Merge pull request #473 from katef/kate/consolidate-unsupported

Consolidate RE_EUNSUPPORTED syntax errors
katef · May 30, 2024 · 201637d · 201637d
2 parents 80e9311 + 61772d0
commit 201637d
Show file tree

Hide file tree

Showing 21 changed files with 104 additions and 58 deletions.
diff --git a/include/re/re.h b/include/re/re.h
@@ -46,8 +46,6 @@ enum re_errno {
 	RE_EERRNO       =  1 | RE_MISC,
 	RE_EBADDIALECT  =  2 | RE_MISC,
 	RE_EBADGROUP    =  3 | RE_MISC,
-	RE_EUNSUPCAPTUR =  4 | RE_MISC,
-	RE_EUNSUPPPCRE  =  5 | RE_MISC,
 
 	RE_ENEGRANGE    =  0 | RE_MARK | RE_GROUP,
 	RE_ENEGCOUNT    =  1 | RE_MARK | RE_GROUP,

diff --git a/src/libre/ast_analysis.c b/src/libre/ast_analysis.c
@@ -783,7 +783,7 @@ analysis_iter_anchoring(struct anchoring_env *env, struct ast_expr *n)
 				    "%s: LITERAL: rejecting non-optional newline match after $ as unsupported\n",
 				    __func__);
 				set_flags(n, AST_FLAG_UNSATISFIABLE);
-				return AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE;
+				return AST_ANALYSIS_ERROR_UNSUPPORTED;
 			}
 		}
 		break;
@@ -1010,8 +1010,7 @@ analysis_iter_anchoring(struct anchoring_env *env, struct ast_expr *n)
 					all_set_past_always_consuming &= child_env.past_always_consuming;
 					any_sat = 1;
 				}
-			} else if (res == AST_ANALYSIS_ERROR_UNSUPPORTED_CAPTURE
-			    || res == AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE) {
+			} else if (res == AST_ANALYSIS_ERROR_UNSUPPORTED) {
 				continue;
 			} else {
 				return res;
@@ -1201,7 +1200,7 @@ analysis_iter_anchoring(struct anchoring_env *env, struct ast_expr *n)
 			    "%s: SUBTRACT: rejecting non-optional newline match after $ as unsupported\n",
 			    __func__);
 			set_flags(n, AST_FLAG_UNSATISFIABLE);
-			return AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE;
+			return AST_ANALYSIS_ERROR_UNSUPPORTED;
 		}
 
 		if (res != AST_ANALYSIS_OK) {
@@ -1287,9 +1286,9 @@ analysis_iter_reverse_anchoring(struct anchoring_env *env, struct ast_expr *n)
 			if (env->followed_by_consuming) {
 				if (env->followed_by_consuming_newline) {
 					LOG(3 - LOG_ANCHORING,
-					    "%s: END anchor & followed_by_consuming, returning UNSUPPORTED_PCRE\n",
+					    "%s: END anchor & followed_by_consuming, returning UNSUPPORTED\n",
 					    __func__);
-					return AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE;
+					return AST_ANALYSIS_ERROR_UNSUPPORTED;
 				} else {
 					LOG(3 - LOG_ANCHORING,
 					    "%s: END anchor & followed_by_consuming, setting UNSATISFIABLE\n",
@@ -1446,9 +1445,8 @@ analysis_iter_reverse_anchoring(struct anchoring_env *env, struct ast_expr *n)
 					any_set_followed_by_consuming_newline |= child_env.followed_by_consuming_newline;
 					any_sat = 1;
 				}
-			} else if (res == AST_ANALYSIS_ERROR_UNSUPPORTED_CAPTURE
-			    || res == AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE) {
-				LOG(3 - LOG_ANCHORING, "%s: got res of UNSUPPORTED_*, bubbling up\n", __func__);
+			} else if (res == AST_ANALYSIS_ERROR_UNSUPPORTED) {
+				LOG(3 - LOG_ANCHORING, "%s: got res of UNSUPPORTED, bubbling up\n", __func__);
 				return res;
 			} else {
 				return res;
@@ -1883,7 +1881,7 @@ analysis_iter_repetition(struct ast_expr *n, struct ast_expr *outermost_repeat_p
 					 *
 					 * An example input that triggers this is '^(($)|)+$' . */
 					set_flags(n, AST_FLAG_UNSATISFIABLE);
-					return AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE;
+					return AST_ANALYSIS_ERROR_UNSUPPORTED;
 				}
 			}
 
@@ -1994,7 +1992,7 @@ analysis_iter_repetition(struct ast_expr *n, struct ast_expr *outermost_repeat_p
 			    && repeat_plus_ancestor->u.repeat.max == AST_COUNT_UNBOUNDED);
 			LOG(3 - LOG_REPETITION_CASES,
 			    "%s: not yet implemented, skipping\n", __func__);
-			/* return AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE; */
+			/* return AST_ANALYSIS_ERROR_UNSUPPORTED; */
 		}
 
 		res = analysis_iter_repetition(n->u.group.e, outermost_repeat_parent,

diff --git a/src/libre/ast_analysis.h b/src/libre/ast_analysis.h
@@ -31,8 +31,7 @@ enum ast_analysis_res {
 
 	AST_ANALYSIS_ERROR_NULL   = -1,
 	AST_ANALYSIS_ERROR_MEMORY = -2,
-	AST_ANALYSIS_ERROR_UNSUPPORTED_CAPTURE = -3,
-	AST_ANALYSIS_ERROR_UNSUPPORTED_PCRE = -4
+	AST_ANALYSIS_ERROR_UNSUPPORTED = -3
 };
 
 enum ast_analysis_res

diff --git a/src/libre/dialect/comp.h b/src/libre/dialect/comp.h
@@ -17,7 +17,7 @@ typedef struct ast *
 re_dialect_parse_fun(re_getchar_fun *getchar, void *opaque,
 	const struct fsm_options *opt,
 	enum re_flags flags, int overlap,
-	struct re_err *err);
+	struct re_err *err, struct re_pos *end);
 
 re_dialect_parse_fun parse_re_literal;
 re_dialect_parse_fun parse_re_glob;

diff --git a/src/libre/dialect/glob/parser.c b/src/libre/dialect/glob/parser.c
diff --git a/src/libre/dialect/glob/parser.h b/src/libre/dialect/glob/parser.h
diff --git a/src/libre/dialect/like/parser.c b/src/libre/dialect/like/parser.c
diff --git a/src/libre/dialect/like/parser.h b/src/libre/dialect/like/parser.h
diff --git a/src/libre/dialect/literal/parser.c b/src/libre/dialect/literal/parser.c
diff --git a/src/libre/dialect/literal/parser.h b/src/libre/dialect/literal/parser.h
diff --git a/src/libre/dialect/native/parser.c b/src/libre/dialect/native/parser.c
diff --git a/src/libre/dialect/native/parser.h b/src/libre/dialect/native/parser.h
diff --git a/src/libre/dialect/pcre/parser.c b/src/libre/dialect/pcre/parser.c
diff --git a/src/libre/dialect/pcre/parser.h b/src/libre/dialect/pcre/parser.h
diff --git a/src/libre/dialect/sql/parser.c b/src/libre/dialect/sql/parser.c
diff --git a/src/libre/dialect/sql/parser.h b/src/libre/dialect/sql/parser.h
diff --git a/src/libre/parser.act b/src/libre/parser.act
@@ -1070,7 +1070,7 @@
 	DIALECT_PARSE(re_getchar_fun *f, void *opaque,
 		const struct fsm_options *opt,
 		enum re_flags flags, int overlap,
-		struct re_err *err)
+		struct re_err *err, struct re_pos *end)
 	{
 		struct ast *ast;
 
@@ -1083,6 +1083,7 @@
 		struct LX_STATE *lx;
 
 		assert(f != NULL);
+		assert(end != NULL);
 
 		ast = ast_new();
 
@@ -1136,6 +1137,9 @@
 
 		DIALECT_ENTRY(&flags, lex_state, act_state, err, &ast->expr);
 
+		/* we output this for reporting errors found through AST analysis */
+		mark(end, &lx->end);
+
 		lx->free(lx->buf_opaque);
 
 		if (err->e != RE_ESUCCESS) {