Reject nodes with AST flags set for re_is_literal()

One concession here is that I'm allowing `(?n)` because we don't currently indicate this in the AST. I'm also explicitly permitting `RE_END_NL`, because by the API contract this is the caller's responsibility.
katef · Oct 30, 2023 · 450d4f4 · 450d4f4
1 parent de237ff
commit 450d4f4
Show file tree

Hide file tree

Showing 25 changed files with 48 additions and 2 deletions.
diff --git a/src/libre/ast.c b/src/libre/ast.c
@@ -997,13 +997,31 @@ ast_expr_is_literal(const struct ast_expr *e,
 		*anchor_end = 1;
 	}
 
+done:
+
 	for (i = 0; i < count; i++) {
 		if (nodes[i] == NULL || nodes[i]->type != AST_EXPR_LITERAL) {
 			return 0;
 		}
-	}
 
-done:
+		/* mask out AST flags that we don't care about; these may be present or not */
+		int ast_flags = nodes[i]->flags
+			& ~(AST_FLAG_FIRST | AST_FLAG_LAST | AST_FLAG_ALWAYS_CONSUMES | AST_FLAG_CAN_CONSUME);
+
+		/* we reject anything else */
+		if (ast_flags != 0) {
+			return 0;
+		}
+
+		/* mask out re flags that we don't handle here yet (the caller is responsible for these) */
+		int re_flags = nodes[i]->re_flags
+			& ~(RE_END_NL);
+
+		/* and we don't permit any other re_flags */
+		if (re_flags != 0) {
+			return 0;
+		}
+	}
 
 	*n = count + is_end_nl;
 

diff --git a/tests/re_literal/in13.re b/tests/re_literal/in13.re
@@ -0,0 +1 @@
+^(?i)a$
diff --git a/tests/re_literal/in14.re b/tests/re_literal/in14.re
@@ -0,0 +1 @@
+(?:x(?i)a)
diff --git a/tests/re_literal/in15.re b/tests/re_literal/in15.re
@@ -0,0 +1 @@
+(?i)
diff --git a/tests/re_literal/in16.re b/tests/re_literal/in16.re
@@ -0,0 +1 @@
+(?i)xy
diff --git a/tests/re_literal/in17.re b/tests/re_literal/in17.re
@@ -0,0 +1 @@
+a(?i)xy
diff --git a/tests/re_literal/in18.re b/tests/re_literal/in18.re
@@ -0,0 +1 @@
+a(?:x(?i)xy)
diff --git a/tests/re_literal/in19.re b/tests/re_literal/in19.re
@@ -0,0 +1 @@
+a(?:x(?-i)xy)
diff --git a/tests/re_literal/in20.re b/tests/re_literal/in20.re
@@ -0,0 +1 @@
+(?i)(?:(?-i)a)(?:x(?-i)xy)
diff --git a/tests/re_literal/in21.re b/tests/re_literal/in21.re
@@ -0,0 +1 @@
+(?i)(?:(?-i)a)(?:(?-i)x(?-i)xy)
diff --git a/tests/re_literal/in22.re b/tests/re_literal/in22.re
@@ -0,0 +1 @@
+(?s)a
diff --git a/tests/re_literal/in23.re b/tests/re_literal/in23.re
@@ -0,0 +1 @@
+(?x)a
diff --git a/tests/re_literal/in24.re b/tests/re_literal/in24.re
@@ -0,0 +1 @@
+(?n)a
diff --git a/tests/re_literal/out13.txt b/tests/re_literal/out13.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out14.txt b/tests/re_literal/out14.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out15.txt b/tests/re_literal/out15.txt
@@ -0,0 +1,2 @@
+anchors: 
+literal: 
diff --git a/tests/re_literal/out16.txt b/tests/re_literal/out16.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out17.txt b/tests/re_literal/out17.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out18.txt b/tests/re_literal/out18.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out19.txt b/tests/re_literal/out19.txt
@@ -0,0 +1,2 @@
+anchors: 
+literal: axxy
diff --git a/tests/re_literal/out20.txt b/tests/re_literal/out20.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out21.txt b/tests/re_literal/out21.txt
@@ -0,0 +1,2 @@
+anchors: 
+literal: axxy
diff --git a/tests/re_literal/out22.txt b/tests/re_literal/out22.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out23.txt b/tests/re_literal/out23.txt
@@ -0,0 +1 @@
+non-literal
diff --git a/tests/re_literal/out24.txt b/tests/re_literal/out24.txt
@@ -0,0 +1,2 @@
+anchors: 
+literal: a