ide: add builtin operators & parser fix

crates/squawk_ide/src/generated/keywords.rs

@@ -1,3 +1,6 @@
+// Generated via:
+//   cargo xtask codegen
+
 pub(crate) const RESERVED_KEYWORDS: &[&str] = &[
     "all",
     "analyse",

crates/squawk_ide/src/quote.rs

@@ -87,27 +87,29 @@ pub(crate) fn normalize_identifier(text: &str) -> String {
 
 #[cfg(test)]
 mod tests {
+    use insta::assert_snapshot;
+
     use super::*;
 
     #[test]
     fn quote_column_alias_handles_embedded_quotes() {
-        assert_eq!(quote_column_alias(r#"foo"bar"#), r#""foo""bar""#);
+        assert_snapshot!(quote_column_alias(r#"foo"bar"#), @r#""foo""bar""#);
     }
 
     #[test]
     fn quote_column_alias_doesnt_quote_reserved_words() {
         // Keywords are allowed as column labels in AS clauses
-        assert_eq!(quote_column_alias("case"), "case");
-        assert_eq!(quote_column_alias("array"), "array");
+        assert_snapshot!(quote_column_alias("case"), @"case");
+        assert_snapshot!(quote_column_alias("array"), @"array");
     }
 
     #[test]
     fn quote_column_alias_doesnt_quote_simple_identifiers() {
-        assert_eq!(quote_column_alias("col_name"), "col_name");
+        assert_snapshot!(quote_column_alias("col_name"), @"col_name");
     }
 
     #[test]
     fn quote_column_alias_handles_special_column_name() {
-        assert_eq!(quote_column_alias("?column?"), r#""?column?""#);
+        assert_snapshot!(quote_column_alias("?column?"), @r#""?column?""#);
     }
 }

crates/squawk_parser/src/grammar.rs

@@ -1653,7 +1653,7 @@ fn path_segment(p: &mut Parser<'_>, kind: SyntaxKind) {
     // TODO: does this need to be flagged?
     // Might want to disallow operators in some paths.
     // Like `create table +()` doesn't make sense.
-    if !p.at(OPERATOR_KW) && current_operator(p).is_some() {
+    if p.at_ts(OPERATOR_FIRST) && !p.at(OPERATOR_KW) && current_operator(p).is_some() {
         // check for operator kw so we can parse things like:
         //   create table operator();
 

crates/squawk_parser/tests/data/ok/create_function.sql

@@ -415,3 +415,9 @@ create function f (foo.bar.buzz.boo, t.u[], bool)
 returns void
 as '' language sql;
 
+-- overlaps doesn't need quoting
+create function overlaps() returns int8 as 'select 1' language sql;
+
+-- like doesn't need quoting
+create function like() returns int8 as 'select 1' language sql;
+

crates/squawk_parser/tests/snapshots/tests__create_function_ok.snap

@@ -3225,3 +3225,77 @@ SOURCE_FILE
         SQL_KW "sql"
   SEMICOLON ";"
   WHITESPACE "\n\n"
+  COMMENT "-- overlaps doesn't need quoting"
+  WHITESPACE "\n"
+  CREATE_FUNCTION
+    CREATE_KW "create"
+    WHITESPACE " "
+    FUNCTION_KW "function"
+    WHITESPACE " "
+    PATH
+      PATH_SEGMENT
+        NAME
+          OVERLAPS_KW "overlaps"
+    PARAM_LIST
+      L_PAREN "("
+      R_PAREN ")"
+    WHITESPACE " "
+    RET_TYPE
+      RETURNS_KW "returns"
+      WHITESPACE " "
+      PATH_TYPE
+        PATH
+          PATH_SEGMENT
+            NAME_REF
+              IDENT "int8"
+    WHITESPACE " "
+    FUNC_OPTION_LIST
+      AS_FUNC_OPTION
+        AS_KW "as"
+        WHITESPACE " "
+        LITERAL
+          STRING "'select 1'"
+      WHITESPACE " "
+      LANGUAGE_FUNC_OPTION
+        LANGUAGE_KW "language"
+        WHITESPACE " "
+        SQL_KW "sql"
+  SEMICOLON ";"
+  WHITESPACE "\n\n"
+  COMMENT "-- like doesn't need quoting"
+  WHITESPACE "\n"
+  CREATE_FUNCTION
+    CREATE_KW "create"
+    WHITESPACE " "
+    FUNCTION_KW "function"
+    WHITESPACE " "
+    PATH
+      PATH_SEGMENT
+        NAME
+          LIKE_KW "like"
+    PARAM_LIST
+      L_PAREN "("
+      R_PAREN ")"
+    WHITESPACE " "
+    RET_TYPE
+      RETURNS_KW "returns"
+      WHITESPACE " "
+      PATH_TYPE
+        PATH
+          PATH_SEGMENT
+            NAME_REF
+              IDENT "int8"
+    WHITESPACE " "
+    FUNC_OPTION_LIST
+      AS_FUNC_OPTION
+        AS_KW "as"
+        WHITESPACE " "
+        LITERAL
+          STRING "'select 1'"
+      WHITESPACE " "
+      LANGUAGE_FUNC_OPTION
+        LANGUAGE_KW "language"
+        WHITESPACE " "
+        SQL_KW "sql"
+  SEMICOLON ";"
+  WHITESPACE "\n\n"

crates/xtask/src/codegen.rs

@@ -217,21 +217,29 @@ fn generate_kind_src(
     }
 }
 
+const PRELUDE: &str = "\
+// Generated via:
+//   cargo xtask codegen
+
+";
+
 fn generate_reserved_keywords_array(reserved_keywords: &[String]) -> Result<String> {
     let mut reserved_keywords = reserved_keywords
         .iter()
         .map(|x| x.to_lowercase())
         .collect::<Vec<_>>();
     reserved_keywords.sort();
 
-    Ok(reformat(
+    let output = reformat(
         quote! {
             pub(crate) const RESERVED_KEYWORDS: &[&str] = &[
                 #(#reserved_keywords),*
             ];
         }
         .to_string(),
-    ))
+    );
+
+    Ok(format!("{PRELUDE}{output}"))
 }
 
 fn generate_syntax_kinds(grammar: KindsSrc) -> Result<String> {
@@ -291,7 +299,7 @@ fn generate_syntax_kinds(grammar: KindsSrc) -> Result<String> {
         .map(|name| format_ident!("{}", name))
         .collect::<Vec<_>>();
 
-    Ok(reformat(reformat(
+    let output = reformat(reformat(
         quote! {
             #![allow(bad_style, missing_docs, clippy::upper_case_acronyms)]
 
@@ -326,7 +334,9 @@ fn generate_syntax_kinds(grammar: KindsSrc) -> Result<String> {
             }
         }
         .to_string(),
-    ).replace("#[space_hack]", "")))
+    ).replace("#[space_hack]", ""));
+
+    Ok(format!("{PRELUDE}{output}"))
 }
 
 fn generate_token_sets(keyword_kinds: &KeywordKinds) -> Result<String> {
@@ -363,7 +373,7 @@ fn generate_token_sets(keyword_kinds: &KeywordKinds) -> Result<String> {
         .map(|key| format_ident!("{}_KW", key.to_case(Case::UpperSnake)))
         .collect::<Vec<_>>();
 
-    Ok(reformat(
+    let output = reformat(
         quote! {
             use crate::syntax_kind::SyntaxKind;
             use crate::token_set::TokenSet;
@@ -394,7 +404,9 @@ fn generate_token_sets(keyword_kinds: &KeywordKinds) -> Result<String> {
         }
         .to_string(),
     )
-    .replace("pub(crate)", "\npub(crate)"))
+    .replace("pub(crate)", "\npub(crate)");
+
+    Ok(format!("{PRELUDE}{output}"))
 }
 
 #[derive(Debug, Default)]
@@ -879,7 +891,8 @@ fn generate_nodes(nodes: &[AstNodeSrc], enums: &[AstEnumSrc]) -> String {
         #(#node_boilerplate_impls)*
         #(#enums_boilierplate_impls)*
     };
-    reformat(file.to_string()).replace("#[derive", "\n#[derive")
+    let output = reformat(file.to_string()).replace("#[derive", "\n#[derive");
+    format!("{PRELUDE}{output}")
 }
 
 fn generate_tokens(tokens: &[(&'static str, &'static str)]) -> String {
@@ -920,5 +933,6 @@ fn generate_tokens(tokens: &[(&'static str, &'static str)]) -> String {
         #(#tokens)*
     };
 
-    reformat(file.to_string()).replace("#[derive", "\n#[derive")
+    let output = reformat(file.to_string()).replace("#[derive", "\n#[derive");
+    format!("{PRELUDE}{output}")
 }

crates/xtask/src/sync_builtins.rs

@@ -55,6 +55,18 @@ where n.nspname not like 'pg_temp%'
 order by n.nspname, p.proname;
 ";
 
+const BUILTIN_OPERATORS_QUERY: &str = r"
+select n.nspname, o.oprname,
+  format_type(o.oprleft, null) as left_type,
+  format_type(o.oprright, null) as right_type
+from pg_operator o
+join pg_namespace n on n.oid = o.oprnamespace
+where n.nspname not like 'pg_temp%'
+  and n.nspname not like 'pg_toast%'
+  and n.nspname <> 'public'
+order by n.nspname, o.oprname;
+";
+
 const PG_VERSION_QUERY: &str = "show server_version;";
 
 fn write_table(sql: &mut String, schema: &str, table_name: &str, columns: &[(String, String)]) {
@@ -68,7 +80,14 @@ fn write_table(sql: &mut String, schema: &str, table_name: &str, columns: &[(Str
 
 fn run_sql(query: &str) -> Result<String> {
     let output = Command::new("psql")
-        .args(["--tuples-only", "--no-align", "--command", query])
+        .args([
+            "--tuples-only",
+            "--no-align",
+            "--field-separator",
+            "\t",
+            "--command",
+            query,
+        ])
         .stdout(Stdio::piped())
         .stderr(Stdio::piped())
         .output()
@@ -112,7 +131,7 @@ pub(crate) fn sync_builtins() -> Result<()> {
         .lines()
         .filter(|line| !line.is_empty())
     {
-        let mut parts = line.split('|');
+        let mut parts = line.split('\t');
         let schema = parts.next().context("expected schema name")?;
         let type_name = parts.next().context("expected type name")?;
         let type_size = parts.next().context("expected type size")?;
@@ -132,7 +151,7 @@ pub(crate) fn sync_builtins() -> Result<()> {
         .lines()
         .filter(|line| !line.is_empty())
     {
-        let mut parts = line.split('|');
+        let mut parts = line.split('\t');
         let schema = parts.next().context("expected schema name")?;
         let table_name = parts.next().context("expected table name")?;
         let col_name = parts.next().context("expected column name")?;
@@ -161,7 +180,7 @@ pub(crate) fn sync_builtins() -> Result<()> {
         .lines()
         .filter(|line| !line.is_empty())
     {
-        let mut parts = line.split('|');
+        let mut parts = line.split('\t');
         let schema = parts.next().context("expected schema name")?;
         let func_name = parts.next().context("expected function name")?;
         let args = parts.next().context("expected function arguments")?;
@@ -171,6 +190,24 @@ pub(crate) fn sync_builtins() -> Result<()> {
         ));
     }
 
+    for line in run_sql(BUILTIN_OPERATORS_QUERY)?
+        .lines()
+        .filter(|line| !line.is_empty())
+    {
+        let mut parts = line.split('\t');
+        let schema = parts.next().context("expected schema name")?;
+        let op_name = parts.next().context("expected operator name")?;
+        let left_type = parts.next().context("expected left type")?;
+        let right_type = parts.next().context("expected right type")?;
+
+        let args = match (left_type, right_type) {
+            ("-", r) => format!("rightarg = {r}"),
+            (l, "-") => format!("leftarg = {l}"),
+            (l, r) => format!("leftarg = {l}, rightarg = {r}"),
+        };
+        sql.push_str(&format!("create operator {schema}.{op_name} ({args});\n\n"));
+    }
+
     let builtins_path = project_root().join("crates/squawk_ide/src/builtins.sql");
     std::fs::write(&builtins_path, sql).context("Failed to write builtins.sql")?;