parser: validation for uescape

crates/squawk_parser/src/plpgsql_grammar.rs

@@ -80,7 +80,6 @@ fn opt_block(p: &mut Parser) {
     opt_exception_section(p);
     p.expect(END_KW);
     opt_label_name_ref(p);
-    // TODO: add validation, sometimes this is required
     p.eat(SEMICOLON);
     m.complete(p, PLPGSQL_BLOCK);
 }
@@ -236,16 +235,26 @@ fn decl_datatype(p: &mut Parser) {
 fn path_name_ref(p: &mut Parser) {
     assert!(at_path(p).is_some());
     let m = p.start();
-    name(p, PATH_SEGMENT_REF);
+    path_segment_ref(p);
     let mut path = m.complete(p, PATH_REF);
     while !p.at(EOF) && p.at(DOT) {
         let m = path.precede(p);
         p.bump(DOT);
-        name(p, PATH_SEGMENT_REF);
+        path_segment_ref(p);
         path = m.complete(p, PATH_REF);
     }
 }
 
+fn path_segment_ref(p: &mut Parser) {
+    let m = p.start();
+    let unicode_ident = p.at(IDENT);
+    p.bump_any();
+    if unicode_ident && p.eat(UESCAPE_KW) {
+        p.expect(STRING);
+    }
+    m.complete(p, PATH_SEGMENT_REF);
+}
+
 fn at_percent_datatype(p: &Parser) -> bool {
     let Some(n) = at_path(p) else {
         return false;
@@ -256,7 +265,11 @@ fn at_percent_datatype(p: &Parser) -> bool {
 fn at_path(p: &Parser) -> Option<usize> {
     let mut n = 0;
     while !p.nth_at(n, EOF) && at_name(p, n) {
+        let unicode_ident = p.nth_at(n, IDENT);
         n += 1;
+        if unicode_ident && p.nth_at(n, UESCAPE_KW) && p.nth_at(n + 1, STRING) {
+            n += 2;
+        }
         if !p.nth_at(n, DOT) {
             return Some(n);
         }

crates/squawk_parser/tests/data/plpgsql/err/block.sql

@@ -0,0 +1,16 @@
+do $$
+begin
+  begin
+    null;
+  end
+end
+$$;
+
+do $$
+begin
+  <<inner>>
+  begin
+    null;
+  end inner
+end
+$$;

crates/squawk_parser/tests/data/plpgsql/err/percent_type.sql

@@ -0,0 +1,8 @@
+do $$
+declare
+  a U&"tbl" UESCAPE '!'%type;
+  b U&"tbl" UESCAPE '!'%rowtype;
+begin
+  null;
+end
+$$;

crates/squawk_parser/tests/data/plpgsql/ok/percent_type.sql

@@ -13,6 +13,7 @@ declare
   j table%rowtype;
   k select%type;
   l $1%type;
+  m U&"int4" UESCAPE '!';
 begin
   null;
 end

crates/squawk_syntax/src/snapshots/squawk_syntax__test__plpgsql_block_err.snap

@@ -0,0 +1,59 @@
+---
+source: crates/squawk_syntax/src/test.rs
+input_file: crates/squawk_parser/tests/data/plpgsql/err/block.sql
+---
+PLPGSQL@0..35
+  WHITESPACE@0..1 "\n"
+  PLPGSQL_BLOCK@1..34
+    BEGIN_KW@1..6 "begin"
+    WHITESPACE@6..9 "\n  "
+    PLPGSQL_BODY@9..30
+      PLPGSQL_BLOCK@9..30
+        BEGIN_KW@9..14 "begin"
+        WHITESPACE@14..19 "\n    "
+        PLPGSQL_BODY@19..24
+          PLPGSQL_NULL_STMT@19..24
+            NULL_KW@19..23 "null"
+            SEMICOLON@23..24 ";"
+        WHITESPACE@24..27 "\n  "
+        END_KW@27..30 "end"
+    WHITESPACE@30..31 "\n"
+    END_KW@31..34 "end"
+  WHITESPACE@34..35 "\n"
+---
+PLPGSQL@0..53
+  WHITESPACE@0..1 "\n"
+  PLPGSQL_BLOCK@1..52
+    BEGIN_KW@1..6 "begin"
+    WHITESPACE@6..9 "\n  "
+    PLPGSQL_BODY@9..48
+      PLPGSQL_BLOCK@9..48
+        PLPGSQL_LABEL@9..18
+          LESS_LESS@9..11 "<<"
+          PLPGSQL_LABEL_NAME@11..16
+            INNER_KW@11..16 "inner"
+          GREATER_GREATER@16..18 ">>"
+        WHITESPACE@18..21 "\n  "
+        BEGIN_KW@21..26 "begin"
+        WHITESPACE@26..31 "\n    "
+        PLPGSQL_BODY@31..36
+          PLPGSQL_NULL_STMT@31..36
+            NULL_KW@31..35 "null"
+            SEMICOLON@35..36 ";"
+        WHITESPACE@36..39 "\n  "
+        END_KW@39..42 "end"
+        WHITESPACE@42..43 " "
+        PLPGSQL_LABEL_NAME_REF@43..48
+          INNER_KW@43..48 "inner"
+    WHITESPACE@48..49 "\n"
+    END_KW@49..52 "end"
+  WHITESPACE@52..53 "\n"
+
+error[syntax-error]: Missing semicolon after block
+  ╭▸ 
+5 │   end
+  ╰╴     ━
+error[syntax-error]: Missing semicolon after block
+   ╭▸ 
+14 │   end inner
+   ╰╴           ━

crates/squawk_syntax/src/snapshots/squawk_syntax__test__plpgsql_percent_type_err.snap

@@ -0,0 +1,61 @@
+---
+source: crates/squawk_syntax/src/test.rs
+input_file: crates/squawk_parser/tests/data/plpgsql/err/percent_type.sql
+---
+PLPGSQL@0..90
+  WHITESPACE@0..1 "\n"
+  PLPGSQL_BLOCK@1..89
+    PLPGSQL_DECLARE_SECTION@1..71
+      DECLARE_KW@1..8 "declare"
+      WHITESPACE@8..11 "\n  "
+      PLPGSQL_VAR_DECL@11..38
+        PLPGSQL_VAR_NAME@11..12
+          IDENT@11..12 "a"
+        WHITESPACE@12..13 " "
+        PERCENT_TYPE@13..37
+          PATH_REF@13..32
+            PATH_SEGMENT_REF@13..32
+              IDENT@13..20 "U&\"tbl\""
+              WHITESPACE@20..21 " "
+              UESCAPE_KW@21..28 "UESCAPE"
+              WHITESPACE@28..29 " "
+              STRING@29..32 "'!'"
+          PERCENT_TYPE_CLAUSE@32..37
+            PERCENT@32..33 "%"
+            TYPE_KW@33..37 "type"
+        SEMICOLON@37..38 ";"
+      WHITESPACE@38..41 "\n  "
+      PLPGSQL_VAR_DECL@41..71
+        PLPGSQL_VAR_NAME@41..42
+          IDENT@41..42 "b"
+        WHITESPACE@42..43 " "
+        PLPGSQL_PERCENT_ROWTYPE@43..70
+          PATH_REF@43..62
+            PATH_SEGMENT_REF@43..62
+              IDENT@43..50 "U&\"tbl\""
+              WHITESPACE@50..51 " "
+              UESCAPE_KW@51..58 "UESCAPE"
+              WHITESPACE@58..59 " "
+              STRING@59..62 "'!'"
+          PERCENT@62..63 "%"
+          ROWTYPE_KW@63..70 "rowtype"
+        SEMICOLON@70..71 ";"
+    WHITESPACE@71..72 "\n"
+    BEGIN_KW@72..77 "begin"
+    WHITESPACE@77..80 "\n  "
+    PLPGSQL_BODY@80..85
+      PLPGSQL_NULL_STMT@80..85
+        NULL_KW@80..84 "null"
+        SEMICOLON@84..85 ";"
+    WHITESPACE@85..86 "\n"
+    END_KW@86..89 "end"
+  WHITESPACE@89..90 "\n"
+
+error[syntax-error]: UESCAPE is not allowed before %TYPE
+  ╭▸ 
+3 │   a U&"tbl" UESCAPE '!'%type;
+  ╰╴                       ━
+error[syntax-error]: UESCAPE is not allowed before %ROWTYPE
+  ╭▸ 
+4 │   b U&"tbl" UESCAPE '!'%rowtype;
+  ╰╴                       ━

crates/squawk_syntax/src/snapshots/squawk_syntax__test__plpgsql_percent_type_ok.snap

@@ -2,10 +2,10 @@
 source: crates/squawk_syntax/src/test.rs
 input_file: crates/squawk_parser/tests/data/plpgsql/ok/percent_type.sql
 ---
-PLPGSQL@0..265
+PLPGSQL@0..291
   WHITESPACE@0..1 "\n"
-  PLPGSQL_BLOCK@1..264
-    PLPGSQL_DECLARE_SECTION@1..246
+  PLPGSQL_BLOCK@1..290
+    PLPGSQL_DECLARE_SECTION@1..272
       DECLARE_KW@1..8 "declare"
       WHITESPACE@8..11 "\n  "
       PLPGSQL_VAR_DECL@11..17
@@ -205,13 +205,27 @@ PLPGSQL@0..265
             PERCENT@240..241 "%"
             TYPE_KW@241..245 "type"
         SEMICOLON@245..246 ";"
-    WHITESPACE@246..247 "\n"
-    BEGIN_KW@247..252 "begin"
-    WHITESPACE@252..255 "\n  "
-    PLPGSQL_BODY@255..260
-      PLPGSQL_NULL_STMT@255..260
-        NULL_KW@255..259 "null"
-        SEMICOLON@259..260 ";"
-    WHITESPACE@260..261 "\n"
-    END_KW@261..264 "end"
-  WHITESPACE@264..265 "\n"
+      WHITESPACE@246..249 "\n  "
+      PLPGSQL_VAR_DECL@249..272
+        PLPGSQL_VAR_NAME@249..250
+          IDENT@249..250 "m"
+        WHITESPACE@250..251 " "
+        PATH_TYPE@251..271
+          PATH_REF@251..271
+            PATH_SEGMENT_REF@251..271
+              IDENT@251..259 "U&\"int4\""
+              WHITESPACE@259..260 " "
+              UESCAPE_KW@260..267 "UESCAPE"
+              WHITESPACE@267..268 " "
+              STRING@268..271 "'!'"
+        SEMICOLON@271..272 ";"
+    WHITESPACE@272..273 "\n"
+    BEGIN_KW@273..278 "begin"
+    WHITESPACE@278..281 "\n  "
+    PLPGSQL_BODY@281..286
+      PLPGSQL_NULL_STMT@281..286
+        NULL_KW@281..285 "null"
+        SEMICOLON@285..286 ";"
+    WHITESPACE@286..287 "\n"
+    END_KW@287..290 "end"
+  WHITESPACE@290..291 "\n"

crates/squawk_syntax/src/validation.rs

@@ -49,6 +49,13 @@ pub(crate) fn validate(root: &SyntaxNode, errors: &mut Vec<SyntaxError>) {
                 ast::NonStandardParam(it) => validate_non_standard_param(it, errors),
                 ast::ParenFromItem(it) => validate_paren_from_item(it, errors),
                 ast::PartitionForValuesWith(it) => validate_hash_partition_bounds(it, errors),
+                ast::PercentType(it) => validate_plpgsql_percent_type(
+                    it.syntax(),
+                    it.percent_type_clause().and_then(|it| it.percent_token()),
+                    "%TYPE",
+                    errors,
+                ),
+                ast::PlpgsqlBlock(it) => validate_plpgsql_block(it, errors),
                 ast::PlpgsqlCaseStmt(it) => validate_no_bare_case(it.subject(), errors),
                 ast::PlpgsqlCompOptionPrintStrictParams(it) => {
                     validate_print_strict_params(it, errors)
@@ -63,6 +70,12 @@ pub(crate) fn validate(root: &SyntaxNode, errors: &mut Vec<SyntaxError>) {
                 ast::PlpgsqlForIStmt(it) => validate_for_i_single_var(it, errors),
                 ast::PlpgsqlForQueryStmt(it) => validate_for_query_no_reverse(it, errors),
                 ast::PlpgsqlIfStmt(it) => validate_no_bare_case(it.cond(), errors),
+                ast::PlpgsqlPercentRowtype(it) => validate_plpgsql_percent_type(
+                    it.syntax(),
+                    it.percent_token(),
+                    "%ROWTYPE",
+                    errors,
+                ),
                 ast::RelationFromItem(it) => validate_relation_from_item(it, errors),
                 ast::RuleStmtList(it) => validate_rule_stmt_list(it, errors),
                 ast::Select(it) => validate_select(it, errors),
@@ -357,6 +370,51 @@ fn validate_atomic_body(it: ast::AtomicBody, acc: &mut Vec<SyntaxError>) {
     }
 }
 
+// -- err
+// declare x U&"tbl" UESCAPE '!'%type;
+// declare x U&"tbl" UESCAPE '!'%rowtype;
+fn validate_plpgsql_percent_type(
+    syntax: &SyntaxNode,
+    percent: Option<SyntaxToken>,
+    percent_type: &str,
+    acc: &mut Vec<SyntaxError>,
+) {
+    if !syntax
+        .parent()
+        .is_some_and(|parent| matches!(parent.kind(), PLPGSQL_VAR_DECL | PLPGSQL_CURSOR_ARG))
+    {
+        return;
+    }
+    let has_uescape = syntax
+        .descendants_with_tokens()
+        .any(|element| element.kind() == UESCAPE_KW);
+    if !has_uescape {
+        return;
+    }
+    let Some(percent) = percent else {
+        return;
+    };
+    acc.push(SyntaxError::new(
+        format!("UESCAPE is not allowed before {percent_type}"),
+        percent.text_range(),
+    ));
+}
+
+fn validate_plpgsql_block(it: ast::PlpgsqlBlock, acc: &mut Vec<SyntaxError>) {
+    if it.semicolon_token().is_some()
+        || it
+            .syntax()
+            .parent()
+            .is_none_or(|parent| parent.kind() != PLPGSQL_BODY)
+    {
+        return;
+    }
+    acc.push(SyntaxError::new(
+        "Missing semicolon after block",
+        TextRange::empty(it.syntax().text_range().end()),
+    ));
+}
+
 fn validate_print_strict_params(
     it: ast::PlpgsqlCompOptionPrintStrictParams,
     acc: &mut Vec<SyntaxError>,