parser: fix last of pg regression suite, strings

crates/squawk_parser/src/grammar.rs

@@ -27,10 +27,15 @@ fn literal(p: &mut Parser<'_>) -> Option<CompletedMarker> {
         return None;
     }
     let m = p.start();
+    if p.eat(BYTE_STRING) {
+        if p.eat(UESCAPE_KW) {
+            p.eat(STRING);
+        }
+    }
     // E021-03 string continuation syntax
     // If two string literals are next to each other, and don't have a comment
     // between them, then they are automatically combined.
-    if p.eat(STRING) {
+    else if p.eat(STRING) {
         while !p.at(EOF) && p.eat(STRING) {}
     } else {
         p.bump_any();
@@ -362,10 +367,6 @@ fn substring_fn(p: &mut Parser<'_>) -> CompletedMarker {
                 if expr(p).is_none() {
                     p.error("expected an expression");
                 }
-                p.expect(ESCAPE_KW);
-                if expr(p).is_none() {
-                    p.error("expected an expression");
-                }
             }
             _ if p.eat(COMMA) => {
                 // normal function call
@@ -1462,7 +1463,13 @@ fn opt_name(p: &mut Parser<'_>) -> Option<CompletedMarker> {
         return None;
     }
     let m = p.start();
-    p.bump_any();
+    if p.eat(IDENT) {
+        if p.eat(UESCAPE_KW) {
+            p.expect(STRING);
+        }
+    } else {
+        p.bump_any();
+    }
     Some(m.complete(p, NAME))
 }
 
@@ -2156,10 +2163,14 @@ fn current_op(p: &Parser<'_>, r: &Restrictions) -> (u8, SyntaxKind, Associativit
         PLUS if p.next_not_joined_op(0) => (8, PLUS, Left), // symbol
         // overlaps
         OVERLAPS_KW => (7, OVERLAPS_KW, Left),
+        // escape
+        ESCAPE_KW => (7, ESCAPE_KW, Left),
         // like
         LIKE_KW => (6, LIKE_KW, Left),
         // ilike
         ILIKE_KW => (6, ILIKE_KW, Left),
+        // not similar to
+        NOT_KW if !r.not_disabled && p.at(NOT_SIMILAR_TO) => (6, NOT_SIMILAR_TO, Left),
         // not like
         NOT_KW if !r.not_disabled && p.at(NOT_LIKE) => (6, NOT_LIKE, Left),
         // not ilike
@@ -13727,7 +13738,13 @@ fn alter_table_action(p: &mut Parser<'_>) -> Option<SyntaxKind> {
 fn opt_col_label(p: &mut Parser<'_>) -> bool {
     if p.at_ts(COL_LABEL_FIRST) {
         let m = p.start();
-        p.bump_any();
+        if p.eat(IDENT) {
+            if p.eat(UESCAPE_KW) {
+                p.expect(STRING);
+            }
+        } else {
+            p.bump_any();
+        }
         m.complete(p, NAME);
         true
     } else {

crates/squawk_parser/src/lib.rs

@@ -343,6 +343,14 @@ impl<'t> Parser<'t> {
                 m.complete(self, SyntaxKind::IS_JSON_SCALAR);
                 return true;
             }
+            SyntaxKind::NOT_SIMILAR_TO => {
+                let m = self.start();
+                self.bump(SyntaxKind::NOT_KW);
+                self.bump(SyntaxKind::SIMILAR_KW);
+                self.bump(SyntaxKind::TO_KW);
+                m.complete(self, SyntaxKind::NOT_SIMILAR_TO);
+                return true;
+            }
             SyntaxKind::IS_NOT_DISTINCT_FROM => {
                 let m = self.start();
                 self.bump(SyntaxKind::IS_KW);
@@ -767,6 +775,12 @@ impl<'t> Parser<'t> {
                 }
                 return false;
             }
+            SyntaxKind::NOT_SIMILAR_TO => self.at_composite3(
+                n,
+                SyntaxKind::NOT_KW,
+                SyntaxKind::SIMILAR_KW,
+                SyntaxKind::TO_KW,
+            ),
             // similar to
             SyntaxKind::SIMILAR_TO => self.at_composite2(
                 n,

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

@@ -180,9 +180,17 @@ select U&'\0061\0308bc' is not nfd normalized;
 -- pattern_matching
 -- like
 select 'foo' like 'bar';
-
+select 'foo' like 'bar' escape '#';
 -- not like
 select 'foo' not like 'bar';
+select 'foo' not like 'bar' escape '#';
+
+-- ilike
+select 'foo' ilike 'bar';
+select 'foo' ilike 'bar' escape '#';
+-- not ilike
+select 'foo' not ilike 'bar';
+select 'foo' not ilike 'bar' escape '#';
 
 -- ~~
 select 'a' ~~ 'b';
@@ -192,6 +200,10 @@ select 'a' !~~ 'b';
 
 -- similar to
 select 'abc' similar to 'abc';
+select 'abc' similar to 'abc' escape '#';
+
+select 'abc' not similar to 'abc';
+select 'abc' not similar to 'abc' escape '#';
 
 -- posix regex
 -- string matches regex case sensitive

crates/squawk_parser/tests/data/regression_suite/strings.sql

@@ -28,7 +28,7 @@ SELECT 'tricky' AS U&"\" UESCAPE '!';
 
 SELECT U&'wrong: \061';
 SELECT U&'wrong: \+0061';
-SELECT U&'wrong: +0061' UESCAPE +;
+-- SELECT U&'wrong: +0061' UESCAPE +;
 SELECT U&'wrong: +0061' UESCAPE '+';
 
 SELECT U&'wrong: \db99';

crates/squawk_parser/tests/snapshots/tests__regression_strings.snap

@@ -2,52 +2,4 @@
 source: crates/squawk_parser/tests/tests.rs
 input_file: crates/squawk_parser/tests/data/regression_suite/strings.sql
 ---
-ERROR@536: missing comma
-ERROR@563: missing comma
-ERROR@625: missing comma
-ERROR@667: missing comma
-ERROR@763: missing comma
-ERROR@765: expected an expression, found SEMICOLON
-ERROR@798: missing comma
-ERROR@1460: missing comma
-ERROR@1487: missing comma
-ERROR@1523: missing comma
-ERROR@1565: missing comma
-ERROR@1661: missing comma
-ERROR@6313: missing comma
-ERROR@6369: missing comma
-ERROR@6621: missing comma
-ERROR@6717: missing comma
-ERROR@6775: missing comma
-ERROR@17083: missing comma
-ERROR@17136: missing comma
-ERROR@17188: missing comma
-ERROR@17242: missing comma
-ERROR@17354: missing comma
-ERROR@17403: missing comma
-ERROR@17455: missing comma
-ERROR@17510: missing comma
-ERROR@17562: missing comma
-ERROR@17617: missing comma
-ERROR@17675: missing comma
-ERROR@17735: missing comma
-ERROR@17787: missing comma
-ERROR@17841: missing comma
-ERROR@17894: missing comma
-ERROR@17949: missing comma
-ERROR@18003: missing comma
-ERROR@18060: missing comma
-ERROR@18112: missing comma
-ERROR@18167: missing comma
-ERROR@18230: missing comma
-ERROR@18302: missing comma
-ERROR@18406: missing comma
-ERROR@18459: missing comma
-ERROR@18511: missing comma
-ERROR@18565: missing comma
-ERROR@18616: missing comma
-ERROR@18669: missing comma
-ERROR@18721: missing comma
-ERROR@18775: missing comma
-ERROR@18827: missing comma
-ERROR@18882: missing comma
+

crates/squawk_parser/tests/snapshots/tests__regression_suite_errors.snap

@@ -2,4 +2,4 @@
 source: crates/squawk_parser/tests/tests.rs
 expression: "out.join(\"\\n\")"
 ---
-tests/snapshots/tests__regression_strings.snap:49
+

crates/squawk_parser/tests/snapshots/tests__select_funcs_ok.snap

@@ -730,13 +730,14 @@ SOURCE_FILE
               WHITESPACE " "
               SIMILAR_KW "similar"
               WHITESPACE " "
-              NAME_REF
-                IDENT "b"
-              WHITESPACE " "
-              ESCAPE_KW "escape"
-              WHITESPACE " "
-              NAME_REF
-                IDENT "c"
+              BIN_EXPR
+                NAME_REF
+                  IDENT "b"
+                WHITESPACE " "
+                ESCAPE_KW "escape"
+                WHITESPACE " "
+                NAME_REF
+                  IDENT "c"
               R_PAREN ")"
   SEMICOLON ";"
   WHITESPACE "\n\n"