1 //===-- NSString.cpp ------------------------------------------------------===//
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 #include "NSString.h"
10 
11 #include "Plugins/TypeSystem/Clang/TypeSystemClang.h"
12 #include "lldb/Core/ValueObject.h"
13 #include "lldb/Core/ValueObjectConstResult.h"
14 #include "lldb/DataFormatters/FormattersHelpers.h"
15 #include "lldb/DataFormatters/StringPrinter.h"
16 #include "lldb/Target/Language.h"
17 #include "lldb/Target/ProcessStructReader.h"
18 #include "lldb/Target/Target.h"
19 #include "lldb/Utility/DataBufferHeap.h"
20 #include "lldb/Utility/Endian.h"
21 #include "lldb/Utility/Status.h"
22 #include "lldb/Utility/Stream.h"
23 
24 using namespace lldb;
25 using namespace lldb_private;
26 using namespace lldb_private::formatters;
27 
28 std::map<ConstString, CXXFunctionSummaryFormat::Callback> &
29 NSString_Additionals::GetAdditionalSummaries() {
30   static std::map<ConstString, CXXFunctionSummaryFormat::Callback> g_map;
31   return g_map;
32 }
33 
34 static CompilerType GetNSPathStore2Type(Target &target) {
35   static ConstString g_type_name("__lldb_autogen_nspathstore2");
36 
37   TypeSystemClang *ast_ctx = TypeSystemClang::GetScratch(target);
38 
39   if (!ast_ctx)
40     return CompilerType();
41 
42   CompilerType voidstar =
43       ast_ctx->GetBasicType(lldb::eBasicTypeVoid).GetPointerType();
44   CompilerType uint32 =
45       ast_ctx->GetBuiltinTypeForEncodingAndBitSize(eEncodingUint, 32);
46 
47   return ast_ctx->GetOrCreateStructForIdentifier(
48       g_type_name,
49       {{"isa", voidstar}, {"lengthAndRef", uint32}, {"buffer", voidstar}});
50 }
51 
52 bool lldb_private::formatters::NSStringSummaryProvider(
53     ValueObject &valobj, Stream &stream,
54     const TypeSummaryOptions &summary_options) {
55   static ConstString g_TypeHint("NSString");
56 
57   ProcessSP process_sp = valobj.GetProcessSP();
58   if (!process_sp)
59     return false;
60 
61   ObjCLanguageRuntime *runtime = ObjCLanguageRuntime::Get(*process_sp);
62 
63   if (!runtime)
64     return false;
65 
66   ObjCLanguageRuntime::ClassDescriptorSP descriptor(
67       runtime->GetClassDescriptor(valobj));
68 
69   if (!descriptor.get() || !descriptor->IsValid())
70     return false;
71 
72   uint32_t ptr_size = process_sp->GetAddressByteSize();
73 
74   lldb::addr_t valobj_addr = valobj.GetValueAsUnsigned(0);
75 
76   if (!valobj_addr)
77     return false;
78 
79   ConstString class_name_cs = descriptor->GetClassName();
80   llvm::StringRef class_name = class_name_cs.GetStringRef();
81 
82   if (class_name.empty())
83     return false;
84 
85   bool is_tagged_ptr = class_name == "NSTaggedPointerString" &&
86                        descriptor->GetTaggedPointerInfo();
87   // for a tagged pointer, the descriptor has everything we need
88   if (is_tagged_ptr)
89     return NSTaggedString_SummaryProvider(valobj, descriptor, stream,
90                                           summary_options);
91 
92   auto &additionals_map(NSString_Additionals::GetAdditionalSummaries());
93   auto iter = additionals_map.find(class_name_cs), end = additionals_map.end();
94   if (iter != end)
95     return iter->second(valobj, stream, summary_options);
96 
97   // if not a tagged pointer that we know about, try the normal route
98   uint64_t info_bits_location = valobj_addr + ptr_size;
99   if (process_sp->GetByteOrder() != lldb::eByteOrderLittle)
100     info_bits_location += 3;
101 
102   Status error;
103 
104   uint8_t info_bits = process_sp->ReadUnsignedIntegerFromMemory(
105       info_bits_location, 1, 0, error);
106   if (error.Fail())
107     return false;
108 
109   bool is_mutable = (info_bits & 1) == 1;
110   bool is_inline = (info_bits & 0x60) == 0;
111   bool has_explicit_length = (info_bits & (1 | 4)) != 4;
112   bool is_unicode = (info_bits & 0x10) == 0x10;
113   bool is_path_store = class_name == "NSPathStore2";
114   bool has_null = (info_bits & 8) == 8;
115 
116   size_t explicit_length = 0;
117   if (!has_null && has_explicit_length && !is_path_store) {
118     lldb::addr_t explicit_length_offset = 2 * ptr_size;
119     if (is_mutable && !is_inline)
120       explicit_length_offset =
121           explicit_length_offset + ptr_size; //  notInlineMutable.length;
122     else if (is_inline)
123       explicit_length = explicit_length + 0; // inline1.length;
124     else if (!is_inline && !is_mutable)
125       explicit_length_offset =
126           explicit_length_offset + ptr_size; // notInlineImmutable1.length;
127     else
128       explicit_length_offset = 0;
129 
130     if (explicit_length_offset) {
131       explicit_length_offset = valobj_addr + explicit_length_offset;
132       explicit_length = process_sp->ReadUnsignedIntegerFromMemory(
133           explicit_length_offset, 4, 0, error);
134     }
135   }
136 
137   const llvm::StringSet<> supported_string_classes = {
138       "NSString",     "CFMutableStringRef",
139       "CFStringRef",  "__NSCFConstantString",
140       "__NSCFString", "NSCFConstantString",
141       "NSCFString",   "NSPathStore2"};
142   if (supported_string_classes.count(class_name) == 0) {
143     // not one of us - but tell me class name
144     stream.Printf("class name = %s", class_name_cs.GetCString());
145     return true;
146   }
147 
148   std::string prefix, suffix;
149   if (Language *language =
150           Language::FindPlugin(summary_options.GetLanguage())) {
151     if (!language->GetFormatterPrefixSuffix(valobj, g_TypeHint, prefix,
152                                             suffix)) {
153       prefix.clear();
154       suffix.clear();
155     }
156   }
157 
158   StringPrinter::ReadStringAndDumpToStreamOptions options(valobj);
159   options.SetPrefixToken(prefix);
160   options.SetSuffixToken(suffix);
161 
162   if (is_mutable) {
163     uint64_t location = 2 * ptr_size + valobj_addr;
164     location = process_sp->ReadPointerFromMemory(location, error);
165     if (error.Fail())
166       return false;
167     if (has_explicit_length && is_unicode) {
168       options.SetLocation(location);
169       options.SetProcessSP(process_sp);
170       options.SetStream(&stream);
171       options.SetQuote('"');
172       options.SetSourceSize(explicit_length);
173       options.SetHasSourceSize(has_explicit_length);
174       options.SetNeedsZeroTermination(false);
175       options.SetIgnoreMaxLength(summary_options.GetCapping() ==
176                                  TypeSummaryCapping::eTypeSummaryUncapped);
177       options.SetBinaryZeroIsTerminator(false);
178       return StringPrinter::ReadStringAndDumpToStream<
179           StringPrinter::StringElementType::UTF16>(options);
180     } else {
181       options.SetLocation(location + 1);
182       options.SetProcessSP(process_sp);
183       options.SetStream(&stream);
184       options.SetSourceSize(explicit_length);
185       options.SetHasSourceSize(has_explicit_length);
186       options.SetNeedsZeroTermination(false);
187       options.SetIgnoreMaxLength(summary_options.GetCapping() ==
188                                  TypeSummaryCapping::eTypeSummaryUncapped);
189       options.SetBinaryZeroIsTerminator(false);
190       return StringPrinter::ReadStringAndDumpToStream<
191           StringPrinter::StringElementType::ASCII>(options);
192     }
193   } else if (is_inline && has_explicit_length && !is_unicode &&
194              !is_path_store && !is_mutable) {
195     uint64_t location = 3 * ptr_size + valobj_addr;
196 
197     options.SetLocation(location);
198     options.SetProcessSP(process_sp);
199     options.SetStream(&stream);
200     options.SetQuote('"');
201     options.SetSourceSize(explicit_length);
202     options.SetHasSourceSize(has_explicit_length);
203     options.SetIgnoreMaxLength(summary_options.GetCapping() ==
204                                TypeSummaryCapping::eTypeSummaryUncapped);
205     return StringPrinter::ReadStringAndDumpToStream<
206         StringPrinter::StringElementType::ASCII>(options);
207   } else if (is_unicode) {
208     uint64_t location = valobj_addr + 2 * ptr_size;
209     if (is_inline) {
210       if (!has_explicit_length) {
211         return false;
212       } else
213         location += ptr_size;
214     } else {
215       location = process_sp->ReadPointerFromMemory(location, error);
216       if (error.Fail())
217         return false;
218     }
219     options.SetLocation(location);
220     options.SetProcessSP(process_sp);
221     options.SetStream(&stream);
222     options.SetQuote('"');
223     options.SetSourceSize(explicit_length);
224     options.SetHasSourceSize(has_explicit_length);
225     options.SetNeedsZeroTermination(!has_explicit_length);
226     options.SetIgnoreMaxLength(summary_options.GetCapping() ==
227                                TypeSummaryCapping::eTypeSummaryUncapped);
228     options.SetBinaryZeroIsTerminator(!has_explicit_length);
229     return StringPrinter::ReadStringAndDumpToStream<
230         StringPrinter::StringElementType::UTF16>(options);
231   } else if (is_path_store) {
232     ProcessStructReader reader(valobj.GetProcessSP().get(),
233                                valobj.GetValueAsUnsigned(0),
234                                GetNSPathStore2Type(*valobj.GetTargetSP()));
235     explicit_length =
236         reader.GetField<uint32_t>(ConstString("lengthAndRef")) >> 20;
237     lldb::addr_t location = valobj.GetValueAsUnsigned(0) + ptr_size + 4;
238 
239     options.SetLocation(location);
240     options.SetProcessSP(process_sp);
241     options.SetStream(&stream);
242     options.SetQuote('"');
243     options.SetSourceSize(explicit_length);
244     options.SetHasSourceSize(has_explicit_length);
245     options.SetNeedsZeroTermination(!has_explicit_length);
246     options.SetIgnoreMaxLength(summary_options.GetCapping() ==
247                                TypeSummaryCapping::eTypeSummaryUncapped);
248     options.SetBinaryZeroIsTerminator(!has_explicit_length);
249     return StringPrinter::ReadStringAndDumpToStream<
250         StringPrinter::StringElementType::UTF16>(options);
251   } else if (is_inline) {
252     uint64_t location = valobj_addr + 2 * ptr_size;
253     if (!has_explicit_length) {
254       // in this kind of string, the byte before the string content is a length
255       // byte so let's try and use it to handle the embedded NUL case
256       Status error;
257       explicit_length =
258           process_sp->ReadUnsignedIntegerFromMemory(location, 1, 0, error);
259       has_explicit_length = !(error.Fail() || explicit_length == 0);
260       location++;
261     }
262     options.SetLocation(location);
263     options.SetProcessSP(process_sp);
264     options.SetStream(&stream);
265     options.SetSourceSize(explicit_length);
266     options.SetHasSourceSize(has_explicit_length);
267     options.SetNeedsZeroTermination(!has_explicit_length);
268     options.SetIgnoreMaxLength(summary_options.GetCapping() ==
269                                TypeSummaryCapping::eTypeSummaryUncapped);
270     options.SetBinaryZeroIsTerminator(!has_explicit_length);
271     if (has_explicit_length)
272       return StringPrinter::ReadStringAndDumpToStream<
273           StringPrinter::StringElementType::UTF8>(options);
274     else
275       return StringPrinter::ReadStringAndDumpToStream<
276           StringPrinter::StringElementType::ASCII>(options);
277   } else {
278     uint64_t location = valobj_addr + 2 * ptr_size;
279     location = process_sp->ReadPointerFromMemory(location, error);
280     if (error.Fail())
281       return false;
282     if (has_explicit_length && !has_null)
283       explicit_length++; // account for the fact that there is no NULL and we
284                          // need to have one added
285     options.SetLocation(location);
286     options.SetProcessSP(process_sp);
287     options.SetStream(&stream);
288     options.SetSourceSize(explicit_length);
289     options.SetHasSourceSize(has_explicit_length);
290     options.SetIgnoreMaxLength(summary_options.GetCapping() ==
291                                TypeSummaryCapping::eTypeSummaryUncapped);
292     return StringPrinter::ReadStringAndDumpToStream<
293         StringPrinter::StringElementType::ASCII>(options);
294   }
295 }
296 
297 bool lldb_private::formatters::NSAttributedStringSummaryProvider(
298     ValueObject &valobj, Stream &stream, const TypeSummaryOptions &options) {
299   TargetSP target_sp(valobj.GetTargetSP());
300   if (!target_sp)
301     return false;
302   uint32_t addr_size = target_sp->GetArchitecture().GetAddressByteSize();
303   uint64_t pointer_value = valobj.GetValueAsUnsigned(0);
304   if (!pointer_value)
305     return false;
306   pointer_value += addr_size;
307   CompilerType type(valobj.GetCompilerType());
308   ExecutionContext exe_ctx(target_sp, false);
309   ValueObjectSP child_ptr_sp(valobj.CreateValueObjectFromAddress(
310       "string_ptr", pointer_value, exe_ctx, type));
311   if (!child_ptr_sp)
312     return false;
313   DataExtractor data;
314   Status error;
315   child_ptr_sp->GetData(data, error);
316   if (error.Fail())
317     return false;
318   ValueObjectSP child_sp(child_ptr_sp->CreateValueObjectFromData(
319       "string_data", data, exe_ctx, type));
320   child_sp->GetValueAsUnsigned(0);
321   if (child_sp)
322     return NSStringSummaryProvider(*child_sp, stream, options);
323   return false;
324 }
325 
326 bool lldb_private::formatters::NSMutableAttributedStringSummaryProvider(
327     ValueObject &valobj, Stream &stream, const TypeSummaryOptions &options) {
328   return NSAttributedStringSummaryProvider(valobj, stream, options);
329 }
330 
331 bool lldb_private::formatters::NSTaggedString_SummaryProvider(
332     ValueObject &valobj, ObjCLanguageRuntime::ClassDescriptorSP descriptor,
333     Stream &stream, const TypeSummaryOptions &summary_options) {
334   static ConstString g_TypeHint("NSString");
335 
336   if (!descriptor)
337     return false;
338   uint64_t len_bits = 0, data_bits = 0;
339   if (!descriptor->GetTaggedPointerInfo(&len_bits, &data_bits, nullptr))
340     return false;
341 
342   static const int g_MaxNonBitmaskedLen = 7; // TAGGED_STRING_UNPACKED_MAXLEN
343   static const int g_SixbitMaxLen = 9;
344   static const int g_fiveBitMaxLen = 11;
345 
346   static const char *sixBitToCharLookup = "eilotrm.apdnsIc ufkMShjTRxgC4013"
347                                           "bDNvwyUL2O856P-B79AFKEWV_zGJ/HYX";
348 
349   if (len_bits > g_fiveBitMaxLen)
350     return false;
351 
352   std::string prefix, suffix;
353   if (Language *language =
354           Language::FindPlugin(summary_options.GetLanguage())) {
355     if (!language->GetFormatterPrefixSuffix(valobj, g_TypeHint, prefix,
356                                             suffix)) {
357       prefix.clear();
358       suffix.clear();
359     }
360   }
361 
362   // this is a fairly ugly trick - pretend that the numeric value is actually a
363   // char* this works under a few assumptions: little endian architecture
364   // sizeof(uint64_t) > g_MaxNonBitmaskedLen
365   if (len_bits <= g_MaxNonBitmaskedLen) {
366     stream.Printf("%s", prefix.c_str());
367     stream.Printf("\"%s\"", (const char *)&data_bits);
368     stream.Printf("%s", suffix.c_str());
369     return true;
370   }
371 
372   // if the data is bitmasked, we need to actually process the bytes
373   uint8_t bitmask = 0;
374   uint8_t shift_offset = 0;
375 
376   if (len_bits <= g_SixbitMaxLen) {
377     bitmask = 0x03f;
378     shift_offset = 6;
379   } else {
380     bitmask = 0x01f;
381     shift_offset = 5;
382   }
383 
384   std::vector<uint8_t> bytes;
385   bytes.resize(len_bits);
386   for (; len_bits > 0; data_bits >>= shift_offset, --len_bits) {
387     uint8_t packed = data_bits & bitmask;
388     bytes.insert(bytes.begin(), sixBitToCharLookup[packed]);
389   }
390 
391   stream.Printf("%s", prefix.c_str());
392   stream.Printf("\"%s\"", &bytes[0]);
393   stream.Printf("%s", suffix.c_str());
394   return true;
395 }
396