Parser: Improve extract.c to strip <% # comments %> with space

src/extract.c

@@ -39,7 +39,26 @@ void herb_extract_ruby_to_buffer_with_semicolons(const char* source, hb_buffer_T
 
       case TOKEN_ERB_CONTENT: {
         if (skip_erb_content == false) {
-          hb_buffer_append(output, token->value);
+          bool is_inline_comment = false;
+
+          if (!is_comment_tag && token->value != NULL) {
+            const char* content = token->value;
+
+            while (*content == ' ' || *content == '\t') {
+              content++;
+            }
+
+            if (*content == '#' && token->location.start.line == token->location.end.line) {
+              is_comment_tag = true;
+              is_inline_comment = true;
+            }
+          }
+
+          if (is_inline_comment) {
+            hb_buffer_append_whitespace(output, range_length(token->range));
+          } else {
+            hb_buffer_append(output, token->value);
+          }
         } else {
           hb_buffer_append_whitespace(output, range_length(token->range));
         }

test/c/test_extract.c

@@ -89,6 +89,62 @@ TEST(extract_ruby_issue_135_if_without_condition)
   free(result);
 END
 
+TEST(extract_ruby_inline_comment_same_line)
+  char* source = "<% if true %><% # Comment here %><% end %>";
+  char* result = herb_extract_ruby_with_semicolons(source);
+
+  ck_assert_str_eq(result, "   if true  ;                       end   ");
+
+  free(result);
+END
+
+TEST(extract_ruby_inline_comment_with_newline)
+  char* source = "<% if true %><% # Comment here %>\n<% end %>";
+  char* result = herb_extract_ruby_with_semicolons(source);
+
+  char expected[] = "   if true  ;                    \n   end   ";
+  ck_assert_str_eq(result, expected);
+
+  free(result);
+END
+
+TEST(extract_ruby_inline_comment_with_spaces)
+  char* source = "<%  # Comment %> <% code %>";
+  char* result = herb_extract_ruby_with_semicolons(source);
+
+  ck_assert_str_eq(result, "                    code   ");
+
+  free(result);
+END
+
+TEST(extract_ruby_inline_comment_multiline)
+  char* source = "<% # Comment\nmore %> <% code %>";
+  char* result = herb_extract_ruby_with_semicolons(source);
+
+  char expected[] = "   # Comment\nmore  ;    code   ";
+  ck_assert_str_eq(result, expected);
+
+  free(result);
+END
+
+TEST(extract_ruby_inline_comment_between_code)
+  char* source = "<% if true %><% # Comment here %><%= hello %><% end %>";
+  char* result = herb_extract_ruby_with_semicolons(source);
+
+  ck_assert_str_eq(result, "   if true  ;                        hello  ;   end   ");
+
+  free(result);
+END
+
+TEST(extract_ruby_inline_comment_complex)
+  char* source = "<% # Comment here %><% if true %><% # Comment here %><%= hello %><% end %>";
+  char* result = herb_extract_ruby_with_semicolons(source);
+
+  ck_assert_str_eq(result, "                       if true  ;                        hello  ;   end   ");
+
+  free(result);
+END
+
 TCase *extract_tests(void) {
   TCase *extract = tcase_create("Extract");
 
@@ -101,6 +157,12 @@ TCase *extract_tests(void) {
   tcase_add_test(extract, extract_ruby_empty_erb_same_line);
   tcase_add_test(extract, extract_ruby_comments_skipped);
   tcase_add_test(extract, extract_ruby_issue_135_if_without_condition);
+  tcase_add_test(extract, extract_ruby_inline_comment_same_line);
+  tcase_add_test(extract, extract_ruby_inline_comment_with_newline);
+  tcase_add_test(extract, extract_ruby_inline_comment_with_spaces);
+  tcase_add_test(extract, extract_ruby_inline_comment_multiline);
+  tcase_add_test(extract, extract_ruby_inline_comment_between_code);
+  tcase_add_test(extract, extract_ruby_inline_comment_complex);
 
   return extract;
 }

test/parser/erb_test.rb

@@ -228,5 +228,35 @@ class ERBTest < Minitest::Spec
         <% end %>
       HTML
     end
+
+    test "inline ruby comment on same line" do
+      assert_parsed_snapshot(%(<% if true %><% # Comment here %><% end %>))
+    end
+
+    test "inline ruby comment with newline" do
+      assert_parsed_snapshot(<<~HTML)
+        <% if true %><% # Comment here %>
+        <% end %>
+      HTML
+    end
+
+    test "inline ruby comment between code" do
+      assert_parsed_snapshot(%(<% if true %><% # Comment here %><%= hello %><% end %>))
+    end
+
+    test "inline ruby comment before and between code" do
+      assert_parsed_snapshot(%(<% # Comment here %><% if true %><% # Comment here %><%= hello %><% end %>))
+    end
+
+    test "inline ruby comment with spaces" do
+      assert_parsed_snapshot(%(<%  # Comment %> <% code %>))
+    end
+
+    test "inline ruby comment multiline" do
+      assert_parsed_snapshot(<<~HTML)
+        <% # Comment
+        more %> <% code %>
+      HTML
+    end
   end
 end