Function ielr::compute_table

source · 
pub fn compute_table(
    algorithm: Algorithm,
    grammar: &Grammar,
    start_nodes: impl IntoIterator<Item = Node>
) -> Result<(Table, Statistics), Error>
Expand description

Main function of this crate.

This computes the LR table for a given grammar.

Parameters

  • algorithm: which Algorithm to use to build the LR table.

    The actual computation may use a less powerful algorithm: in this case, you need to check algorithm_needed field of Statistics.

  • max_states: the maximum number of states the algorithm is allowed to compute.

    This is to prevent the algorithm from taking too many time. Most real-life programming language generate no more than 5000 states. Anything higher than 1_000_000 will certainly take way too many time and memory in practice.

  • grammar: the specification of the grammar used. See input to build it.

  • start_nodes: which nodes may be used to start parsing.

Panics

At the moment, any algorithm different from Algorithm::Lalr(1) or Algorithm::Lr(1) will result in a panic.

Examples found in repository?
examples/grammar_with_precedence.rs (lines 161-167)
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169

fn main() {
    use Symbol::{Node as N, Token as T};

    let mut grammar = Grammar::new();
    // productions that do not need precedence annotations
    for (lhs, rhs) in [
        (START, vec![N(STATEMENTS)]),
        (STATEMENTS, vec![N(STATEMENT), N(STATEMENTS)]),
        (STATEMENTS, vec![]),
        (STATEMENT, vec![N(FUNCTION)]),
        (STATEMENT, vec![T(RETURN_KW), N(EXPRESSION), T(SEMICOLON)]),
        (
            STATEMENT,
            vec![T(LET_KW), T(IDENT), T(EQUAL), N(EXPRESSION), T(SEMICOLON)],
        ),
        (STATEMENT, vec![N(EXPRESSION), T(SEMICOLON)]),
        (
            FUNCTION,
            vec![
                T(FN_KW),
                T(IDENT),
                T(PARENTHESIS_LEFT),
                N(FUNCTION_ARGS),
                T(PARENTHESIS_RIGHT),
                T(BRACE_LEFT),
                N(STATEMENTS),
                T(BRACE_RIGHT),
            ],
        ),
        (
            FUNCTION_ARGS,
            vec![N(FUNCTION_ARG), T(COMMA), N(FUNCTION_ARGS)],
        ),
        (FUNCTION_ARGS, vec![N(FUNCTION_ARG)]),
        (FUNCTION_ARGS, vec![]),
        (FUNCTION_ARG, vec![T(IDENT)]),
        (EXPRESSION, vec![T(INT)]),
        (EXPRESSION, vec![T(IDENT)]),
        (ARGS, vec![N(EXPRESSION), T(COMMA), N(ARGS)]),
        (ARGS, vec![N(EXPRESSION)]),
        (ARGS, vec![]),
    ] {
        grammar.add_production(lhs, rhs).unwrap();
    }

    let precedence_family = grammar.add_precedence_family();

    // productions that need precedence annotations
    for (lhs, rhs, left, right) in [
        (
            EXPRESSION,
            vec![N(EXPRESSION), T(PLUS), N(EXPRESSION)],
            Some(1),
            Some(2),
        ),
        (
            EXPRESSION,
            vec![N(EXPRESSION), T(MINUS), N(EXPRESSION)],
            Some(1),
            Some(2),
        ),
        (
            EXPRESSION,
            vec![N(EXPRESSION), T(STAR), N(EXPRESSION)],
            Some(3),
            Some(4),
        ),
        (
            EXPRESSION,
            vec![N(EXPRESSION), T(SLASH), N(EXPRESSION)],
            Some(3),
            Some(4),
        ),
        (EXPRESSION, vec![T(MINUS), N(EXPRESSION)], None, Some(5)),
    ] {
        let production = grammar.add_production(lhs, rhs).unwrap();
        let production = grammar.get_production_mut(production).unwrap();
        if let Some(left) = left {
            production.set_left_precedence(precedence_family, left);
        }
        if let Some(right) = right {
            production.set_right_precedence(precedence_family, right);
        }
    }

    // Now we build the parsing table !
    let (_tables, _statistics) = ielr::compute_table(
        // This grammar only requires LALR(1)
        ielr::Algorithm::Lalr(std::num::NonZeroU8::new(1).unwrap()),
        // We do not care about a maximum number of states
        &grammar,
        [START],
    )
    .unwrap();
}
More examples
Hide additional examples
examples/grammar_with_conflicts.rs (lines 187-193)
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195

fn main() {
    use Symbol::{Node as N, Token as T};

    let mut grammar = Grammar::new();
    // productions that do not need precedence annotations
    for (lhs, rhs) in [
        (START, vec![N(STATEMENTS)]),
        (STATEMENTS, vec![N(STATEMENT), N(STATEMENTS)]),
        (STATEMENTS, vec![]),
        (STATEMENT, vec![N(FUNCTION)]),
        (STATEMENT, vec![T(RETURN_KW), N(EXPRESSION), T(SEMICOLON)]),
        (
            STATEMENT,
            vec![T(LET_KW), T(IDENT), T(EQUAL), N(EXPRESSION), T(SEMICOLON)],
        ),
        (STATEMENT, vec![N(EXPRESSION), T(SEMICOLON)]),
        (
            FUNCTION,
            vec![
                T(FN_KW),
                T(IDENT),
                T(PARENTHESIS_LEFT),
                N(FUNCTION_ARGS),
                T(PARENTHESIS_RIGHT),
                T(BRACE_LEFT),
                N(STATEMENTS),
                T(BRACE_RIGHT),
            ],
        ),
        (
            FUNCTION_ARGS,
            vec![N(FUNCTION_ARG), T(COMMA), N(FUNCTION_ARGS)],
        ),
        (FUNCTION_ARGS, vec![N(FUNCTION_ARG)]),
        (FUNCTION_ARGS, vec![]),
        (FUNCTION_ARG, vec![T(IDENT)]),
        (EXPRESSION, vec![T(INT)]),
        (EXPRESSION, vec![T(IDENT)]),
        (ARGS, vec![N(EXPRESSION), T(COMMA), N(ARGS)]),
        (ARGS, vec![N(EXPRESSION)]),
        (ARGS, vec![]),
    ] {
        grammar.add_production(lhs, rhs).unwrap();
    }

    let add_prod = grammar
        .add_production(EXPRESSION, vec![N(EXPRESSION), T(PLUS), N(EXPRESSION)])
        .unwrap();
    let sub_prod = grammar
        .add_production(EXPRESSION, vec![N(EXPRESSION), T(MINUS), N(EXPRESSION)])
        .unwrap();
    let mul_prod = grammar
        .add_production(EXPRESSION, vec![N(EXPRESSION), T(STAR), N(EXPRESSION)])
        .unwrap();
    let div_prod = grammar
        .add_production(EXPRESSION, vec![N(EXPRESSION), T(SLASH), N(EXPRESSION)])
        .unwrap();
    let neg_prod = grammar
        .add_production(EXPRESSION, vec![T(MINUS), N(EXPRESSION)])
        .unwrap();

    // Add conflicts solutions.

    let operator_to_production = HashMap::from([
        (PLUS, add_prod),
        (MINUS, sub_prod),
        (STAR, mul_prod),
        (SLASH, div_prod),
    ]);

    // - all binary operators are left-associative here
    // - '-' has precedence over all binary operators
    for operator in [PLUS, MINUS, STAR, SLASH] {
        grammar.add_conflict_solution(ConflictSolution {
            prefer: ConflictingAction::Reduce(operator_to_production[&operator]),
            over: ConflictingAction::Shift(Lookahead::Token(operator)),
        });
        grammar.add_conflict_solution(ConflictSolution {
            prefer: ConflictingAction::Reduce(neg_prod),
            over: ConflictingAction::Shift(Lookahead::Token(operator)),
        });
    }

    // operator with the same precedence
    for (operator1, operator2) in [(PLUS, MINUS), (STAR, SLASH)] {
        grammar.add_conflict_solution(ConflictSolution {
            prefer: ConflictingAction::Reduce(operator_to_production[&operator1]),
            over: ConflictingAction::Shift(Lookahead::Token(operator2)),
        });
        grammar.add_conflict_solution(ConflictSolution {
            prefer: ConflictingAction::Reduce(operator_to_production[&operator2]),
            over: ConflictingAction::Shift(Lookahead::Token(operator1)),
        });
    }

    // operator with the different precedence
    for (prefer, over) in [(STAR, PLUS), (STAR, MINUS), (SLASH, PLUS), (SLASH, MINUS)] {
        grammar.add_conflict_solution(ConflictSolution {
            prefer: ConflictingAction::Shift(Lookahead::Token(prefer)),
            over: ConflictingAction::Reduce(operator_to_production[&over]),
        });
        grammar.add_conflict_solution(ConflictSolution {
            prefer: ConflictingAction::Reduce(operator_to_production[&prefer]),
            over: ConflictingAction::Shift(Lookahead::Token(over)),
        });
    }

    // Now we build the parsing table !
    let (_tables, _statistics) = ielr::compute_table(
        // We are required to specify LR(1) when using conflict resolution.
        ielr::Algorithm::Lr(std::num::NonZeroU8::new(1).unwrap()),
        // We do not care about a maximum number of states
        &grammar,
        [START],
    )
    .unwrap();
}