Merge for 2.26.6 release

RELEASE_NOTES_LATEST.md

@@ -3,7 +3,7 @@
 
 Available through https://crates.io/crates/typeql.
 ```
-cargo add [email protected]-rc0
+cargo add [email protected]
 ```
 
 ## TypeQL Grammar and Language Library distributions for Java
@@ -20,12 +20,12 @@ cargo add [email protected]
     <dependency>
         <groupId>com.vaticle.typeql</groupId>
         <artifactId>typeql-grammar</artifactId>
-        <version>2.26.6-rc0</version>
+        <version>2.26.6</version>
     </dependency>
     <dependency>
         <groupId>com.vaticle.typeql</groupId>
         <artifactId>typeql-lang</artifactId>
-        <version>2.26.6-rc0</version>
+        <version>2.26.6</version>
     </dependency>
 </dependencies>
 ```
@@ -35,7 +35,7 @@ cargo add [email protected]
 Available through https://pypi.org
 
 ```
-pip install typeql-grammar==2.26.6-rc0
+pip install typeql-grammar==2.26.6
 ```
 
 
@@ -62,6 +62,21 @@ pip install typeql-grammar==2.26.6-rc0
 
 
 ## Code Refactors
+- **Allow variables to have a leading digit**
+
+  We modify the behaviour of #310 which unified variables and labels to have the same valid identifier syntax, which eliminated the capability of variables to have a leading number. For example, the variable `$0` was banned.
+
+  This PR reverts this specific behaviour, and enables usage of variables with leading digits:
+  ```
+  match
+  $1_a isa entity;
+  get;
+  ```
+  is made valid again.
+
+  Testing specified in https://github.com/vaticle/typedb-behaviour/pull/281
+
+
 
 - **Merge typedb-common repository into typeql**
 

VERSION

@@ -1 +1 @@
-2.26.6-rc0
+2.26.6

dependencies/vaticle/repositories.bzl

@@ -21,12 +21,12 @@ def vaticle_dependencies():
     git_repository(
         name = "vaticle_dependencies",
         remote = "https://github.com/vaticle/dependencies",
-        commit = "cd00aa9bc16bc2eb857b9b5e4d7a301bf19908dc", # sync-marker: do not remove this comment, this is used for sync-dependencies by @vaticle_dependencies
+        commit = "08d4e3bafb5a90db76eb5188be6f90bee564eb01", # sync-marker: do not remove this comment, this is used for sync-dependencies by @vaticle_dependencies
     )
 
 def vaticle_typedb_behaviour():
     git_repository(
         name = "vaticle_typedb_behaviour",
         remote = "https://github.com/vaticle/typedb-behaviour",
-        commit = "483eac8b122cbafe5ea154580b98aa671e51ea35", # sync-marker: do not remove this comment, this is used for sync-dependencies by @vaticle_typedb_behaviour
+        commit = "f996fb7322eabfaafdbbf203a59fdc27fd6babf0", # sync-marker: do not remove this comment, this is used for sync-dependencies by @vaticle_typedb_behaviour
     )

grammar/TypeQL.g4

@@ -332,35 +332,47 @@ DATETIME_       : DATE_FRAGMENT_ 'T' TIME_              ;
 
 VAR_CONCEPT_            : VAR_CONCEPT_ANONYMOUS_ | VAR_CONCEPT_NAMED_   ;
 VAR_CONCEPT_ANONYMOUS_  : '$_' ;
-VAR_CONCEPT_NAMED_      : '$'  TYPE_CHAR_H_ TYPE_CHAR_T_* ;
-VAR_VALUE_              : '?'  TYPE_CHAR_H_ TYPE_CHAR_T_* ;
+VAR_CONCEPT_NAMED_      : '$'  IDENTIFIER_VAR_H_ IDENTIFIER_VAR_T_* ;
+VAR_VALUE_              : '?'  IDENTIFIER_VAR_H_ IDENTIFIER_VAR_T_* ;
 IID_                    : '0x' [0-9a-f]+                  ;
-LABEL_                  : TYPE_CHAR_H_ TYPE_CHAR_T_*      ;
+LABEL_                  : IDENTIFIER_LABEL_H_ IDENTIFIER_LABEL_T_*      ;
 LABEL_SCOPED_           : LABEL_ ':' LABEL_               ;
 
 // FRAGMENTS OF KEYWORDS =======================================================
 
-fragment TYPE_CHAR_H_     : 'A'..'Z' | 'a'..'z'
-                          | '\u00C0'..'\u00D6'
-                          | '\u00D8'..'\u00F6'
-                          | '\u00F8'..'\u02FF'
-                          | '\u0370'..'\u037D'
-                          | '\u037F'..'\u1FFF'
-                          | '\u200C'..'\u200D'
-                          | '\u2070'..'\u218F'
-                          | '\u2C00'..'\u2FEF'
-                          | '\u3001'..'\uD7FF'
-                          | '\uF900'..'\uFDCF'
-                          | '\uFDF0'..'\uFFFD'
-                          ;
-fragment TYPE_CHAR_T_     : TYPE_CHAR_H_
-                          | '0'..'9'
-                          | '_'
-                          | '-'
-                          | '\u00B7'
-                          | '\u0300'..'\u036F'
-                          | '\u203F'..'\u2040'
-                          ;
+fragment IDENTIFIER_DIGIT_     : '0'..'9' ;
+fragment IDENTIFIER_CONNECTOR_ : '_'
+                               | '-'
+                               | '\u00B7'
+                               | '\u0300'..'\u036F'
+                               | '\u203F'..'\u2040'
+                               ;
+fragment IDENTIFIER_CHAR_      :'A'..'Z' | 'a'..'z'
+                               | '\u00C0'..'\u00D6'
+                               | '\u00D8'..'\u00F6'
+                               | '\u00F8'..'\u02FF'
+                               | '\u0370'..'\u037D'
+                               | '\u037F'..'\u1FFF'
+                               | '\u200C'..'\u200D'
+                               | '\u2070'..'\u218F'
+                               | '\u2C00'..'\u2FEF'
+                               | '\u3001'..'\uD7FF'
+                               | '\uF900'..'\uFDCF'
+                               | '\uFDF0'..'\uFFFD'
+                               ;
+fragment IDENTIFIER_VAR_H_     : IDENTIFIER_DIGIT_
+                               | IDENTIFIER_CHAR_
+                               ;
+fragment IDENTIFIER_VAR_T_     : IDENTIFIER_DIGIT_
+                               | IDENTIFIER_CHAR_
+                               | IDENTIFIER_CONNECTOR_
+                               ;
+fragment IDENTIFIER_LABEL_H_   : IDENTIFIER_CHAR_ ;
+fragment IDENTIFIER_LABEL_T_   : IDENTIFIER_CHAR_
+                               | IDENTIFIER_DIGIT_
+                               | IDENTIFIER_CONNECTOR_
+                               ;
+
 fragment DATE_FRAGMENT_   : YEAR_ '-' MONTH_ '-' DAY_ ;
 fragment MONTH_           : [0-1][0-9] ;
 fragment DAY_             : [0-3][0-9] ;

java/common/Reference.java

@@ -36,7 +36,7 @@
 
 public abstract class Reference {
 
-    private static final String IDENTIFIER_START = "A-Za-z" +
+    private static final String IDENTIFIER_CHAR = "A-Za-z" +
             "\\u00C0-\\u00D6" +
             "\\u00D8-\\u00F6" +
             "\\u00F8-\\u02FF" +
@@ -48,16 +48,18 @@ public abstract class Reference {
             "\\u3001-\\uD7FF" +
             "\\uF900-\\uFDCF" +
             "\\uFDF0-\\uFFFD";
-    private static final String IDENTIFIER_TAIL = IDENTIFIER_START +
-            "0-9" +
-            "_" +
+    private static final String IDENTIFIER_DIGIT = "0-9";
+    private static final String IDENTIFIER_CONNECTOR = "_" +
             "\\-" +
             "\\u00B7" +
             "\\u0300-\\u036F" +
             "\\u203F-\\u2040";
 
-    public static final Pattern IDENTIFIER_REGEX = Pattern.compile(
-            "^[" + IDENTIFIER_START + "][" + IDENTIFIER_TAIL + "]*$"
+    public static final Pattern IDENTIFIER_VAR = Pattern.compile(
+            "^[" + IDENTIFIER_CHAR + IDENTIFIER_DIGIT + "][" + IDENTIFIER_CHAR + IDENTIFIER_DIGIT + IDENTIFIER_CONNECTOR + "]*$"
+    );
+    public static final Pattern IDENTIFIER_LABEL = Pattern.compile(
+            "^[" + IDENTIFIER_CHAR  + "][" + IDENTIFIER_CHAR + IDENTIFIER_DIGIT + IDENTIFIER_CONNECTOR + "]*$"
     );
 
     final Type type;
@@ -155,7 +157,7 @@ public static abstract class Name extends Reference {
 
         Name(Type type, String name, boolean isVisible) {
             super(type, isVisible);
-            if (!IDENTIFIER_REGEX.matcher(name).matches()) {
+            if (!IDENTIFIER_VAR.matcher(name).matches()) {
                 throw TypeQLException.of(INVALID_VARIABLE_NAME.message(name));
             }
             this.name = name;
@@ -241,9 +243,9 @@ public static class Label extends Reference {
 
         Label(String label, @Nullable String scope) {
             super(Type.LABEL, false);
-            if (!IDENTIFIER_REGEX.matcher(label).matches()) {
+            if (!IDENTIFIER_LABEL.matcher(label).matches()) {
                 throw TypeQLException.of(INVALID_TYPE_LABEL.message(label));
-            } else if (scope != null && !IDENTIFIER_REGEX.matcher(scope).matches()) {
+            } else if (scope != null && !IDENTIFIER_LABEL.matcher(scope).matches()) {
                 throw TypeQLException.of(INVALID_TYPE_LABEL.message(scope));
             }
             this.label = label;

java/pattern/statement/builder/ConceptStatementBuilder.java

@@ -21,7 +21,6 @@
 
 package com.vaticle.typeql.lang.pattern.statement.builder;
 
-import com.vaticle.typeql.lang.common.Reference;
 import com.vaticle.typeql.lang.common.TypeQLVariable;
 import com.vaticle.typeql.lang.pattern.constraint.ConceptConstraint;
 import com.vaticle.typeql.lang.pattern.statement.ConceptStatement;

java/pattern/statement/builder/PredicateBuilder.java

@@ -22,9 +22,9 @@
 package com.vaticle.typeql.lang.pattern.statement.builder;
 
 import com.vaticle.typeql.lang.common.TypeQLToken;
+import com.vaticle.typeql.lang.common.TypeQLVariable;
 import com.vaticle.typeql.lang.pattern.constraint.Predicate;
 import com.vaticle.typeql.lang.pattern.statement.Statement;
-import com.vaticle.typeql.lang.common.TypeQLVariable;
 
 import java.time.LocalDateTime;
 import java.util.function.BiFunction;

java/pattern/statement/builder/ValueStatementBuilder.java

@@ -21,8 +21,8 @@
 
 package com.vaticle.typeql.lang.pattern.statement.builder;
 
-import com.vaticle.typeql.lang.pattern.expression.Expression;
 import com.vaticle.typeql.lang.pattern.constraint.ValueConstraint;
+import com.vaticle.typeql.lang.pattern.expression.Expression;
 import com.vaticle.typeql.lang.pattern.statement.ValueStatement;
 
 public interface ValueStatementBuilder extends PredicateBuilder<ValueStatement> {

java/query/TypeQLFetch.java

@@ -330,7 +330,7 @@ private Label(Either<String, String> quotedOrUnquoted) {
             }
 
             public static Label of(String label) {
-                if (Reference.IDENTIFIER_REGEX.matcher(label).matches()) {
+                if (Reference.IDENTIFIER_LABEL.matcher(label).matches()) {
                     return unquoted(label);
                 } else {
                     return quoted(label);

rust/common/identifier.rs

@@ -23,10 +23,7 @@ use std::sync::OnceLock;
 
 use regex::{Regex, RegexBuilder};
 
-pub fn is_valid_identifier(identifier: &str) -> bool {
-    static REGEX: OnceLock<Regex> = OnceLock::new();
-    let regex = REGEX.get_or_init(|| {
-        let identifier_start = "A-Za-z\
+const IDENTIFIER_CHAR: &str = "A-Za-z\
             \\u00C0-\\u00D6\
             \\u00D8-\\u00F6\
             \\u00F8-\\u02FF\
@@ -38,17 +35,29 @@ pub fn is_valid_identifier(identifier: &str) -> bool {
             \\u3001-\\uD7FF\
             \\uF900-\\uFDCF\
             \\uFDF0-\\uFFFD";
-        let identifier_tail = format!(
-            "{}\
-            0-9\
-            _\
+const IDENTIFIER_CONNECTOR: &str = "_\
             \\-\
             \\u00B7\
             \\u0300-\\u036F\
-            \\u203F-\\u2040",
-            identifier_start
-        );
-        let identifier_pattern = format!("^[{}][{}]*$", identifier_start, identifier_tail);
+            \\u203F-\\u2040";
+const IDENTIFIER_DIGIT: &str = "0-9";
+
+pub fn is_valid_label_identifier(identifier: &str) -> bool {
+    static REGEX: OnceLock<Regex> = OnceLock::new();
+    let regex = REGEX.get_or_init(|| {
+        let identifier_tail = format!("{}{}{}", IDENTIFIER_CHAR, IDENTIFIER_CONNECTOR, IDENTIFIER_DIGIT);
+        let identifier_pattern = format!("^[{}][{}]*$", IDENTIFIER_CHAR, identifier_tail);
+        RegexBuilder::new(&identifier_pattern).build().unwrap()
+    });
+    regex.is_match(identifier)
+}
+
+pub fn is_valid_var_identifier(identifier: &str) -> bool {
+    static REGEX: OnceLock<Regex> = OnceLock::new();
+    let regex = REGEX.get_or_init(|| {
+        let identifier_head = format!("{}{}", IDENTIFIER_CHAR, IDENTIFIER_DIGIT);
+        let identifier_tail = format!("{}{}{}", IDENTIFIER_CHAR, IDENTIFIER_DIGIT, IDENTIFIER_CONNECTOR);
+        let identifier_pattern = format!("^[{}][{}]*$", identifier_head, identifier_tail);
         RegexBuilder::new(&identifier_pattern).build().unwrap()
     });
     regex.is_match(identifier)

rust/parser/typeql.pest

@@ -354,37 +354,41 @@ DATETIME_ = @{ DATE_FRAGMENT_ ~ "T" ~ TIME_ ~ WB }
 VAR_ = ${ VAR_CONCEPT_ | VAR_VALUE_ }
 VAR_CONCEPT_ = @{ VAR_CONCEPT_ANONYMOUS_ | VAR_CONCEPT_NAMED_ }
 VAR_CONCEPT_ANONYMOUS_ = @{ "$_" ~ WB }
-VAR_CONCEPT_NAMED_ = @{ "$" ~ IDENTIFIER }
-VAR_VALUE_ = @{ "?" ~ IDENTIFIER }
+VAR_CONCEPT_NAMED_ = @{ "$" ~ IDENTIFIER_VAR_H_ ~ IDENTIFIER_VAR_T_*  ~ WB }
+VAR_VALUE_ = @{ "?" ~ (IDENTIFIER_VAR_H_ ~ IDENTIFIER_VAR_T_* ~ WB) }
 IID_ = @{ "0x" ~ ASCII_HEX_DIGIT+ ~ WB }
-LABEL_ = @{ IDENTIFIER }
-LABEL_SCOPED_ = @{ IDENTIFIER ~ ":" ~ IDENTIFIER }
-IDENTIFIER = @{ TYPE_CHAR_H_ ~ TYPE_CHAR_T_* ~ WB }
+LABEL_ = @{ (IDENTIFIER_LABEL_H_ ~ IDENTIFIER_LABEL_T_* ~ WB) }
+LABEL_SCOPED_ = @{ (IDENTIFIER_LABEL_H_ ~ IDENTIFIER_LABEL_T_*) ~ ":" ~ (IDENTIFIER_LABEL_H_ ~ IDENTIFIER_LABEL_T_*)  ~ WB }
 
 
 // FRAGMENTS OF KEYWORDS =======================================================
 
-TYPE_CHAR_H_ = @{ ASCII_ALPHA
-               | '\u{00C0}'..'\u{00D6}'
-               | '\u{00D8}'..'\u{00F6}'
-               | '\u{00F8}'..'\u{02FF}'
-               | '\u{0370}'..'\u{037D}'
-               | '\u{037F}'..'\u{1FFF}'
-               | '\u{200C}'..'\u{200D}'
-               | '\u{2070}'..'\u{218F}'
-               | '\u{2C00}'..'\u{2FEF}'
-               | '\u{3001}'..'\u{D7FF}'
-               | '\u{F900}'..'\u{FDCF}'
-               | '\u{FDF0}'..'\u{FFFD}'
-               }
-TYPE_CHAR_T_ = @{ TYPE_CHAR_H_
-               | ASCII_DIGIT
-               | "_"
-               | "-"
-               | "\u{00B7}"
-               | '\u{0300}'..'\u{036F}'
-               | '\u{203F}'..'\u{2040}'
-               }
+IDENTIFIER_CHAR_ = @{ ASCII_ALPHA
+                   | '\u{00C0}'..'\u{00D6}'
+                   | '\u{00D8}'..'\u{00F6}'
+                   | '\u{00F8}'..'\u{02FF}'
+                   | '\u{0370}'..'\u{037D}'
+                   | '\u{037F}'..'\u{1FFF}'
+                   | '\u{200C}'..'\u{200D}'
+                   | '\u{2070}'..'\u{218F}'
+                   | '\u{2C00}'..'\u{2FEF}'
+                   | '\u{3001}'..'\u{D7FF}'
+                   | '\u{F900}'..'\u{FDCF}'
+                   | '\u{FDF0}'..'\u{FFFD}'
+                   }
+IDENTIFIER_CONNECTOR_ = @{ "_"
+                        | "-"
+                        | "\u{00B7}"
+                        | '\u{0300}'..'\u{036F}'
+                        | '\u{203F}'..'\u{2040}'
+                        }
+
+IDENTIFIER_LABEL_H_ = @{ IDENTIFIER_CHAR_ }
+IDENTIFIER_LABEL_T_ = @{ IDENTIFIER_LABEL_H_ | ASCII_DIGIT | IDENTIFIER_CONNECTOR_ }
+
+IDENTIFIER_VAR_H_ = @{ IDENTIFIER_CHAR_ | ASCII_DIGIT }
+IDENTIFIER_VAR_T_ = @{ IDENTIFIER_VAR_H_ | IDENTIFIER_CONNECTOR_ }
+
 DATE_FRAGMENT_ = @{ YEAR_ ~ "-" ~ MONTH_ ~ "-" ~ DAY_ }
 MONTH_ = @{ ('0'..'1') ~ ASCII_DIGIT }
 DAY_ = @{ ('0'..'3') ~ ASCII_DIGIT }

rust/pattern/label.rs

@@ -22,7 +22,9 @@
 
 use std::fmt;
 
-use crate::common::{error::TypeQLError, identifier::is_valid_identifier, token, validatable::Validatable, Result};
+use crate::common::{
+    error::TypeQLError, identifier::is_valid_label_identifier, token, validatable::Validatable, Result,
+};
 
 #[derive(Debug, Clone, Eq, PartialEq)]
 pub struct Label {
@@ -71,7 +73,7 @@ impl Validatable for Label {
 }
 
 fn validate_label(label: &str) -> Result {
-    if !is_valid_identifier(label) {
+    if !is_valid_label_identifier(label) {
         Err(TypeQLError::InvalidTypeLabel { label: label.to_owned() })?
     } else {
         Ok(())

rust/query/typeql_fetch.rs

@@ -26,7 +26,7 @@ use itertools::Itertools;
 use crate::{
     common::{
         error::{collect_err, TypeQLError},
-        identifier::is_valid_identifier,
+        identifier::is_valid_label_identifier,
         string::indent,
         token,
         validatable::Validatable,
@@ -164,7 +164,7 @@ impl ProjectionKeyLabel {
     }
 
     fn must_quote(s: &str) -> bool {
-        !is_valid_identifier(s)
+        !is_valid_label_identifier(s)
     }
 }