From 8b63780a9d6b7debd4de656f68a0c666cfc2bda1 Mon Sep 17 00:00:00 2001
From: Earlopain <14981592+Earlopain@users.noreply.github.com>
Date: Tue, 10 Feb 2026 18:28:18 +0100
Subject: [PATCH] Remove unneeded exit conditions for `parse_stream`

As I understand it, the code operates line-by-line. When finding `__END__`, it has to decide if it is
the end marker or just appears as part of of a multi-line literal (see tests).

For a future PR I plan to make, `lex_modes` may be empty after parsing finished, even if it was syntax invalid.
Currently there will _always_ be errors when lex modes are not popped. So these extra checks are redundant.

Basically, if we found `__END__` on a line by itself and there are syntax errors, it's impossible to be the end marker.
---
 src/prism.c                         | 24 +-------------
 test/prism/api/parse_stream_test.rb | 51 +++++++++++++++++++++++++----
 2 files changed, 45 insertions(+), 30 deletions(-)

diff --git a/src/prism.c b/src/prism.c
index 8f3617adce..34e5d38b0a 100644
--- a/src/prism.c
+++ b/src/prism.c
@@ -22282,28 +22282,6 @@ pm_parse_stream_read(pm_buffer_t *buffer, void *stream, pm_parse_stream_fgets_t
 #undef LINE_SIZE
 }
 
-/**
- * Determine if there was an unterminated heredoc at the end of the input, which
- * would mean the stream isn't finished and we should keep reading.
- *
- * For the other lex modes we can check if the lex mode has been closed, but for
- * heredocs when we hit EOF we close the lex mode and then go back to parse the
- * rest of the line after the heredoc declaration so that we get more of the
- * syntax tree.
- */
-static bool
-pm_parse_stream_unterminated_heredoc_p(pm_parser_t *parser) {
-    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) parser->error_list.head;
-
-    for (; diagnostic != NULL; diagnostic = (pm_diagnostic_t *) diagnostic->node.next) {
-        if (diagnostic->diag_id == PM_ERR_HEREDOC_TERM) {
-            return true;
-        }
-    }
-
-    return false;
-}
-
 /**
  * Parse a stream of Ruby source and return the tree.
  *
@@ -22319,7 +22297,7 @@ pm_parse_stream(pm_parser_t *parser, pm_buffer_t *buffer, void *stream, pm_parse
     pm_parser_init(parser, (const uint8_t *) pm_buffer_value(buffer), pm_buffer_length(buffer), options);
     pm_node_t *node = pm_parse(parser);
 
-    while (!eof && parser->error_list.size > 0 && (parser->lex_modes.index > 0 || pm_parse_stream_unterminated_heredoc_p(parser))) {
+    while (!eof && parser->error_list.size > 0) {
         pm_node_destroy(parser, node);
         eof = pm_parse_stream_read(buffer, stream, stream_fgets, stream_feof);
 
diff --git a/test/prism/api/parse_stream_test.rb b/test/prism/api/parse_stream_test.rb
index 1c068c617c..3bc86fbd61 100644
--- a/test/prism/api/parse_stream_test.rb
+++ b/test/prism/api/parse_stream_test.rb
@@ -30,16 +30,28 @@ def test_multi_read
     end
 
     def test___END__
-      io = StringIO.new("1 + 2\n3 + 4\n__END__\n5 + 6")
+      io = StringIO.new(<<~RUBY)
+        1 + 2
+        3 + 4
+        __END__
+        5 + 6
+      RUBY
       result = Prism.parse_stream(io)
 
       assert result.success?
       assert_equal 2, result.value.statements.body.length
-      assert_equal "5 + 6", io.read
+      assert_equal "5 + 6\n", io.read
     end
 
     def test_false___END___in_string
-      io = StringIO.new("1 + 2\n3 + 4\n\"\n__END__\n\"\n5 + 6")
+      io = StringIO.new(<<~RUBY)
+        1 + 2
+        3 + 4
+        "
+        __END__
+        "
+        5 + 6
+      RUBY
       result = Prism.parse_stream(io)
 
       assert result.success?
@@ -47,7 +59,14 @@ def test_false___END___in_string
     end
 
     def test_false___END___in_regexp
-      io = StringIO.new("1 + 2\n3 + 4\n/\n__END__\n/\n5 + 6")
+      io = StringIO.new(<<~RUBY)
+        1 + 2
+        3 + 4
+        /
+        __END__
+        /
+        5 + 6
+      RUBY
       result = Prism.parse_stream(io)
 
       assert result.success?
@@ -55,7 +74,14 @@ def test_false___END___in_regexp
     end
 
     def test_false___END___in_list
-      io = StringIO.new("1 + 2\n3 + 4\n%w[\n__END__\n]\n5 + 6")
+      io = StringIO.new(<<~RUBY)
+        1 + 2
+        3 + 4
+        %w[
+        __END__
+        ]
+        5 + 6
+      RUBY
       result = Prism.parse_stream(io)
 
       assert result.success?
@@ -63,7 +89,14 @@ def test_false___END___in_list
     end
 
     def test_false___END___in_heredoc
-      io = StringIO.new("1 + 2\n3 + 4\n<<-EOF\n__END__\nEOF\n5 + 6")
+      io = StringIO.new(<<~RUBY)
+        1 + 2
+        3 + 4
+        <<-EOF
+        __END__
+        EOF
+        5 + 6
+      RUBY
       result = Prism.parse_stream(io)
 
       assert result.success?
@@ -71,7 +104,11 @@ def test_false___END___in_heredoc
     end
 
     def test_nul_bytes
-      io = StringIO.new("1 # \0\0\0 \n2 # \0\0\0\n3")
+      io = StringIO.new(<<~RUBY)
+        1 # \0\0\0\t
+        2 # \0\0\0
+        3
+      RUBY
       result = Prism.parse_stream(io)
 
       assert result.success?