xref: /llvm-project/clang-tools-extra/modularize/PreprocessorTracker.cpp (revision 546943f9f1ab8d7885b54fc44f95620458967d7e)
1 //===--- PreprocessorTracker.cpp - Preprocessor tracking -*- C++ -*------===//
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 // The Basic Idea (Macro and Conditional Checking)
11 //
12 // Basically we install a PPCallbacks-derived object to track preprocessor
13 // activity, namely when a header file is entered/exited, when a macro
14 // is expanded, when "defined" is used, and when #if, #elif, #ifdef,
15 // and #ifndef are used.  We save the state of macro and "defined"
16 // expressions in a map, keyed on a name/file/line/column quadruple.
17 // The map entries store the different states (values) that a macro expansion,
18 // "defined" expression, or condition expression has in the course of
19 // processing for the one location in the one header containing it,
20 // plus a list of the nested include stacks for the states.  When a macro
21 // or "defined" expression evaluates to the same value, which is the
22 // desired case, only one state is stored.  Similarly, for conditional
23 // directives, we save the condition expression states in a separate map.
24 //
25 // This information is collected as modularize compiles all the headers
26 // given to it to process.  After all the compilations are performed,
27 // a check is performed for any entries in the maps that contain more
28 // than one different state, and for these an output message is generated.
29 //
30 // For example:
31 //
32 //   (...)/SubHeader.h:11:5:
33 //   #if SYMBOL == 1
34 //       ^
35 //   error: Macro instance 'SYMBOL' has different values in this header,
36 //          depending on how it was included.
37 //     'SYMBOL' expanded to: '1' with respect to these inclusion paths:
38 //       (...)/Header1.h
39 //         (...)/SubHeader.h
40 //   (...)/SubHeader.h:3:9:
41 //   #define SYMBOL 1
42 //             ^
43 //   Macro defined here.
44 //     'SYMBOL' expanded to: '2' with respect to these inclusion paths:
45 //       (...)/Header2.h
46 //           (...)/SubHeader.h
47 //   (...)/SubHeader.h:7:9:
48 //   #define SYMBOL 2
49 //             ^
50 //   Macro defined here.
51 //
52 // The Basic Idea ('Extern "C/C++" {}' Or 'namespace {}') With Nested
53 // '#include' Checking)
54 //
55 // To check for '#include' directives nested inside 'Extern "C/C++" {}'
56 // or 'namespace {}' blocks, we keep track of the '#include' directives
57 // while running the preprocessor, and later during a walk of the AST
58 // we call a function to check for any '#include' directies inside
59 // an 'Extern "C/C++" {}' or 'namespace {}' block, given its source
60 // range.
61 //
62 // Design and Implementation Details (Macro and Conditional Checking)
63 //
64 // A PreprocessorTrackerImpl class implements the PreprocessorTracker
65 // interface. It uses a PreprocessorCallbacks class derived from PPCallbacks
66 // to track preprocessor activity, namely entering/exiting a header, macro
67 // expansions, use of "defined" expressions, and #if, #elif, #ifdef, and
68 // #ifndef conditional directives. PreprocessorTrackerImpl stores a map
69 // of MacroExpansionTracker objects keyed on a name/file/line/column
70 // value represented by a light-weight PPItemKey value object. This
71 // is the key top-level data structure tracking the values of macro
72 // expansion instances.  Similarly, it stores a map of ConditionalTracker
73 // objects with the same kind of key, for tracking preprocessor conditional
74 // directives.
75 //
76 // The MacroExpansionTracker object represents one macro reference or use
77 // of a "defined" expression in a header file. It stores a handle to a
78 // string representing the unexpanded macro instance, a handle to a string
79 // representing the unpreprocessed source line containing the unexpanded
80 // macro instance, and a vector of one or more MacroExpansionInstance
81 // objects.
82 //
83 // The MacroExpansionInstance object represents one or more expansions
84 // of a macro reference, for the case where the macro expands to the same
85 // value. MacroExpansionInstance stores a handle to a string representing
86 // the expanded macro value, a PPItemKey representing the file/line/column
87 // where the macro was defined, a handle to a string representing the source
88 // line containing the macro definition, and a vector of InclusionPathHandle
89 // values that represents the hierarchies of include files for each case
90 // where the particular header containing the macro reference was referenced
91 // or included.
92 
93 // In the normal case where a macro instance always expands to the same
94 // value, the MacroExpansionTracker object will only contain one
95 // MacroExpansionInstance representing all the macro expansion instances.
96 // If a case was encountered where a macro instance expands to a value
97 // that is different from that seen before, or the macro was defined in
98 // a different place, a new MacroExpansionInstance object representing
99 // that case will be added to the vector in MacroExpansionTracker. If a
100 // macro instance expands to a value already seen before, the
101 // InclusionPathHandle representing that case's include file hierarchy
102 // will be added to the existing MacroExpansionInstance object.
103 
104 // For checking conditional directives, the ConditionalTracker class
105 // functions similarly to MacroExpansionTracker, but tracks an #if,
106 // #elif, #ifdef, or #ifndef directive in a header file.  It stores
107 // a vector of one or two ConditionalExpansionInstance objects,
108 // representing the cases where the conditional expression evaluates
109 // to true or false.  This latter object stores the evaluated value
110 // of the condition expression (a bool) and a vector of
111 // InclusionPathHandles.
112 //
113 // To reduce the instances of string and object copying, the
114 // PreprocessorTrackerImpl class uses a StringPool to save all stored
115 // strings, and defines a StringHandle type to abstract the references
116 // to the strings.
117 //
118 // PreprocessorTrackerImpl also maintains a list representing the unique
119 // headers, which is just a vector of StringHandle's for the header file
120 // paths. A HeaderHandle abstracts a reference to a header, and is simply
121 // the index of the stored header file path.
122 //
123 // A HeaderInclusionPath class abstracts a unique hierarchy of header file
124 // inclusions. It simply stores a vector of HeaderHandles ordered from the
125 // top-most header (the one from the header list passed to modularize) down
126 // to the header containing the macro reference. PreprocessorTrackerImpl
127 // stores a vector of these objects. An InclusionPathHandle typedef
128 // abstracts a reference to one of the HeaderInclusionPath objects, and is
129 // simply the index of the stored HeaderInclusionPath object. The
130 // MacroExpansionInstance object stores a vector of these handles so that
131 // the reporting function can display the include hierarchies for the macro
132 // expansion instances represented by that object, to help the user
133 // understand how the header was included. (A future enhancement might
134 // be to associate a line number for the #include directives, but I
135 // think not doing so is good enough for the present.)
136 //
137 // A key reason for using these opaque handles was to try to keep all the
138 // internal objects light-weight value objects, in order to reduce string
139 // and object copying overhead, and to abstract this implementation detail.
140 //
141 // The key data structures are built up while modularize runs the headers
142 // through the compilation. A PreprocessorTracker instance is created and
143 // passed down to the AST action and consumer objects in modularize. For
144 // each new compilation instance, the consumer calls the
145 // PreprocessorTracker's handleNewPreprocessorEntry function, which sets
146 // up a PreprocessorCallbacks object for the preprocessor. At the end of
147 // the compilation instance, the PreprocessorTracker's
148 // handleNewPreprocessorExit function handles cleaning up with respect
149 // to the preprocessing instance.
150 //
151 // The PreprocessorCallbacks object uses an overidden FileChanged callback
152 // to determine when a header is entered and exited (including exiting the
153 // header during #include directives). It calls PreprocessorTracker's
154 // handleHeaderEntry and handleHeaderExit functions upon entering and
155 // exiting a header. These functions manage a stack of header handles
156 // representing by a vector, pushing and popping header handles as headers
157 // are entered and exited. When a HeaderInclusionPath object is created,
158 // it simply copies this stack.
159 //
160 // The PreprocessorCallbacks object uses an overridden MacroExpands callback
161 // to track when a macro expansion is performed. It calls a couple of helper
162 // functions to get the unexpanded and expanded macro values as strings, but
163 // then calls PreprocessorTrackerImpl's addMacroExpansionInstance function to
164 // do the rest of the work. The getMacroExpandedString function uses the
165 // preprocessor's getSpelling to convert tokens to strings using the
166 // information passed to the MacroExpands callback, and simply concatenates
167 // them. It makes recursive calls to itself to handle nested macro
168 // definitions, and also handles function-style macros.
169 //
170 // PreprocessorTrackerImpl's addMacroExpansionInstance function looks for
171 // an existing MacroExpansionTracker entry in its map of MacroExampleTracker
172 // objects. If none exists, it adds one with one MacroExpansionInstance and
173 // returns. If a MacroExpansionTracker object already exists, it looks for
174 // an existing MacroExpansionInstance object stored in the
175 // MacroExpansionTracker object, one that matches the macro expanded value
176 // and the macro definition location. If a matching MacroExpansionInstance
177 // object is found, it just adds the current HeaderInclusionPath object to
178 // it. If not found, it creates and stores a new MacroExpantionInstance
179 // object. The addMacroExpansionInstance function calls a couple of helper
180 // functions to get the pre-formatted location and source line strings for
181 // the macro reference and the macro definition stored as string handles.
182 // These helper functions use the current source manager from the
183 // preprocessor. This is done in advance at this point in time because the
184 // source manager doesn't exist at the time of the reporting.
185 //
186 // For conditional check, the PreprocessorCallbacks class overrides the
187 // PPCallbacks handlers for #if, #elif, #ifdef, and #ifndef.  These handlers
188 // call the addConditionalExpansionInstance method of
189 // PreprocessorTrackerImpl.  The process is similar to that of macros, but
190 // with some different data and error messages.  A lookup is performed for
191 // the conditional, and if a ConditionalTracker object doesn't yet exist for
192 // the conditional, a new one is added, including adding a
193 // ConditionalExpansionInstance object to it to represent the condition
194 // expression state.  If a ConditionalTracker for the conditional does
195 // exist, a lookup is made for a ConditionalExpansionInstance object
196 // matching the condition expression state.  If one exists, a
197 // HeaderInclusionPath is added to it.  Otherwise a new
198 // ConditionalExpansionInstance  entry is made.  If a ConditionalTracker
199 // has two ConditionalExpansionInstance objects, it means there was a
200 // conflict, meaning the conditional expression evaluated differently in
201 // one or more cases.
202 //
203 // After modularize has performed all the compilations, it enters a phase
204 // of error reporting. This new feature adds to this reporting phase calls
205 // to the PreprocessorTracker's reportInconsistentMacros and
206 // reportInconsistentConditionals functions. These functions walk the maps
207 // of MacroExpansionTracker's and ConditionalTracker's respectively. If
208 // any of these objects have more than one MacroExpansionInstance or
209 // ConditionalExpansionInstance objects, it formats and outputs an error
210 // message like the example shown previously, using the stored data.
211 //
212 // A potential issue is that there is some overlap between the #if/#elif
213 // conditional and macro reporting.  I could disable the #if and #elif,
214 // leaving just the #ifdef and #ifndef, since these don't overlap.  Or,
215 // to make clearer the separate reporting phases, I could add an output
216 // message marking the phases.
217 //
218 // Design and Implementation Details ('Extern "C/C++" {}' Or
219 // 'namespace {}') With Nested '#include' Checking)
220 //
221 // We override the InclusionDirective in PPCallbacks to record information
222 // about each '#include' directive encountered during preprocessing.
223 // We co-opt the PPItemKey class to store the information about each
224 // '#include' directive, including the source file name containing the
225 // directive, the name of the file being included, and the source line
226 // and column of the directive.  We store these object in a vector,
227 // after first check to see if an entry already exists.
228 //
229 // Later, while the AST is being walked for other checks, we provide
230 // visit handlers for 'extern "C/C++" {}' and 'namespace (name) {}'
231 // blocks, checking to see if any '#include' directives occurred
232 // within the blocks, reporting errors if any found.
233 //
234 // Future Directions
235 //
236 // We probably should add options to disable any of the checks, in case
237 // there is some problem with them, or the messages get too verbose.
238 //
239 // With the map of all the macro and conditional expansion instances,
240 // it might be possible to add to the existing modularize error messages
241 // (the second part referring to definitions being different), attempting
242 // to tie them to the last macro conflict encountered with respect to the
243 // order of the code encountered.
244 //
245 //===--------------------------------------------------------------------===//
246 
247 #include "clang/Lex/LexDiagnostic.h"
248 #include "PreprocessorTracker.h"
249 #include "clang/Lex/MacroArgs.h"
250 #include "clang/Lex/PPCallbacks.h"
251 #include "llvm/ADT/SmallSet.h"
252 #include "llvm/Support/StringPool.h"
253 #include "llvm/Support/raw_ostream.h"
254 #include "ModularizeUtilities.h"
255 
256 namespace Modularize {
257 
258 // Some handle types
259 typedef llvm::PooledStringPtr StringHandle;
260 
261 typedef int HeaderHandle;
262 const HeaderHandle HeaderHandleInvalid = -1;
263 
264 typedef int InclusionPathHandle;
265 const InclusionPathHandle InclusionPathHandleInvalid = -1;
266 
267 // Some utility functions.
268 
269 // Get a "file:line:column" source location string.
270 static std::string getSourceLocationString(clang::Preprocessor &PP,
271                                            clang::SourceLocation Loc) {
272   if (Loc.isInvalid())
273     return std::string("(none)");
274   else
275     return Loc.printToString(PP.getSourceManager());
276 }
277 
278 // Get just the file name from a source location.
279 static std::string getSourceLocationFile(clang::Preprocessor &PP,
280                                          clang::SourceLocation Loc) {
281   std::string Source(getSourceLocationString(PP, Loc));
282   size_t Offset = Source.find(':', 2);
283   if (Offset == std::string::npos)
284     return Source;
285   return Source.substr(0, Offset);
286 }
287 
288 // Get just the line and column from a source location.
289 static void getSourceLocationLineAndColumn(clang::Preprocessor &PP,
290                                            clang::SourceLocation Loc, int &Line,
291                                            int &Column) {
292   clang::PresumedLoc PLoc = PP.getSourceManager().getPresumedLoc(Loc);
293   if (PLoc.isInvalid()) {
294     Line = 0;
295     Column = 0;
296     return;
297   }
298   Line = PLoc.getLine();
299   Column = PLoc.getColumn();
300 }
301 
302 // Retrieve source snippet from file image.
303 static std::string getSourceString(clang::Preprocessor &PP,
304                                    clang::SourceRange Range) {
305   clang::SourceLocation BeginLoc = Range.getBegin();
306   clang::SourceLocation EndLoc = Range.getEnd();
307   const char *BeginPtr = PP.getSourceManager().getCharacterData(BeginLoc);
308   const char *EndPtr = PP.getSourceManager().getCharacterData(EndLoc);
309   size_t Length = EndPtr - BeginPtr;
310   return llvm::StringRef(BeginPtr, Length).trim().str();
311 }
312 
313 // Retrieve source line from file image given a location.
314 static std::string getSourceLine(clang::Preprocessor &PP,
315                                  clang::SourceLocation Loc) {
316   const llvm::MemoryBuffer *MemBuffer =
317       PP.getSourceManager().getBuffer(PP.getSourceManager().getFileID(Loc));
318   const char *Buffer = MemBuffer->getBufferStart();
319   const char *BufferEnd = MemBuffer->getBufferEnd();
320   const char *BeginPtr = PP.getSourceManager().getCharacterData(Loc);
321   const char *EndPtr = BeginPtr;
322   while (BeginPtr > Buffer) {
323     if (*BeginPtr == '\n') {
324       BeginPtr++;
325       break;
326     }
327     BeginPtr--;
328   }
329   while (EndPtr < BufferEnd) {
330     if (*EndPtr == '\n') {
331       break;
332     }
333     EndPtr++;
334   }
335   size_t Length = EndPtr - BeginPtr;
336   return llvm::StringRef(BeginPtr, Length).str();
337 }
338 
339 // Retrieve source line from file image given a file ID and line number.
340 static std::string getSourceLine(clang::Preprocessor &PP, clang::FileID FileID,
341                                  int Line) {
342   const llvm::MemoryBuffer *MemBuffer = PP.getSourceManager().getBuffer(FileID);
343   const char *Buffer = MemBuffer->getBufferStart();
344   const char *BufferEnd = MemBuffer->getBufferEnd();
345   const char *BeginPtr = Buffer;
346   const char *EndPtr = BufferEnd;
347   int LineCounter = 1;
348   if (Line == 1)
349     BeginPtr = Buffer;
350   else {
351     while (Buffer < BufferEnd) {
352       if (*Buffer == '\n') {
353         if (++LineCounter == Line) {
354           BeginPtr = Buffer++ + 1;
355           break;
356         }
357       }
358       Buffer++;
359     }
360   }
361   while (Buffer < BufferEnd) {
362     if (*Buffer == '\n') {
363       EndPtr = Buffer;
364       break;
365     }
366     Buffer++;
367   }
368   size_t Length = EndPtr - BeginPtr;
369   return llvm::StringRef(BeginPtr, Length).str();
370 }
371 
372 // Get the string for the Unexpanded macro instance.
373 // The soureRange is expected to end at the last token
374 // for the macro instance, which in the case of a function-style
375 // macro will be a ')', but for an object-style macro, it
376 // will be the macro name itself.
377 static std::string getMacroUnexpandedString(clang::SourceRange Range,
378                                             clang::Preprocessor &PP,
379                                             llvm::StringRef MacroName,
380                                             const clang::MacroInfo *MI) {
381   clang::SourceLocation BeginLoc(Range.getBegin());
382   const char *BeginPtr = PP.getSourceManager().getCharacterData(BeginLoc);
383   size_t Length;
384   std::string Unexpanded;
385   if (MI->isFunctionLike()) {
386     clang::SourceLocation EndLoc(Range.getEnd());
387     const char *EndPtr = PP.getSourceManager().getCharacterData(EndLoc) + 1;
388     Length = (EndPtr - BeginPtr) + 1; // +1 is ')' width.
389   } else
390     Length = MacroName.size();
391   return llvm::StringRef(BeginPtr, Length).trim().str();
392 }
393 
394 // Get the expansion for a macro instance, given the information
395 // provided by PPCallbacks.
396 // FIXME: This doesn't support function-style macro instances
397 // passed as arguments to another function-style macro. However,
398 // since it still expands the inner arguments, it still
399 // allows modularize to effectively work with respect to macro
400 // consistency checking, although it displays the incorrect
401 // expansion in error messages.
402 static std::string getMacroExpandedString(clang::Preprocessor &PP,
403                                           llvm::StringRef MacroName,
404                                           const clang::MacroInfo *MI,
405                                           const clang::MacroArgs *Args) {
406   std::string Expanded;
407   // Walk over the macro Tokens.
408   for (const auto &T : MI->tokens()) {
409     clang::IdentifierInfo *II = T.getIdentifierInfo();
410     int ArgNo = (II && Args ? MI->getParameterNum(II) : -1);
411     if (ArgNo == -1) {
412       // This isn't an argument, just add it.
413       if (II == nullptr)
414         Expanded += PP.getSpelling(T); // Not an identifier.
415       else {
416         // Token is for an identifier.
417         std::string Name = II->getName().str();
418         // Check for nexted macro references.
419         clang::MacroInfo *MacroInfo = PP.getMacroInfo(II);
420         if (MacroInfo && (Name != MacroName))
421           Expanded += getMacroExpandedString(PP, Name, MacroInfo, nullptr);
422         else
423           Expanded += Name;
424       }
425       continue;
426     }
427     // We get here if it's a function-style macro with arguments.
428     const clang::Token *ResultArgToks;
429     const clang::Token *ArgTok = Args->getUnexpArgument(ArgNo);
430     if (Args->ArgNeedsPreexpansion(ArgTok, PP))
431       ResultArgToks = &(const_cast<clang::MacroArgs *>(Args))
432           ->getPreExpArgument(ArgNo, PP)[0];
433     else
434       ResultArgToks = ArgTok; // Use non-preexpanded Tokens.
435     // If the arg token didn't expand into anything, ignore it.
436     if (ResultArgToks->is(clang::tok::eof))
437       continue;
438     unsigned NumToks = clang::MacroArgs::getArgLength(ResultArgToks);
439     // Append the resulting argument expansions.
440     for (unsigned ArgumentIndex = 0; ArgumentIndex < NumToks; ++ArgumentIndex) {
441       const clang::Token &AT = ResultArgToks[ArgumentIndex];
442       clang::IdentifierInfo *II = AT.getIdentifierInfo();
443       if (II == nullptr)
444         Expanded += PP.getSpelling(AT); // Not an identifier.
445       else {
446         // It's an identifier.  Check for further expansion.
447         std::string Name = II->getName().str();
448         clang::MacroInfo *MacroInfo = PP.getMacroInfo(II);
449         if (MacroInfo)
450           Expanded += getMacroExpandedString(PP, Name, MacroInfo, nullptr);
451         else
452           Expanded += Name;
453       }
454     }
455   }
456   return Expanded;
457 }
458 
459 namespace {
460 
461 // ConditionValueKind strings.
462 const char *
463 ConditionValueKindStrings[] = {
464   "(not evaluated)", "false", "true"
465 };
466 
467 bool operator<(const StringHandle &H1, const StringHandle &H2) {
468   const char *S1 = (H1 ? *H1 : "");
469   const char *S2 = (H2 ? *H2 : "");
470   int Diff = strcmp(S1, S2);
471   return Diff < 0;
472 }
473 bool operator>(const StringHandle &H1, const StringHandle &H2) {
474   const char *S1 = (H1 ? *H1 : "");
475   const char *S2 = (H2 ? *H2 : "");
476   int Diff = strcmp(S1, S2);
477   return Diff > 0;
478 }
479 
480 // Preprocessor item key.
481 //
482 // This class represents a location in a source file, for use
483 // as a key representing a unique name/file/line/column quadruplet,
484 // which in this case is used to identify a macro expansion instance,
485 // but could be used for other things as well.
486 // The file is a header file handle, the line is a line number,
487 // and the column is a column number.
488 class PPItemKey {
489 public:
490   PPItemKey(clang::Preprocessor &PP, StringHandle Name, HeaderHandle File,
491             clang::SourceLocation Loc)
492       : Name(Name), File(File) {
493     getSourceLocationLineAndColumn(PP, Loc, Line, Column);
494   }
495   PPItemKey(StringHandle Name, HeaderHandle File, int Line, int Column)
496       : Name(Name), File(File), Line(Line), Column(Column) {}
497   PPItemKey(const PPItemKey &Other)
498       : Name(Other.Name), File(Other.File), Line(Other.Line),
499         Column(Other.Column) {}
500   PPItemKey() : File(HeaderHandleInvalid), Line(0), Column(0) {}
501   bool operator==(const PPItemKey &Other) const {
502     if (Name != Other.Name)
503       return false;
504     if (File != Other.File)
505       return false;
506     if (Line != Other.Line)
507       return false;
508     return Column == Other.Column;
509   }
510   bool operator<(const PPItemKey &Other) const {
511     if (Name < Other.Name)
512       return true;
513     else if (Name > Other.Name)
514       return false;
515     if (File < Other.File)
516       return true;
517     else if (File > Other.File)
518       return false;
519     if (Line < Other.Line)
520       return true;
521     else if (Line > Other.Line)
522       return false;
523     return Column < Other.Column;
524   }
525   StringHandle Name;
526   HeaderHandle File;
527   int Line;
528   int Column;
529 };
530 
531 // Header inclusion path.
532 class HeaderInclusionPath {
533 public:
534   HeaderInclusionPath(std::vector<HeaderHandle> HeaderInclusionPath)
535       : Path(HeaderInclusionPath) {}
536   HeaderInclusionPath(const HeaderInclusionPath &Other) : Path(Other.Path) {}
537   HeaderInclusionPath() {}
538   std::vector<HeaderHandle> Path;
539 };
540 
541 // Macro expansion instance.
542 //
543 // This class represents an instance of a macro expansion with a
544 // unique value.  It also stores the unique header inclusion paths
545 // for use in telling the user the nested include path to the header.
546 class MacroExpansionInstance {
547 public:
548   MacroExpansionInstance(StringHandle MacroExpanded,
549                          PPItemKey &DefinitionLocation,
550                          StringHandle DefinitionSourceLine,
551                          InclusionPathHandle H)
552       : MacroExpanded(MacroExpanded), DefinitionLocation(DefinitionLocation),
553         DefinitionSourceLine(DefinitionSourceLine) {
554     InclusionPathHandles.push_back(H);
555   }
556   MacroExpansionInstance() {}
557 
558   // Check for the presence of a header inclusion path handle entry.
559   // Return false if not found.
560   bool haveInclusionPathHandle(InclusionPathHandle H) {
561     for (auto I = InclusionPathHandles.begin(), E = InclusionPathHandles.end();
562          I != E; ++I) {
563       if (*I == H)
564         return true;
565     }
566     return InclusionPathHandleInvalid;
567   }
568   // Add a new header inclusion path entry, if not already present.
569   void addInclusionPathHandle(InclusionPathHandle H) {
570     if (!haveInclusionPathHandle(H))
571       InclusionPathHandles.push_back(H);
572   }
573 
574   // A string representing the macro instance after preprocessing.
575   StringHandle MacroExpanded;
576   // A file/line/column triplet representing the macro definition location.
577   PPItemKey DefinitionLocation;
578   // A place to save the macro definition line string.
579   StringHandle DefinitionSourceLine;
580   // The header inclusion path handles for all the instances.
581   std::vector<InclusionPathHandle> InclusionPathHandles;
582 };
583 
584 // Macro expansion instance tracker.
585 //
586 // This class represents one macro expansion, keyed by a PPItemKey.
587 // It stores a string representing the macro reference in the source,
588 // and a list of ConditionalExpansionInstances objects representing
589 // the unique values the condition expands to in instances of the header.
590 class MacroExpansionTracker {
591 public:
592   MacroExpansionTracker(StringHandle MacroUnexpanded,
593                         StringHandle MacroExpanded,
594                         StringHandle InstanceSourceLine,
595                         PPItemKey &DefinitionLocation,
596                         StringHandle DefinitionSourceLine,
597                         InclusionPathHandle InclusionPathHandle)
598       : MacroUnexpanded(MacroUnexpanded),
599         InstanceSourceLine(InstanceSourceLine) {
600     addMacroExpansionInstance(MacroExpanded, DefinitionLocation,
601                               DefinitionSourceLine, InclusionPathHandle);
602   }
603   MacroExpansionTracker() {}
604 
605   // Find a matching macro expansion instance.
606   MacroExpansionInstance *
607   findMacroExpansionInstance(StringHandle MacroExpanded,
608                              PPItemKey &DefinitionLocation) {
609     for (auto I = MacroExpansionInstances.begin(),
610               E = MacroExpansionInstances.end();
611          I != E; ++I) {
612       if ((I->MacroExpanded == MacroExpanded) &&
613           (I->DefinitionLocation == DefinitionLocation)) {
614         return &*I; // Found.
615       }
616     }
617     return nullptr; // Not found.
618   }
619 
620   // Add a macro expansion instance.
621   void addMacroExpansionInstance(StringHandle MacroExpanded,
622                                  PPItemKey &DefinitionLocation,
623                                  StringHandle DefinitionSourceLine,
624                                  InclusionPathHandle InclusionPathHandle) {
625     MacroExpansionInstances.push_back(
626         MacroExpansionInstance(MacroExpanded, DefinitionLocation,
627                                DefinitionSourceLine, InclusionPathHandle));
628   }
629 
630   // Return true if there is a mismatch.
631   bool hasMismatch() { return MacroExpansionInstances.size() > 1; }
632 
633   // A string representing the macro instance without expansion.
634   StringHandle MacroUnexpanded;
635   // A place to save the macro instance source line string.
636   StringHandle InstanceSourceLine;
637   // The macro expansion instances.
638   // If all instances of the macro expansion expand to the same value,
639   // This vector will only have one instance.
640   std::vector<MacroExpansionInstance> MacroExpansionInstances;
641 };
642 
643 // Conditional expansion instance.
644 //
645 // This class represents an instance of a condition exoression result
646 // with a unique value.  It also stores the unique header inclusion paths
647 // for use in telling the user the nested include path to the header.
648 class ConditionalExpansionInstance {
649 public:
650   ConditionalExpansionInstance(clang::PPCallbacks::ConditionValueKind ConditionValue, InclusionPathHandle H)
651       : ConditionValue(ConditionValue) {
652     InclusionPathHandles.push_back(H);
653   }
654   ConditionalExpansionInstance() {}
655 
656   // Check for the presence of a header inclusion path handle entry.
657   // Return false if not found.
658   bool haveInclusionPathHandle(InclusionPathHandle H) {
659     for (auto I = InclusionPathHandles.begin(), E = InclusionPathHandles.end();
660          I != E; ++I) {
661       if (*I == H)
662         return true;
663     }
664     return InclusionPathHandleInvalid;
665   }
666   // Add a new header inclusion path entry, if not already present.
667   void addInclusionPathHandle(InclusionPathHandle H) {
668     if (!haveInclusionPathHandle(H))
669       InclusionPathHandles.push_back(H);
670   }
671 
672   // A flag representing the evaluated condition value.
673   clang::PPCallbacks::ConditionValueKind ConditionValue;
674   // The header inclusion path handles for all the instances.
675   std::vector<InclusionPathHandle> InclusionPathHandles;
676 };
677 
678 // Conditional directive instance tracker.
679 //
680 // This class represents one conditional directive, keyed by a PPItemKey.
681 // It stores a string representing the macro reference in the source,
682 // and a list of ConditionExpansionInstance objects representing
683 // the unique value the condition expression expands to in instances of
684 // the header.
685 class ConditionalTracker {
686 public:
687   ConditionalTracker(clang::tok::PPKeywordKind DirectiveKind,
688                      clang::PPCallbacks::ConditionValueKind ConditionValue,
689                      StringHandle ConditionUnexpanded,
690                      InclusionPathHandle InclusionPathHandle)
691       : DirectiveKind(DirectiveKind), ConditionUnexpanded(ConditionUnexpanded) {
692     addConditionalExpansionInstance(ConditionValue, InclusionPathHandle);
693   }
694   ConditionalTracker() {}
695 
696   // Find a matching condition expansion instance.
697   ConditionalExpansionInstance *
698   findConditionalExpansionInstance(clang::PPCallbacks::ConditionValueKind ConditionValue) {
699     for (auto I = ConditionalExpansionInstances.begin(),
700               E = ConditionalExpansionInstances.end();
701          I != E; ++I) {
702       if (I->ConditionValue == ConditionValue) {
703         return &*I; // Found.
704       }
705     }
706     return nullptr; // Not found.
707   }
708 
709   // Add a conditional expansion instance.
710   void
711   addConditionalExpansionInstance(clang::PPCallbacks::ConditionValueKind ConditionValue,
712                                   InclusionPathHandle InclusionPathHandle) {
713     ConditionalExpansionInstances.push_back(
714         ConditionalExpansionInstance(ConditionValue, InclusionPathHandle));
715   }
716 
717   // Return true if there is a mismatch.
718   bool hasMismatch() { return ConditionalExpansionInstances.size() > 1; }
719 
720   // The kind of directive.
721   clang::tok::PPKeywordKind DirectiveKind;
722   // A string representing the macro instance without expansion.
723   StringHandle ConditionUnexpanded;
724   // The condition expansion instances.
725   // If all instances of the conditional expression expand to the same value,
726   // This vector will only have one instance.
727   std::vector<ConditionalExpansionInstance> ConditionalExpansionInstances;
728 };
729 
730 class PreprocessorTrackerImpl;
731 
732 // Preprocessor callbacks for modularize.
733 //
734 // This class derives from the Clang PPCallbacks class to track preprocessor
735 // actions, such as changing files and handling preprocessor directives and
736 // macro expansions.  It has to figure out when a new header file is entered
737 // and left, as the provided handler is not particularly clear about it.
738 class PreprocessorCallbacks : public clang::PPCallbacks {
739 public:
740   PreprocessorCallbacks(PreprocessorTrackerImpl &ppTracker,
741                         clang::Preprocessor &PP, llvm::StringRef rootHeaderFile)
742       : PPTracker(ppTracker), PP(PP), RootHeaderFile(rootHeaderFile) {}
743   ~PreprocessorCallbacks() override {}
744 
745   // Overridden handlers.
746   void InclusionDirective(clang::SourceLocation HashLoc,
747                           const clang::Token &IncludeTok,
748                           llvm::StringRef FileName, bool IsAngled,
749                           clang::CharSourceRange FilenameRange,
750                           const clang::FileEntry *File,
751                           llvm::StringRef SearchPath,
752                           llvm::StringRef RelativePath,
753                           const clang::Module *Imported,
754                           clang::SrcMgr::CharacteristicKind FileType) override;
755   void FileChanged(clang::SourceLocation Loc,
756                    clang::PPCallbacks::FileChangeReason Reason,
757                    clang::SrcMgr::CharacteristicKind FileType,
758                    clang::FileID PrevFID = clang::FileID()) override;
759   void MacroExpands(const clang::Token &MacroNameTok,
760                     const clang::MacroDefinition &MD, clang::SourceRange Range,
761                     const clang::MacroArgs *Args) override;
762   void Defined(const clang::Token &MacroNameTok,
763                const clang::MacroDefinition &MD,
764                clang::SourceRange Range) override;
765   void If(clang::SourceLocation Loc, clang::SourceRange ConditionRange,
766           clang::PPCallbacks::ConditionValueKind ConditionResult) override;
767   void Elif(clang::SourceLocation Loc, clang::SourceRange ConditionRange,
768             clang::PPCallbacks::ConditionValueKind ConditionResult,
769             clang::SourceLocation IfLoc) override;
770   void Ifdef(clang::SourceLocation Loc, const clang::Token &MacroNameTok,
771              const clang::MacroDefinition &MD) override;
772   void Ifndef(clang::SourceLocation Loc, const clang::Token &MacroNameTok,
773               const clang::MacroDefinition &MD) override;
774 
775 private:
776   PreprocessorTrackerImpl &PPTracker;
777   clang::Preprocessor &PP;
778   std::string RootHeaderFile;
779 };
780 
781 // Preprocessor macro expansion item map types.
782 typedef std::map<PPItemKey, MacroExpansionTracker> MacroExpansionMap;
783 typedef std::map<PPItemKey, MacroExpansionTracker>::iterator
784 MacroExpansionMapIter;
785 
786 // Preprocessor conditional expansion item map types.
787 typedef std::map<PPItemKey, ConditionalTracker> ConditionalExpansionMap;
788 typedef std::map<PPItemKey, ConditionalTracker>::iterator
789 ConditionalExpansionMapIter;
790 
791 // Preprocessor tracker for modularize.
792 //
793 // This class stores information about all the headers processed in the
794 // course of running modularize.
795 class PreprocessorTrackerImpl : public PreprocessorTracker {
796 public:
797   PreprocessorTrackerImpl(llvm::SmallVector<std::string, 32> &Headers,
798         bool DoBlockCheckHeaderListOnly)
799       : BlockCheckHeaderListOnly(DoBlockCheckHeaderListOnly),
800         CurrentInclusionPathHandle(InclusionPathHandleInvalid),
801         InNestedHeader(false) {
802     // Use canonical header path representation.
803     for (llvm::ArrayRef<std::string>::iterator I = Headers.begin(),
804       E = Headers.end();
805       I != E; ++I) {
806       HeaderList.push_back(getCanonicalPath(*I));
807     }
808   }
809 
810   ~PreprocessorTrackerImpl() override {}
811 
812   // Handle entering a preprocessing session.
813   void handlePreprocessorEntry(clang::Preprocessor &PP,
814                                llvm::StringRef rootHeaderFile) override {
815     HeadersInThisCompile.clear();
816     assert((HeaderStack.size() == 0) && "Header stack should be empty.");
817     pushHeaderHandle(addHeader(rootHeaderFile));
818     PP.addPPCallbacks(llvm::make_unique<PreprocessorCallbacks>(*this, PP,
819                                                                rootHeaderFile));
820   }
821   // Handle exiting a preprocessing session.
822   void handlePreprocessorExit() override { HeaderStack.clear(); }
823 
824   // Handle include directive.
825   // This function is called every time an include directive is seen by the
826   // preprocessor, for the purpose of later checking for 'extern "" {}' or
827   // "namespace {}" blocks containing #include directives.
828   void handleIncludeDirective(llvm::StringRef DirectivePath, int DirectiveLine,
829                               int DirectiveColumn,
830                               llvm::StringRef TargetPath) override {
831     // If it's not a header in the header list, ignore it with respect to
832     // the check.
833     if (BlockCheckHeaderListOnly && !isHeaderListHeader(TargetPath))
834       return;
835     HeaderHandle CurrentHeaderHandle = findHeaderHandle(DirectivePath);
836     StringHandle IncludeHeaderHandle = addString(TargetPath);
837     for (std::vector<PPItemKey>::const_iterator I = IncludeDirectives.begin(),
838                                                 E = IncludeDirectives.end();
839          I != E; ++I) {
840       // If we already have an entry for this directive, return now.
841       if ((I->File == CurrentHeaderHandle) && (I->Line == DirectiveLine))
842         return;
843     }
844     PPItemKey IncludeDirectiveItem(IncludeHeaderHandle, CurrentHeaderHandle,
845                                    DirectiveLine, DirectiveColumn);
846     IncludeDirectives.push_back(IncludeDirectiveItem);
847   }
848 
849   // Check for include directives within the given source line range.
850   // Report errors if any found.  Returns true if no include directives
851   // found in block.
852   bool checkForIncludesInBlock(clang::Preprocessor &PP,
853                                clang::SourceRange BlockSourceRange,
854                                const char *BlockIdentifierMessage,
855                                llvm::raw_ostream &OS) override {
856     clang::SourceLocation BlockStartLoc = BlockSourceRange.getBegin();
857     clang::SourceLocation BlockEndLoc = BlockSourceRange.getEnd();
858     // Use block location to get FileID of both the include directive
859     // and block statement.
860     clang::FileID FileID = PP.getSourceManager().getFileID(BlockStartLoc);
861     std::string SourcePath = getSourceLocationFile(PP, BlockStartLoc);
862     SourcePath = ModularizeUtilities::getCanonicalPath(SourcePath);
863     HeaderHandle SourceHandle = findHeaderHandle(SourcePath);
864     if (SourceHandle == -1)
865       return true;
866     int BlockStartLine, BlockStartColumn, BlockEndLine, BlockEndColumn;
867     bool returnValue = true;
868     getSourceLocationLineAndColumn(PP, BlockStartLoc, BlockStartLine,
869                                    BlockStartColumn);
870     getSourceLocationLineAndColumn(PP, BlockEndLoc, BlockEndLine,
871                                    BlockEndColumn);
872     for (std::vector<PPItemKey>::const_iterator I = IncludeDirectives.begin(),
873                                                 E = IncludeDirectives.end();
874          I != E; ++I) {
875       // If we find an entry within the block, report an error.
876       if ((I->File == SourceHandle) && (I->Line >= BlockStartLine) &&
877           (I->Line < BlockEndLine)) {
878         returnValue = false;
879         OS << SourcePath << ":" << I->Line << ":" << I->Column << ":\n";
880         OS << getSourceLine(PP, FileID, I->Line) << "\n";
881         if (I->Column > 0)
882           OS << std::string(I->Column - 1, ' ') << "^\n";
883         OS << "error: Include directive within " << BlockIdentifierMessage
884            << ".\n";
885         OS << SourcePath << ":" << BlockStartLine << ":" << BlockStartColumn
886            << ":\n";
887         OS << getSourceLine(PP, BlockStartLoc) << "\n";
888         if (BlockStartColumn > 0)
889           OS << std::string(BlockStartColumn - 1, ' ') << "^\n";
890         OS << "The \"" << BlockIdentifierMessage << "\" block is here.\n";
891       }
892     }
893     return returnValue;
894   }
895 
896   // Handle entering a header source file.
897   void handleHeaderEntry(clang::Preprocessor &PP, llvm::StringRef HeaderPath) {
898     // Ignore <built-in> and <command-line> to reduce message clutter.
899     if (HeaderPath.startswith("<"))
900       return;
901     HeaderHandle H = addHeader(HeaderPath);
902     if (H != getCurrentHeaderHandle())
903       pushHeaderHandle(H);
904     // Check for nested header.
905     if (!InNestedHeader)
906       InNestedHeader = !HeadersInThisCompile.insert(H).second;
907   }
908 
909   // Handle exiting a header source file.
910   void handleHeaderExit(llvm::StringRef HeaderPath) {
911     // Ignore <built-in> and <command-line> to reduce message clutter.
912     if (HeaderPath.startswith("<"))
913       return;
914     HeaderHandle H = findHeaderHandle(HeaderPath);
915     HeaderHandle TH;
916     if (isHeaderHandleInStack(H)) {
917       do {
918         TH = getCurrentHeaderHandle();
919         popHeaderHandle();
920       } while ((TH != H) && (HeaderStack.size() != 0));
921     }
922     InNestedHeader = false;
923   }
924 
925   // Lookup/add string.
926   StringHandle addString(llvm::StringRef Str) { return Strings.intern(Str); }
927 
928   // Convert to a canonical path.
929   std::string getCanonicalPath(llvm::StringRef path) const {
930     std::string CanonicalPath(path);
931     std::replace(CanonicalPath.begin(), CanonicalPath.end(), '\\', '/');
932     return CanonicalPath;
933   }
934 
935   // Return true if the given header is in the header list.
936   bool isHeaderListHeader(llvm::StringRef HeaderPath) const {
937     std::string CanonicalPath = getCanonicalPath(HeaderPath);
938     for (llvm::ArrayRef<std::string>::iterator I = HeaderList.begin(),
939         E = HeaderList.end();
940         I != E; ++I) {
941       if (*I == CanonicalPath)
942         return true;
943     }
944     return false;
945   }
946 
947   // Get the handle of a header file entry.
948   // Return HeaderHandleInvalid if not found.
949   HeaderHandle findHeaderHandle(llvm::StringRef HeaderPath) const {
950     std::string CanonicalPath = getCanonicalPath(HeaderPath);
951     HeaderHandle H = 0;
952     for (auto I = HeaderPaths.begin(), E = HeaderPaths.end(); I != E;
953          ++I, ++H) {
954       if (**I == CanonicalPath)
955         return H;
956     }
957     return HeaderHandleInvalid;
958   }
959 
960   // Add a new header file entry, or return existing handle.
961   // Return the header handle.
962   HeaderHandle addHeader(llvm::StringRef HeaderPath) {
963     std::string CanonicalPath = getCanonicalPath(HeaderPath);
964     HeaderHandle H = findHeaderHandle(CanonicalPath);
965     if (H == HeaderHandleInvalid) {
966       H = HeaderPaths.size();
967       HeaderPaths.push_back(addString(CanonicalPath));
968     }
969     return H;
970   }
971 
972   // Return a header file path string given its handle.
973   StringHandle getHeaderFilePath(HeaderHandle H) const {
974     if ((H >= 0) && (H < (HeaderHandle)HeaderPaths.size()))
975       return HeaderPaths[H];
976     return StringHandle();
977   }
978 
979   // Returns a handle to the inclusion path.
980   InclusionPathHandle pushHeaderHandle(HeaderHandle H) {
981     HeaderStack.push_back(H);
982     return CurrentInclusionPathHandle = addInclusionPathHandle(HeaderStack);
983   }
984   // Pops the last header handle from the stack;
985   void popHeaderHandle() {
986     // assert((HeaderStack.size() != 0) && "Header stack already empty.");
987     if (HeaderStack.size() != 0) {
988       HeaderStack.pop_back();
989       CurrentInclusionPathHandle = addInclusionPathHandle(HeaderStack);
990     }
991   }
992   // Get the top handle on the header stack.
993   HeaderHandle getCurrentHeaderHandle() const {
994     if (HeaderStack.size() != 0)
995       return HeaderStack.back();
996     return HeaderHandleInvalid;
997   }
998 
999   // Check for presence of header handle in the header stack.
1000   bool isHeaderHandleInStack(HeaderHandle H) const {
1001     for (auto I = HeaderStack.begin(), E = HeaderStack.end(); I != E; ++I) {
1002       if (*I == H)
1003         return true;
1004     }
1005     return false;
1006   }
1007 
1008   // Get the handle of a header inclusion path entry.
1009   // Return InclusionPathHandleInvalid if not found.
1010   InclusionPathHandle
1011   findInclusionPathHandle(const std::vector<HeaderHandle> &Path) const {
1012     InclusionPathHandle H = 0;
1013     for (auto I = InclusionPaths.begin(), E = InclusionPaths.end(); I != E;
1014          ++I, ++H) {
1015       if (I->Path == Path)
1016         return H;
1017     }
1018     return HeaderHandleInvalid;
1019   }
1020   // Add a new header inclusion path entry, or return existing handle.
1021   // Return the header inclusion path entry handle.
1022   InclusionPathHandle
1023   addInclusionPathHandle(const std::vector<HeaderHandle> &Path) {
1024     InclusionPathHandle H = findInclusionPathHandle(Path);
1025     if (H == HeaderHandleInvalid) {
1026       H = InclusionPaths.size();
1027       InclusionPaths.push_back(HeaderInclusionPath(Path));
1028     }
1029     return H;
1030   }
1031   // Return the current inclusion path handle.
1032   InclusionPathHandle getCurrentInclusionPathHandle() const {
1033     return CurrentInclusionPathHandle;
1034   }
1035 
1036   // Return an inclusion path given its handle.
1037   const std::vector<HeaderHandle> &
1038   getInclusionPath(InclusionPathHandle H) const {
1039     if ((H >= 0) && (H <= (InclusionPathHandle)InclusionPaths.size()))
1040       return InclusionPaths[H].Path;
1041     static std::vector<HeaderHandle> Empty;
1042     return Empty;
1043   }
1044 
1045   // Add a macro expansion instance.
1046   void addMacroExpansionInstance(clang::Preprocessor &PP, HeaderHandle H,
1047                                  clang::SourceLocation InstanceLoc,
1048                                  clang::SourceLocation DefinitionLoc,
1049                                  clang::IdentifierInfo *II,
1050                                  llvm::StringRef MacroUnexpanded,
1051                                  llvm::StringRef MacroExpanded,
1052                                  InclusionPathHandle InclusionPathHandle) {
1053     if (InNestedHeader)
1054       return;
1055     StringHandle MacroName = addString(II->getName());
1056     PPItemKey InstanceKey(PP, MacroName, H, InstanceLoc);
1057     PPItemKey DefinitionKey(PP, MacroName, H, DefinitionLoc);
1058     auto I = MacroExpansions.find(InstanceKey);
1059     // If existing instance of expansion not found, add one.
1060     if (I == MacroExpansions.end()) {
1061       std::string InstanceSourceLine =
1062           getSourceLocationString(PP, InstanceLoc) + ":\n" +
1063           getSourceLine(PP, InstanceLoc) + "\n";
1064       std::string DefinitionSourceLine =
1065           getSourceLocationString(PP, DefinitionLoc) + ":\n" +
1066           getSourceLine(PP, DefinitionLoc) + "\n";
1067       MacroExpansions[InstanceKey] = MacroExpansionTracker(
1068           addString(MacroUnexpanded), addString(MacroExpanded),
1069           addString(InstanceSourceLine), DefinitionKey,
1070           addString(DefinitionSourceLine), InclusionPathHandle);
1071     } else {
1072       // We've seen the macro before.  Get its tracker.
1073       MacroExpansionTracker &CondTracker = I->second;
1074       // Look up an existing instance value for the macro.
1075       MacroExpansionInstance *MacroInfo =
1076           CondTracker.findMacroExpansionInstance(addString(MacroExpanded),
1077                                                  DefinitionKey);
1078       // If found, just add the inclusion path to the instance.
1079       if (MacroInfo)
1080         MacroInfo->addInclusionPathHandle(InclusionPathHandle);
1081       else {
1082         // Otherwise add a new instance with the unique value.
1083         std::string DefinitionSourceLine =
1084             getSourceLocationString(PP, DefinitionLoc) + ":\n" +
1085             getSourceLine(PP, DefinitionLoc) + "\n";
1086         CondTracker.addMacroExpansionInstance(
1087             addString(MacroExpanded), DefinitionKey,
1088             addString(DefinitionSourceLine), InclusionPathHandle);
1089       }
1090     }
1091   }
1092 
1093   // Add a conditional expansion instance.
1094   void
1095   addConditionalExpansionInstance(clang::Preprocessor &PP, HeaderHandle H,
1096                                   clang::SourceLocation InstanceLoc,
1097                                   clang::tok::PPKeywordKind DirectiveKind,
1098                                   clang::PPCallbacks::ConditionValueKind ConditionValue,
1099                                   llvm::StringRef ConditionUnexpanded,
1100                                   InclusionPathHandle InclusionPathHandle) {
1101     // Ignore header guards, assuming the header guard is the only conditional.
1102     if (InNestedHeader)
1103       return;
1104     StringHandle ConditionUnexpandedHandle(addString(ConditionUnexpanded));
1105     PPItemKey InstanceKey(PP, ConditionUnexpandedHandle, H, InstanceLoc);
1106     auto I = ConditionalExpansions.find(InstanceKey);
1107     // If existing instance of condition not found, add one.
1108     if (I == ConditionalExpansions.end()) {
1109       std::string InstanceSourceLine =
1110           getSourceLocationString(PP, InstanceLoc) + ":\n" +
1111           getSourceLine(PP, InstanceLoc) + "\n";
1112       ConditionalExpansions[InstanceKey] =
1113           ConditionalTracker(DirectiveKind, ConditionValue,
1114                              ConditionUnexpandedHandle, InclusionPathHandle);
1115     } else {
1116       // We've seen the conditional before.  Get its tracker.
1117       ConditionalTracker &CondTracker = I->second;
1118       // Look up an existing instance value for the condition.
1119       ConditionalExpansionInstance *MacroInfo =
1120           CondTracker.findConditionalExpansionInstance(ConditionValue);
1121       // If found, just add the inclusion path to the instance.
1122       if (MacroInfo)
1123         MacroInfo->addInclusionPathHandle(InclusionPathHandle);
1124       else {
1125         // Otherwise add a new instance with the unique value.
1126         CondTracker.addConditionalExpansionInstance(ConditionValue,
1127                                                     InclusionPathHandle);
1128       }
1129     }
1130   }
1131 
1132   // Report on inconsistent macro instances.
1133   // Returns true if any mismatches.
1134   bool reportInconsistentMacros(llvm::raw_ostream &OS) override {
1135     bool ReturnValue = false;
1136     // Walk all the macro expansion trackers in the map.
1137     for (auto I = MacroExpansions.begin(), E = MacroExpansions.end(); I != E;
1138          ++I) {
1139       const PPItemKey &ItemKey = I->first;
1140       MacroExpansionTracker &MacroExpTracker = I->second;
1141       // If no mismatch (only one instance value) continue.
1142       if (!MacroExpTracker.hasMismatch())
1143         continue;
1144       // Tell caller we found one or more errors.
1145       ReturnValue = true;
1146       // Start the error message.
1147       OS << *MacroExpTracker.InstanceSourceLine;
1148       if (ItemKey.Column > 0)
1149         OS << std::string(ItemKey.Column - 1, ' ') << "^\n";
1150       OS << "error: Macro instance '" << *MacroExpTracker.MacroUnexpanded
1151          << "' has different values in this header, depending on how it was "
1152             "included.\n";
1153       // Walk all the instances.
1154       for (auto IMT = MacroExpTracker.MacroExpansionInstances.begin(),
1155                 EMT = MacroExpTracker.MacroExpansionInstances.end();
1156            IMT != EMT; ++IMT) {
1157         MacroExpansionInstance &MacroInfo = *IMT;
1158         OS << "  '" << *MacroExpTracker.MacroUnexpanded << "' expanded to: '"
1159            << *MacroInfo.MacroExpanded
1160            << "' with respect to these inclusion paths:\n";
1161         // Walk all the inclusion path hierarchies.
1162         for (auto IIP = MacroInfo.InclusionPathHandles.begin(),
1163                   EIP = MacroInfo.InclusionPathHandles.end();
1164              IIP != EIP; ++IIP) {
1165           const std::vector<HeaderHandle> &ip = getInclusionPath(*IIP);
1166           auto Count = (int)ip.size();
1167           for (int Index = 0; Index < Count; ++Index) {
1168             HeaderHandle H = ip[Index];
1169             OS << std::string((Index * 2) + 4, ' ') << *getHeaderFilePath(H)
1170                << "\n";
1171           }
1172         }
1173         // For a macro that wasn't defined, we flag it by using the
1174         // instance location.
1175         // If there is a definition...
1176         if (MacroInfo.DefinitionLocation.Line != ItemKey.Line) {
1177           OS << *MacroInfo.DefinitionSourceLine;
1178           if (MacroInfo.DefinitionLocation.Column > 0)
1179             OS << std::string(MacroInfo.DefinitionLocation.Column - 1, ' ')
1180                << "^\n";
1181           OS << "Macro defined here.\n";
1182         } else
1183           OS << "(no macro definition)"
1184              << "\n";
1185       }
1186     }
1187     return ReturnValue;
1188   }
1189 
1190   // Report on inconsistent conditional instances.
1191   // Returns true if any mismatches.
1192   bool reportInconsistentConditionals(llvm::raw_ostream &OS) override {
1193     bool ReturnValue = false;
1194     // Walk all the conditional trackers in the map.
1195     for (auto I = ConditionalExpansions.begin(),
1196               E = ConditionalExpansions.end();
1197          I != E; ++I) {
1198       const PPItemKey &ItemKey = I->first;
1199       ConditionalTracker &CondTracker = I->second;
1200       if (!CondTracker.hasMismatch())
1201         continue;
1202       // Tell caller we found one or more errors.
1203       ReturnValue = true;
1204       // Start the error message.
1205       OS << *HeaderPaths[ItemKey.File] << ":" << ItemKey.Line << ":"
1206          << ItemKey.Column << "\n";
1207       OS << "#" << getDirectiveSpelling(CondTracker.DirectiveKind) << " "
1208          << *CondTracker.ConditionUnexpanded << "\n";
1209       OS << "^\n";
1210       OS << "error: Conditional expression instance '"
1211          << *CondTracker.ConditionUnexpanded
1212          << "' has different values in this header, depending on how it was "
1213             "included.\n";
1214       // Walk all the instances.
1215       for (auto IMT = CondTracker.ConditionalExpansionInstances.begin(),
1216                 EMT = CondTracker.ConditionalExpansionInstances.end();
1217            IMT != EMT; ++IMT) {
1218         ConditionalExpansionInstance &MacroInfo = *IMT;
1219         OS << "  '" << *CondTracker.ConditionUnexpanded << "' expanded to: '"
1220            << ConditionValueKindStrings[MacroInfo.ConditionValue]
1221            << "' with respect to these inclusion paths:\n";
1222         // Walk all the inclusion path hierarchies.
1223         for (auto IIP = MacroInfo.InclusionPathHandles.begin(),
1224                   EIP = MacroInfo.InclusionPathHandles.end();
1225              IIP != EIP; ++IIP) {
1226           const std::vector<HeaderHandle> &ip = getInclusionPath(*IIP);
1227           auto Count = (int)ip.size();
1228           for (int Index = 0; Index < Count; ++Index) {
1229             HeaderHandle H = ip[Index];
1230             OS << std::string((Index * 2) + 4, ' ') << *getHeaderFilePath(H)
1231                << "\n";
1232           }
1233         }
1234       }
1235     }
1236     return ReturnValue;
1237   }
1238 
1239   // Get directive spelling.
1240   static const char *getDirectiveSpelling(clang::tok::PPKeywordKind kind) {
1241     switch (kind) {
1242     case clang::tok::pp_if:
1243       return "if";
1244     case clang::tok::pp_elif:
1245       return "elif";
1246     case clang::tok::pp_ifdef:
1247       return "ifdef";
1248     case clang::tok::pp_ifndef:
1249       return "ifndef";
1250     default:
1251       return "(unknown)";
1252     }
1253   }
1254 
1255 private:
1256   llvm::SmallVector<std::string, 32> HeaderList;
1257   // Only do extern, namespace check for headers in HeaderList.
1258   bool BlockCheckHeaderListOnly;
1259   llvm::StringPool Strings;
1260   std::vector<StringHandle> HeaderPaths;
1261   std::vector<HeaderHandle> HeaderStack;
1262   std::vector<HeaderInclusionPath> InclusionPaths;
1263   InclusionPathHandle CurrentInclusionPathHandle;
1264   llvm::SmallSet<HeaderHandle, 32> HeadersInThisCompile;
1265   std::vector<PPItemKey> IncludeDirectives;
1266   MacroExpansionMap MacroExpansions;
1267   ConditionalExpansionMap ConditionalExpansions;
1268   bool InNestedHeader;
1269 };
1270 
1271 } // namespace
1272 
1273 // PreprocessorTracker functions.
1274 
1275 // PreprocessorTracker desctructor.
1276 PreprocessorTracker::~PreprocessorTracker() {}
1277 
1278 // Create instance of PreprocessorTracker.
1279 PreprocessorTracker *PreprocessorTracker::create(
1280     llvm::SmallVector<std::string, 32> &Headers,
1281     bool DoBlockCheckHeaderListOnly) {
1282   return new PreprocessorTrackerImpl(Headers, DoBlockCheckHeaderListOnly);
1283 }
1284 
1285 // Preprocessor callbacks for modularize.
1286 
1287 // Handle include directive.
1288 void PreprocessorCallbacks::InclusionDirective(
1289     clang::SourceLocation HashLoc, const clang::Token &IncludeTok,
1290     llvm::StringRef FileName, bool IsAngled,
1291     clang::CharSourceRange FilenameRange, const clang::FileEntry *File,
1292     llvm::StringRef SearchPath, llvm::StringRef RelativePath,
1293     const clang::Module *Imported, clang::SrcMgr::CharacteristicKind FileType) {
1294   int DirectiveLine, DirectiveColumn;
1295   std::string HeaderPath = getSourceLocationFile(PP, HashLoc);
1296   getSourceLocationLineAndColumn(PP, HashLoc, DirectiveLine, DirectiveColumn);
1297   PPTracker.handleIncludeDirective(HeaderPath, DirectiveLine, DirectiveColumn,
1298                                    FileName);
1299 }
1300 
1301 // Handle file entry/exit.
1302 void PreprocessorCallbacks::FileChanged(
1303     clang::SourceLocation Loc, clang::PPCallbacks::FileChangeReason Reason,
1304     clang::SrcMgr::CharacteristicKind FileType, clang::FileID PrevFID) {
1305   switch (Reason) {
1306   case EnterFile:
1307     PPTracker.handleHeaderEntry(PP, getSourceLocationFile(PP, Loc));
1308     break;
1309   case ExitFile: {
1310     const clang::FileEntry *F =
1311         PP.getSourceManager().getFileEntryForID(PrevFID);
1312     if (F)
1313       PPTracker.handleHeaderExit(F->getName());
1314   } break;
1315   case SystemHeaderPragma:
1316   case RenameFile:
1317     break;
1318   }
1319 }
1320 
1321 // Handle macro expansion.
1322 void PreprocessorCallbacks::MacroExpands(const clang::Token &MacroNameTok,
1323                                          const clang::MacroDefinition &MD,
1324                                          clang::SourceRange Range,
1325                                          const clang::MacroArgs *Args) {
1326   clang::SourceLocation Loc = Range.getBegin();
1327   // Ignore macro argument expansions.
1328   if (!Loc.isFileID())
1329     return;
1330   clang::IdentifierInfo *II = MacroNameTok.getIdentifierInfo();
1331   const clang::MacroInfo *MI = MD.getMacroInfo();
1332   std::string MacroName = II->getName().str();
1333   std::string Unexpanded(getMacroUnexpandedString(Range, PP, MacroName, MI));
1334   std::string Expanded(getMacroExpandedString(PP, MacroName, MI, Args));
1335   PPTracker.addMacroExpansionInstance(
1336       PP, PPTracker.getCurrentHeaderHandle(), Loc, MI->getDefinitionLoc(), II,
1337       Unexpanded, Expanded, PPTracker.getCurrentInclusionPathHandle());
1338 }
1339 
1340 void PreprocessorCallbacks::Defined(const clang::Token &MacroNameTok,
1341                                     const clang::MacroDefinition &MD,
1342                                     clang::SourceRange Range) {
1343   clang::SourceLocation Loc(Range.getBegin());
1344   clang::IdentifierInfo *II = MacroNameTok.getIdentifierInfo();
1345   const clang::MacroInfo *MI = MD.getMacroInfo();
1346   std::string MacroName = II->getName().str();
1347   std::string Unexpanded(getSourceString(PP, Range));
1348   PPTracker.addMacroExpansionInstance(
1349       PP, PPTracker.getCurrentHeaderHandle(), Loc,
1350       (MI ? MI->getDefinitionLoc() : Loc), II, Unexpanded,
1351       (MI ? "true" : "false"), PPTracker.getCurrentInclusionPathHandle());
1352 }
1353 
1354 void PreprocessorCallbacks::If(clang::SourceLocation Loc,
1355                                clang::SourceRange ConditionRange,
1356                                clang::PPCallbacks::ConditionValueKind ConditionResult) {
1357   std::string Unexpanded(getSourceString(PP, ConditionRange));
1358   PPTracker.addConditionalExpansionInstance(
1359       PP, PPTracker.getCurrentHeaderHandle(), Loc, clang::tok::pp_if,
1360       ConditionResult, Unexpanded, PPTracker.getCurrentInclusionPathHandle());
1361 }
1362 
1363 void PreprocessorCallbacks::Elif(clang::SourceLocation Loc,
1364                                  clang::SourceRange ConditionRange,
1365                                  clang::PPCallbacks::ConditionValueKind ConditionResult,
1366                                  clang::SourceLocation IfLoc) {
1367   std::string Unexpanded(getSourceString(PP, ConditionRange));
1368   PPTracker.addConditionalExpansionInstance(
1369       PP, PPTracker.getCurrentHeaderHandle(), Loc, clang::tok::pp_elif,
1370       ConditionResult, Unexpanded, PPTracker.getCurrentInclusionPathHandle());
1371 }
1372 
1373 void PreprocessorCallbacks::Ifdef(clang::SourceLocation Loc,
1374                                   const clang::Token &MacroNameTok,
1375                                   const clang::MacroDefinition &MD) {
1376   clang::PPCallbacks::ConditionValueKind IsDefined =
1377     (MD ? clang::PPCallbacks::CVK_True : clang::PPCallbacks::CVK_False );
1378   PPTracker.addConditionalExpansionInstance(
1379       PP, PPTracker.getCurrentHeaderHandle(), Loc, clang::tok::pp_ifdef,
1380       IsDefined, PP.getSpelling(MacroNameTok),
1381       PPTracker.getCurrentInclusionPathHandle());
1382 }
1383 
1384 void PreprocessorCallbacks::Ifndef(clang::SourceLocation Loc,
1385                                    const clang::Token &MacroNameTok,
1386                                    const clang::MacroDefinition &MD) {
1387   clang::PPCallbacks::ConditionValueKind IsNotDefined =
1388     (!MD ? clang::PPCallbacks::CVK_True : clang::PPCallbacks::CVK_False );
1389   PPTracker.addConditionalExpansionInstance(
1390       PP, PPTracker.getCurrentHeaderHandle(), Loc, clang::tok::pp_ifndef,
1391       IsNotDefined, PP.getSpelling(MacroNameTok),
1392       PPTracker.getCurrentInclusionPathHandle());
1393 }
1394 } // end namespace Modularize
1395