fix: parsing modulemaps with comments (#1105)

Fixes the tokenizer and modulemap parser to support C99 formatted
comments. This is probably not amazing but we'll be able to move off
this soon: bazelbuild/rules_swift#1212

Author: luispadron

swiftpkg/internal/modulemap_parser/parser.bzl

@@ -26,7 +26,7 @@ def _parse(text):
         collect_result = None
         err = None
 
-        if token.type == tts.newline:
+        if token.type == tts.newline or token.type == tts.comment:
             pass
 
         elif token.type == tts.reserved:

swiftpkg/internal/modulemap_parser/tokenizer.bzl

@@ -36,6 +36,41 @@ def _error(char, msg):
         msg = msg,
     )
 
+def _collect_single_line_comment(chars):
+    collected_chars = []
+
+    for char in chars:
+        if char == "\n":
+            break
+        collected_chars.append(char)
+
+    return _collection_result(
+        chars = collected_chars,
+    )
+
+def _collect_multi_line_comment(chars):
+    chars = chars[2:]
+    collected_chars = ["/", "*"]
+    terminated = False
+
+    for idx, char in enumerate(chars):
+        collected_chars.append(char)
+
+        if char == "*" and idx + 1 < len(chars) and chars[idx + 1] == "/":
+            collected_chars.append("/")
+            terminated = True
+            break
+
+    if not terminated:
+        return _collection_result(
+            chars = collected_chars,
+            errors = [_error("".join(collected_chars), "Unclosed multi-line comment")],
+        )
+    else:
+        return _collection_result(
+            chars = collected_chars,
+        )
+
 def _collect_chars_in_set(chars, target_set):
     collected_chars = []
     for char in chars:
@@ -114,6 +149,14 @@ def _tokenize(text):
         elif sets.contains(character_sets.whitespaces, char):
             pass
 
+        elif char == "/":
+            if idx + 1 < chars_len and chars[idx + 1] == "*":
+                collect_result = _collect_multi_line_comment(chars[idx:])
+            else:
+                collect_result = _collect_single_line_comment(chars[idx:])
+
+            collected_tokens.append(tokens.comment(collect_result.value))
+
         elif sets.contains(character_sets.newlines, char):
             collect_result = _collect_chars_in_set(chars[idx:], character_sets.newlines)
             collected_tokens.append(tokens.newline())

swiftpkg/tests/modulemap_parser/tokenizer_tests.bzl

@@ -82,6 +82,28 @@ def _tokenize_test(ctx):
     result = tokenizer.tokenize(text)
     asserts.equals(env, expected, result, "consume string literals")
 
+    text = "// single line comment 1\n/* single line comment 2 */"
+    expected = tokenizer.result(
+        tokens = [
+            tokens.comment("// single line comment 1"),
+            tokens.newline(),
+            tokens.comment("/* single line comment 2 */"),
+        ],
+    )
+    result = tokenizer.tokenize(text)
+    asserts.equals(env, expected, result, "consume single line comments")
+
+    text = "/*\nmulti line comment\nline 1\n // line 2\n*/\n// single line comment"
+    expected = tokenizer.result(
+        tokens = [
+            tokens.comment("/*\nmulti line comment\nline 1\n // line 2\n*/"),
+            tokens.newline(),
+            tokens.comment("// single line comment"),
+        ],
+    )
+    result = tokenizer.tokenize(text)
+    asserts.equals(env, expected, result, "consume multi line comments")
+
     return unittest.end(env)
 
 tokenize_test = unittest.make(_tokenize_test)