[GR-23265] Make test_traceback.test_encoded_file pass

PullRequest: graalpython/1076

Author: msimacek

graalpython/com.oracle.graal.python.test/src/tests/unittest_tags/test_traceback.txt

@@ -47,6 +47,7 @@
 *graalpython.lib-python.3.test.test_traceback.TracebackCases.test_bad_indentation
 *graalpython.lib-python.3.test.test_traceback.TracebackCases.test_base_exception
 *graalpython.lib-python.3.test.test_traceback.TracebackCases.test_caret
+*graalpython.lib-python.3.test.test_traceback.TracebackCases.test_encoded_file
 *graalpython.lib-python.3.test.test_traceback.TracebackCases.test_format_exception_only_bad__str__
 *graalpython.lib-python.3.test.test_traceback.TracebackCases.test_nocaret
 *graalpython.lib-python.3.test.test_traceback.TracebackCases.test_print_exception

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/PythonFileDetector.java

@@ -58,6 +58,7 @@ public final class PythonFileDetector implements TruffleFile.FileTypeDetector {
 
     private static final String UTF_8_BOM_IN_LATIN_1 = new String(new byte[]{(byte) 0xEF, (byte) 0xBB, (byte) 0xBF}, StandardCharsets.ISO_8859_1);
     private static final Pattern ENCODING_COMMENT = Pattern.compile("^[ \t\f]*#.*?coding[:=][ \t]*([-_.a-zA-Z0-9]+).*");
+    private static final Pattern BLANK_LINE = Pattern.compile("^[ \t\f]*(?:#.*)?");
 
     @Override
     public String findMimeType(TruffleFile file) throws IOException {
@@ -106,19 +107,23 @@ private static Charset tryGetCharsetFromLine(String line, boolean hasBOM) {
 
     @TruffleBoundary
     public static Charset findEncodingStrict(BufferedReader reader) throws IOException {
-        Charset charset;
         // Read first two lines like CPython
         String firstLine = reader.readLine();
-        boolean hasBOM = false;
-        if (firstLine != null && firstLine.startsWith(UTF_8_BOM_IN_LATIN_1)) {
-            hasBOM = true;
-            firstLine = firstLine.substring(UTF_8_BOM_IN_LATIN_1.length());
-        }
-        if ((charset = tryGetCharsetFromLine(firstLine, hasBOM)) != null) {
-            return charset;
-        }
-        if ((charset = tryGetCharsetFromLine(reader.readLine(), hasBOM)) != null) {
-            return charset;
+        if (firstLine != null) {
+            boolean hasBOM = false;
+            if (firstLine.startsWith(UTF_8_BOM_IN_LATIN_1)) {
+                hasBOM = true;
+                firstLine = firstLine.substring(UTF_8_BOM_IN_LATIN_1.length());
+            }
+            Charset charset;
+            if ((charset = tryGetCharsetFromLine(firstLine, hasBOM)) != null) {
+                return charset;
+            }
+            if (BLANK_LINE.matcher(firstLine).matches()) {
+                if ((charset = tryGetCharsetFromLine(reader.readLine(), hasBOM)) != null) {
+                    return charset;
+                }
+            }
         }
         return StandardCharsets.UTF_8;
     }

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/modules/PosixSubprocessModuleBuiltins.java

@@ -175,6 +175,10 @@ private synchronized int forkExec(PList args, @SuppressWarnings("unused") PList
                 pb.redirectError(Redirect.INHERIT);
             }
 
+            if (errwrite == c2pwrite) {
+                pb.redirectErrorStream(true);
+            }
+
             try {
                 if (getContext().getEnv().getPublicTruffleFile(cwd).exists()) {
                     pb.directory(new File(cwd));

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/parser/PythonParserImpl.java

@@ -79,7 +79,7 @@ public final class PythonParserImpl implements PythonParser, PythonCodeSerialize
     private long numberOfFiles = 0;
     private static final boolean IN_IMAGE_BUILD_TIME = ImageInfo.inImageBuildtimeCode();
 
-    private static final Pattern START_INDENT_REGEX = Pattern.compile("^([ \t]+)[^#\r\n\f]");
+    private static final Pattern START_INDENT_REGEX = Pattern.compile("^([ \t]+)[^#\r\n\f\t ]");
 
     public static final DescriptiveBailErrorListener ERROR_LISTENER = new DescriptiveBailErrorListener();