feat: Give joins their own token kind

Author: lu-zero

src/formatter.rs

@@ -101,6 +101,10 @@ pub(crate) fn format(
                 formatter.format_newline_reserved_word(token, &mut formatted_query);
                 formatter.previous_reserved_word = Some(token);
             }
+            TokenKind::Join => {
+                formatter.format_newline_reserved_word(token, &mut formatted_query);
+                formatter.previous_reserved_word = Some(token);
+            }
             TokenKind::Reserved => {
                 formatter.format_with_spaces(token, &mut formatted_query);
                 formatter.previous_reserved_word = Some(token);
@@ -210,8 +214,9 @@ impl<'a> Formatter<'a> {
         self.add_new_line(query);
         self.indentation.increase_top_level(span_len);
         query.push_str(&self.equalize_whitespace(&self.format_reserved_word(token.value)));
-        if !(!["select", "from"].contains(&token.value.to_lowercase().as_str())
-            && self.options.inline_first_top_level)
+        let new_line = ["select", "from"].contains(&token.value.to_lowercase().as_str())
+            || !self.options.inline_first_top_level;
+        if new_line
             && self
                 .options
                 .max_inline_top_level

src/tokenizer.rs

@@ -69,6 +69,7 @@ pub(crate) enum TokenKind {
     Number,
     Placeholder,
     Word,
+    Join,
 }
 
 #[derive(Debug, Clone)]
@@ -379,6 +380,7 @@ fn get_reserved_word_token<'a>(
     alt((
         get_top_level_reserved_token(last_reserved_top_level_token),
         get_newline_reserved_token(last_reserved_token),
+        get_join_token(),
         get_top_level_reserved_token_no_indent,
         get_plain_reserved_token,
     ))
@@ -497,16 +499,11 @@ fn get_top_level_reserved_token<'a>(
     }
 }
 
-fn get_newline_reserved_token<'a>(
-    last_reserved_token: Option<Token<'a>>,
-) -> impl Parser<&'a str, Token<'a>, ContextError> {
+fn get_join_token<'a>() -> impl Parser<&'a str, Token<'a>, ContextError> {
     move |input: &mut &'a str| {
         let uc_input: String = get_uc_words(input, 3);
         let mut uc_input = uc_input.as_str();
 
-        // We have to break up the alternatives into multiple subsets
-        // to avoid exceeding the alt() 21 element limit.
-
         // Standard SQL joins
         let standard_joins = alt((
             terminated("JOIN", end_of_word),
@@ -544,6 +541,37 @@ fn get_newline_reserved_token<'a>(
             terminated("GLOBAL FULL JOIN", end_of_word),
         ));
 
+        // Combine all parsers
+        let result: PResult<&str> =
+            alt((standard_joins, specific_joins, special_joins)).parse_next(&mut uc_input);
+
+        if let Ok(token) = result {
+            let final_word = token.split(' ').last().unwrap();
+            let input_end_pos =
+                input.to_ascii_uppercase().find(final_word).unwrap() + final_word.len();
+            let token = input.next_slice(input_end_pos);
+            let kind = TokenKind::Join;
+            Ok(Token {
+                kind,
+                value: token,
+                key: None,
+            })
+        } else {
+            Err(ErrMode::from_error_kind(input, ErrorKind::Alt))
+        }
+    }
+}
+
+fn get_newline_reserved_token<'a>(
+    last_reserved_token: Option<Token<'a>>,
+) -> impl Parser<&'a str, Token<'a>, ContextError> {
+    move |input: &mut &'a str| {
+        let uc_input: String = get_uc_words(input, 3);
+        let mut uc_input = uc_input.as_str();
+
+        // We have to break up the alternatives into multiple subsets
+        // to avoid exceeding the alt() 21 element limit.
+
         // Legacy and logical operators
         let operators = alt((
             terminated("CROSS APPLY", end_of_word),
@@ -565,14 +593,7 @@ fn get_newline_reserved_token<'a>(
         ));
 
         // Combine all parsers
-        let result: PResult<&str> = alt((
-            standard_joins,
-            specific_joins,
-            special_joins,
-            operators,
-            alter_table_actions,
-        ))
-        .parse_next(&mut uc_input);
+        let result: PResult<&str> = alt((operators, alter_table_actions)).parse_next(&mut uc_input);
 
         if let Ok(token) = result {
             let final_word = token.split(' ').last().unwrap();