From 92e6e827e2d69eb28110a26628b15d353ec498a6 Mon Sep 17 00:00:00 2001
From: Scott Vokes <svokes@fastly.com>
Date: Mon, 16 Sep 2024 16:00:21 -0400
Subject: [PATCH] Add re_is_anchored interface.

---
 include/re/re.h      | 16 ++++++++++++++++
 src/libre/libre.syms |  1 +
 src/libre/re.c       | 37 +++++++++++++++++++++++++++++++++++++
 3 files changed, 54 insertions(+)

diff --git a/include/re/re.h b/include/re/re.h
index 20408e98a..841e4e946 100644
--- a/include/re/re.h
+++ b/include/re/re.h
@@ -136,6 +136,22 @@ re_comp(enum re_dialect dialect,
 	const struct fsm_alloc *alloc,
 	enum re_flags flags, struct re_err *err);
 
+/* Parse and analyze the regex enough to determine whether it is
+ * anchored at the start and/or end.
+ *
+ * As long as the result is checked for RE_IS_ANCHORED_ERROR first,
+ * the result can be used like a bitset. */
+enum re_is_anchored_res {
+	RE_IS_ANCHORED_NONE = 0x00,
+	RE_IS_ANCHORED_START = 0x01,
+	RE_IS_ANCHORED_END = 0x02,
+	RE_IS_ANCHORED_BOTH = 0x03,
+	RE_IS_ANCHORED_ERROR = 0xFFFF,
+};
+enum re_is_anchored_res
+re_is_anchored(enum re_dialect dialect, re_getchar_fun *f, void *opaque,
+	enum re_flags flags, struct re_err *err);
+
 /*
  * Return a human-readable string describing a given error code. The string
  * returned has static storage, and must not be freed.
diff --git a/src/libre/libre.syms b/src/libre/libre.syms
index a4f1a223b..9d381cb0f 100644
--- a/src/libre/libre.syms
+++ b/src/libre/libre.syms
@@ -3,6 +3,7 @@ re_is_literal
 re_flags
 re_strerror
 re_perror
+re_is_anchored
 
 ast_print
 ast_print_dot
diff --git a/src/libre/re.c b/src/libre/re.c
index 15af848b5..013e2b58c 100644
--- a/src/libre/re.c
+++ b/src/libre/re.c
@@ -335,3 +335,40 @@ re_is_literal(enum re_dialect dialect, int (*getc)(void *opaque), void *opaque,
 	return -1;
 }
 
+enum re_is_anchored_res
+re_is_anchored(enum re_dialect dialect, re_getchar_fun *getc, void *opaque,
+	enum re_flags flags, struct re_err *err)
+{
+	/* FIXME: copy/pasted from above, factor out common code later. */
+
+	struct ast *ast;
+	const struct dialect *m;
+	int unsatisfiable;
+
+	assert(getc != NULL);
+
+	m = re_dialect(dialect);
+	if (m == NULL) {
+		if (err != NULL) { err->e = RE_EBADDIALECT; }
+		return RE_IS_ANCHORED_ERROR;
+	}
+
+	flags |= m->flags;
+
+	ast = re_parse(dialect, getc, opaque, flags, err, &unsatisfiable);
+	if (ast == NULL) {
+		return RE_IS_ANCHORED_ERROR;
+	}
+
+	/* Copy anchoring flags, ending up with NONE, START, END, or BOTH. */
+	enum re_is_anchored_res res = RE_IS_ANCHORED_NONE;
+	if (ast->expr->flags & AST_FLAG_ANCHORED_START) {
+		res |= RE_IS_ANCHORED_START;
+	}
+	if (ast->expr->flags & AST_FLAG_ANCHORED_END) {
+		res |= RE_IS_ANCHORED_END;
+	}
+
+	ast_free(ast);
+	return res;
+}