Fix spurious rebuilds of pegparser

Author: msimacek

graalpython/com.oracle.graal.python.pegparser.generator/main_asdl_gen.py

@@ -65,7 +65,7 @@ def main():
 
     old_contents = stamp(args.sst_path, args.ast_path)
     generate(args.input_file, args.sst_path, args.ast_path)
-    if old_contents != stamp(args.sst_path, args.ast_path):
+    if old_contents != stamp(args.sst_path, args.ast_path) or not args.stamp.exists():
         args.stamp.touch()
     else:
         print(f"{args.sst_path} and {args.ast_path} not modified")

graalpython/com.oracle.graal.python.pegparser.generator/main_parser_gen.py

@@ -41,6 +41,7 @@
 import os
 import sys
 import tokenize
+import io
 
 from pegen.build import generate_token_definitions
 from pegen.grammar_parser import GeneratedParser as GrammarParser
@@ -71,34 +72,29 @@ def main():
 
     # Determine which line separator to use
     if os.path.exists(args.output_file):
-        stat_result = os.stat(args.output_file)
-        atime, mtime = stat_result.st_atime, stat_result.st_mtime
-        with open(args.output_file, "r", encoding="utf-8", newline=os.linesep) as f:
-            content = f.read()
-        if os.linesep != "\n":
-            unix_content = content.replace(os.linesep, "\n")
-            if unix_content == content:
-                # Windows file has Unix line endings
-                linesep = "\n"
-                content = unix_content
+        with open(args.output_file, "rb") as f:
+            raw_old_content = f.read()
+            if b'\r\n' in raw_old_content:
+                linesep = '\r\n'
             else:
-                linesep = os.linesep
-        else:
-            linesep = "\n"
+                linesep = '\n'
+        with open(args.output_file, "r", encoding="utf-8") as f:
+            old_content = f.read()
     else:
-        content = None
+        old_content = None
         linesep = os.linesep
 
-    with open(args.output_file, "w", encoding="utf-8", newline=linesep) as file:
-        gen = JavaParserGenerator(grammar, all_tokens, exact_tokens, non_exact_tokens, file, debug=args.debug)
-        gen.generate(os.path.basename(args.grammar_file))
+    out = io.StringIO()
+    class_name = os.path.splitext(os.path.basename(args.output_file))[0]
+    gen = JavaParserGenerator(grammar, all_tokens, exact_tokens, non_exact_tokens, out, class_name, debug=args.debug)
+    gen.generate(os.path.basename(args.grammar_file))
 
-    with open(args.output_file, "r", encoding="utf-8", newline=linesep) as file:
-        new_content = file.read()
-
-    if content == new_content:
+    if out.getvalue() != old_content:
+        print(f"Writing new {args.output_file}")
+        with open(args.output_file, "w", encoding="utf-8", newline=linesep) as f:
+            f.write(out.getvalue())
+    else:
         print(f"{args.output_file} not modified")
-        os.utime(args.output_file, (atime, mtime))
 
 
 if __name__ == '__main__':

graalpython/com.oracle.graal.python.pegparser.generator/pegjava/java_generator.py

@@ -874,10 +874,12 @@ def __init__(
         exact_tokens: Dict[str, int],
         non_exact_tokens: Set[str],
         file: Optional[IO[Text]],
+        class_name: str,
         debug: bool = True,
         skip_actions: bool = False,
     ):
         super().__init__(grammar, set(tokens.values()), file)
+        self.class_name = class_name
         self.typingvisitor = TypingVisitor(self) # Java type hack
         self.callmakervisitor = JavaCallMakerVisitor(self, exact_tokens, non_exact_tokens, self.print)
         self.lookahead_functions: Dict[str, FunctionCall] = {}
@@ -930,10 +932,9 @@ def generate(self, filename: str) -> None:
         self.print(f"// Generated from {filename} by pegen")
         self.print("package com.oracle.graal.python.pegparser;")
         self.print(IMPORTS)
-        className = os.path.splitext(os.path.basename(self.file.name))[0]
         self.print('@SuppressWarnings({"all", "cast"})')
         self.print('@SuppressFBWarnings')
-        self.print("public final class %s extends AbstractParser {" % className)
+        self.print("public final class %s extends AbstractParser {" % self.class_name)
         # Java needs a few fields declarations. Also, we're now in a class
         self.level += 1
         self.print()
@@ -945,12 +946,12 @@ def generate(self, filename: str) -> None:
             self.print(f"private static final int {rulename.upper()}_ID = {i};{comment}")
         self.print()
         # Java needs a constructor
-        self.print("public %s(String source, SourceRange sourceRange, ParserCallbacks parserCb, InputType startRule, EnumSet<Flags> flags, int featureVersion) {" % className)
+        self.print("public %s(String source, SourceRange sourceRange, ParserCallbacks parserCb, InputType startRule, EnumSet<Flags> flags, int featureVersion) {" % self.class_name)
         with self.indent():
             self.print("super(source, sourceRange, parserCb, startRule, flags, featureVersion);")
         self.print("}")
         self.print()
-        self.print("public %s(String source, ParserCallbacks parserCb, InputType startRule, EnumSet<Flags> flags, int featureVersion) {" % className)
+        self.print("public %s(String source, ParserCallbacks parserCb, InputType startRule, EnumSet<Flags> flags, int featureVersion) {" % self.class_name)
         with self.indent():
             self.print("super(source, null, parserCb, startRule, flags, featureVersion);")
         self.print("}")