gh-93351: Validate the position information when compiling AST nodes

Author: pablogsal

Lib/test/test_ast.py

@@ -335,6 +335,33 @@ def test_ast_validation(self):
         for snippet in snippets_to_validate:
             tree = ast.parse(snippet)
             compile(tree, '<string>', 'exec')
+    
+    def test_invalid_position_information(self):
+        invalid_linenos = [
+            (10, 1), (-10, -11), (10, -11), (-5, -2), (-5, 1)
+        ]
+
+        for lineno, end_lineno in invalid_linenos:
+            with self.subTest(f"Check invalid linenos {lineno}:{end_lineno}"):
+                snippet = "a = 1"
+                tree = ast.parse(snippet)
+                tree.body[0].lineno = lineno
+                tree.body[0].end_lineno = end_lineno
+                with self.assertRaises(ValueError):
+                    compile(tree, '<string>', 'exec')
+
+        invalid_col_offsets = [
+            (10, 1), (-10, -11), (10, -11), (-5, -2), (-5, 1)
+        ]
+        for col_offset, end_col_offset in invalid_col_offsets:
+            with self.subTest(f"Check invalid col_offset {col_offset}:{end_col_offset}"):
+                snippet = "a = 1"
+                tree = ast.parse(snippet)
+                tree.body[0].col_offset = col_offset
+                tree.body[0].end_col_offset = end_col_offset
+                with self.assertRaises(ValueError):
+                    compile(tree, '<string>', 'exec')
+
 
     def test_slice(self):
         slc = ast.parse("x[::]").body[0].value.slice

Python/ast.c

@@ -22,6 +22,27 @@ static int validate_stmt(struct validator *, stmt_ty);
 static int validate_expr(struct validator *, expr_ty, expr_context_ty);
 static int validate_pattern(struct validator *, pattern_ty, int);
 
+#define VALIDATE_POSITIONS(node) \
+    if (node->lineno > node->end_lineno) { \
+        PyErr_Format(PyExc_ValueError, \
+                     "line %d-%d is not a valid range", \
+                     node->lineno, node->end_lineno); \
+        return 0; \
+    } \
+    if ((node->lineno < 0 && node->end_lineno != node->lineno) || \
+        (node->col_offset < 0 &&  node->col_offset != node->end_col_offset)) { \
+        PyErr_Format(PyExc_ValueError, \
+                     "line %d-%d, column %d-%d is not a valid range", \
+                     node->lineno, node->end_lineno, node->col_offset, node->end_col_offset); \
+        return 0; \
+    } \
+    if (node->lineno == node->end_lineno && node->col_offset > node->end_col_offset) { \
+        PyErr_Format(PyExc_ValueError, \
+                     "line %d, column %d-%d is not a valid range", \
+                     node->lineno, node->col_offset, node->end_col_offset); \
+        return 0; \
+    }
+
 static int
 validate_name(PyObject *name)
 {
@@ -183,6 +204,7 @@ validate_constant(struct validator *state, PyObject *value)
 static int
 validate_expr(struct validator *state, expr_ty exp, expr_context_ty ctx)
 {
+    VALIDATE_POSITIONS(exp);
     int ret = -1;
     if (++state->recursion_depth > state->recursion_limit) {
         PyErr_SetString(PyExc_RecursionError,
@@ -505,6 +527,7 @@ validate_capture(PyObject *name)
 static int
 validate_pattern(struct validator *state, pattern_ty p, int star_ok)
 {
+    VALIDATE_POSITIONS(p);
     int ret = -1;
     if (++state->recursion_depth > state->recursion_limit) {
         PyErr_SetString(PyExc_RecursionError,
@@ -674,6 +697,7 @@ validate_body(struct validator *state, asdl_stmt_seq *body, const char *owner)
 static int
 validate_stmt(struct validator *state, stmt_ty stmt)
 {
+    VALIDATE_POSITIONS(stmt);
     int ret = -1;
     Py_ssize_t i;
     if (++state->recursion_depth > state->recursion_limit) {