Merge pull request #28 from PeerDB-io/merge-upstream

Merge 0.40 + apache/datafusion-sqlparser-rs#1040

Upstream implemented END parsing as COMMIT on postgresql: apache/datafusion-sqlparser-rs#1035 allowing some convergence

Author: serprex

.tool-versions

@@ -0,0 +1 @@
+rust 1.73.0

CHANGELOG.md

@@ -8,6 +8,29 @@ Given that the parser produces a typed AST, any changes to the AST will technica
 ## [Unreleased]
 Check https://github.com/sqlparser-rs/sqlparser-rs/commits/main for undocumented changes.
 
+
+## [0.40.0] 2023-11-27
+
+### Added
+* Add `{pre,post}_visit_query` to `Visitor` (#1044) - Thanks @jmhain
+* Support generated virtual columns with expression (#1051) - Thanks @takluyver
+* Support PostgreSQL `END` (#1035) - Thanks @tobyhede
+* Support `INSERT INTO ... DEFAULT VALUES ...` (#1036) - Thanks @CDThomas
+* Support `RELEASE` and `ROLLBACK TO SAVEPOINT` (#1045) - Thanks @CDThomas
+* Support `CONVERT` expressions (#1048) - Thanks @lovasoa
+* Support `GLOBAL` and `SESSION` parts in `SHOW VARIABLES` for mysql and generic - Thanks @emin100
+* Support snowflake `PIVOT` on derived table factors (#1027) - Thanks @lustefaniak
+* Support mssql json and xml extensions (#1043) - Thanks @lovasoa
+* Support for `MAX` as a character length (#1038) - Thanks @lovasoa
+* Support `IN ()` syntax of SQLite (#1028) - Thanks @alamb
+
+### Fixed
+* Fix extra whitespace printed before `ON CONFLICT` (#1037) - Thanks @CDThomas
+
+### Changed
+* Document round trip ability (#1052) - Thanks @alamb
+* Add PRQL to list of users (#1031) - Thanks @vanillajonathan
+
 ## [0.39.0] 2023-10-27
 
 ### Added

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "sqlparser"
 description = "Extensible SQL Lexer and Parser with support for ANSI SQL:2011"
-version = "0.39.0"
+version = "0.40.0"
 authors = ["Andy Grove <andygrove73@gmail.com>"]
 homepage = "https://github.com/sqlparser-rs/sqlparser-rs"
 documentation = "https://docs.rs/sqlparser/"
@@ -34,7 +34,7 @@ serde = { version = "1.0", features = ["derive"], optional = true }
 # of dev-dependencies because of
 # https://github.com/rust-lang/cargo/issues/1596
 serde_json = { version = "1.0", optional = true }
-sqlparser_derive = { version = "0.1.1", path = "derive", optional = true }
+sqlparser_derive = { version = "0.2.0", path = "derive", optional = true }
 
 [dev-dependencies]
 simple_logger = "4.0"

README.md

@@ -59,6 +59,28 @@ This crate avoids semantic analysis because it varies drastically
 between dialects and implementations. If you want to do semantic
 analysis, feel free to use this project as a base.
 
+## Preserves Syntax Round Trip 
+
+This crate allows users to recover the original SQL text (with normalized
+whitespace and keyword capitalization), which is useful for tools that
+analyze and manipulate SQL.
+
+This means that other than whitespace and the capitalization of keywords, the
+following should hold true for all SQL:
+
+```rust
+// Parse SQL
+let ast = Parser::parse_sql(&GenericDialect, sql).unwrap();
+
+// The original SQL text can be generated from the AST
+assert_eq!(ast[0].to_string(), sql);
+```
+
+There are still some cases in this crate where different SQL with seemingly
+similar semantics are represented with the same AST. We welcome PRs to fix such
+issues and distinguish different syntaxes in the AST.
+
+
 ## SQL compliance
 
 SQL was first standardized in 1987, and revisions of the standard have been
@@ -93,7 +115,7 @@ $ cargo run --features json_example --example cli FILENAME.sql [--dialectname]
 ## Users
 
 This parser is currently being used by the [DataFusion] query engine,
-[LocustDB], [Ballista], [GlueSQL], [Opteryx], and [JumpWire].
+[LocustDB], [Ballista], [GlueSQL], [Opteryx], [PRQL], and [JumpWire].
 
 If your project is using sqlparser-rs feel free to make a PR to add it
 to this list.
@@ -188,6 +210,7 @@ licensed as above, without any additional terms or conditions.
 [Ballista]: https://github.com/apache/arrow-ballista
 [GlueSQL]: https://github.com/gluesql/gluesql
 [Opteryx]: https://github.com/mabel-dev/opteryx
+[PRQL]: https://github.com/PRQL/prql
 [JumpWire]: https://github.com/extragoodlabs/jumpwire
 [Pratt Parser]: https://tdop.github.io/
 [sql-2016-grammar]: https://jakewheat.github.io/sql-overview/sql-2016-foundation-grammar.html

derive/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "sqlparser_derive"
 description = "proc macro for sqlparser"
-version = "0.1.1"
+version = "0.2.1"
 authors = ["sqlparser-rs authors"]
 homepage = "https://github.com/sqlparser-rs/sqlparser-rs"
 documentation = "https://docs.rs/sqlparser_derive/"
@@ -18,6 +18,6 @@ edition = "2021"
 proc-macro = true
 
 [dependencies]
-syn = "1.0"
+syn = { version = "2.0", default-features = false, features = ["printing", "parsing", "derive", "proc-macro"] }
 proc-macro2 = "1.0"
 quote = "1.0"

derive/README.md

@@ -48,33 +48,102 @@ impl Visit for Bar {
 }
 ```
 
-Additionally certain types may wish to call a corresponding method on visitor before recursing
+Some types may wish to call a corresponding method on the visitor:
 
 ```rust
 #[derive(Visit, VisitMut)]
 #[visit(with = "visit_expr")]
 enum Expr {
-    A(),
-    B(String, #[cfg_attr(feature = "visitor", visit(with = "visit_relation"))] ObjectName, bool),
+    IsNull(Box<Expr>),
+    ..
 }
 ```
 
-Will generate
+This will result in the following sequence of visitor calls when an `IsNull`
+expression is visited
+
+```
+visitor.pre_visit_expr(<is null expr>)
+visitor.pre_visit_expr(<is null operand>)
+visitor.post_visit_expr(<is null operand>)
+visitor.post_visit_expr(<is null expr>)
+```
+
+For some types it is only appropriate to call a particular visitor method in
+some contexts. For example, not every `ObjectName` refers to a relation.
+
+In these cases, the `visit` attribute can be used on the field for which we'd
+like to call the method:
 
 ```rust
-impl Visit for Bar {
+#[derive(Visit, VisitMut)]
+#[visit(with = "visit_table_factor")]
+pub enum TableFactor {
+    Table {
+        #[visit(with = "visit_relation")]
+        name: ObjectName,
+        alias: Option<TableAlias>,
+    },
+    ..
+}
+```
+
+This will generate
+
+```rust
+impl Visit for TableFactor {
     fn visit<V: Visitor>(&self, visitor: &mut V) -> ControlFlow<V::Break> {
-        visitor.visit_expr(self)?;
+        visitor.pre_visit_table_factor(self)?;
         match self {
-            Self::A() => {}
-            Self::B(_1, _2, _3) => {
-                _1.visit(visitor)?;
-                visitor.visit_relation(_3)?;
-                _2.visit(visitor)?;
-                _3.visit(visitor)?;
+            Self::Table { name, alias } => {
+                visitor.pre_visit_relation(name)?;
+                alias.visit(name)?;
+                visitor.post_visit_relation(name)?;
+                alias.visit(visitor)?;
             }
         }
+        visitor.post_visit_table_factor(self)?;
         ControlFlow::Continue(())
     }
 }
 ```
+
+Note that annotating both the type and the field is incorrect as it will result
+in redundant calls to the method. For example
+
+```rust
+#[derive(Visit, VisitMut)]
+#[visit(with = "visit_expr")]
+enum Expr {
+    IsNull(#[visit(with = "visit_expr")] Box<Expr>),
+    ..
+}
+```
+
+will result in these calls to the visitor
+
+
+```
+visitor.pre_visit_expr(<is null expr>)
+visitor.pre_visit_expr(<is null operand>)
+visitor.pre_visit_expr(<is null operand>)
+visitor.post_visit_expr(<is null operand>)
+visitor.post_visit_expr(<is null operand>)
+visitor.post_visit_expr(<is null expr>)
+```
+
+## Releasing
+
+This crate's release is not automated. Instead it is released manually as needed
+
+Steps:
+1. Update the version in `Cargo.toml`
+2. Update the corresponding version in `../Cargo.toml`
+3. Commit via PR
+4. Publish to crates.io:
+
+```shell
+# update to latest checked in main branch and publish via
+cargo publish 
+```
+

derive/src/lib.rs

@@ -2,8 +2,9 @@ use proc_macro2::TokenStream;
 use quote::{format_ident, quote, quote_spanned, ToTokens};
 use syn::spanned::Spanned;
 use syn::{
-    parse_macro_input, parse_quote, Attribute, Data, DeriveInput, Fields, GenericParam, Generics,
-    Ident, Index, Lit, Meta, MetaNameValue, NestedMeta,
+    parse::{Parse, ParseStream},
+    parse_macro_input, parse_quote, Attribute, Data, DeriveInput,
+    Fields, GenericParam, Generics, Ident, Index, LitStr, Meta, Token
 };
 
 /// Implementation of `[#derive(Visit)]`
@@ -84,38 +85,43 @@ struct Attributes {
     with: Option<Ident>,
 }
 
+struct WithIdent {
+    with: Option<Ident>,
+}
+impl Parse for WithIdent {
+    fn parse(input: ParseStream) -> Result<Self, syn::Error> {
+        let mut result = WithIdent { with: None };
+        let ident = input.parse::<Ident>()?;
+        if ident != "with" {
+            return Err(syn::Error::new(ident.span(), "Expected identifier to be `with`"));
+        }
+        input.parse::<Token!(=)>()?;
+        let s = input.parse::<LitStr>()?;
+        result.with = Some(format_ident!("{}", s.value(), span = s.span()));
+        Ok(result)
+    }
+}
+
 impl Attributes {
     fn parse(attrs: &[Attribute]) -> Self {
         let mut out = Self::default();
-        for attr in attrs.iter().filter(|a| a.path.is_ident("visit")) {
-            let meta = attr.parse_meta().expect("visit attribute");
-            match meta {
-                Meta::List(l) => {
-                    for nested in &l.nested {
-                        match nested {
-                            NestedMeta::Meta(Meta::NameValue(v)) => out.parse_name_value(v),
-                            _ => panic!("Expected #[visit(key = \"value\")]"),
+        for attr in attrs {
+            if let Meta::List(ref metalist) = attr.meta {
+                if metalist.path.is_ident("visit") {
+                    match syn::parse2::<WithIdent>(metalist.tokens.clone()) {
+                        Ok(with_ident) => {
+                            out.with = with_ident.with;
+                        }
+                        Err(e) => {
+                            panic!("{}", e);
                         }
                     }
                 }
-                _ => panic!("Expected #[visit(...)]"),
             }
         }
         out
     }
 
-    /// Updates self with a name value attribute
-    fn parse_name_value(&mut self, v: &MetaNameValue) {
-        if v.path.is_ident("with") {
-            match &v.lit {
-                Lit::Str(s) => self.with = Some(format_ident!("{}", s.value(), span = s.span())),
-                _ => panic!("Expected a string value, got {}", v.lit.to_token_stream()),
-            }
-            return;
-        }
-        panic!("Unrecognised kv attribute {}", v.path.to_token_stream())
-    }
-
     /// Returns the pre and post visit token streams
     fn visit(&self, s: TokenStream) -> (Option<TokenStream>, Option<TokenStream>) {
         let pre_visit = self.with.as_ref().map(|m| {

src/ast/data_type.rs

@@ -374,14 +374,14 @@ impl fmt::Display for DataType {
                 }
                 write!(f, ")")
             }
-            DataType::SnowflakeTimestamp => write!(f, "TIMESTAMP_NTZ"),
             DataType::Struct(fields) => {
                 if !fields.is_empty() {
                     write!(f, "STRUCT<{}>", display_comma_separated(fields))
                 } else {
                     write!(f, "STRUCT")
                 }
             }
+            DataType::SnowflakeTimestamp => write!(f, "TIMESTAMP_NTZ"),
         }
     }
 }
@@ -521,18 +521,29 @@ impl fmt::Display for ExactNumberInfo {
 #[derive(Debug, Copy, Clone, PartialEq, PartialOrd, Eq, Ord, Hash)]
 #[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
 #[cfg_attr(feature = "visitor", derive(Visit, VisitMut))]
-pub struct CharacterLength {
-    /// Default (if VARYING) or maximum (if not VARYING) length
-    pub length: u64,
-    /// Optional unit. If not informed, the ANSI handles it as CHARACTERS implicitly
-    pub unit: Option<CharLengthUnits>,
+pub enum CharacterLength {
+    IntegerLength {
+        /// Default (if VARYING) or maximum (if not VARYING) length
+        length: u64,
+        /// Optional unit. If not informed, the ANSI handles it as CHARACTERS implicitly
+        unit: Option<CharLengthUnits>,
+    },
+    /// VARCHAR(MAX) or NVARCHAR(MAX), used in T-SQL (Miscrosoft SQL Server)
+    Max,
 }
 
 impl fmt::Display for CharacterLength {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        write!(f, "{}", self.length)?;
-        if let Some(unit) = &self.unit {
-            write!(f, " {unit}")?;
+        match self {
+            CharacterLength::IntegerLength { length, unit } => {
+                write!(f, "{}", length)?;
+                if let Some(unit) = unit {
+                    write!(f, " {unit}")?;
+                }
+            }
+            CharacterLength::Max => {
+                write!(f, "MAX")?;
+            }
         }
         Ok(())
     }