fix #468: support snowflake // comments

CHANGELOG.md

@@ -11,6 +11,7 @@ All notable changes to this project will be documented in this file.
 ### Formatting Changes and Bug Fixes
 
 -   `any()` and `all()` will no longer get spaces between the function name and the parenthesis, unless they are a part of a `like any ()` or `like all ()` operator ([#483](https://github.com/tconbeer/sqlfmt/issues/483) - thank you [@damirbk](https://github.com/damirbk)!).
+-   Snowflake's `//` comment markers are now parsed as comments and rewritten to `--` on formatting ([#468](https://github.com/tconbeer/sqlfmt/issues/468) - thank you [@nilsonavp](https://github.com/nilsonavp)!).
 
 ## [0.19.2] - 2023-07-31
 

src/sqlfmt/api.py

@@ -374,17 +374,13 @@ def _perform_safety_check(analyzer: Analyzer, raw_query: Query, result: str) ->
         )
 
     raw_comments = [
-        comment.token.token for line in raw_query.lines for comment in line.comments
+        comment.body for line in raw_query.lines for comment in line.comments
     ]
     result_comments = [
-        comment.token.token for line in result_query.lines for comment in line.comments
+        comment.body for line in result_query.lines for comment in line.comments
     ]
-    stripped_raw = "".join(
-        ["".join(c.split()).replace("--", "").replace("#", "") for c in raw_comments]
-    )
-    stripped_res = "".join(
-        ["".join(c.split()).replace("--", "").replace("#", "") for c in result_comments]
-    )
+    stripped_raw = "".join(["".join(c.split()) for c in raw_comments])
+    stripped_res = "".join(["".join(c.split()) for c in result_comments])
     try:
         assert stripped_raw == stripped_res
     except AssertionError:

src/sqlfmt/comment.py

@@ -16,7 +16,7 @@ class Comment:
     token: Token
     is_standalone: bool
     previous_node: Optional[Node]
-    comment_marker: ClassVar[re.Pattern] = re.compile(r"(--|#|/\*|\{#-?)([^\S\n]*)")
+    comment_marker: ClassVar[re.Pattern] = re.compile(r"(--|#|//|/\*|\{#-?)([^\S\n]*)")
 
     def __str__(self) -> str:
         """
@@ -52,6 +52,18 @@ def _get_marker(self) -> Tuple[str, int]:
         _, len = match.span(2)
         return self.token.token[:epos], len
 
+    def _rewrite_marker(self, marker: str) -> str:
+        """
+        Rewrites the comment marker to the standard --
+
+        The following markers are rewritten:
+            //
+        """
+        if marker == "//":
+            return "--"
+        else:
+            return marker
+
     def _comment_parts(self) -> Tuple[str, str]:
         """
         For a comment, returns a tuple of the comment's marker and its contents
@@ -60,7 +72,7 @@ def _comment_parts(self) -> Tuple[str, str]:
         assert not self.is_multiline
         marker, skipchars = self._get_marker()
         comment_text = self.token.token[skipchars:]
-        return marker, comment_text
+        return self._rewrite_marker(marker), comment_text
 
     @property
     def is_multiline(self) -> bool:
@@ -77,6 +89,13 @@ def is_c_style(self) -> bool:
     def is_inline(self) -> bool:
         return not self.is_standalone and not self.is_multiline and not self.is_c_style
 
+    @property
+    def body(self) -> str:
+        if self.is_multiline:
+            return self.token.token.strip()
+        else:
+            return self._comment_parts()[1].strip()
+
     @property
     def formatting_disabled(self) -> bool:
         if self.previous_node is None:

src/sqlfmt/rules/common.py

@@ -25,6 +25,7 @@ def group(*choices: str) -> str:
 SQL_COMMENT = group(
     r"--[^\r\n]*",
     r"#[^\r\n]*",
+    r"//[^\r\n]*",  # snowflake's js-style double-slash comment
     r"/\*[^*]*\*+(?:[^/*][^*]*\*+)*/",  # simple block comment
 )
 

tests/data/unformatted/128_double_slash_comments.sql

@@ -0,0 +1,17 @@
+-- source: https://github.com/tconbeer/sqlfmt/issues/468
+select * from  {{ ref("events") }} e
+// join from events table
+left join {{ ref("users") }} u
+on u.id = e.user_id
+;
+select "https://sqlfmt.com" as url_not_a_comment
+from foo
+)))))__SQLFMT_OUTPUT__(((((
+-- source: https://github.com/tconbeer/sqlfmt/issues/468
+select *
+from {{ ref("events") }} e
+-- join from events table
+left join {{ ref("users") }} u on u.id = e.user_id
+;
+select "https://sqlfmt.com" as url_not_a_comment
+from foo

tests/functional_tests/test_general_formatting.py

@@ -44,6 +44,7 @@
         "unformatted/125_numeric_constants.sql",
         "unformatted/126_blank_lines.sql",
         "unformatted/127_more_comments.sql",
+        "unformatted/128_double_slash_comments.sql",
         "unformatted/200_base_model.sql",
         "unformatted/201_basic_snapshot.sql",
         "unformatted/202_unpivot_macro.sql",

tests/unit_tests/test_comment.py

@@ -25,6 +25,15 @@ def short_mysql_comment() -> Comment:
     return comment
 
 
+@pytest.fixture
+def short_js_comment() -> Comment:
+    t = Token(
+        type=TokenType.COMMENT, prefix=" ", token="// short comment", spos=0, epos=16
+    )
+    comment = Comment(t, is_standalone=False, previous_node=None)
+    return comment
+
+
 @pytest.fixture
 def nospace_comment() -> Comment:
     t = Token(
@@ -69,38 +78,54 @@ def fmt_disabled_comment() -> Comment:
 
 
 def test_get_marker(
-    short_comment: Comment, short_mysql_comment: Comment, nospace_comment: Comment
+    short_comment: Comment,
+    short_mysql_comment: Comment,
+    nospace_comment: Comment,
+    short_js_comment: Comment,
 ) -> None:
     assert short_comment._get_marker() == ("--", 3)
     assert short_mysql_comment._get_marker() == ("#", 2)
+    assert short_js_comment._get_marker() == ("//", 3)
     assert nospace_comment._get_marker() == ("--", 2)
 
 
 def test_comment_parts(
-    short_comment: Comment, short_mysql_comment: Comment, nospace_comment: Comment
+    short_comment: Comment,
+    short_mysql_comment: Comment,
+    nospace_comment: Comment,
+    short_js_comment: Comment,
 ) -> None:
     assert short_comment._comment_parts() == ("--", "short comment")
     assert short_mysql_comment._comment_parts() == ("#", "short comment")
+    assert short_js_comment._comment_parts() == ("--", "short comment")
     assert nospace_comment._comment_parts() == ("--", "short comment")
 
 
 def test_str_len(
-    short_comment: Comment, short_mysql_comment: Comment, nospace_comment: Comment
+    short_comment: Comment,
+    short_mysql_comment: Comment,
+    nospace_comment: Comment,
+    short_js_comment: Comment,
 ) -> None:
     assert str(short_comment) == short_comment.token.token
     assert str(short_mysql_comment) == short_mysql_comment.token.token
+    assert str(short_js_comment) == f"-- {short_js_comment.body}"
     assert str(nospace_comment) == str(short_comment)
 
     assert len(short_comment) == 16
     assert len(short_mysql_comment) == 15
+    assert len(short_js_comment) == 16
     assert len(nospace_comment) == 16
 
 
 def test_render_inline(
-    short_comment: Comment, nospace_comment: Comment, standalone_comment: Comment
+    short_comment: Comment,
+    nospace_comment: Comment,
+    short_js_comment: Comment,
 ) -> None:
     expected = "  -- short comment"
     assert short_comment.render_inline() == expected
+    assert short_js_comment.render_inline() == expected
     assert nospace_comment.render_inline() == expected
 
 

tests/unit_tests/test_rule.py

@@ -51,6 +51,8 @@ def get_rule(ruleset: List[Rule], rule_name: str) -> Rule:
         (CORE, "comment", "--no-space comment"),
         (CORE, "comment", "# mysql-style # comments"),
         (CORE, "comment", "#nospace"),
+        (CORE, "comment", "// why is this a comment??!"),
+        (CORE, "comment", "//whywhywhy??"),
         (CORE, "comment_start", "/*"),
         (CORE, "comment_end", "*/"),
         (MAIN, "statement_start", "case"),