| blob | cc122cdcac927e8edb8f2e9c3dd1479451193891 |
1 //===--- PPDirectives.cpp - Directive Handling for Preprocessor -----------===//
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 /// \file
10 /// Implements # directive processing for the Preprocessor.
11 ///
12 //===----------------------------------------------------------------------===//
46 #include <algorithm>
47 #include <cassert>
48 #include <cstring>
49 #include <new>
50 #include <string>
51 #include <utility>
55 //===----------------------------------------------------------------------===//
56 // Utility Methods for Preprocessor Directive Handling.
57 //===----------------------------------------------------------------------===//
63 }
66 SourceLocation Loc) {
68 }
70 UndefMacroDirective *
73 }
75 VisibilityMacroDirective *
79 }
81 /// Read and discard all tokens remaining on the current line until
82 /// the tok::eod token is found.
84 Token Tmp;
85 SourceRange Res;
92 }
95 }
97 /// Enumerates possible cases of #define/#undef a reserved identifier.
101 MD_ReservedMacro //> #define of #undef reserved id, disabled by default
102 };
104 /// Enumerates possible %select values for the pp_err_elif_after_else and
105 /// pp_err_elif_without_if diagnostics.
107 PED_Elif,
108 PED_Elifdef,
109 PED_Elifndef
110 };
112 // The -fmodule-name option tells the compiler to textually include headers in
113 // the specified module, meaning clang won't build the specified module. This is
114 // useful in a number of situations, for instance, when building a library that
115 // vends a module map, one might want to avoid hitting intermediate build
116 // products containimg the module map or avoid finding the system installed
117 // modulemap for that library.
119 StringRef ModuleName) {
122 // When building framework Foo, we wanna make sure that Foo *and* Foo_Private
123 // are textually included and no modules are built for both.
129 }
134 // list from:
135 // - https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_macros.html
136 // - https://docs.microsoft.com/en-us/cpp/c-runtime-library/security-features-in-the-crt?view=msvc-160
137 // - man 7 feature_test_macros
138 // The list must be sorted for correct binary search.
169 };
175 }
182 }
186 // Do not warn on keyword undef. It is generally harmless and widely used.
190 }
192 // Return true if we want to issue a diagnostic by default if we
193 // encounter this name in a #include with the wrong case. For now,
194 // this includes the standard C and C++ headers, Posix headers,
195 // and Boost headers. Improper case for these #includes is a
196 // potential portability issue.
198 // If the first component of the path is "boost", treat this like a standard header
199 // for the purposes of diagnostics.
203 // "condition_variable" is the longest standard header name at 18 characters.
204 // If the include file name is longer than that, it can't be a standard header.
209 // Lowercase and normalize the search string.
212 // In the ASCII range?
215 // ASCII lowercase:
218 // Normalize path separators for comparison purposes.
221 }
223 // The standard C/C++ and Posix headers
225 // C library headers
233 // C++ headers for C library facilities
241 // C++ library headers
254 // POSIX headers (which aren't also C headers)
268 }
270 /// Find a similar string in `Candidates`.
271 ///
272 /// \param LHS a string for a similar string in `Candidates`
273 ///
274 /// \param Candidates the candidates to find a similar string.
275 ///
276 /// \returns a similar string if exists. If no similar string exists,
277 /// returns None.
280 // We need to check if `Candidates` has the exact case-insensitive string
281 // because the Levenshtein distance match does not care about it.
285 }
286 }
288 // Keep going with the Levenshtein distance match.
289 // If the LHS size is less than 3, use the LHS size minus 1 and if not,
290 // use the LHS size divided by 3.
299 // The first similar string found.
302 // More similar string found.
304 }
305 }
306 }
312 }
313 }
317 // Missing macro name?
326 // C++ 2.5p2: Alternative tokens behave the same as its primary token
327 // except for their spellings.
332 // Allow #defining |and| and friends for Microsoft compatibility or
333 // recovery when legacy C headers are included in C++.
334 }
337 // Error if defining "defined": C99 6.10.8/4, C++ [cpp.predefined]p4.
339 }
344 // Warn if undefining "__LINE__" and other builtins, per C99 6.10.8/4
345 // and C++ [cpp.predefined]p4], but allow it as an extension.
347 }
348 }
350 // If defining/undefining reserved identifier or a keyword, we need to issue
351 // a warning.
360 }
364 // We do not want to warn on some patterns widely used in configuration
365 // scripts. This requires analyzing next tokens, so do not issue warnings
366 // now, only inform caller.
369 }
372 }
374 // Okay, we got a good identifier.
376 }
378 /// Lex and validate a macro name, which occurs after a
379 /// \#define or \#undef.
380 ///
381 /// This sets the token kind to eod and discards the rest of the macro line if
382 /// the macro name is invalid.
383 ///
384 /// \param MacroNameTok Token that is expected to be a macro name.
385 /// \param isDefineUndef Context in which macro is used.
386 /// \param ShadowFlag Points to a flag that is set if macro shadows a keyword.
389 // Read the token, don't allow macro expansion on it.
397 }
402 // Invalid macro name, read and discard the rest of the line and set the
403 // token kind to tok::eod if necessary.
407 }
408 }
410 /// Ensure that the next token is a tok::eod token.
411 ///
412 /// If not, emit a diagnostic and consume up until the eod. If EnableMacros is
413 /// true, then we consider macros that expand to zero tokens as being ok.
414 ///
415 /// Returns the location of the end of the directive.
418 Token Tmp;
419 // Lex unexpanded tokens for most directives: macros might expand to zero
420 // tokens, causing us to miss diagnosing invalid lines. Some directives (like
421 // #line) allow empty macros.
424 else
427 // There should be no tokens after the directive, but we allow them as an
428 // extension.
435 // Add a fixit in GNU/C99/C++ mode. Don't offer a fixit for strict-C89,
436 // or if this is a macro-style preprocessing directive, because it is more
437 // trouble than it is worth to insert /**/ and check that there is no /**/
438 // in the range also.
439 FixItHint Hint;
445 }
449 // If this is a `.S` file, treat unknown # directives as non-preprocessor
450 // directives.
455 };
460 // Directive cannot be coming from macro.
469 }
470 }
472 /// SkipExcludedConditionalBlock - We just read a \#if or related directive and
473 /// decided that the subsequent tokens are in the \#if'd out portion of the
474 /// file. Lex the rest of the file, until we see an \#endif. If
475 /// FoundNonSkipPortion is true, then we have already emitted code for part of
476 /// this \#if directive, so \#else/\#elif blocks should never be entered.
477 /// If ElseOk is true, then \#else directives are ok, if not, then we have
478 /// already seen one so a \#else directive is a duplicate. When this returns,
479 /// the caller can lex the first valid token.
481 SourceLocation IfTokenLoc,
484 SourceLocation ElseLoc) {
485 // In SkippingRangeStateTy we are depending on SkipExcludedConditionalBlock()
486 // not getting called recursively by storing the RecordedSkippedRanges
487 // DenseMap lookup pointer (field SkipRangePtr). SkippingRangeStateTy expects
488 // that RecordedSkippedRanges won't get modified and SkipRangePtr won't be
489 // invalidated. If this changes and there is a need to call
490 // SkipExcludedConditionalBlock() recursively, SkippingRangeStateTy should
491 // change to do a second lookup in endLexPass function instead of reusing the
492 // lookup pointer.
503 else
507 // Enter raw mode to disable identifier lookup (and thus macro expansion),
508 // disabling warnings, etc.
510 Token Tok;
511 SourceLocation endLoc;
513 /// Keeps track and caches skipped ranges and also retrieves a prior skipped
514 /// range if the same block is re-visited.
527 // Initiate a skipping block and adjust the lexer if we already skipped it
528 // before.
534 }
535 }
539 // Not doing normal lexing.
542 }
544 // Finished skipping a block, record the range if it's first time visited.
547 }
551 }
567 }
569 // If this is the end of the buffer, we have an error.
571 // We don't emit errors for unterminated conditionals here,
572 // Lexer::LexEndOfFile can do that properly.
573 // Just return and let the caller lex after this #include.
576 FoundNonSkipPortion,
579 }
581 // If this token is not a preprocessor directive, just skip it.
586 }
587 }
591 // We just parsed a # character at the start of a line, so we're in
592 // directive mode. Tell the lexer this so any newlines we see will be
593 // converted into an EOD token (this terminates the macro).
601 // Read the next token, the directive flavor.
604 // If this isn't an identifier directive (e.g. is "# 1\n" or "#\n", or
605 // something bogus), skip it.
608 // Restore comment saving mode.
611 }
613 // If the first letter isn't i or e, it isn't intesting to us. We know that
614 // this is safe in the face of spelling differences, because there is no way
615 // to spell an i/e in a strange way that is another letter. Skipping this
616 // allows us to avoid looking up the identifier info for #define/#undef and
617 // other common directives.
624 // Restore comment saving mode.
627 }
629 // Get the identifier name without trigraphs or embedded newlines. Note
630 // that we can't use Tok.getIdentifierInfo() because its lookup is disabled
631 // when skipping.
633 StringRef Directive;
641 // Restore comment saving mode.
644 }
647 }
654 // We know the entire #if/#ifdef/#ifndef block will be skipped, don't
655 // bother parsing the condition.
662 }
666 PPConditionalInfo CondInfo;
672 // If we popped the outermost skipping block, we're done skipping!
675 // Restore the value of LexingRawMode so that trailing comments
676 // are handled correctly, if we've reached the outermost block.
685 }
687 // #else directive in a skipping conditional. If not in some other
688 // skipping conditional, and if #else hasn't already been seen, enter it
689 // as a non-skipping conditional.
695 // If this is a #else with a #else before it, report the error.
699 // Note that we've seen a #else in this conditional.
702 // If the conditional is at the top level, and the #if block wasn't
703 // entered, enter the #else block now.
706 // Restore the value of LexingRawMode so that trailing comments
707 // are handled correctly.
716 }
723 // If this is a #elif with a #else before it, report the error.
727 // If this is in a skipping block or if we're already handled this #if
728 // block, don't bother parsing the condition.
730 // FIXME: We should probably do at least some minimal parsing of the
731 // condition to verify that it is well-formed. The current state
732 // allows #elif* directives with completely malformed (or missing)
733 // conditions.
736 // Restore the value of LexingRawMode so that identifiers are
737 // looked up, etc, inside the #elif expression.
742 // Stop if Lexer became invalid after hitting code completion token.
752 }
753 // If this condition is true, enter it!
757 }
758 }
768 // Warn if using `#elifdef` & `#elifndef` in not C2x & C++2b mode even
769 // if this branch is in a skipping block.
774 else
779 // If this is a #elif with a #else before it, report the error.
784 // If this is in a skipping block or if we're already handled this #if
785 // block, don't bother parsing the condition.
787 // FIXME: We should probably do at least some minimal parsing of the
788 // condition to verify that it is well-formed. The current state
789 // allows #elif* directives with completely malformed (or missing)
790 // conditions.
793 // Restore the value of LexingRawMode so that identifiers are
794 // looked up, etc, inside the #elif[n]def expression.
797 Token MacroNameTok;
801 // If the macro name token is tok::eod, there was an error that was
802 // already reported.
804 // Skip code until we get to #endif. This helps with recovery by
805 // not emitting an error when the #endif is reached.
807 }
820 MD);
823 MD);
824 }
825 }
826 // If this condition is true, enter it!
830 }
831 }
834 }
837 }
840 // Restore comment saving mode.
842 }
844 // Finally, if we are out of the conditional (saw an #endif or ran off the end
845 // of the file, just stop skipping and return to lexing whatever came after
846 // the #if block.
849 // The last skipped range isn't actually skipped yet if it's truncated
850 // by the end of the preamble; we'll resume parsing after the preamble.
854 ? endLoc
857 }
861 // Try to determine the module of the include directive.
862 // FIXME: Look into directly passing the FileEntry from LookupFile instead.
865 // The include comes from an included file.
869 }
870 }
872 // This is either in the main file or not in a file at all. It belongs
873 // to the current module, if there is one.
877 }
881 SourceLocation Loc) {
884 // Walk up through the include stack, looking through textual headers of M
885 // until we hit a non-textual header that we can #include. (We assume textual
886 // headers of a module with non-textual headers aren't meant to be used to
887 // import entities from the module.)
895 // We want to find all possible modules that might contain this header, so
896 // search all enclosing directories for module maps and load them.
903 // It's in a private header; we can't #include it.
904 // FIXME: If there's a public header in some module that re-exports it,
905 // then we could suggest including that, but it's not clear that's the
906 // expected way to make this entity visible.
909 }
911 // We'll suggest including textual headers below if they're
912 // include-guarded.
916 // If we have a module import syntax, we shouldn't include a header to
917 // make a particular module visible. Let the caller know they should
918 // suggest an import instead.
923 // If this is an accessible, non-textual header of M's top-level module
924 // that transitively includes the given location and makes the
925 // corresponding module visible, this is the thing to #include.
927 }
929 // FIXME: If we're bailing out due to a private header, we shouldn't suggest
930 // an import either.
934 // If the header is includable and has an include guard, assume the
935 // intended way to expose its contents is by #include, not by importing a
936 // module that transitively includes it.
941 }
944 }
959 // If the header lookup mechanism may be relative to the current inclusion
960 // stack, record the parent #includes.
962 Includers;
968 // If there is no file entry associated with this file, it must be the
969 // predefines buffer or the module includes buffer. Any other file is not
970 // lexed with a normal lexer, so it won't be scanned for preprocessor
971 // directives.
972 //
973 // If we have the predefines buffer, resolve #include references (which come
974 // from the -include command line argument) from the current working
975 // directory instead of relative to the main file.
976 //
977 // If we have the module includes buffer, resolve #include references (which
978 // come from header declarations in the module map) relative to the module
979 // map file.
989 }
991 // MSVC searches the current include stack from top to bottom for
992 // headers included by quoted include directives.
993 // See: http://msdn.microsoft.com/en-us/library/36k2cdd4.aspx
999 }
1000 }
1001 }
1006 // We're supposed to start looking from after a particular file. Search
1007 // the include path until we find that file or run out of files.
1015 // Keep looking as if this file did a #include_next.
1019 // Found it.
1023 }
1024 }
1025 }
1027 // Do a standard file entry lookup.
1038 }
1041 // Otherwise, see if this is a subframework header. If so, this is relative
1042 // to one of the headers on the #include stack. Walk the list of the current
1043 // headers on the #include stack and pass them to HeaderInfo.
1048 SuggestedModule)) {
1054 }
1055 }
1056 }
1069 }
1070 }
1071 }
1072 }
1074 // Otherwise, we really couldn't find the file.
1076 }
1078 //===----------------------------------------------------------------------===//
1079 // Preprocessor Directive Handling.
1080 //===----------------------------------------------------------------------===//
1088 }
1092 }
1097 };
1099 /// Process a directive while looking for the through header or a #pragma
1100 /// hdrstop. The following directives are handled:
1101 /// #include (to check if it is the through header)
1102 /// #define (to warn about macros that don't match the PCH)
1103 /// #pragma (to check for pragma hdrstop).
1104 /// All other directives are completely discarded.
1106 SourceLocation HashLoc) {
1111 }
1115 }
1121 }
1122 }
1124 }
1126 /// HandleDirective - This callback is invoked when the lexer sees a # token
1127 /// at the start of a line. This consumes the directive, modifies the
1128 /// lexer/preprocessor state, and advances the lexer(s) so that the next token
1129 /// read is the correct one.
1131 // FIXME: Traditional: # with whitespace before it not recognized by K&R?
1133 // We just parsed a # character at the start of a line, so we're in directive
1134 // mode. Tell the lexer this so any newlines we see will be converted into an
1135 // EOD token (which terminates the directive).
1145 // We are about to read a token. For the multiple-include optimization FA to
1146 // work, we have to remember if we had read any tokens *before* this
1147 // pp-directive.
1150 // Save the '#' token in case we need to return it later.
1153 // Read the next token, the directive flavor. This isn't expanded due to
1154 // C99 6.10.3p8.
1157 // C99 6.10.3p11: Is this preprocessor directive in macro invocation? e.g.:
1158 // #define A(x) #x
1159 // A(abc
1160 // #warning blah
1161 // def)
1162 // If so, the user is relying on undefined behavior, emit a diagnostic. Do
1163 // not support this for #include-like directives, since that can result in
1164 // terrible diagnostics, and does not work in GCC.
1180 }
1181 }
1183 }
1185 // Temporarily enable macro expansion if set so
1186 // and reset to previous state when returning from this function.
1209 // Ask what the preprocessor keyword ID is.
1212 // C99 6.10.1 - Conditional Inclusion.
1220 ReadAnyTokensBeforeDirective);
1231 // C99 6.10.2 - Source File Inclusion.
1233 // Handle #include.
1236 // Handle -imacros.
1239 // C99 6.10.3 - Macro Replacement.
1245 // C99 6.10.4 - Line Control.
1249 // C99 6.10.5 - Error Directive.
1253 // C99 6.10.6 - Pragma Directive.
1257 // GNU Extensions.
1269 else
1280 //isExtension = true; // FIXME: implement #assert
1283 //isExtension = true; // FIXME: implement #unassert
1295 }
1297 }
1299 // If this is a .S file, treat unknown # directives as non-preprocessor
1300 // directives. This is important because # may be a comment or introduce
1301 // various pseudo-ops. Just return the # token and push back the following
1302 // token to be lexed next time.
1305 // Return the # and the token after it.
1309 // If the second token is a hashhash token, then we need to translate it to
1310 // unknown so the token lexer doesn't try to perform token pasting.
1314 // Enter this token stream so that we re-lex the tokens. Make sure to
1315 // enable macro expansion, in case the token after the # is an identifier
1316 // that is expanded.
1319 }
1321 // If we reached here, the preprocessing token is not valid!
1322 // Start suggesting if a similar directive found.
1325 // Read the rest of the PP line.
1328 // Okay, we're done parsing the directive.
1329 }
1331 /// GetLineValue - Convert a numeric token into an unsigned value, emitting
1332 /// Diagnostic DiagID if it is invalid, and returning the value in Val.
1342 }
1352 // Verify that we have a simple digit-sequence, and compute the value. This
1353 // is always a simple digit string computed in decimal, so we do this manually
1354 // here.
1357 // C++1y [lex.fcon]p1:
1358 // Optional separating single quotes in a digit-sequence are ignored
1367 }
1374 }
1376 }
1383 }
1385 /// Handle a \#line directive: C99 6.10.4.
1386 ///
1387 /// The two acceptable forms are:
1388 /// \verbatim
1389 /// # line digit-sequence
1390 /// # line digit-sequence "s-char-sequence"
1391 /// \endverbatim
1393 // Read the line # and string argument. Per C99 6.10.4p5, these tokens are
1394 // expanded.
1395 Token DigitTok;
1398 // Validate the number and convert it to an unsigned.
1406 // Enforce C99 6.10.4p3: "The digit sequence shall not specify ... a
1407 // number greater than 2147483647". C90 requires that the line # be <= 32767.
1417 Token StrTok;
1420 // If the StrTok is "eod", then it wasn't present. Otherwise, it must be a
1421 // string followed by eod.
1433 // Parse and validate the string, converting it into a unique ID.
1439 }
1444 }
1447 // Verify that there is nothing after the string, other than EOD. Because
1448 // of C99 6.10.4p5, macros that expand to empty tokens are ok.
1450 }
1452 // Take the file kind of the file containing the #line directive. #line
1453 // directives are often used for generated sources from the same codebase, so
1454 // the new file should generally be classified the same way as the current
1455 // file. This is visible in GCC's pre-processed output, which rewrites #line
1456 // to GNU line markers.
1466 }
1468 /// ReadLineMarkerFlags - Parse and validate any flags at the end of a GNU line
1469 /// marker directive.
1474 Token FlagTok;
1491 // If we are leaving the current presumed file, check to make sure the
1492 // presumed include stack isn't empty!
1493 FileID CurFileID =
1499 // If there is no include loc (main file) or if the include loc is in a
1500 // different physical file, then we aren't in a "1" line marker flag region.
1507 }
1513 }
1515 // We must have 3 if there are still flags.
1520 }
1529 // We must have 4 if there is yet another flag.
1534 }
1541 // There are no more valid flags here.
1545 }
1547 /// HandleDigitDirective - Handle a GNU line marker directive, whose syntax is
1548 /// one of the following forms:
1549 ///
1550 /// # 42
1551 /// # 42 "file" ('1' | '2')?
1552 /// # 42 "file" ('1' | '2')? '3' '4'?
1553 ///
1555 // Validate the number and convert it to an unsigned. GNU does not have a
1556 // line # limit other than it fit in 32-bits.
1562 Token StrTok;
1569 // If the StrTok is "eod", then it wasn't present. Otherwise, it must be a
1570 // string followed by eod.
1573 // Treat this like "#line NN", which doesn't change file characteristics.
1584 // Parse and validate the string, converting it into a unique ID.
1590 }
1595 }
1597 // If a filename was present, read any flags that are present.
1604 // Exiting to an empty string means pop to the including file, so leave
1605 // FilenameID as -1 in that case.
1608 }
1610 // Create a line note with this information.
1614 // If the preprocessor has callbacks installed, notify them of the #line
1615 // change. This is used so that the line marker comes out in -E mode for
1616 // example.
1625 }
1626 }
1628 /// HandleUserDiagnosticDirective - Handle a #warning or #error directive.
1629 ///
1632 // Read the rest of the line raw. We do this because we don't want macros
1633 // to be expanded and we don't require that the tokens be valid preprocessing
1634 // tokens. For example, this is allowed: "#warning ` 'foo". GCC does
1635 // collapse multiple consecutive white space between tokens, but this isn't
1636 // specified by the standard.
1640 // Find the first non-whitespace character, so that we can make the
1641 // diagnostic more succinct.
1646 else
1648 }
1650 /// HandleIdentSCCSDirective - Handle a #ident/#sccs directive.
1651 ///
1653 // Yes, this directive is an extension.
1656 // Read the string argument.
1657 Token StrTok;
1660 // If the token kind isn't a string, it's a malformed directive.
1667 }
1673 }
1675 // Verify that there is nothing after the string, other than EOD.
1683 }
1684 }
1686 /// Handle a #public directive.
1688 Token MacroNameTok;
1691 // Error reading macro name? If so, diagnostic already issued.
1695 // Check to see if this is the last token on the #__public_macro line.
1699 // Okay, we finally have a valid identifier to undef.
1702 // If the macro is not defined, this is an error.
1706 }
1708 // Note that this macro has now been exported.
1711 }
1713 /// Handle a #private directive.
1715 Token MacroNameTok;
1718 // Error reading macro name? If so, diagnostic already issued.
1722 // Check to see if this is the last token on the #__private_macro line.
1726 // Okay, we finally have a valid identifier to undef.
1729 // If the macro is not defined, this is an error.
1733 }
1735 // Note that this macro has now been marked private.
1738 }
1740 //===----------------------------------------------------------------------===//
1741 // Preprocessor Include Directive Handling.
1742 //===----------------------------------------------------------------------===//
1744 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully
1745 /// checked and spelled filename, e.g. as an operand of \#include. This returns
1746 /// true if the input filename was in <>'s or false if it were in ""'s. The
1747 /// caller is expected to provide a buffer that is large enough to hold the
1748 /// spelling of the filename, but is also expected to handle the case when
1749 /// this method decides to use a different buffer.
1752 // Get the text form of the filename.
1755 // FIXME: Consider warning on some of the cases described in C11 6.4.7/3 and
1756 // C++20 [lex.header]/2:
1757 //
1758 // If `"`, `'`, `\`, `/*`, or `//` appears in a header-name, then
1759 // in C: behavior is undefined
1760 // in C++: program is conditionally-supported with implementation-defined
1761 // semantics
1763 // Make sure the filename is <x> or "x".
1770 }
1777 }
1783 }
1785 // Diagnose #include "" as invalid.
1790 }
1792 // Skip the brackets.
1795 }
1797 /// Push a token onto the token stream containing an annotation.
1801 // FIXME: Produce this as the current token directly, rather than
1802 // allocating a new token for it.
1810 }
1812 /// Produce a diagnostic informing the user that a #include or similar
1813 /// was implicitly treated as a module import.
1817 SourceLocation PathEnd) {
1823 }
1845 }
1849 }
1851 // Given a vector of path components and a string containing the real
1852 // path to the file, build a properly-cased replacement in the vector,
1853 // and return true if the replacement should be suggested.
1855 StringRef RealPathName) {
1860 // Below is a best-effort to handle ".." in paths. It is admittedly
1861 // not 100% correct in the presence of symlinks.
1870 // If these path components differ by more than just case, then we
1871 // may be looking at symlinked paths. Bail on this diagnostic to avoid
1872 // noisy false positives.
1873 SuggestReplacement =
1878 }
1880 }
1881 }
1883 }
1892 ShadowingModule))
1903 // FIXME: Track the location at which the requirement was specified, and
1904 // use it here.
1907 }
1909 }
1913 // #include_next is like #include, except that we start searching after
1914 // the current found directory. If we can't do this, issue a
1915 // diagnostic.
1920 // If the main file is a header, then it's either for PCH/AST generation,
1921 // or libclang opened it. Either way, handle it as a normal include below
1922 // and do not complain about include_next.
1927 // Start looking up in the directory *after* the one in which the current
1928 // file would be found, if any.
1933 // The current file was not found by walking the include path. Either it
1934 // is the primary file (handled above), or it was found by absolute path,
1935 // or it was found relative to such a file.
1936 // FIXME: Track enough information so we know which case we're in.
1939 // Start looking up in the next directory.
1941 }
1944 }
1946 /// HandleIncludeDirective - The "\#include" tokens have just been read, read
1947 /// the file to be included from the lexer, then include it! This is a common
1948 /// routine with functionality shared between \#include, \#include_next and
1949 /// \#import. LookupFrom is set when this is a \#include_next directive, it
1950 /// specifies the file to start searching from.
1953 ConstSearchDirIterator LookupFrom,
1955 Token FilenameTok;
1964 }
1966 // Verify that there is nothing after the filename, other than EOD. Note
1967 // that we allow macros that expand to nothing after the filename, because
1968 // this falls into the category of "#include pp-tokens new-line" specified
1969 // in C99 6.10.2p4.
1970 SourceLocation EndLoc =
1998 }
1999 }
2020 // If the file could not be located and it was included via angle
2021 // brackets, we can attempt a lookup as though it were a quoted path to
2022 // provide the user with a possible fixit.
2036 }
2037 }
2039 // Check for likely typos due to leading or trailing non-isAlphanumeric
2040 // characters
2043 // A heuristic to correct a typo file name by removing leading and
2044 // trailing non-isAlphanumeric characters.
2049 }
2051 };
2068 // We found the file, so set the Filename to the name after typo
2069 // correction.
2073 }
2074 }
2076 // If the file is still not found, just go with the vanilla diagnostic
2091 }
2094 }
2096 /// Handle either a #include-like directive or an import declaration that names
2097 /// a header file.
2098 ///
2099 /// \param HashLoc The location of the '#' token for an include, or
2100 /// SourceLocation() for an import declaration.
2101 /// \param IncludeTok The include / include_next / import token.
2102 /// \param FilenameTok The header-name token.
2103 /// \param EndLoc The location at which any imported macros become visible.
2104 /// \param LookupFrom For #include_next, the starting directory for the
2105 /// directory lookup.
2106 /// \param LookupFromFile For #include_next, the starting file for the directory
2107 /// lookup.
2116 CharSourceRange FilenameRange
2122 // If GetIncludeFilenameSpelling set the start ptr to null, there was an
2123 // error.
2130 // Complain about attempts to #include files in an audit pragma.
2135 // Immediately leave the pragma.
2137 }
2139 // Complain about attempts to #include files in an assume-nonnull pragma.
2144 // Immediately leave the pragma.
2146 }
2149 // Map the filename with the brackets still attached. If the name doesn't
2150 // map to anything, fall back on the filename we've already gotten the
2151 // spelling for.
2155 }
2157 // Search include directories.
2163 // We get the raw path only if we have 'Callbacks' to which we later pass
2164 // the path.
2169 // Normalize slashes when compiling with -fms-extensions on non-Windows. This
2170 // is unnecessary on Windows since the filesystem there handles backslashes.
2178 }
2189 }
2191 // Should we enter the source file? Set to Skip if either the source file is
2192 // known to have no effect beyond its effect on module visibility -- that is,
2193 // if it's got an include guard that is already defined, set to Import if it
2194 // is a modular header we've already built and should import.
2196 // For C++20 Modules
2197 // [cpp.include]/7 If the header identified by the header-name denotes an
2198 // importable header, it is implementation-defined whether the #include
2199 // preprocessing directive is instead replaced by an import directive.
2200 // For this implementation, the translation is permitted when we are parsing
2201 // the Global Module Fragment, and not otherwise (the cases where it would be
2202 // valid to replace an include with an import are highly constrained once in
2203 // named module purview; this choice avoids considerable complexity in
2204 // determining valid cases).
2211 // If we've reached the max allowed include depth, it is usually due to an
2212 // include cycle. Don't enter already processed files again as it can lead to
2213 // reaching the max allowed include depth again.
2223 // FIXME: We do not have a good way to disambiguate C++ clang modules from
2224 // C++ standard modules (other than use/non-use of Header Units).
2226 // Maybe a usable Header Unit
2232 // This is a Header Unit that we do not include-translate
2235 }
2236 }
2237 // Maybe a usable clang header module.
2242 // Determine whether we should try to import the module for this #include, if
2243 // there is one. Don't do so if precompiled module support is disabled or we
2244 // are processing this module textually (because we're building the module).
2246 // If this include corresponds to a module but that module is
2247 // unavailable, diagnose the situation and bail out.
2248 // FIXME: Remove this; loadModule does the same check (but produces
2249 // slightly worse diagnostics).
2256 }
2258 // Compute the module access path corresponding to this module.
2259 // FIXME: Should we have a second loadModule() overload to avoid this
2260 // extra lookup step?
2267 // Warn that we're replacing the include/import with a module import.
2271 // Load the module to import its macros. We'll make the declarations
2272 // visible when the parser gets here.
2273 // FIXME: Pass SuggestedModule in here rather than converting it to a path
2274 // and making the module loader convert it back again.
2284 // We failed to find a submodule that we assumed would exist (because it
2285 // was in the directory of an umbrella header, for instance), but no
2286 // actual module containing it exists (because the umbrella header is
2287 // incomplete). Treat this as a textual inclusion.
2291 // On a configuration mismatch, enter the header textually. We still know
2292 // that it's part of the corresponding module.
2294 // We hit an error processing the import. Bail out.
2296 // With a fatal failure in the module loader, we abort parsing.
2302 }
2304 }
2305 }
2307 // The #included file will be considered to be a system header if either it is
2308 // in a system include directory, or if the #includer is a system include
2309 // header.
2314 FileCharacter);
2316 // If this is a '#import' or an import-declaration, don't re-enter the file.
2317 //
2318 // FIXME: If we have a suggested module for a '#include', and we've already
2319 // visited this file, don't bother entering it again. We know it has no
2320 // further effect.
2322 IsImportDecl ||
2327 // Ask HeaderInfo if we should enter this #include file. If not, #including
2328 // this file will have no effect.
2332 IsFirstIncludeOfFile)) {
2333 // C++ standard modules:
2334 // If we are not in the GMF, then we textually include only
2335 // clang modules:
2336 // Even if we've already preprocessed this header once and know that we
2337 // don't need to see its contents again, we still need to import it if it's
2338 // modular because we might not have imported it from this submodule before.
2339 //
2340 // FIXME: We don't do this when compiling a PCH because the AST
2341 // serialization layer can't cope with it. This means we get local
2342 // submodule visibility semantics wrong in that case.
2345 else
2347 }
2349 // Check for circular inclusion of the main file.
2350 // We can't generate a consistent preamble with regard to the conditional
2351 // stack if the main file is included again as due to the preamble bounds
2352 // some directives (e.g. #endif of a header guard) will never be seen.
2353 // Since this will lead to confusing errors, avoid the inclusion.
2359 }
2362 // Notify the callback object that we've seen an inclusion directive.
2363 // FIXME: Use a different callback for a pp-import?
2368 FileCharacter);
2371 }
2376 // If this is a C++20 pp-import declaration, diagnose if we didn't find any
2377 // module corresponding to the named header.
2382 }
2384 // Issue a diagnostic if the name of the file on disk has a different case
2385 // than the one we're about to open.
2392 #if defined(_WIN32)
2393 // Skip UNC prefix if present. (tryGetRealPathName() always
2394 // returns a path with the prefix skipped.)
2397 #endif
2401 #if defined(_WIN32)
2402 // -Wnonportable-include-path is designed to diagnose includes using
2403 // case even on systems with a case-insensitive file system.
2404 // On Windows, RealPathName always starts with an upper-case drive
2405 // letter for absolute paths, but Name might start with either
2406 // case depending on if `cd c:\foo` or `cd C:\foo` was used in the shell.
2407 // ("foo" will always have on-disk case, no matter which case was
2408 // used in the cd command). To not emit this warning solely for
2409 // the drive letter, whose case is dependent on if `cd` is used
2410 // with upper- or lower-case drive letters, always consider the
2411 // given drive letter case as correct for the purpose of this warning.
2422 }
2423 #endif
2432 };
2435 // On POSIX, Components will contain a single '/' as first element
2436 // exactly if Name is an absolute path.
2437 // On Windows, it will contain "C:" followed by '\' for absolute paths.
2438 // The drive letter is optional for absolute paths on Windows, but
2439 // clang currently cannot process absolute paths in #include lines that
2440 // don't have a drive.
2441 // If the first entry in Components is a directory separator,
2442 // then the code at the bottom of this loop that keeps the original
2443 // directory separator style copies it. If the second entry is
2444 // a directory separator (the C:\ case), then that separator already
2445 // got copied when the C: was processed and we want to skip that entry.
2451 // Append the separator(s) the user used, or the close quote
2455 }
2457 do
2461 }
2463 #if defined(_WIN32)
2464 // Restore UNC prefix if it was there.
2467 #endif
2469 // For user files and known standard headers, issue a diagnostic.
2470 // For other system headers, don't. They can be controlled separately.
2477 }
2478 }
2482 // If we don't need to enter the file, stop now.
2488 // If we reached our include limit and don't want to enter any more files,
2489 // don't go any further.
2493 // If this is a module import, make it visible if needed.
2503 }
2507 }
2509 // Check that we don't have infinite #include recursion.
2514 }
2516 // Look up the file, create a File ID for it.
2518 // If the filename string was the result of macro expansions, set the include
2519 // position on the file where it will be included and after the expansions.
2526 }
2528 // If all is good, enter the new file!
2530 IsFirstIncludeOfFile))
2533 // Determine if we're switching to building a new submodule, and which one.
2534 // This does not apply for C++20 modules header units.
2537 // We are building a submodule that belongs to a shadowed module. This
2538 // means we find header files in the shadowed module.
2544 }
2545 // When building a pch, -fmodule-name tells the compiler to textually
2546 // include headers in the specified module. We are not building the
2547 // specified module.
2548 //
2549 // FIXME: This is the wrong way to handle this. We should produce a PCH
2550 // that behaves the same as the header would behave in a compilation using
2551 // that PCH, which means we should enter the submodule. We need to teach
2552 // the AST serialization layer to deal with the resulting AST.
2561 // Let the macro handling code know that any future macros are within
2562 // the new submodule.
2565 // Let the parser know that any future declarations are within the new
2566 // submodule.
2567 // FIXME: There's no point doing this if we're handling a #__include_macros
2568 // directive.
2570 }
2574 }
2576 /// HandleIncludeNextDirective - Implements \#include_next.
2577 ///
2587 LookupFromFile);
2588 }
2590 /// HandleMicrosoftImportDirective - Implements \#import for Microsoft Mode
2592 // The Microsoft #import directive takes a type library and generates header
2593 // files from it, and includes those. This is beyond the scope of what clang
2594 // does, so we ignore it and error out. However, #import can optionally have
2595 // trailing attributes that span multiple lines. We're going to eat those
2596 // so we can continue processing from there.
2599 // Read tokens until we get to the end of the directive. Note that the
2600 // directive can be split over multiple lines using the backslash character.
2602 }
2604 /// HandleImportDirective - Implements \#import.
2605 ///
2612 }
2614 }
2616 /// HandleIncludeMacrosDirective - The -imacros command line option turns into a
2617 /// pseudo directive in the predefines buffer. This handles it by sucking all
2618 /// tokens through the preprocessor and discarding them (only keeping the side
2619 /// effects on the preprocessor).
2622 // This directive should only occur in the predefines buffer. If not, emit an
2623 // error and reject it.
2630 }
2632 // Treat this as a normal #include for checking purposes. If this is
2633 // successful, it will push a new lexer onto the include stack.
2636 Token TmpTok;
2641 }
2643 //===----------------------------------------------------------------------===//
2644 // Preprocessor Macro Directive Handling.
2645 //===----------------------------------------------------------------------===//
2647 /// ReadMacroParameterList - The ( starting a parameter list of a macro
2648 /// definition has just been read. Lex the rest of the parameters and the
2649 /// closing ), updating MI with what we learn. Return true if an error occurs
2650 /// parsing the param list.
2658 // Found the end of the parameter list.
2661 // Otherwise we have #define FOO(A,)
2670 // OpenCL v1.2 s6.9.e: variadic macros are not supported.
2673 }
2675 // Lex the token after the identifier.
2680 }
2681 // Add the __VA_ARGS__ identifier as a parameter.
2690 // Handle keywords and identifiers here to accept things like
2691 // #define Foo(for) for.
2694 // #define X(1
2697 }
2699 // If this is already used as a parameter, it is used multiple times (e.g.
2700 // #define X(A,A.
2704 }
2706 // Add the parameter to the macro info.
2709 // Lex the token after the identifier.
2722 // Diagnose extension.
2725 // Lex the token after the identifier.
2730 }
2735 }
2736 }
2737 }
2738 }
2745 // Macro that is identity, like '#define inline inline' is a valid pattern.
2749 // Macro that maps a keyword to the same keyword decorated with leading/
2750 // trailing underscores is a valid pattern:
2751 // #define inline __inline
2752 // #define inline __inline__
2753 // #define inline _inline (in MS compatibility mode)
2763 else
2769 }
2773 }
2774 }
2776 // #define inline
2780 }
2782 // ReadOptionalMacroParameterListAndBody - This consumes all (i.e. the
2783 // entire line) of the macro's tokens and adds them to MacroInfo, and while
2784 // doing so performs certain validity checks including (but not limited to):
2785 // - # (stringization) is followed by a macro parameter
2786 //
2787 // Returns a nullptr if an invalid sequence of tokens is encountered or returns
2788 // a pointer to a MacroInfo object.
2794 // Create the new macro.
2797 Token Tok;
2800 // Ensure we consume the rest of the macro body if errors occur.
2802 // The flag indicates if we are still waiting for 'eod'.
2805 });
2807 // Used to un-poison and then re-poison identifiers of the __VA_ARGS__ ilk
2808 // within their appropriate context.
2811 // If this is a function-like macro definition, parse the argument list,
2812 // marking each of the identifiers as being used as macro arguments. Also,
2813 // check other constraints on the first token of the macro body.
2816 // Save this macro information since it may part of a header guard.
2819 }
2820 // If there is no body to this macro, we have no special handling here.
2822 // This is a normal token with leading space. Clear the leading space
2823 // marker on the first token to get proper expansion.
2826 // This is a function-like macro definition. Read the argument list.
2831 // If this is a definition of an ISO C/C++ variadic function-like macro (not
2832 // using the GNU named varargs extension) inform our variadic scope guard
2833 // which un-poisons and re-poisons certain identifiers (e.g. __VA_ARGS__)
2834 // allowed only within the definition of a variadic macro.
2838 }
2840 // Read the first token after the arg list for down below.
2843 // C99 requires whitespace between the macro definition and the body. Emit
2844 // a diagnostic for something like "#define X+".
2847 // C90 6.8 TC1 says: "In the definition of an object-like macro, if the
2848 // first character of a replacement list is not a character required by
2849 // subclause 5.2.1, then there shall be white-space separation between the
2850 // identifier and the replacement list.". 5.2.1 lists this set:
2851 // "A-Za-z0-9!"#%&'()*+,_./:;<=>?[\]^_{|}~" as well as whitespace, which
2852 // is irrelevant here.
2857 // If we have an unknown token, it is something strange like "`". Since
2858 // all of valid characters would have lexed into a single character
2859 // token of some sort, we know this is not a valid case.
2861 }
2864 else
2866 }
2873 // Read the rest of the macro body.
2875 // Object-like macros are very simple, just read their body.
2879 // Get the next token of the macro.
2881 }
2883 // Otherwise, read the body of a function-like macro. While we are at it,
2884 // check C99 6.10.3.2p1: ensure that # operators are followed by macro
2885 // parameters in function-like macro expansions.
2896 // If we're already within a VAOPT, emit an error.
2900 }
2901 // Ensure VAOPT is followed by a '(' .
2906 }
2913 }
2919 "Must have seen at least __VA_OPT__( "
2924 }
2925 }
2928 }
2929 }
2930 // Get the next token of the macro.
2933 }
2935 // If we're in -traditional mode, then we should ignore stringification
2936 // and token pasting. Mark the tokens as unknown so as not to confuse
2937 // things.
2942 // Get the next token of the macro.
2945 }
2948 // If we see token pasting, check if it looks like the gcc comma
2949 // pasting extension. We'll use this information to suppress
2950 // diagnostics later on.
2952 // Get the next token of the macro.
2958 }
2964 // Things look ok, add the '##' token to the macro.
2967 }
2969 // Our Token is a stringization operator.
2970 // Get the next token of the macro.
2973 // Check for a valid macro arg identifier or __VA_OPT__.
2978 // If this is assembler-with-cpp mode, we accept random gibberish after
2979 // the '#' because '#' is often a comment character. However, change
2980 // the kind of the token to tok::unknown so that the preprocessor isn't
2981 // confused.
2990 }
2991 }
2993 // Things look ok, add the '#' and param name tokens to the macro.
2996 // If the token following '#' is VAOPT, let the next iteration handle it
2997 // and check it for correctness, otherwise add the token and prime the
2998 // loop with the next one.
3003 // Get the next token of the macro.
3005 }
3006 }
3013 }
3014 }
3019 }
3020 /// HandleDefineDirective - Implements \#define. This consumes the entire macro
3021 /// line then lets the caller lex the next real token.
3026 Token MacroNameTok;
3030 // Error reading macro name? If so, diagnostic already issued.
3035 // Issue a final pragma warning if we're defining a macro that was has been
3036 // undefined and is being redefined.
3040 // If we are supposed to keep comments in #defines, reenable comment saving
3041 // mode.
3052 }
3053 // Check that there is no paste (##) operator at the beginning or end of the
3054 // replacement list.
3060 }
3064 }
3065 }
3067 // When skipping just warn about macros that do not match.
3074 // Issue the diagnostic but allow the change if msvc extensions are enabled
3077 }
3079 // Finally, if this identifier already had a macro defined for it, verify that
3080 // the macro bodies are identical, and issue diagnostics if they are not.
3082 // Final macros are hard-mode: they always warn. Even if the bodies are
3083 // identical. Even if they are in system headers. Even if they are things we
3084 // would silently allow in the past.
3088 // In Objective-C, ignore attempts to directly redefine the builtin
3089 // definitions of the ownership qualifiers. It's still possible to
3090 // #undef them.
3096 };
3101 // Warn if it changes the tokens.
3107 }
3110 }
3112 // It is very common for system headers to have tons of macro redefinitions
3113 // and for warnings to be disabled in system headers. If this is the case,
3114 // then don't bother calling MacroInfo::isIdenticalTo.
3121 // Warn if defining "__LINE__" and other builtins, per C99 6.10.8/4 and
3122 // C++ [cpp.predefined]p4, but allow it as an extension.
3125 // Macros must be identical. This means all tokens and whitespace
3126 // separation must be the same. C99 6.10.3p2.
3132 }
3133 }
3136 }
3142 // If we need warning for not using the macro, add its location in the
3143 // warn-because-unused-macro set. If it gets used it will be removed from set.
3151 }
3153 // If the callbacks want to know, tell them about the macro definition.
3157 // If we're in MS compatibility mode and the macro being defined is the
3158 // assert macro, implicitly add a macro definition for static_assert to work
3159 // around their broken assert.h header file in C. Only do so if there isn't
3160 // already a static_assert macro defined.
3166 Token Tok;
3172 }
3173 }
3175 /// HandleUndefDirective - Implements \#undef.
3176 ///
3180 Token MacroNameTok;
3183 // Error reading macro name? If so, diagnostic already issued.
3187 // Check to see if this is the last token on the #undef line.
3190 // Okay, we have a valid identifier to undef.
3198 // If the macro is not defined, this is a noop undef.
3207 }
3209 // If the callbacks want to know, tell them about the macro #undef.
3210 // Note: no matter if the macro was defined or not.
3216 }
3218 //===----------------------------------------------------------------------===//
3219 // Preprocessor Conditional Directive Handling.
3220 //===----------------------------------------------------------------------===//
3222 /// HandleIfdefDirective - Implements the \#ifdef/\#ifndef directive. isIfndef
3223 /// is true when this is a \#ifndef directive. ReadAnyTokensBeforeDirective is
3224 /// true if any tokens have been returned or pp-directives activated before this
3225 /// \#ifndef has been lexed.
3226 ///
3234 Token MacroNameTok;
3237 // Error reading macro name? If so, diagnostic already issued.
3239 // Skip code until we get to #endif. This helps with recovery by not
3240 // emitting an error when the #endif is reached.
3245 }
3249 // Check to see if this is the last token on the #if[n]def line.
3257 // If the start of a top-level #ifdef and if the macro is not defined,
3258 // inform MIOpt that this might be the start of a proper include guard.
3259 // Otherwise it is some other form of unknown conditional which we can't
3260 // handle.
3266 }
3268 // If there is a macro, process it.
3275 else
3277 }
3282 // Should we include the stuff contained by this directive?
3284 // In 'single-file-parse mode' undefined identifiers trigger parsing of all
3285 // the directive blocks.
3290 // Yes, remember that we are inside a conditional, then lex the next token.
3295 // No, skip the contents of this block.
3300 }
3301 }
3303 /// HandleIfDirective - Implements the \#if directive.
3304 ///
3310 // Parse and evaluate the conditional expression.
3314 // Lexer might become invalid if we hit code completion point while evaluating
3315 // expression.
3319 // If this condition is equivalent to #ifndef X, and if this is the first
3320 // directive seen, handle it for the multiple-include optimization.
3323 // FIXME: Pass in the location of the macro name, not the 'if' token.
3325 else
3327 }
3337 // Should we include the stuff contained by this directive?
3339 // In 'single-file-parse mode' undefined identifiers trigger parsing of all
3340 // the directive blocks.
3344 // Yes, remember that we are inside a conditional, then lex the next token.
3348 // No, skip the contents of this block.
3352 }
3353 }
3355 /// HandleEndifDirective - Implements the \#endif directive.
3356 ///
3360 // Check that this is the whole directive.
3363 PPConditionalInfo CondInfo;
3365 // No conditionals on the stack: this is an #endif without an #if.
3368 }
3370 // If this the end of a top-level #endif, inform MIOpt.
3379 }
3381 /// HandleElseDirective - Implements the \#else directive.
3382 ///
3386 // #else directive in a non-skipping conditional... start skipping.
3389 PPConditionalInfo CI;
3393 }
3395 // If this is a top-level #else, inform the MIOpt.
3399 // If this is a #else with a #else before it, report the error.
3409 // In 'single-file-parse mode' undefined identifiers trigger parsing of all
3410 // the directive blocks.
3414 }
3416 // Finally, skip the rest of the contents of this block.
3420 }
3422 /// Implements the \#elif, \#elifdef, and \#elifndef directives.
3431 // Warn if using `#elifdef` & `#elifndef` in not C2x & C++2b mode.
3439 else
3446 }
3448 // #elif directive in a non-skipping conditional... start skipping.
3449 // We don't care what the condition is, because we will always skip it (since
3450 // the block immediately before it was included).
3453 PPConditionalInfo CI;
3457 }
3459 // If this is a top-level #elif, inform the MIOpt.
3463 // If this is a #elif with a #else before it, report the error.
3482 }
3483 }
3489 // In 'single-file-parse mode' undefined identifiers trigger parsing of all
3490 // the directive blocks.
3494 }
3496 // Finally, skip the rest of the contents of this block.
3500 }