refactor: comments

g4/SecLangLexer.g4

@@ -1384,13 +1384,16 @@ OPERATOR_QUOTED_STRING
     : (('\\"') | ~([" @!])) (('\\"')|~('"'))* -> pushMode(DEFAULT_MODE)
     ;
 
-
 mode COMMENT_MODE;
 
 COMMENT
-   : (~[\r\n] | '\\' '\r'? '\n')+
+   : (~[\r\n])+
    ;
 
-NEWLINE_COMMENT
+HASH_COMMENT_BLOCK
    : '\r'? '\n' -> skip,popMode
+   ;
+
+BLOCK_COMMENT_END
+   : '\r'? '\n' '\r'? '\n' -> popMode
    ;

g4/SecLangParser.g4

@@ -25,19 +25,23 @@ configuration
      ;
 
 stmt:
-    comment* engine_config_rule_directive variables operator actions?
-    | comment* rule_script_directive file_path actions?
-    | comment* rule_script_directive QUOTE file_path QUOTE actions?
-    | comment* remove_rule_by_id remove_rule_by_id_values+
-    | comment* string_remove_rules string_remove_rules_values
-    | comment* string_remove_rules QUOTE string_remove_rules_values QUOTE
-    | comment* update_target_rules update_target_rules_values update_variables
-    | comment* update_target_rules QUOTE update_target_rules_values QUOTE update_variables
-    | comment* update_target_rules update_target_rules_values update_variables PIPE new_target
-    | comment* update_target_rules QUOTE update_target_rules_values QUOTE update_variables PIPE new_target
-    | comment* update_action_rule id actions
-    | comment* engine_config_directive
-    | comment+;
+    comment_block? engine_config_rule_directive variables operator actions?
+    | comment_block? rule_script_directive file_path actions?
+    | comment_block? rule_script_directive QUOTE file_path QUOTE actions?
+    | comment_block? remove_rule_by_id remove_rule_by_id_values+
+    | comment_block? string_remove_rules string_remove_rules_values
+    | comment_block? string_remove_rules QUOTE string_remove_rules_values QUOTE
+    | comment_block? update_target_rules update_target_rules_values update_variables
+    | comment_block? update_target_rules QUOTE update_target_rules_values QUOTE update_variables
+    | comment_block? update_target_rules update_target_rules_values update_variables PIPE new_target
+    | comment_block? update_target_rules QUOTE update_target_rules_values QUOTE update_variables PIPE new_target
+    | comment_block? update_action_rule id actions
+    | comment_block? engine_config_directive
+    | comment_block+;
+
+comment_block:
+    comment+ (BLOCK_COMMENT_END | EOF)?
+    ;
 
 comment:
     HASH COMMENT?

listener.go

@@ -14,6 +14,7 @@ import (
 
 type ParserResult struct {
 	Comments              []string `yaml:"comments"`
+	CommentsBlocks        int      `yaml:"comment_blocks"`
 	Variables             []string `yaml:"variables"`
 	NegatedVarCount       int      `yaml:"negated_var_count"`
 	CollectionLengthCount int      `yaml:"collection_length_count"`
@@ -197,3 +198,7 @@ func (l *TreeShapeListener) EnterComment(ctx *parser.CommentContext) {
 func (l *TreeShapeListener) EnterTransformation_action_value(ctx *parser.Transformation_action_valueContext) {
 	l.results.DirectiveValues = append(l.results.DirectiveValues, ctx.GetText())
 }
+
+func (l *TreeShapeListener) EnterComment_block(ctx *parser.Comment_blockContext) {
+	l.results.CommentsBlocks++
+}

src/seclang_parser/SecLangParserListener.py

@@ -26,6 +26,15 @@ def exitStmt(self, ctx:SecLangParser.StmtContext):
         pass
 
 
+    # Enter a parse tree produced by SecLangParser#comment_block.
+    def enterComment_block(self, ctx:SecLangParser.Comment_blockContext):
+        pass
+
+    # Exit a parse tree produced by SecLangParser#comment_block.
+    def exitComment_block(self, ctx:SecLangParser.Comment_blockContext):
+        pass
+
+
     # Enter a parse tree produced by SecLangParser#comment.
     def enterComment(self, ctx:SecLangParser.CommentContext):
         pass

src/seclang_parser/listener.py

@@ -14,6 +14,7 @@ class ParserResult:
     """Stores results from parsing for test validation."""
 
     comments: list[str] = field(default_factory=list)
+    comment_blocks: int = 0
     variables: list[str] = field(default_factory=list)
     negated_var_count: int = 0
     collection_length_count: int = 0
@@ -133,6 +134,10 @@ def enterComment(self, ctx: SecLangParser.CommentContext):
             self.results.comments.append(ctx.COMMENT().getText())
         else:
             self.results.comments.append("")
+        # print(ctx.stop.getLine())
 
     def enterTransformation_action_value(self, ctx: SecLangParser.Transformation_action_valueContext):
-        self.results.directive_values.append(ctx.getText())
+        self.results.directive_values.append(ctx.getText())
+
+    def enterComment_block(self, ctx: SecLangParser.Comment_blockContext):
+        self.results.comment_blocks += 1

testdata/test_44_comments.conf

@@ -0,0 +1,7 @@
+#
+# This is a comment
+#
+# This is a multiline \
+# comment
+#
+# This is a multiline \

testdata/tests.yaml

@@ -214,6 +214,7 @@ check_output_tests:
     expected_result:
       comments:
         - " Test comment"
+      comment_blocks: 1
       variables:
         - REQUEST_FILENAME
         - REQUEST_URI
@@ -273,6 +274,7 @@ check_output_tests:
         - " The variable is a numerical representation of the CRS version number."
         - " E.g., v3.0.0 is represented as 300."
         - ""
+      comment_blocks: 1
       variables:
         - REQUEST_URI
       negated_var_count: 2
@@ -343,6 +345,7 @@ check_output_tests:
       comments:
         - " Force body variable"
         - " Force body processor URLENCODED"
+      comment_blocks: 2
       variables:
         - REQBODY_PROCESSOR
         - REQBODY_PROCESSOR
@@ -420,3 +423,16 @@ check_output_tests:
         - "1"
         - "/%{tx.0}/"
         - "%{tx.critical_anomaly_score}"
+  testdata/test_44_comments.conf:
+    error_count: 0
+    comment: ""
+    expected_result:
+      comments:
+        - ""
+        - " This is a comment"
+        - ""
+        - " This is a multiline \\"
+        - " comment"
+        - ""
+        - " This is a multiline \\"
+      comment_blocks: 1