From 0cf9537286b6031d89185add95bee2b93ea249f1 Mon Sep 17 00:00:00 2001
From: Ben Herzberg <ben@satoricyber.com>
Date: Mon, 14 Sep 2026 16:15:35 +0300
Subject: [PATCH] Postgres: Support IS DOCUMENT predicate

---
 src/ast/mod.rs            | 14 ++++++++++++++
 src/ast/spans.rs          |  1 +
 src/parser/mod.rs         | 35 ++++++++++++++++++++++++++++++++++-
 tests/sqlparser_common.rs | 12 ++++++++++++
 4 files changed, 61 insertions(+), 1 deletion(-)

diff --git a/src/ast/mod.rs b/src/ast/mod.rs
index 20058b83ab..e05e50095d 100644
--- a/src/ast/mod.rs
+++ b/src/ast/mod.rs
@@ -944,6 +944,13 @@ pub enum Expr {
     IsUnknown(Box<Expr>),
     /// `IS NOT UNKNOWN` operator
     IsNotUnknown(Box<Expr>),
+    /// `<expr> IS [NOT] DOCUMENT`
+    IsDocument {
+        /// Expression being tested.
+        expr: Box<Expr>,
+        /// `true` when `NOT` is present.
+        negated: bool,
+    },
     /// `IS DISTINCT FROM` operator
     IsDistinctFrom(Box<Expr>, Box<Expr>),
     /// `IS NOT DISTINCT FROM` operator
@@ -1764,6 +1771,13 @@ impl fmt::Display for Expr {
             Expr::IsNotNull(ast) => write!(f, "{ast} IS NOT NULL"),
             Expr::IsUnknown(ast) => write!(f, "{ast} IS UNKNOWN"),
             Expr::IsNotUnknown(ast) => write!(f, "{ast} IS NOT UNKNOWN"),
+            Expr::IsDocument { expr, negated } => {
+                write!(f, "{expr} IS ")?;
+                if *negated {
+                    write!(f, "NOT ")?;
+                }
+                write!(f, "DOCUMENT")
+            }
             Expr::IsJson {
                 expr,
                 kind,
diff --git a/src/ast/spans.rs b/src/ast/spans.rs
index 7acbd7d0b4..7848660c5e 100644
--- a/src/ast/spans.rs
+++ b/src/ast/spans.rs
@@ -1497,6 +1497,7 @@ impl Spanned for Expr {
             Expr::IsNotNull(expr) => expr.span(),
             Expr::IsUnknown(expr) => expr.span(),
             Expr::IsNotUnknown(expr) => expr.span(),
+            Expr::IsDocument { expr, negated: _ } => expr.span(),
             Expr::IsJson {
                 expr,
                 kind: _,
diff --git a/src/parser/mod.rs b/src/parser/mod.rs
index 15f135fffa..d7ab119d09 100644
--- a/src/parser/mod.rs
+++ b/src/parser/mod.rs
@@ -4091,6 +4091,15 @@ impl<'a> Parser<'a> {
                         Ok(Expr::IsUnknown(Box::new(expr)))
                     } else if self.parse_keywords(&[Keyword::NOT, Keyword::UNKNOWN]) {
                         Ok(Expr::IsNotUnknown(Box::new(expr)))
+                    } else if let Some(negated) = if dialect.supports_xml_expressions() {
+                        self.parse_is_document_predicate()
+                    } else {
+                        None
+                    } {
+                        Ok(Expr::IsDocument {
+                            expr: Box::new(expr),
+                            negated,
+                        })
                     } else if self.parse_keywords(&[Keyword::DISTINCT, Keyword::FROM]) {
                         // The right operand binds no more loosely than `IS`
                         // itself, so that e.g. `a IS DISTINCT FROM b AND c`
@@ -4109,7 +4118,7 @@ impl<'a> Parser<'a> {
                         Ok(is_normalized)
                     } else {
                         self.expected_ref(
-                            "[NOT] NULL | TRUE | FALSE | DISTINCT | [NOT] JSON [VALUE | SCALAR | ARRAY | OBJECT] [WITH | WITHOUT UNIQUE [KEYS]] | [form] NORMALIZED FROM after IS",
+                            "[NOT] NULL | TRUE | FALSE | DISTINCT | [NOT] DOCUMENT | [NOT] JSON [VALUE | SCALAR | ARRAY | OBJECT] [WITH | WITHOUT UNIQUE [KEYS]] | [form] NORMALIZED FROM after IS",
                             self.peek_token_ref(),
                         )
                     }
@@ -4752,6 +4761,16 @@ impl<'a> Parser<'a> {
         }
     }
 
+    fn parse_unquoted_word(&mut self, expected: &str) -> bool {
+        match &self.peek_token_ref().token {
+            Token::Word(w) if w.quote_style.is_none() && w.value.eq_ignore_ascii_case(expected) => {
+                self.advance_token();
+                true
+            }
+            _ => false,
+        }
+    }
+
     #[must_use]
     /// Check if the current token is the expected keyword without consuming it.
     ///
@@ -12742,6 +12761,20 @@ impl<'a> Parser<'a> {
         })
     }
 
+    fn parse_is_document_predicate(&mut self) -> Option<bool> {
+        if self.parse_unquoted_word("document") {
+            return Some(false);
+        }
+
+        let start_index = self.index;
+        if self.parse_keyword(Keyword::NOT) && self.parse_unquoted_word("document") {
+            Some(true)
+        } else {
+            self.index = start_index;
+            None
+        }
+    }
+
     /// Parse a literal unicode normalization clause
     pub fn parse_unicode_is_normalized(&mut self, expr: Expr) -> Result<Expr, ParserError> {
         let neg = self.parse_keyword(Keyword::NOT);
diff --git a/tests/sqlparser_common.rs b/tests/sqlparser_common.rs
index 2de6062b28..dc3c79ee34 100644
--- a/tests/sqlparser_common.rs
+++ b/tests/sqlparser_common.rs
@@ -19837,6 +19837,18 @@ fn parse_xmlparse() {
         .is_err());
 }
 
+#[test]
+fn parse_xml_is_document_predicate() {
+    let dialects = all_dialects_where(|d| d.supports_xml_expressions());
+    dialects.one_statement_parses_to("SELECT '<a/>' IS document", "SELECT '<a/>' IS DOCUMENT");
+    dialects.verified_stmt("SELECT '<a/>' IS NOT DOCUMENT");
+
+    let others = all_dialects_except(|d| d.supports_xml_expressions());
+    assert!(others
+        .parse_sql_statements("SELECT '<a/>' IS DOCUMENT")
+        .is_err());
+}
+
 /// Regression test for the 2^N parse-time blowup in `parse_compound_expr` on
 /// inputs like `IF a0.a1...aN.#`. The parse is run on a worker thread and the
 /// main thread asserts that it reports back within a generous timeout. Post-fix