Add graal_readSparseArray() and add more tests.

Author: rocky

test/bytecode_graal310/02_while1else.graalpy310.pyc

@@ -9,7 +9,6 @@
 import os
 import os.path as osp
 import sys
-from typing import List
 
 import click
 

test/bytecode_graal310/04_withas.graalpy310.pyc

@@ -310,15 +310,21 @@ def load_module_from_file_object(
                     source_size = unpack("<I", fp.read(4))[0]  # size mod 2**32
 
             if get_code:
-                is_graal = magic_int in GRAAL3_MAGICS
                 # Graal uses the same magic int for separate major/minor releases!
-                graal_weirdness_check = not is_graal or PYTHON_VERSION_TRIPLE == version
+                # So we can't get the major minor number using just a magic check.
+                # Instead we'd also need to seek for a "graal number" typically
+                # lower number like 25, 29, or 85. to distinguish.
+                # Furthermore, runnin unmarshal in graal on graal, can cause the
+                # the graal python interpreter to crash!
+                # For these reasons, we are better off using our marshal routines
+                # for Graal Python.
+                is_graal = magic_int in GRAAL3_MAGICS
                 if save_file_offsets and not is_graal:
                     co, file_offsets = xdis.unmarshal.load_code_and_get_file_offsets(
                         fp, magic_int, code_objects
                     )
 
-                elif my_magic_int == magic_int and graal_weirdness_check:
+                elif my_magic_int == magic_int and not is_graal:
                     bytecode = fp.read()
                     co = marshal.loads(bytecode)
                     # Python 3.10 returns a tuple here?

test/bytecode_graal311/02_while1else.graalpy311.pyc

@@ -354,12 +354,21 @@ def graal_readStringArray(self) -> tuple[str, ...]:
         Python equvalent of Python Graal's readObjectArray() from
         MarshalModuleBuiltins.java
         """
-        length: int = int(unpack("<i", self.fp.read(4))[0])
+        length: int = self.graal_readInt()
         return tuple([self.graal_readString() for _ in range(length)])
 
-    def graal_readSparseTable(self):
-        """ """
-        pass
+    def graal_readSparseTable(self) -> Dict[int, tuple]:
+        """
+        Python equvalent of Python Graal's readObjectArray() from
+        MarshalModuleBuiltins.java
+        """
+        self.graal_readInt()  # the length return value isn't used.
+        table = {} # new int[length][];
+        while True:
+            i = self.graal_readInt()
+            if i == -1:
+                return table
+            table[i] = self.graal_readIntArray()
 
     def load(self):
         """
@@ -876,6 +885,8 @@ def t_code_graal(self, save_ref, bytes_for_s: bool = False):
         else:
             code = self.t_graal_CodeUnit(save_ref=False, bytes_for_s=False)
         assert self.graal_code_info["co_flags"] == code.co_flags
+
+        # FIXME: add an assert self.fp.tell() has advanced to save_position?
         self.fp.seek(saved_position)
 
         self.code_to_file_offsets[code] = (
@@ -995,20 +1006,20 @@ def t_graal_CodeUnit(self, save_ref, bytes_for_s: bool = False):
         code. In particular, instructions are JVM bytecode.
         """
 
-        # This is Java code for how a CodeUnit (type "U") is dumped
-        # writeByte(Compiler.BYTECODE_VERSION);
-        # writeString(code.name);
-        # writeString(code.qualname);
-        # writeInt(code.argCount);
-        # writeInt(code.kwOnlyArgCount);
-        # writeInt(code.positionalOnlyArgCount);
-        # writeInt(code.stacksize);
-        # writeBytes(code.code);
-        # writeBytes(code.srcOffsetTable);
-        # writeInt(code.flags);
-
         graal_bytecode_version = self.graal_readByte()
         assert (21000 + graal_bytecode_version * 10) in GRAAL3_MAGICS
+
+        # This is Java code for how a CodeUnit (type "U") is read
+        # TruffleString name = readString();
+        # TruffleString qualname = readString();
+        # int argCount = readInt();
+        # int kwOnlyArgCount = readInt();
+        # int positionalOnlyArgCount = readInt();
+        # int stacksize = readInt();
+        # byte[] code = readBytes();
+        # byte[] srcOffsetTable = readBytes();
+        # int flags = readInt();
+
         co_name = self.graal_readString()
         co_qualname = self.graal_readString()
         co_argcount = self.graal_readInt()
@@ -1031,12 +1042,11 @@ def t_graal_CodeUnit(self, save_ref, bytes_for_s: bool = False):
         co_cellvars = self.graal_readStringArray()
         co_freevars = self.graal_readStringArray()
 
-        # if (code.cell2arg != null) {
-        #         writeIntArray(code.cell2arg);
-        # } else {
-        #     writeIntArray(PythonUtils.EMPTY_INT_ARRAY);
-        # }
-        # writeObjectArray(code.constants);
+        # int[] cell2arg = readIntArray();
+        # if (cell2arg.length == 0) {
+        #     cell2arg = null;
+        #     }
+        # Object[] constants = readObjectArray();
 
         cell2arg = self.graal_readIntArray()
         co_consts = self.graal_readObjectArray()
@@ -1069,23 +1079,23 @@ def t_graal_CodeUnit(self, save_ref, bytes_for_s: bool = False):
             [self.graal_code_info["co_codeunit_position"], co_code_offset_in_file]
         )
 
-        # writeLongArray(code.primitiveConstants);
-        # writeIntArray(code.exceptionHandlerRanges);
-        # writeInt(code.conditionProfileCount);
-        # writeInt(code.startLine);
-        # writeInt(code.startColumn);
-        # writeInt(code.endLine);
-        # writeInt(code.endColumn);
-        # writeBytes(code.outputCanQuicken);
-        # writeBytes(code.variableShouldUnbox);
-        # writeSparseTable(code.generalizeInputsMap);
-        # writeSparseTable(code.generalizeVarsMap);
-
         # The data from the below is not used, but we run the
         # the extraction to keep the self.fp location where it should for
         # the situation that we have code objects inside the codes' co_consts
         # table marshaled as an ObjectArray.
 
+        # long[] primitiveConstants = readLongArray();
+        # int[] exceptionHandlerRanges = readIntArray();
+        # int conditionProfileCount = readInt();
+        # int startLine = readInt();
+        # int startColumn = readInt();
+        # int endLine = readInt();
+        # int endColumn = readInt();
+        # byte[] outputCanQuicken = readBytes();
+        # byte[] variableShouldUnbox = readBytes();
+        # int[][] generalizeInputsMap = readSparseTable();
+        # int[][] generalizeVarsMap = readSparseTable();
+
         primitive_constants = self.graal_readLongArray()
         exception_handler_ranges = self.graal_readIntArray()
         condition_profileCount = self.graal_readInt()