translate-c: Fix performance hazard in transPreprocessorEntities

Fixes O(N^2) behavior of `transPreprocessorEntities` due to repeated calls to `mem.len` Closes #8959
ziglang · Jun 2, 2021 · a9dd8d7 · a9dd8d7
1 parent c6a0a4e
commit a9dd8d7
Show file tree

Hide file tree

Showing 4 changed files with 20 additions and 1 deletion.
diff --git a/src/clang.zig b/src/clang.zig
@@ -607,6 +607,13 @@ pub const IntegerLiteral = opaque {
     extern fn ZigClangIntegerLiteral_isZero(*const IntegerLiteral, *bool, *const ASTContext) bool;
 };
 
+/// This is just used as a namespace for a static method on clang's Lexer class; we don't directly
+/// deal with Lexer objects
+pub const Lexer = struct {
+    pub const getLocForEndOfToken = ZigClangLexer_getLocForEndOfToken;
+    extern fn ZigClangLexer_getLocForEndOfToken(SourceLocation, *const SourceManager, *const ASTUnit) SourceLocation;
+};
+
 pub const MacroDefinitionRecord = opaque {
     pub const getName_getNameStart = ZigClangMacroDefinitionRecord_getName_getNameStart;
     extern fn ZigClangMacroDefinitionRecord_getName_getNameStart(*const MacroDefinitionRecord) [*:0]const u8;

diff --git a/src/translate_c.zig b/src/translate_c.zig
@@ -4688,6 +4688,7 @@ fn transPreprocessorEntities(c: *Context, unit: *clang.ASTUnit) Error!void {
                 const macro = @ptrCast(*clang.MacroDefinitionRecord, entity);
                 const raw_name = macro.getName_getNameStart();
                 const begin_loc = macro.getSourceRange_getBegin();
+                const end_loc = clang.Lexer.getLocForEndOfToken(macro.getSourceRange_getEnd(), c.source_manager, unit);
 
                 const name = try c.str(raw_name);
                 // TODO https://github.com/ziglang/zig/issues/3756
@@ -4698,7 +4699,9 @@ fn transPreprocessorEntities(c: *Context, unit: *clang.ASTUnit) Error!void {
                 }
 
                 const begin_c = c.source_manager.getCharacterData(begin_loc);
-                const slice = begin_c[0..mem.len(begin_c)];
+                const end_c = c.source_manager.getCharacterData(end_loc);
+                const slice_len = @ptrToInt(end_c) - @ptrToInt(begin_c);
+                const slice = begin_c[0..slice_len];
 
                 var tokenizer = std.c.Tokenizer{
                     .buffer = slice,

diff --git a/src/zig_clang.cpp b/src/zig_clang.cpp
@@ -3100,6 +3100,13 @@ struct ZigClangSourceLocation ZigClangMacroDefinitionRecord_getSourceRange_getEn
     return bitcast(casted->getSourceRange().getEnd());
 }
 
+struct ZigClangSourceLocation ZigClangLexer_getLocForEndOfToken(ZigClangSourceLocation loc, const ZigClangSourceManager *sm, const ZigClangASTUnit *unit) {
+    const clang::SourceManager *casted_sm = reinterpret_cast<const clang::SourceManager *>(sm);
+    const clang::ASTUnit *casted_unit = reinterpret_cast<const clang::ASTUnit *>(unit);
+    clang::SourceLocation endloc = clang::Lexer::getLocForEndOfToken(bitcast(loc), 0, *casted_sm, casted_unit->getLangOpts());
+    return bitcast(endloc);
+}
+
 ZigClangRecordDecl_field_iterator ZigClangRecordDecl_field_begin(const struct ZigClangRecordDecl *self) {
     auto casted = reinterpret_cast<const clang::RecordDecl *>(self);
     return bitcast(casted->field_begin());

diff --git a/src/zig_clang.h b/src/zig_clang.h
@@ -966,6 +966,8 @@ ZIG_EXTERN_C const char* ZigClangSourceManager_getCharacterData(const struct Zig
 
 ZIG_EXTERN_C struct ZigClangQualType ZigClangASTContext_getPointerType(const struct ZigClangASTContext*, struct ZigClangQualType T);
 
+ZIG_EXTERN_C struct ZigClangSourceLocation ZigClangLexer_getLocForEndOfToken(struct ZigClangSourceLocation,
+        const ZigClangSourceManager *, const ZigClangASTUnit *);
 
 // Can return null.
 ZIG_EXTERN_C struct ZigClangASTUnit *ZigClangLoadFromCommandLine(const char **args_begin, const char **args_end,