachary Turner void ConstString::SetCStringWithLength(const char *cstr, size_t cstr_len) {
323bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringWithLength(cstr, cstr_len);
324bf9a7730SZachary Turner }
325bf9a7730SZachary Turner 
326bf9a7730SZachary Turner void ConstString::SetTrimmedCStringWithLength(const char *cstr,
327bf9a7730SZachary Turner                                               size_t cstr_len) {
328bf9a7730SZachary Turner   m_string = StringPool().GetConstTrimmedCStringWithLength(cstr, cstr_len);
329bf9a7730SZachary Turner }
330bf9a7730SZachary Turner 
331bf9a7730SZachary Turner size_t ConstString::StaticMemorySize() {
332bf9a7730SZachary Turner   // Get the size of the static string pool
333bf9a7730SZachary Turner   return StringPool().MemorySize();
334bf9a7730SZachary Turner }
3353b7e1981SPavel Labath 
3363b7e1981SPavel Labath void llvm::format_provider<ConstString>::format(const ConstString &CS,
3373b7e1981SPavel Labath                                                 llvm::raw_ostream &OS,
3383b7e1981SPavel Labath                                                 llvm::StringRef Options) {
3393b7e1981SPavel Labath   format_provider<StringRef>::format(CS.AsCString(), OS, Options);
3403b7e1981SPavel Labath }
341