Experimental Regex Strawperson (use Swift in the parser) (#40117)

[regex] Use Swift in the parser

Add in a strawperson use of Swift by the parser, for
future regex support.

Author: milseman

include/swift/AST/DiagnosticsParse.def

@@ -107,6 +107,10 @@ ERROR(lex_unprintable_ascii_character,none,
       "unprintable ASCII character found in source file", ())
 ERROR(lex_invalid_utf8,none,
       "invalid UTF-8 found in source file", ())
+
+NOTE(lex_experimental_regex_strawperson,none,
+     "'%0'", (StringRef))
+
 ERROR(lex_single_quote_string,none,
       "single-quoted string literal found, use '\"'", ())
 ERROR(lex_invalid_curly_quote,none,

include/swift/Basic/LangOptions.h

@@ -144,6 +144,9 @@ namespace swift {
     /// PackageDescription version to compile for.
     version::Version PackageDescriptionVersion;
 
+    /// Enable experimental string processing
+    bool EnableExperimentalRegex = false;
+
     /// Disable API availability checking.
     bool DisableAvailabilityChecking = false;
 

include/swift/Option/FrontendOptions.td

@@ -469,6 +469,10 @@ def disable_deserialization_recovery :
   Flag<["-"], "disable-deserialization-recovery">,
   HelpText<"Don't attempt to recover from missing xrefs (etc) in swiftmodules">;
 
+def enable_experimental_regex :
+  Flag<["-"], "enable-experimental-regex">,
+  HelpText<"Enable experimental string processing">;
+
 def disable_availability_checking : Flag<["-"],
   "disable-availability-checking">,
   HelpText<"Disable checking for potentially unavailable APIs">;

include/swift/Parse/ExperimentalRegexBridging.h

@@ -0,0 +1,20 @@
+#ifndef EXPERIMENTAL_REGEX_BRIDGING
+#define EXPERIMENTAL_REGEX_BRIDGING
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef const char *(* ParseRegexStrawperson)(const char *);
+
+void Parser_registerParseRegexStrawperson(ParseRegexStrawperson fn);
+
+#ifdef __cplusplus
+} // extern "C"
+#endif
+
+#endif // EXPERIMENTAL_REGEX_BRIDGING
+
+
+//const char* experimental_regex_strawperson(const char *in);
+

include/swift/Parse/Lexer.h

@@ -607,6 +607,15 @@ class Lexer {
   bool lexUnknown(bool EmitDiagnosticsIfToken);
 
   NulCharacterKind getNulCharacterKind(const char *Ptr) const;
+
+  /// Emit diagnostics for single-quote string and suggest replacement
+  /// with double-quoted equivalent.
+  ///
+  /// Or, if we're in strawperson mode, we will emit a custom
+  /// error message instead, determined by the Swift library.
+  void diagnoseSingleQuoteStringLiteral(const char *TokStart,
+                                        const char *TokEnd);
+
 };
 
 /// A lexer that can lex trivia into its pieces

include/swift/module.modulemap

@@ -8,3 +8,7 @@ module OptimizerBridging {
   export *
 }
 
+module ExperimentalRegexBridging {
+  header "Parse/ExperimentalRegexBridging.h"
+  export *
+}

lib/Frontend/CompilerInvocation.cpp

@@ -471,6 +471,10 @@ static bool ParseLangArgs(LangOptions &Opts, ArgList &Args,
       = A->getOption().matches(OPT_enable_deserialization_recovery);
   }
 
+  // Experimental string processing
+  Opts.EnableExperimentalRegex |=
+      Args.hasArg(OPT_enable_experimental_regex);
+
   Opts.DisableAvailabilityChecking |=
       Args.hasArg(OPT_disable_availability_checking);
   Opts.CheckAPIAvailabilityOnly |=

lib/Parse/CMakeLists.txt

@@ -31,7 +31,8 @@ _swift_gyb_target_sources(swiftParse PRIVATE
 target_link_libraries(swiftParse PRIVATE
   swiftAST
   swiftSyntax
-  swiftSyntaxParse)
+  swiftSyntaxParse
+)
 
 add_dependencies(swiftParse swift-parse-syntax-generated-headers)
 

lib/Parse/Lexer.cpp

@@ -30,6 +30,13 @@
 // FIXME: Figure out if this can be migrated to LLVM.
 #include "clang/Basic/CharInfo.h"
 
+// Regex parser delivered via libSwift
+#include "swift/Parse/ExperimentalRegexBridging.h"
+static ParseRegexStrawperson parseRegexStrawperson = nullptr;
+void Parser_registerParseRegexStrawperson(ParseRegexStrawperson fn) {
+  parseRegexStrawperson = fn;
+}
+
 #include <limits>
 
 using namespace swift;
@@ -1788,13 +1795,27 @@ static void validateMultilineIndents(const Token &Str,
 
 /// Emit diagnostics for single-quote string and suggest replacement
 /// with double-quoted equivalent.
-static void diagnoseSingleQuoteStringLiteral(const char *TokStart,
-                                             const char *TokEnd,
-                                             DiagnosticEngine *D) {
+///
+/// Or, if we're in experimental regex mode, we will emit a custom
+/// error message instead, determined by the Swift library.
+void Lexer::diagnoseSingleQuoteStringLiteral(const char *TokStart,
+                                             const char *TokEnd) {
   assert(*TokStart == '\'' && TokEnd[-1] == '\'');
-  if (!D)
+  if (!Diags) // or assert?
     return;
 
+  auto startLoc = Lexer::getSourceLoc(TokStart);
+  auto endLoc = Lexer::getSourceLoc(TokEnd);
+
+  if (LangOpts.EnableExperimentalRegex) {
+    if (parseRegexStrawperson) {
+      auto copy = std::string(TokStart, TokEnd-TokStart);
+      auto msg = parseRegexStrawperson(copy.c_str());
+      assert(msg != nullptr);
+      Diags->diagnose(startLoc, diag::lex_experimental_regex_strawperson, msg);
+    }
+  }
+
   SmallString<32> replacement;
   replacement.push_back('"');
   const char *Ptr = TokStart + 1;
@@ -1826,9 +1847,8 @@ static void diagnoseSingleQuoteStringLiteral(const char *TokStart,
   replacement.append(OutputPtr, Ptr - 1);
   replacement.push_back('"');
 
-  D->diagnose(Lexer::getSourceLoc(TokStart), diag::lex_single_quote_string)
-      .fixItReplaceChars(Lexer::getSourceLoc(TokStart),
-                         Lexer::getSourceLoc(TokEnd), replacement);
+  Diags->diagnose(startLoc, diag::lex_single_quote_string)
+      .fixItReplaceChars(startLoc, endLoc, replacement);
 }
 
 /// lexStringLiteral:
@@ -1895,7 +1915,7 @@ void Lexer::lexStringLiteral(unsigned CustomDelimiterLen) {
   if (QuoteChar == '\'') {
     assert(!IsMultilineString && CustomDelimiterLen == 0 &&
            "Single quoted string cannot have custom delimitor, nor multiline");
-    diagnoseSingleQuoteStringLiteral(TokStart, CurPtr, Diags);
+    diagnoseSingleQuoteStringLiteral(TokStart, CurPtr);
   }
 
   if (wasErroneous)

libswift/Sources/CMakeLists.txt

@@ -6,5 +6,6 @@
 # See http://swift.org/LICENSE.txt for license information
 # See http://swift.org/CONTRIBUTORS.txt for Swift project authors
 
+add_subdirectory(ExperimentalRegex)
 add_subdirectory(SIL)
 add_subdirectory(Optimizer)