Better void stmt detection for comments in ripper translation

Author: kddnewton

lib/prism/translation/ripper.rb

@@ -385,7 +385,7 @@ def visit_begin_node(node)
             on_stmts_add(on_stmts_new, on_void_stmt)
           else
             body = node.statements.body
-            body.unshift(nil) if semicolon?(location, node.statements.body[0].location)
+            body.unshift(nil) if void_stmt?(location, node.statements.body[0].location)
 
             bounds(node.statements.location)
             visit_statements_node_body(body)
@@ -399,7 +399,7 @@ def visit_begin_node(node)
                 [nil]
               else
                 body = else_clause_node.statements.body
-                body.unshift(nil) if semicolon?(else_clause_node.else_keyword_loc, else_clause_node.statements.body[0].location)
+                body.unshift(nil) if void_stmt?(else_clause_node.else_keyword_loc, else_clause_node.statements.body[0].location)
                 body
               end
 
@@ -421,7 +421,7 @@ def visit_begin_node(node)
           on_bodystmt(visit_statements_node_body([nil]), nil, nil, nil)
         when StatementsNode
           body = [*node.body]
-          body.unshift(nil) if semicolon?(location, body[0].location)
+          body.unshift(nil) if void_stmt?(location, body[0].location)
           stmts = visit_statements_node_body(body)
 
           bounds(node.body.first.location)
@@ -461,7 +461,7 @@ def visit_block_node(node)
             braces ? stmts : on_bodystmt(stmts, nil, nil, nil)
           when StatementsNode
             stmts = node.body.body
-            stmts.unshift(nil) if semicolon?(node.parameters&.location || node.opening_loc, node.body.location)
+            stmts.unshift(nil) if void_stmt?(node.parameters&.location || node.opening_loc, node.body.location)
             stmts = visit_statements_node_body(stmts)
 
             bounds(node.body.location)
@@ -1148,7 +1148,7 @@ def visit_else_node(node)
             [nil]
           else
             body = node.statements.body
-            body.unshift(nil) if semicolon?(node.else_keyword_loc, node.statements.body[0].location)
+            body.unshift(nil) if void_stmt?(node.else_keyword_loc, node.statements.body[0].location)
             body
           end
 
@@ -1181,7 +1181,7 @@ def visit_ensure_node(node)
             [nil]
           else
             body = node.statements.body
-            body.unshift(nil) if semicolon?(node.ensure_keyword_loc, body[0].location)
+            body.unshift(nil) if void_stmt?(node.ensure_keyword_loc, body[0].location)
             body
           end
 
@@ -1762,7 +1762,7 @@ def visit_lambda_node(node)
             braces ? stmts : on_bodystmt(stmts, nil, nil, nil)
           when StatementsNode
             stmts = node.body.body
-            stmts.unshift(nil) if semicolon?(node.parameters&.location || node.opening_loc, node.body.location)
+            stmts.unshift(nil) if void_stmt?(node.parameters&.location || node.opening_loc, node.body.location)
             stmts = visit_statements_node_body(stmts)
 
             bounds(node.body.location)
@@ -2146,13 +2146,20 @@ def visit_redo_node(node)
       # /foo/
       # ^^^^^
       def visit_regular_expression_node(node)
-        bounds(node.content_loc)
-        tstring_content = on_tstring_content(node.content)
+        if node.content.empty?
+          bounds(node.closing_loc)
+          closing = on_regexp_end(node.closing)
 
-        bounds(node.closing_loc)
-        closing = on_regexp_end(node.closing)
+          on_regexp_literal(on_regexp_new, closing)
+        else
+          bounds(node.content_loc)
+          tstring_content = on_tstring_content(node.content)
 
-        on_regexp_literal(on_regexp_add(on_regexp_new, tstring_content), closing)
+          bounds(node.closing_loc)
+          closing = on_regexp_end(node.closing)
+
+          on_regexp_literal(on_regexp_add(on_regexp_new, tstring_content), closing)
+        end
       end
 
       # def foo(bar:); end
@@ -2676,8 +2683,8 @@ def result
       ##########################################################################
 
       # Returns true if there is a semicolon between the two locations.
-      def semicolon?(left, right)
-        source.byteslice(left.end_offset...right.start_offset).include?(";")
+      def void_stmt?(left, right)
+        source.byteslice(left.end_offset...right.start_offset).match?(/[;#]/)
       end
 
       # Visit the string content of a particular node. This method is used to

test/prism/ripper_test.rb

@@ -10,13 +10,13 @@ class RipperTest < TestCase
     base = File.join(__dir__, "fixtures")
     relatives = ENV["FOCUS"] ? [ENV["FOCUS"]] : Dir["**/*.txt", base: base]
 
-    failures = %w[
+    incorrect = %w[
       whitequark/break_block.txt
       whitequark/next_block.txt
       whitequark/return_block.txt
     ]
 
-    skips = failures | %w[
+    skips = incorrect | %w[
       arrays.txt
       blocks.txt
       case.txt
@@ -61,7 +61,6 @@ class RipperTest < TestCase
       seattlerb/call_block_arg_named.txt
       seattlerb/call_trailing_comma.txt
       seattlerb/case_in.txt
-      seattlerb/class_comments.txt
       seattlerb/defn_oneliner_eq2.txt
       seattlerb/defs_oneliner_eq2.txt
       seattlerb/difficult3_.txt
@@ -106,7 +105,6 @@ class RipperTest < TestCase
       seattlerb/mlhs_front_splat.txt
       seattlerb/mlhs_mid_anonsplat.txt
       seattlerb/mlhs_mid_splat.txt
-      seattlerb/module_comments.txt
       seattlerb/parse_line_dstr_escaped_newline.txt
       seattlerb/parse_line_dstr_soft_newline.txt
       seattlerb/parse_line_evstr_after_break.txt
@@ -122,7 +120,6 @@ class RipperTest < TestCase
       seattlerb/yield_call_assocs.txt
       single_method_call_with_bang.txt
       spanning_heredoc.txt
-      spanning_heredoc_newlines.txt
       strings.txt
       symbols.txt
       ternary_operator.txt