1bf9a7730SZachary Turner //===-- ConstString.cpp -----------------------------------------*- C++ -*-===//
2bf9a7730SZachary Turner //
32946cd70SChandler Carruth // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
42946cd70SChandler Carruth // See https://llvm.org/LICENSE.txt for license information.
52946cd70SChandler Carruth // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6bf9a7730SZachary Turner //
7bf9a7730SZachary Turner //===----------------------------------------------------------------------===//
8bf9a7730SZachary Turner 
9bf9a7730SZachary Turner #include "lldb/Utility/ConstString.h"
10bf9a7730SZachary Turner 
114479ac15SZachary Turner #include "lldb/Utility/Stream.h"
12bf9a7730SZachary Turner 
13bf9a7730SZachary Turner #include "llvm/ADT/StringMap.h"
14672d2c12SJonas Devlieghere #include "llvm/ADT/iterator.h"
15672d2c12SJonas Devlieghere #include "llvm/Support/Allocator.h"
16672d2c12SJonas Devlieghere #include "llvm/Support/DJB.h"
17672d2c12SJonas Devlieghere #include "llvm/Support/FormatProviders.h"
18bf9a7730SZachary Turner #include "llvm/Support/RWMutex.h"
19c5f28e2aSKamil Rytarowski #include "llvm/Support/Threading.h"
204479ac15SZachary Turner 
21672d2c12SJonas Devlieghere #include <algorithm>
224479ac15SZachary Turner #include <array>
23672d2c12SJonas Devlieghere #include <utility>
244479ac15SZachary Turner 
25672d2c12SJonas Devlieghere #include <inttypes.h>
26672d2c12SJonas Devlieghere #include <stdint.h>
27672d2c12SJonas Devlieghere #include <string.h>
28bf9a7730SZachary Turner 
29bf9a7730SZachary Turner using namespace lldb_private;
30bf9a7730SZachary Turner 
31bf9a7730SZachary Turner class Pool {
32bf9a7730SZachary Turner public:
33bf9a7730SZachary Turner   typedef const char *StringPoolValueType;
34bf9a7730SZachary Turner   typedef llvm::StringMap<StringPoolValueType, llvm::BumpPtrAllocator>
35bf9a7730SZachary Turner       StringPool;
36bf9a7730SZachary Turner   typedef llvm::StringMapEntry<StringPoolValueType> StringPoolEntryType;
37bf9a7730SZachary Turner 
38bf9a7730SZachary Turner   static StringPoolEntryType &
39bf9a7730SZachary Turner   GetStringMapEntryFromKeyData(const char *keyData) {
408070bf0aSPavel Labath     return StringPoolEntryType::GetStringMapEntryFromKeyData(keyData);
41bf9a7730SZachary Turner   }
42bf9a7730SZachary Turner 
438070bf0aSPavel Labath   static size_t GetConstCStringLength(const char *ccstr) {
44bf9a7730SZachary Turner     if (ccstr != nullptr) {
4505097246SAdrian Prantl       // Since the entry is read only, and we derive the entry entirely from
4605097246SAdrian Prantl       // the pointer, we don't need the lock.
47bf9a7730SZachary Turner       const StringPoolEntryType &entry = GetStringMapEntryFromKeyData(ccstr);
48bf9a7730SZachary Turner       return entry.getKey().size();
49bf9a7730SZachary Turner     }
50bf9a7730SZachary Turner     return 0;
51bf9a7730SZachary Turner   }
52bf9a7730SZachary Turner 
53bf9a7730SZachary Turner   StringPoolValueType GetMangledCounterpart(const char *ccstr) const {
54bf9a7730SZachary Turner     if (ccstr != nullptr) {
55bf9a7730SZachary Turner       const uint8_t h = hash(llvm::StringRef(ccstr));
56bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
57bf9a7730SZachary Turner       return GetStringMapEntryFromKeyData(ccstr).getValue();
58bf9a7730SZachary Turner     }
59bf9a7730SZachary Turner     return nullptr;
60bf9a7730SZachary Turner   }
61bf9a7730SZachary Turner 
62bf9a7730SZachary Turner   bool SetMangledCounterparts(const char *key_ccstr, const char *value_ccstr) {
63bf9a7730SZachary Turner     if (key_ccstr != nullptr && value_ccstr != nullptr) {
64bf9a7730SZachary Turner       {
65bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(key_ccstr));
66bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
67bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(key_ccstr).setValue(value_ccstr);
68bf9a7730SZachary Turner       }
69bf9a7730SZachary Turner       {
70bf9a7730SZachary Turner         const uint8_t h = hash(llvm::StringRef(value_ccstr));
71bf9a7730SZachary Turner         llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
72bf9a7730SZachary Turner         GetStringMapEntryFromKeyData(value_ccstr).setValue(key_ccstr);
73bf9a7730SZachary Turner       }
74bf9a7730SZachary Turner       return true;
75bf9a7730SZachary Turner     }
76bf9a7730SZachary Turner     return false;
77bf9a7730SZachary Turner   }
78bf9a7730SZachary Turner 
79bf9a7730SZachary Turner   const char *GetConstCString(const char *cstr) {
80bf9a7730SZachary Turner     if (cstr != nullptr)
81bf9a7730SZachary Turner       return GetConstCStringWithLength(cstr, strlen(cstr));
82bf9a7730SZachary Turner     return nullptr;
83bf9a7730SZachary Turner   }
84bf9a7730SZachary Turner 
85bf9a7730SZachary Turner   const char *GetConstCStringWithLength(const char *cstr, size_t cstr_len) {
86bf9a7730SZachary Turner     if (cstr != nullptr)
87bf9a7730SZachary Turner       return GetConstCStringWithStringRef(llvm::StringRef(cstr, cstr_len));
88bf9a7730SZachary Turner     return nullptr;
89bf9a7730SZachary Turner   }
90bf9a7730SZachary Turner 
91bf9a7730SZachary Turner   const char *GetConstCStringWithStringRef(const llvm::StringRef &string_ref) {
92bf9a7730SZachary Turner     if (string_ref.data()) {
93bf9a7730SZachary Turner       const uint8_t h = hash(string_ref);
94bf9a7730SZachary Turner 
95bf9a7730SZachary Turner       {
96bf9a7730SZachary Turner         llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex);
97bf9a7730SZachary Turner         auto it = m_string_pools[h].m_string_map.find(string_ref);
98bf9a7730SZachary Turner         if (it != m_string_pools[h].m_string_map.end())
99bf9a7730SZachary Turner           return it->getKeyData();
100bf9a7730SZachary Turner       }
101bf9a7730SZachary Turner 
102bf9a7730SZachary Turner       llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
103bf9a7730SZachary Turner       StringPoolEntryType &entry =
104bf9a7730SZachary Turner           *m_string_pools[h]
105bf9a7730SZachary Turner                .m_string_map.insert(std::make_pair(string_ref, nullptr))
106bf9a7730SZachary Turner                .first;
107bf9a7730SZachary Turner       return entry.getKeyData();
108bf9a7730SZachary Turner     }
109bf9a7730SZachary Turner     return nullptr;
110bf9a7730SZachary Turner   }
111bf9a7730SZachary Turner 
112bf9a7730SZachary Turner   const char *
11319a357adSPavel Labath   GetConstCStringAndSetMangledCounterPart(llvm::StringRef demangled,
114bf9a7730SZachary Turner                                           const char *mangled_ccstr) {
115bf9a7730SZachary Turner     const char *demangled_ccstr = nullptr;
116bf9a7730SZachary Turner 
117bf9a7730SZachary Turner     {
11819a357adSPavel Labath       const uint8_t h = hash(demangled);
119bf9a7730SZachary Turner       llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
120bf9a7730SZachary Turner 
1212397a2b6SStefan Granitz       // Make or update string pool entry with the mangled counterpart
1222397a2b6SStefan Granitz       StringPool &map = m_string_pools[h].m_string_map;
1232397a2b6SStefan Granitz       StringPoolEntryType &entry = *map.try_emplace(demangled).first;
1242397a2b6SStefan Granitz 
1252397a2b6SStefan Granitz       entry.second = mangled_ccstr;
126bf9a7730SZachary Turner 
127bf9a7730SZachary Turner       // Extract the const version of the demangled_cstr
128bf9a7730SZachary Turner       demangled_ccstr = entry.getKeyData();
129bf9a7730SZachary Turner     }
130bf9a7730SZachary Turner 
131bf9a7730SZachary Turner     {
132bf9a7730SZachary Turner       // Now assign the demangled const string as the counterpart of the
133bf9a7730SZachary Turner       // mangled const string...
134bf9a7730SZachary Turner       const uint8_t h = hash(llvm::StringRef(mangled_ccstr));
135bf9a7730SZachary Turner       llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex);
136bf9a7730SZachary Turner       GetStringMapEntryFromKeyData(mangled_ccstr).setValue(demangled_ccstr);
137bf9a7730SZachary Turner     }
138bf9a7730SZachary Turner 
139bf9a7730SZachary Turner     // Return the constant demangled C string
140bf9a7730SZachary Turner     return demangled_ccstr;
141bf9a7730SZachary Turner   }
142bf9a7730SZachary Turner 
143bf9a7730SZachary Turner   const char *GetConstTrimmedCStringWithLength(const char *cstr,
144bf9a7730SZachary Turner                                                size_t cstr_len) {
145bf9a7730SZachary Turner     if (cstr != nullptr) {
146*bb3609e4SJan Kratochvil       const size_t trimmed_len = strnlen(cstr, cstr_len);
147bf9a7730SZachary Turner       return GetConstCStringWithLength(cstr, trimmed_len);
148bf9a7730SZachary Turner     }
149bf9a7730SZachary Turner     return nullptr;
150bf9a7730SZachary Turner   }
151bf9a7730SZachary Turner 
152bf9a7730SZachary Turner   //------------------------------------------------------------------
15305097246SAdrian Prantl   // Return the size in bytes that this object and any items in its collection
15405097246SAdrian Prantl   // of uniqued strings + data count values takes in memory.
155bf9a7730SZachary Turner   //------------------------------------------------------------------
156bf9a7730SZachary Turner   size_t MemorySize() const {
157bf9a7730SZachary Turner     size_t mem_size = sizeof(Pool);
158bf9a7730SZachary Turner     for (const auto &pool : m_string_pools) {
159bf9a7730SZachary Turner       llvm::sys::SmartScopedReader<false> rlock(pool.m_mutex);
160bf9a7730SZachary Turner       for (const auto &entry : pool.m_string_map)
161bf9a7730SZachary Turner         mem_size += sizeof(StringPoolEntryType) + entry.getKey().size();
162bf9a7730SZachary Turner     }
163bf9a7730SZachary Turner     return mem_size;
164bf9a7730SZachary Turner   }
165bf9a7730SZachary Turner 
166bf9a7730SZachary Turner protected:
167bf9a7730SZachary Turner   uint8_t hash(const llvm::StringRef &s) const {
168560ce2c7SJonas Devlieghere     uint32_t h = llvm::djbHash(s);
169bf9a7730SZachary Turner     return ((h >> 24) ^ (h >> 16) ^ (h >> 8) ^ h) & 0xff;
170bf9a7730SZachary Turner   }
171bf9a7730SZachary Turner 
172bf9a7730SZachary Turner   struct PoolEntry {
173bf9a7730SZachary Turner     mutable llvm::sys::SmartRWMutex<false> m_mutex;
174bf9a7730SZachary Turner     StringPool m_string_map;
175bf9a7730SZachary Turner   };
176bf9a7730SZachary Turner 
177bf9a7730SZachary Turner   std::array<PoolEntry, 256> m_string_pools;
178bf9a7730SZachary Turner };
179bf9a7730SZachary Turner 
180bf9a7730SZachary Turner //----------------------------------------------------------------------
18105097246SAdrian Prantl // Frameworks and dylibs aren't supposed to have global C++ initializers so we
18205097246SAdrian Prantl // hide the string pool in a static function so that it will get initialized on
18305097246SAdrian Prantl // the first call to this static function.
184bf9a7730SZachary Turner //
18505097246SAdrian Prantl // Note, for now we make the string pool a pointer to the pool, because we
18605097246SAdrian Prantl // can't guarantee that some objects won't get destroyed after the global
18705097246SAdrian Prantl // destructor chain is run, and trying to make sure no destructors touch
18805097246SAdrian Prantl // ConstStrings is difficult.  So we leak the pool instead.
189bf9a7730SZachary Turner //----------------------------------------------------------------------
190bf9a7730SZachary Turner static Pool &StringPool() {
191c5f28e2aSKamil Rytarowski   static llvm::once_flag g_pool_initialization_flag;
192bf9a7730SZachary Turner   static Pool *g_string_pool = nullptr;
193bf9a7730SZachary Turner 
194c5f28e2aSKamil Rytarowski   llvm::call_once(g_pool_initialization_flag,
195bf9a7730SZachary Turner                  []() { g_string_pool = new Pool(); });
196bf9a7730SZachary Turner 
197bf9a7730SZachary Turner   return *g_string_pool;
198bf9a7730SZachary Turner }
199bf9a7730SZachary Turner 
200bf9a7730SZachary Turner ConstString::ConstString(const char *cstr)
201bf9a7730SZachary Turner     : m_string(StringPool().GetConstCString(cstr)) {}
202bf9a7730SZachary Turner 
203bf9a7730SZachary Turner ConstString::ConstString(const char *cstr, size_t cstr_len)
204bf9a7730SZachary Turner     : m_string(StringPool().GetConstCStringWithLength(cstr, cstr_len)) {}
205bf9a7730SZachary Turner 
206bf9a7730SZachary Turner ConstString::ConstString(const llvm::StringRef &s)
207bf9a7730SZachary Turner     : m_string(StringPool().GetConstCStringWithLength(s.data(), s.size())) {}
208bf9a7730SZachary Turner 
209bf9a7730SZachary Turner bool ConstString::operator<(const ConstString &rhs) const {
210bf9a7730SZachary Turner   if (m_string == rhs.m_string)
211bf9a7730SZachary Turner     return false;
212bf9a7730SZachary Turner 
2138070bf0aSPavel Labath   llvm::StringRef lhs_string_ref(GetStringRef());
2148070bf0aSPavel Labath   llvm::StringRef rhs_string_ref(rhs.GetStringRef());
215bf9a7730SZachary Turner 
216bf9a7730SZachary Turner   // If both have valid C strings, then return the comparison
217bf9a7730SZachary Turner   if (lhs_string_ref.data() && rhs_string_ref.data())
218bf9a7730SZachary Turner     return lhs_string_ref < rhs_string_ref;
219bf9a7730SZachary Turner 
220bf9a7730SZachary Turner   // Else one of them was nullptr, so if LHS is nullptr then it is less than
221bf9a7730SZachary Turner   return lhs_string_ref.data() == nullptr;
222bf9a7730SZachary Turner }
223bf9a7730SZachary Turner 
224bf9a7730SZachary Turner Stream &lldb_private::operator<<(Stream &s, const ConstString &str) {
225bf9a7730SZachary Turner   const char *cstr = str.GetCString();
226bf9a7730SZachary Turner   if (cstr != nullptr)
227bf9a7730SZachary Turner     s << cstr;
228bf9a7730SZachary Turner 
229bf9a7730SZachary Turner   return s;
230bf9a7730SZachary Turner }
231bf9a7730SZachary Turner 
232bf9a7730SZachary Turner size_t ConstString::GetLength() const {
2338070bf0aSPavel Labath   return Pool::GetConstCStringLength(m_string);
234bf9a7730SZachary Turner }
235bf9a7730SZachary Turner 
236bf9a7730SZachary Turner bool ConstString::Equals(const ConstString &lhs, const ConstString &rhs,
237bf9a7730SZachary Turner                          const bool case_sensitive) {
238bf9a7730SZachary Turner   if (lhs.m_string == rhs.m_string)
239bf9a7730SZachary Turner     return true;
240bf9a7730SZachary Turner 
241bf9a7730SZachary Turner   // Since the pointers weren't equal, and identical ConstStrings always have
24205097246SAdrian Prantl   // identical pointers, the result must be false for case sensitive equality
24305097246SAdrian Prantl   // test.
244bf9a7730SZachary Turner   if (case_sensitive)
245bf9a7730SZachary Turner     return false;
246bf9a7730SZachary Turner 
247bf9a7730SZachary Turner   // perform case insensitive equality test
2488070bf0aSPavel Labath   llvm::StringRef lhs_string_ref(lhs.GetStringRef());
2498070bf0aSPavel Labath   llvm::StringRef rhs_string_ref(rhs.GetStringRef());
250bf9a7730SZachary Turner   return lhs_string_ref.equals_lower(rhs_string_ref);
251bf9a7730SZachary Turner }
252bf9a7730SZachary Turner 
253bf9a7730SZachary Turner int ConstString::Compare(const ConstString &lhs, const ConstString &rhs,
254bf9a7730SZachary Turner                          const bool case_sensitive) {
255bf9a7730SZachary Turner   // If the iterators are the same, this is the same string
256bf9a7730SZachary Turner   const char *lhs_cstr = lhs.m_string;
257bf9a7730SZachary Turner   const char *rhs_cstr = rhs.m_string;
258bf9a7730SZachary Turner   if (lhs_cstr == rhs_cstr)
259bf9a7730SZachary Turner     return 0;
260bf9a7730SZachary Turner   if (lhs_cstr && rhs_cstr) {
2618070bf0aSPavel Labath     llvm::StringRef lhs_string_ref(lhs.GetStringRef());
2628070bf0aSPavel Labath     llvm::StringRef rhs_string_ref(rhs.GetStringRef());
263bf9a7730SZachary Turner 
264bf9a7730SZachary Turner     if (case_sensitive) {
265bf9a7730SZachary Turner       return lhs_string_ref.compare(rhs_string_ref);
266bf9a7730SZachary Turner     } else {
267bf9a7730SZachary Turner       return lhs_string_ref.compare_lower(rhs_string_ref);
268bf9a7730SZachary Turner     }
269bf9a7730SZachary Turner   }
270bf9a7730SZachary Turner 
271bf9a7730SZachary Turner   if (lhs_cstr)
272bf9a7730SZachary Turner     return +1; // LHS isn't nullptr but RHS is
273bf9a7730SZachary Turner   else
274bf9a7730SZachary Turner     return -1; // LHS is nullptr but RHS isn't
275bf9a7730SZachary Turner }
276bf9a7730SZachary Turner 
277bf9a7730SZachary Turner void ConstString::Dump(Stream *s, const char *fail_value) const {
278bf9a7730SZachary Turner   if (s != nullptr) {
279bf9a7730SZachary Turner     const char *cstr = AsCString(fail_value);
280bf9a7730SZachary Turner     if (cstr != nullptr)
281bf9a7730SZachary Turner       s->PutCString(cstr);
282bf9a7730SZachary Turner   }
283bf9a7730SZachary Turner }
284bf9a7730SZachary Turner 
285bf9a7730SZachary Turner void ConstString::DumpDebug(Stream *s) const {
286bf9a7730SZachary Turner   const char *cstr = GetCString();
287bf9a7730SZachary Turner   size_t cstr_len = GetLength();
288bf9a7730SZachary Turner   // Only print the parens if we have a non-nullptr string
289bf9a7730SZachary Turner   const char *parens = cstr ? "\"" : "";
290bf9a7730SZachary Turner   s->Printf("%*p: ConstString, string = %s%s%s, length = %" PRIu64,
291bf9a7730SZachary Turner             static_cast<int>(sizeof(void *) * 2),
292bf9a7730SZachary Turner             static_cast<const void *>(this), parens, cstr, parens,
293bf9a7730SZachary Turner             static_cast<uint64_t>(cstr_len));
294bf9a7730SZachary Turner }
295bf9a7730SZachary Turner 
296bf9a7730SZachary Turner void ConstString::SetCString(const char *cstr) {
297bf9a7730SZachary Turner   m_string = StringPool().GetConstCString(cstr);
298bf9a7730SZachary Turner }
299bf9a7730SZachary Turner 
300bf9a7730SZachary Turner void ConstString::SetString(const llvm::StringRef &s) {
301bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringWithLength(s.data(), s.size());
302bf9a7730SZachary Turner }
303bf9a7730SZachary Turner 
30419a357adSPavel Labath void ConstString::SetStringWithMangledCounterpart(llvm::StringRef demangled,
305bf9a7730SZachary Turner                                                    const ConstString &mangled) {
306bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringAndSetMangledCounterPart(
307bf9a7730SZachary Turner       demangled, mangled.m_string);
308bf9a7730SZachary Turner }
309bf9a7730SZachary Turner 
310bf9a7730SZachary Turner bool ConstString::GetMangledCounterpart(ConstString &counterpart) const {
311bf9a7730SZachary Turner   counterpart.m_string = StringPool().GetMangledCounterpart(m_string);
312bf9a7730SZachary Turner   return (bool)counterpart;
313bf9a7730SZachary Turner }
314bf9a7730SZachary Turner 
315bf9a7730SZachary Turner void ConstString::SetCStringWithLength(const char *cstr, size_t cstr_len) {
316bf9a7730SZachary Turner   m_string = StringPool().GetConstCStringWithLength(cstr, cstr_len);
317bf9a7730SZachary Turner }
318bf9a7730SZachary Turner 
319bf9a7730SZachary Turner void ConstString::SetTrimmedCStringWithLength(const char *cstr,
320bf9a7730SZachary Turner                                               size_t cstr_len) {
321bf9a7730SZachary Turner   m_string = StringPool().GetConstTrimmedCStringWithLength(cstr, cstr_len);
322bf9a7730SZachary Turner }
323bf9a7730SZachary Turner 
324bf9a7730SZachary Turner size_t ConstString::StaticMemorySize() {
325bf9a7730SZachary Turner   // Get the size of the static string pool
326bf9a7730SZachary Turner   return StringPool().MemorySize();
327bf9a7730SZachary Turner }
3283b7e1981SPavel Labath 
3293b7e1981SPavel Labath void llvm::format_provider<ConstString>::format(const ConstString &CS,
3303b7e1981SPavel Labath                                                 llvm::raw_ostream &OS,
3313b7e1981SPavel Labath                                                 llvm::StringRef Options) {
3323b7e1981SPavel Labath   format_provider<StringRef>::format(CS.AsCString(), OS, Options);
3333b7e1981SPavel Labath }
334