bpo-43950: use 0-indexed column offsets for bytecode positions (GH-27011)

Author: isidentical

Lib/test/test_code.py

@@ -344,7 +344,7 @@ def test_co_positions_artificial_instructions(self):
             # get assigned the first_lineno but they don't have other positions.
             # There is no easy way of inferring them at that stage, so for now
             # we don't support it.
-            self.assertTrue(all(positions) or not any(positions))
+            self.assertTrue(positions.count(None) in [0, 4])
 
             if not any(positions):
                 artificial_instructions.append(instr)

Lib/test/test_compile.py

@@ -1006,8 +1006,8 @@ def generic_visit(self, node):
                     return
                 lines.add(node.lineno)
                 end_lines.add(node.end_lineno)
-                columns.add(node.col_offset + 1)
-                end_columns.add(node.end_col_offset + 1)
+                columns.add(node.col_offset)
+                end_columns.add(node.end_col_offset)
 
         SourceOffsetVisitor().visit(ast_tree)
 
@@ -1058,10 +1058,10 @@ def test_compiles_to_extended_op_arg(self):
 
         self.assertOpcodeSourcePositionIs(compiled_code, 'INPLACE_SUBTRACT',
             line=10_000 + 2, end_line=10_000 + 2,
-            column=3, end_column=9)
+            column=2, end_column=8)
         self.assertOpcodeSourcePositionIs(compiled_code, 'INPLACE_ADD',
             line=10_000 + 4, end_line=10_000 + 4,
-            column=3, end_column=10)
+            column=2, end_column=9)
 
     def test_multiline_expression(self):
         snippet = """\
@@ -1071,7 +1071,7 @@ def test_multiline_expression(self):
 """
         compiled_code, _ = self.check_positions_against_ast(snippet)
         self.assertOpcodeSourcePositionIs(compiled_code, 'CALL_FUNCTION',
-            line=1, end_line=3, column=1, end_column=2)
+            line=1, end_line=3, column=0, end_column=1)
 
     def test_very_long_line_end_offset(self):
         # Make sure we get None for when the column offset is too large to
@@ -1088,15 +1088,15 @@ def test_complex_single_line_expression(self):
 
         compiled_code, _ = self.check_positions_against_ast(snippet)
         self.assertOpcodeSourcePositionIs(compiled_code, 'BINARY_SUBSCR',
-            line=1, end_line=1, column=14, end_column=22)
+            line=1, end_line=1, column=13, end_column=21)
         self.assertOpcodeSourcePositionIs(compiled_code, 'BINARY_MULTIPLY',
-            line=1, end_line=1, column=10, end_column=22)
+            line=1, end_line=1, column=9, end_column=21)
         self.assertOpcodeSourcePositionIs(compiled_code, 'BINARY_ADD',
-            line=1, end_line=1, column=10, end_column=27)
+            line=1, end_line=1, column=9, end_column=26)
         self.assertOpcodeSourcePositionIs(compiled_code, 'BINARY_MATRIX_MULTIPLY',
-            line=1, end_line=1, column=5, end_column=28)
+            line=1, end_line=1, column=4, end_column=27)
         self.assertOpcodeSourcePositionIs(compiled_code, 'BINARY_SUBTRACT',
-            line=1, end_line=1, column=1, end_column=28)
+            line=1, end_line=1, column=0, end_column=27)
 
 
 class TestExpressionStackSize(unittest.TestCase):

Objects/codeobject.c

@@ -610,9 +610,6 @@ PyCode_Addr2Location(PyCodeObject *co, int addrq,
                      int *end_line, int *end_column)
 {
     *start_line = PyCode_Addr2Line(co, addrq);
-    if (*start_line == -1) {
-        *start_line = 0;
-    }
     *start_column = _PyCode_Addr2Offset(co, addrq);
     *end_line = _PyCode_Addr2EndLine(co, addrq);
     *end_column = _PyCode_Addr2EndOffset(co, addrq);
@@ -626,7 +623,7 @@ _PyCode_Addr2EndLine(PyCodeObject* co, int addrq)
         return co->co_firstlineno;
     }
     else if (co->co_endlinetable == Py_None) {
-        return 0;
+        return -1;
     }
 
     assert(addrq >= 0 && addrq < PyBytes_GET_SIZE(co->co_code));
@@ -639,34 +636,34 @@ int
 _PyCode_Addr2Offset(PyCodeObject* co, int addrq)
 {
     if (co->co_columntable == Py_None || addrq < 0) {
-        return 0;
+        return -1;
     }
     if (addrq % 2 == 1) {
         --addrq;
     }
     if (addrq >= PyBytes_GET_SIZE(co->co_columntable)) {
-        return 0;
+        return -1;
     }
 
     unsigned char* bytes = (unsigned char*)PyBytes_AS_STRING(co->co_columntable);
-    return bytes[addrq];
+    return bytes[addrq] - 1;
 }
 
 int
 _PyCode_Addr2EndOffset(PyCodeObject* co, int addrq)
 {
     if (co->co_columntable == Py_None || addrq < 0) {
-        return 0;
+        return -1;
     }
     if (addrq % 2 == 0) {
         ++addrq;
     }
     if (addrq >= PyBytes_GET_SIZE(co->co_columntable)) {
-        return 0;
+        return -1;
     }
 
     unsigned char* bytes = (unsigned char*)PyBytes_AS_STRING(co->co_columntable);
-    return bytes[addrq];
+    return bytes[addrq] - 1;
 }
 
 void
@@ -980,7 +977,7 @@ positionsiter_dealloc(positionsiterator* pi)
 
 static PyObject*
 _source_offset_converter(int* value) {
-    if (*value <= 0) {
+    if (*value == -1) {
         Py_RETURN_NONE;
     }
     return PyLong_FromLong(*value);