Return a single token for OpenQASM version statement from lexer

crates/oq3_lexer/src/lib.rs

@@ -76,6 +76,11 @@ pub enum TokenKind {
     /// Like the above, but containing invalid unicode codepoints.
     InvalidIdent,
 
+    OpenQasmVersionStmt {
+        major: bool,
+        minor: bool,
+    },
+
     Pragma,
 
     Dim,
@@ -301,6 +306,22 @@ impl Cursor<'_> {
 
             'p' => self.pragma_or_ident_or_unknown_prefix(),
 
+            // This implementation stores the entire "OPENQASM x.y" string as a single token.
+            // There are several options, but others seem to come with a cost.
+            // At present the tokenizer has in a sense no context, very little state.
+            // There are no "modes". It might be possible to do this in a way that does not
+            // affect performance. But I don't want to take the time to experiment. Breaking
+            // an abstraction for one small feature seems unwise.
+            'O' => {
+                if self.have_openqasm() {
+                    self.eat_while(is_whitespace);
+                    let (major, minor) = self.openqasm_version();
+                    OpenQasmVersionStmt { major, minor }
+                } else {
+                    self.ident_or_unknown_prefix()
+                }
+            }
+
             // Identifier (this should be checked after other variant that can
             // start as identifier).
             c if is_id_start(c) => self.ident_or_unknown_prefix(),
@@ -483,6 +504,7 @@ impl Cursor<'_> {
         }
     }
 
+    /// Yikes!! assumes the *previous* token was also whitespace.
     fn whitespace(&mut self) -> TokenKind {
         debug_assert!(is_whitespace(self.prev()));
         self.eat_while(is_whitespace);
@@ -528,6 +550,55 @@ impl Cursor<'_> {
         false
     }
 
+    /// This is called if we just consumed 'O'
+    /// If we consume "OPENQASM" + whitespace, then is is a version statement.
+    /// Otherwise, an identifier, valid or not.
+    fn have_openqasm(&mut self) -> bool {
+        if self.first() == 'P' {
+            self.bump();
+            if self.first() == 'E' {
+                self.bump();
+                if self.first() == 'N' {
+                    self.bump();
+                    if self.first() == 'Q' {
+                        self.bump();
+                        if self.first() == 'A' {
+                            self.bump();
+                            if self.first() == 'S' {
+                                self.bump();
+                                if self.first() == 'M' {
+                                    self.bump();
+                                    return is_whitespace(self.first());
+                                }
+                            }
+                        }
+                    }
+                }
+            }
+        }
+        false
+    }
+
+    fn openqasm_version(&mut self) -> (bool, bool) {
+        if !self.eat_decimal_digits() {
+            return (false, false);
+        }
+        let c = self.first();
+        if c == '.' {
+            self.bump();
+
+            if !self.eat_decimal_digits() {
+                // Do not allow "3."
+                return (true, false);
+            }
+        }
+        let c = self.first();
+        if c != ';' && !is_whitespace(c) {
+            return (false, false);
+        }
+        (true, true)
+    }
+
     fn pragma_or_ident_or_unknown_prefix(&mut self) -> TokenKind {
         if self.have_pragma() {
             Pragma
@@ -856,6 +927,7 @@ impl Cursor<'_> {
     }
 
     fn eat_decimal_digits(&mut self) -> bool {
+        //
         let mut has_digits = false;
         loop {
             match self.first() {

crates/oq3_parser/src/grammar/expressions.rs

@@ -101,6 +101,14 @@ pub(crate) fn stmt(p: &mut Parser<'_>) {
         return q_or_c_reg_declaration(p, m);
     }
 
+    if p.at(VERSION_STRING) {
+        p.bump_any();
+        if !p.eat(T![;]) {
+            p.error("Expecting semicolon terminating statement");
+        }
+        m.complete(p, VERSION_STRING);
+        return;
+    }
     // FIXME: straighten out logic
     if !(p.current().is_classical_type() && (p.nth(1) == T!['('] || p.nth(1) == T!['[']))
         && !p.at_ts(EXPR_FIRST)

crates/oq3_parser/src/lexed_str.rs

@@ -225,6 +225,19 @@ fn inner_extend_token<'a>(
                 COMMENT
             }
 
+            oq3_lexer::TokenKind::OpenQasmVersionStmt { major, minor } => {
+                if *major {
+                    if *minor {
+                        // All good
+                    } else {
+                        err = "Invalid minor version in OpenQASM version statement";
+                    }
+                } else {
+                    err = "Invalid version number in OpenQASM version statement";
+                }
+                VERSION_STRING
+            }
+
             oq3_lexer::TokenKind::Whitespace => WHITESPACE,
             oq3_lexer::TokenKind::Ident if token_text == "_" => UNDERSCORE,
 

crates/oq3_semantics/src/syntax_to_semantics.rs

@@ -536,8 +536,10 @@ fn stmt_to_asg_stmt(stmt: synast::Stmt, context: &mut Context) -> Option<asg::St
         synast::Stmt::ExprStmt(expr_stmt) => expr_stmt_to_asg_stmt(expr_stmt, context),
 
         synast::Stmt::VersionString(version_string) => {
-            let version = version_string.version().unwrap().version().unwrap();
-            let _ = version.split_into_parts();
+            // let version = version_string.version().unwrap().version().unwrap();
+            //            not_impl!(context, version_string)
+            context.insert_error(NotImplementedError, &version_string);
+            // Better None than NullStmt.
             None
         }
 

crates/oq3_semantics/tests/from_string_tests.rs

@@ -515,7 +515,7 @@ bit[2] out;
 out[0] = measure $0;
 "#;
     let (program, errors, _symbol_table, _have_syntax_errors) = parse_string(code);
-    assert!(errors.is_empty());
+    assert_eq!(errors.len(), 1);
     assert_eq!(program.len(), 2);
 }
 

crates/oq3_semantics/tests/spec.rs

@@ -56,7 +56,7 @@ include "stdgates.inc";
 // Rest of QASM program
 "#;
     let (program, errors, _symbol_table) = parse_string(code);
-    assert!(errors.is_empty());
+    assert_eq!(errors.len(), 1);
     assert!(program.is_empty());
 }
 

crates/pipeline-tests/tests/snapshots/runner__tests__snippets__invalid__statements__headers.qasm-lex.snap

@@ -18,43 +18,38 @@ include "hello;
 ok: false
 errors: 1
 [0] Whitespace "\n" @0..1
-[1] Ident "OPENQASM" @1..9
-[2] Whitespace " " @9..10
-[3] Ident "int" @10..13
-[4] Semi ";" @13..14
-[5] Whitespace "\n" @14..15
-[6] Ident "OPENQASM" @15..23
-[7] Whitespace " " @23..24
-[8] Literal { kind: Str { terminated: true }, suffix_start: 14 } "'hello, world'" @24..38
-[9] Semi ";" @38..39
-[10] Whitespace "\n" @39..40
-[11] Ident "OPENQASM" @40..48
-[12] Whitespace " " @48..49
-[13] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "3" @49..50
-[14] Whitespace " " @50..51
-[15] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "3" @51..52
-[16] Semi ";" @52..53
-[17] Whitespace "\n" @53..54
-[18] Ident "OPENQASM" @54..62
-[19] Whitespace " " @62..63
-[20] Literal { kind: Float { base: Decimal, empty_exponent: false }, suffix_start: 2 } "3.x" @63..66
-[21] Semi ";" @66..67
-[22] Whitespace "\n" @67..68
-[23] Ident "include" @68..75
-[24] Whitespace " " @75..76
-[25] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "3" @76..77
-[26] Semi ";" @77..78
-[27] Whitespace "\n" @78..79
-[28] Ident "include" @79..86
-[29] Whitespace " " @86..87
-[30] Ident "include" @87..94
-[31] Semi ";" @94..95
-[32] Whitespace "\n" @95..96
-[33] Ident "include" @96..103
-[34] Whitespace " " @103..104
-[35] Ident "def" @104..107
-[36] Semi ";" @107..108
-[37] Whitespace "\n" @108..109
-[38] Ident "include" @109..116
-[39] Whitespace " " @116..117
-[40] Literal { kind: Str { terminated: false }, suffix_start: 8 } ""hello;\n" @117..125
+[1] OpenQasmVersionStmt { major: false, minor: false } "OPENQASM " @1..10
+[2] Ident "int" @10..13
+[3] Semi ";" @13..14
+[4] Whitespace "\n" @14..15
+[5] OpenQasmVersionStmt { major: false, minor: false } "OPENQASM " @15..24
+[6] Literal { kind: Str { terminated: true }, suffix_start: 14 } "'hello, world'" @24..38
+[7] Semi ";" @38..39
+[8] Whitespace "\n" @39..40
+[9] OpenQasmVersionStmt { major: true, minor: true } "OPENQASM 3" @40..50
+[10] Whitespace " " @50..51
+[11] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "3" @51..52
+[12] Semi ";" @52..53
+[13] Whitespace "\n" @53..54
+[14] OpenQasmVersionStmt { major: true, minor: false } "OPENQASM 3." @54..65
+[15] Ident "x" @65..66
+[16] Semi ";" @66..67
+[17] Whitespace "\n" @67..68
+[18] Ident "include" @68..75
+[19] Whitespace " " @75..76
+[20] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "3" @76..77
+[21] Semi ";" @77..78
+[22] Whitespace "\n" @78..79
+[23] Ident "include" @79..86
+[24] Whitespace " " @86..87
+[25] Ident "include" @87..94
+[26] Semi ";" @94..95
+[27] Whitespace "\n" @95..96
+[28] Ident "include" @96..103
+[29] Whitespace " " @103..104
+[30] Ident "def" @104..107
+[31] Semi ";" @107..108
+[32] Whitespace "\n" @108..109
+[33] Ident "include" @109..116
+[34] Whitespace " " @116..117
+[35] Literal { kind: Str { terminated: false }, suffix_start: 8 } ""hello;\n" @117..125

crates/pipeline-tests/tests/snapshots/runner__tests__snippets__invalid__statements__headers.qasm-parse.snap

@@ -7,6 +7,6 @@ expect-parse: Diag
 --- parser ---
 ok: false
 panicked: false
-errors: 1
+errors: 4
 --- ast ---
 (no ast)

crates/pipeline-tests/tests/snapshots/runner__tests__snippets__reference__comments__comments.qasm-lex.snap

@@ -21,57 +21,55 @@ cx q[0], q[1];
 ok: true
 errors: 0
 [0] Whitespace "\n" @0..1
-[1] Ident "OPENQASM" @1..9
-[2] Whitespace " " @9..10
-[3] Literal { kind: Float { base: Decimal, empty_exponent: false }, suffix_start: 3 } "3.0" @10..13
-[4] Semi ";" @13..14
-[5] Whitespace "\n" @14..15
-[6] LineComment "// Line comment before include" @15..45
-[7] Whitespace "\n" @45..46
-[8] Ident "include" @46..53
-[9] Whitespace " " @53..54
-[10] Literal { kind: Str { terminated: true }, suffix_start: 14 } ""stdgates.inc"" @54..68
-[11] Semi ";" @68..69
-[12] Whitespace " " @69..70
-[13] LineComment "// Inline comment" @70..87
-[14] Whitespace "\n" @87..88
-[15] BlockComment { terminated: true } "/* Block comment before declaration */" @88..126
-[16] Whitespace "\n" @126..127
-[17] Ident "qubit" @127..132
-[18] OpenBracket "[" @132..133
-[19] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "2" @133..134
-[20] CloseBracket "]" @134..135
-[21] Whitespace " " @135..136
-[22] Ident "q" @136..137
-[23] Semi ";" @137..138
-[24] Whitespace " " @138..139
-[25] BlockComment { terminated: true } "/* Inline block comment */" @139..165
-[26] Whitespace "\n\n" @165..167
-[27] LineComment "// Comment before gate" @167..189
-[28] Whitespace "\n" @189..190
-[29] Ident "h" @190..191
-[30] Whitespace " " @191..192
-[31] Ident "q" @192..193
-[32] OpenBracket "[" @193..194
-[33] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "0" @194..195
-[34] CloseBracket "]" @195..196
-[35] Semi ";" @196..197
-[36] Whitespace " " @197..198
-[37] LineComment "// Gate with comment" @198..218
-[38] Whitespace "\n" @218..219
-[39] BlockComment { terminated: true } "/* Multi-line block comment\n   spanning multiple lines */" @219..276
-[40] Whitespace "\n" @276..277
-[41] Ident "cx" @277..279
-[42] Whitespace " " @279..280
-[43] Ident "q" @280..281
-[44] OpenBracket "[" @281..282
-[45] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "0" @282..283
-[46] CloseBracket "]" @283..284
-[47] Comma "," @284..285
-[48] Whitespace " " @285..286
-[49] Ident "q" @286..287
-[50] OpenBracket "[" @287..288
-[51] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "1" @288..289
-[52] CloseBracket "]" @289..290
-[53] Semi ";" @290..291
-[54] Whitespace "\n" @291..292
+[1] OpenQasmVersionStmt { major: true, minor: true } "OPENQASM 3.0" @1..13
+[2] Semi ";" @13..14
+[3] Whitespace "\n" @14..15
+[4] LineComment "// Line comment before include" @15..45
+[5] Whitespace "\n" @45..46
+[6] Ident "include" @46..53
+[7] Whitespace " " @53..54
+[8] Literal { kind: Str { terminated: true }, suffix_start: 14 } ""stdgates.inc"" @54..68
+[9] Semi ";" @68..69
+[10] Whitespace " " @69..70
+[11] LineComment "// Inline comment" @70..87
+[12] Whitespace "\n" @87..88
+[13] BlockComment { terminated: true } "/* Block comment before declaration */" @88..126
+[14] Whitespace "\n" @126..127
+[15] Ident "qubit" @127..132
+[16] OpenBracket "[" @132..133
+[17] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "2" @133..134
+[18] CloseBracket "]" @134..135
+[19] Whitespace " " @135..136
+[20] Ident "q" @136..137
+[21] Semi ";" @137..138
+[22] Whitespace " " @138..139
+[23] BlockComment { terminated: true } "/* Inline block comment */" @139..165
+[24] Whitespace "\n\n" @165..167
+[25] LineComment "// Comment before gate" @167..189
+[26] Whitespace "\n" @189..190
+[27] Ident "h" @190..191
+[28] Whitespace " " @191..192
+[29] Ident "q" @192..193
+[30] OpenBracket "[" @193..194
+[31] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "0" @194..195
+[32] CloseBracket "]" @195..196
+[33] Semi ";" @196..197
+[34] Whitespace " " @197..198
+[35] LineComment "// Gate with comment" @198..218
+[36] Whitespace "\n" @218..219
+[37] BlockComment { terminated: true } "/* Multi-line block comment\n   spanning multiple lines */" @219..276
+[38] Whitespace "\n" @276..277
+[39] Ident "cx" @277..279
+[40] Whitespace " " @279..280
+[41] Ident "q" @280..281
+[42] OpenBracket "[" @281..282
+[43] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "0" @282..283
+[44] CloseBracket "]" @283..284
+[45] Comma "," @284..285
+[46] Whitespace " " @285..286
+[47] Ident "q" @286..287
+[48] OpenBracket "[" @287..288
+[49] Literal { kind: Int { base: Decimal, empty_int: false }, suffix_start: 1 } "1" @288..289
+[50] CloseBracket "]" @289..290
+[51] Semi ";" @290..291
+[52] Whitespace "\n" @291..292

crates/pipeline-tests/tests/snapshots/runner__tests__snippets__reference__comments__comments.qasm-parse.snap

@@ -23,7 +23,6 @@ h q[0]; // Gate with comment
 cx q[0], q[1];
 
 VERSION_STRING@1..14: OPENQASM 3.0;
-VERSION@10..14: 3.0;
 INCLUDE@46..69: include "stdgates.inc";
 FILE_PATH@54..68: "stdgates.inc"
 QUANTUM_DECLARATION_STATEMENT@127..138: qubit[2] q;

crates/pipeline-tests/tests/snapshots/runner__tests__snippets__reference__comments__comments.qasm-sema.snap

@@ -3,11 +3,11 @@ source: crates/pipeline-tests/tests/runner.rs
 expression: sema_snap
 ---
 id: tests/snippets/reference/comments/comments.qasm
-expect-sema: Ok
+expect-sema: Todo
 --- sema ---
-ok: true
+ok: false
 panicked: false
-errors: 0
+errors: 1
 --- asg ---
 DeclareQuantum(DeclareQuantum { name: Ok(SymbolId(39)) })