dhaka 0.0.2 → 0.0.3
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/parser/parse_tree.rb +24 -0
- data/lib/parser/parser_methods.rb +1 -2
- data/lib/parser/parser_run.rb +19 -10
- data/test/arithmetic_evaluator.rb +0 -4
- data/test/arithmetic_evaluator_test.rb +5 -5
- data/test/arithmetic_grammar.rb +0 -3
- data/test/arithmetic_tokenizer.rb +2 -2
- data/test/arithmetic_tokenizer_test.rb +7 -7
- data/test/compiled_arithmetic_parser.rb +167 -176
- data/test/compiled_parser_test.rb +5 -6
- data/test/nullable_grammar.rb +1 -1
- data/test/parser_test.rb +5 -8
- metadata +2 -2
data/lib/parser/parse_tree.rb
CHANGED
@@ -11,6 +11,23 @@ module Dhaka
|
|
11
11
|
def to_s
|
12
12
|
"CompositeNode: #{production.symbol} --> [#{child_nodes.join(", ")}]"
|
13
13
|
end
|
14
|
+
def dot_name
|
15
|
+
"Node#{object_id}"
|
16
|
+
end
|
17
|
+
|
18
|
+
def to_dot
|
19
|
+
result = []
|
20
|
+
result << ["digraph x {", "node [fontsize=\"10\" shape=box size=\"5\"]"] if head_node?
|
21
|
+
label = self.production
|
22
|
+
result << "#{dot_name} [label=\"#{label}\"]"
|
23
|
+
child_nodes.each do |child|
|
24
|
+
result << "#{dot_name} -> #{child.dot_name}"
|
25
|
+
result << "#{child.to_dot}"
|
26
|
+
end
|
27
|
+
result << ['}'] if head_node?
|
28
|
+
result.join("\n")
|
29
|
+
end
|
30
|
+
|
14
31
|
def head_node?
|
15
32
|
production.symbol.name == START_SYMBOL_NAME
|
16
33
|
end
|
@@ -27,6 +44,13 @@ module Dhaka
|
|
27
44
|
def to_s
|
28
45
|
"LeafNode: #{token}"
|
29
46
|
end
|
47
|
+
def dot_name
|
48
|
+
"Node#{object_id}"
|
49
|
+
end
|
50
|
+
def to_dot
|
51
|
+
label = "#{token}#{' : '+token.value.to_s if token.value}"
|
52
|
+
"#{dot_name} [label=\"#{label}\"]"
|
53
|
+
end
|
30
54
|
def head_node?
|
31
55
|
false
|
32
56
|
end
|
@@ -2,8 +2,7 @@
|
|
2
2
|
module Dhaka
|
3
3
|
module ParserMethods
|
4
4
|
def parse token_stream
|
5
|
-
|
6
|
-
parser_run = ParserRun.new(grammar, start_state, token_stream+[Token.new(@grammar.end_symbol, nil)])
|
5
|
+
parser_run = ParserRun.new(grammar, start_state, token_stream)
|
7
6
|
parser_run.run
|
8
7
|
end
|
9
8
|
end
|
data/lib/parser/parser_run.rb
CHANGED
@@ -1,6 +1,6 @@
|
|
1
1
|
module Dhaka
|
2
2
|
class ParserRun
|
3
|
-
|
3
|
+
|
4
4
|
def initialize(grammar, start_state, token_stream)
|
5
5
|
@grammar = grammar
|
6
6
|
@node_stack = []
|
@@ -8,21 +8,29 @@ module Dhaka
|
|
8
8
|
@token_stream = token_stream
|
9
9
|
@current_token_index = 0
|
10
10
|
end
|
11
|
-
|
12
|
-
@token_stream[@current_token_index]
|
13
|
-
end
|
14
|
-
def advance
|
15
|
-
node_stack << ParseTreeLeafNode.new(current_token)
|
16
|
-
@current_token_index += 1
|
17
|
-
end
|
11
|
+
|
18
12
|
def run
|
19
|
-
|
13
|
+
for_each_token_with_end do
|
20
14
|
error = execute_action current_token.grammar_symbol.name
|
21
15
|
return error if error
|
22
|
-
|
16
|
+
node_stack << ParseTreeLeafNode.new(current_token)
|
17
|
+
@current_token_index += 1
|
23
18
|
end
|
24
19
|
ParseSuccessResult.new(node_stack[0])
|
25
20
|
end
|
21
|
+
|
22
|
+
private
|
23
|
+
|
24
|
+
attr_reader :state_stack, :token_stream, :node_stack, :current_token
|
25
|
+
|
26
|
+
def for_each_token_with_end
|
27
|
+
token_stream.each do |@current_token|
|
28
|
+
yield
|
29
|
+
end
|
30
|
+
@current_token = Token.new(@grammar.end_symbol, nil)
|
31
|
+
yield
|
32
|
+
end
|
33
|
+
|
26
34
|
def execute_action symbol_name
|
27
35
|
action = state_stack[-1].actions[symbol_name]
|
28
36
|
return ParseErrorResult.new(@current_token_index) unless action
|
@@ -31,5 +39,6 @@ module Dhaka
|
|
31
39
|
end
|
32
40
|
nil
|
33
41
|
end
|
42
|
+
|
34
43
|
end
|
35
44
|
end
|
@@ -14,7 +14,7 @@ class TestArithmeticEvaluator < Test::Unit::TestCase
|
|
14
14
|
|
15
15
|
def test_results_simple_arithmetic_given_tokens_and_syntax_tree_1
|
16
16
|
|
17
|
-
token_stream = [token('n', 2), token('-', nil), token('n', 4)
|
17
|
+
token_stream = [token('n', 2), token('-', nil), token('n', 4)]
|
18
18
|
syntax_tree = get_syntax_tree_with_compiled_arithmetic_parser(token_stream)
|
19
19
|
assert_equal -2, ArithmeticEvaluator.new(syntax_tree, @min_func, @max_func).result
|
20
20
|
|
@@ -22,7 +22,7 @@ class TestArithmeticEvaluator < Test::Unit::TestCase
|
|
22
22
|
|
23
23
|
def test_results_simple_arithmetic_given_tokens_and_syntax_tree_2
|
24
24
|
|
25
|
-
token_stream = [token('n', 2), token('-', nil), token('(', nil), token('n', 3), token('/', nil), token('n', 4), token(')', nil)
|
25
|
+
token_stream = [token('n', 2), token('-', nil), token('(', nil), token('n', 3), token('/', nil), token('n', 4), token(')', nil)]
|
26
26
|
syntax_tree = get_syntax_tree_with_compiled_arithmetic_parser(token_stream)
|
27
27
|
assert_equal 1.25, ArithmeticEvaluator.new(syntax_tree, @min_func, @max_func).result
|
28
28
|
|
@@ -30,7 +30,7 @@ class TestArithmeticEvaluator < Test::Unit::TestCase
|
|
30
30
|
|
31
31
|
def test_results_simple_arithmetic_given_tokens_and_syntax_tree_3
|
32
32
|
|
33
|
-
token_stream = [token('n', 2), token('+', nil), token('(', nil), token('n', 3), token('/', nil), token('(', nil), token('n', 7), token('-', nil), token('n', 5), token(')', nil) , token(')', nil)
|
33
|
+
token_stream = [token('n', 2), token('+', nil), token('(', nil), token('n', 3), token('/', nil), token('(', nil), token('n', 7), token('-', nil), token('n', 5), token(')', nil) , token(')', nil)]
|
34
34
|
syntax_tree = get_syntax_tree_with_compiled_arithmetic_parser(token_stream)
|
35
35
|
assert_equal 3.5, ArithmeticEvaluator.new(syntax_tree, @min_func, @max_func).result
|
36
36
|
|
@@ -38,7 +38,7 @@ class TestArithmeticEvaluator < Test::Unit::TestCase
|
|
38
38
|
|
39
39
|
def test_results_simple_arithmetic_given_tokens_and_syntax_tree_4
|
40
40
|
|
41
|
-
token_stream = [token('n', 2), token('+', nil), token('h', nil), token('(', nil), token('n', 3), token(',', nil), token('n', 4), token(')', nil)
|
41
|
+
token_stream = [token('n', 2), token('+', nil), token('h', nil), token('(', nil), token('n', 3), token(',', nil), token('n', 4), token(')', nil)]
|
42
42
|
syntax_tree = get_syntax_tree_with_compiled_arithmetic_parser(token_stream)
|
43
43
|
assert_equal 6, ArithmeticEvaluator.new(syntax_tree, @min_func, @max_func).result
|
44
44
|
|
@@ -46,7 +46,7 @@ class TestArithmeticEvaluator < Test::Unit::TestCase
|
|
46
46
|
|
47
47
|
def test_results_simple_arithmetic_given_tokens_and_syntax_tree_5
|
48
48
|
|
49
|
-
token_stream = [token('n', 2), token('+', nil), token('l', nil), token('(', nil), token('n', 3), token(',', nil), token('n', 4), token(')', nil)
|
49
|
+
token_stream = [token('n', 2), token('+', nil), token('l', nil), token('(', nil), token('n', 3), token(',', nil), token('n', 4), token(')', nil)]
|
50
50
|
syntax_tree = get_syntax_tree_with_compiled_arithmetic_parser(token_stream)
|
51
51
|
assert_equal 5, ArithmeticEvaluator.new(syntax_tree, @min_func, @max_func).result
|
52
52
|
|
data/test/arithmetic_grammar.rb
CHANGED
@@ -9,10 +9,9 @@ class ArithmeticTokenizer < Dhaka::Tokenizer
|
|
9
9
|
operators = ['-', '+', '/', '*']
|
10
10
|
functions = ['h', 'l']
|
11
11
|
arg_separator = [',']
|
12
|
-
sentinel = ['#']
|
13
12
|
whitespace = [' ']
|
14
13
|
|
15
|
-
all_characters = digits + parenths + operators + functions + arg_separator +
|
14
|
+
all_characters = digits + parenths + operators + functions + arg_separator + whitespace
|
16
15
|
|
17
16
|
for_state :idle_state do
|
18
17
|
for_characters(all_characters - (digits + whitespace)) do
|
@@ -36,6 +35,7 @@ class ArithmeticTokenizer < Dhaka::Tokenizer
|
|
36
35
|
for_characters digits do
|
37
36
|
self.accumulator += curr_char
|
38
37
|
advance
|
38
|
+
tokens << Dhaka::Token.new(ArithmeticGrammar.symbol_for_name('n'), accumulator.to_i) unless curr_char
|
39
39
|
end
|
40
40
|
end
|
41
41
|
|
@@ -6,24 +6,24 @@ class TestArithmeticTokenizer < Test::Unit::TestCase
|
|
6
6
|
assert_equal([], ArithmeticTokenizer.tokenize([]))
|
7
7
|
end
|
8
8
|
def test_tokenizes_given_a_string_input
|
9
|
-
assert_equal([token('n', 2), token('-', nil), token('n', 4)
|
9
|
+
assert_equal([token('n', 2), token('-', nil), token('n', 4)], ArithmeticTokenizer.tokenize('2 - 4'))
|
10
10
|
end
|
11
11
|
def test_a_longer_input
|
12
|
-
actual = ArithmeticTokenizer.tokenize('2+(3 / (7 - 5))
|
13
|
-
assert_equal([token('n', 2), token('+', nil), token('(', nil), token('n', 3), token('/', nil), token('(', nil), token('n', 7), token('-', nil), token('n', 5), token(')', nil) , token(')', nil)
|
12
|
+
actual = ArithmeticTokenizer.tokenize('2+(3 / (7 - 5))')
|
13
|
+
assert_equal([token('n', 2), token('+', nil), token('(', nil), token('n', 3), token('/', nil), token('(', nil), token('n', 7), token('-', nil), token('n', 5), token(')', nil) , token(')', nil)], actual)
|
14
14
|
end
|
15
15
|
|
16
16
|
def test_another_input_with_multi_digit_numbers
|
17
|
-
actual = ArithmeticTokenizer.tokenize('2034 +(3433 / (7 - 5))
|
18
|
-
assert_equal([token('n', 2034), token('+', nil), token('(', nil), token('n', 3433), token('/', nil), token('(', nil), token('n', 7), token('-', nil), token('n', 5), token(')', nil) , token(')', nil)
|
17
|
+
actual = ArithmeticTokenizer.tokenize('2034 +(3433 / (7 - 5))')
|
18
|
+
assert_equal([token('n', 2034), token('+', nil), token('(', nil), token('n', 3433), token('/', nil), token('(', nil), token('n', 7), token('-', nil), token('n', 5), token(')', nil) , token(')', nil)], actual)
|
19
19
|
end
|
20
20
|
|
21
21
|
def test_an_input_with_unrecognized_characters
|
22
|
-
assert_raise(Dhaka::UnrecognizedInputCharacterException) {ArithmeticTokenizer.tokenize('2+(3 / (7 -& 5))
|
22
|
+
assert_raise(Dhaka::UnrecognizedInputCharacterException) {ArithmeticTokenizer.tokenize('2+(3 / (7 -& 5))')}
|
23
23
|
end
|
24
24
|
|
25
25
|
def test_another_input_with_illegal_characters
|
26
|
-
assert_raise(Dhaka::UnrecognizedInputCharacterException) {ArithmeticTokenizer.tokenize('2034 +(34b3 / (7 - 5))
|
26
|
+
assert_raise(Dhaka::UnrecognizedInputCharacterException) {ArithmeticTokenizer.tokenize('2034 +(34b3 / (7 - 5))')}
|
27
27
|
end
|
28
28
|
|
29
29
|
def token(symbol_name, value)
|
@@ -5,248 +5,239 @@ class CompiledArithmeticParser < Dhaka::CompiledParser
|
|
5
5
|
|
6
6
|
self.grammar = ArithmeticGrammar
|
7
7
|
|
8
|
-
start_with
|
8
|
+
start_with 11
|
9
|
+
|
10
|
+
at_state(21) {
|
11
|
+
for_symbol('l') { shift_to 16 }
|
12
|
+
for_symbol('FunctionName') { shift_to 23 }
|
13
|
+
for_symbol('n') { shift_to 12 }
|
14
|
+
for_symbol('Function') { shift_to 31 }
|
15
|
+
for_symbol('F') { shift_to 22 }
|
16
|
+
for_symbol('(') { shift_to 17 }
|
17
|
+
for_symbol('h') { shift_to 13 }
|
18
|
+
}
|
19
|
+
|
20
|
+
at_state(28) {
|
21
|
+
for_symbol('l') { shift_to 16 }
|
22
|
+
for_symbol('FunctionName') { shift_to 23 }
|
23
|
+
for_symbol('n') { shift_to 12 }
|
24
|
+
for_symbol('E') { shift_to 27 }
|
25
|
+
for_symbol('Function') { shift_to 31 }
|
26
|
+
for_symbol('F') { shift_to 30 }
|
27
|
+
for_symbol('Args') { shift_to 29 }
|
28
|
+
for_symbol('(') { shift_to 17 }
|
29
|
+
for_symbol('h') { shift_to 13 }
|
30
|
+
for_symbol(')') { reduce_with 'empty_args' }
|
31
|
+
for_symbol('T') { shift_to 14 }
|
32
|
+
}
|
9
33
|
|
10
|
-
at_state(
|
11
|
-
for_symbol('l') { shift_to
|
12
|
-
for_symbol('
|
13
|
-
for_symbol('
|
14
|
-
for_symbol('
|
15
|
-
for_symbol('
|
16
|
-
for_symbol('
|
17
|
-
for_symbol('(') { shift_to
|
34
|
+
at_state(17) {
|
35
|
+
for_symbol('l') { shift_to 16 }
|
36
|
+
for_symbol('FunctionName') { shift_to 23 }
|
37
|
+
for_symbol('n') { shift_to 12 }
|
38
|
+
for_symbol('E') { shift_to 18 }
|
39
|
+
for_symbol('Function') { shift_to 31 }
|
40
|
+
for_symbol('F') { shift_to 30 }
|
41
|
+
for_symbol('(') { shift_to 17 }
|
42
|
+
for_symbol('h') { shift_to 13 }
|
43
|
+
for_symbol('T') { shift_to 14 }
|
18
44
|
}
|
19
45
|
|
20
|
-
at_state(
|
21
|
-
for_symbol('
|
22
|
-
for_symbol(',') { reduce_with 'subtraction' }
|
23
|
-
for_symbol('-') { reduce_with 'subtraction' }
|
24
|
-
for_symbol('#') { reduce_with 'subtraction' }
|
25
|
-
for_symbol('/') { shift_to 44 }
|
26
|
-
for_symbol(')') { reduce_with 'subtraction' }
|
27
|
-
for_symbol('*') { shift_to 48 }
|
46
|
+
at_state(23) {
|
47
|
+
for_symbol('(') { shift_to 24 }
|
28
48
|
}
|
29
49
|
|
30
|
-
at_state(
|
31
|
-
for_symbol('
|
32
|
-
|
33
|
-
|
34
|
-
|
50
|
+
at_state(16) {
|
51
|
+
for_symbol('(') { reduce_with 'min_function' }
|
52
|
+
}
|
53
|
+
|
54
|
+
at_state(13) {
|
55
|
+
for_symbol('(') { reduce_with 'max_function' }
|
35
56
|
}
|
36
57
|
|
37
|
-
at_state(
|
58
|
+
at_state(20) {
|
38
59
|
for_symbol('+') { reduce_with 'addition' }
|
39
60
|
for_symbol(',') { reduce_with 'addition' }
|
40
61
|
for_symbol('-') { reduce_with 'addition' }
|
41
|
-
for_symbol('
|
42
|
-
for_symbol('
|
62
|
+
for_symbol('/') { shift_to 15 }
|
63
|
+
for_symbol('_End_') { reduce_with 'addition' }
|
43
64
|
for_symbol(')') { reduce_with 'addition' }
|
44
|
-
for_symbol('*') { shift_to
|
65
|
+
for_symbol('*') { shift_to 21 }
|
45
66
|
}
|
46
67
|
|
47
|
-
at_state(
|
48
|
-
for_symbol('+') { reduce_with '
|
49
|
-
for_symbol(',') { reduce_with '
|
50
|
-
for_symbol('-') { reduce_with '
|
51
|
-
for_symbol('
|
52
|
-
for_symbol('
|
53
|
-
for_symbol(')') { reduce_with '
|
54
|
-
for_symbol('*') { reduce_with '
|
68
|
+
at_state(34) {
|
69
|
+
for_symbol('+') { reduce_with 'unpacking_parenthetized_expression' }
|
70
|
+
for_symbol(',') { reduce_with 'unpacking_parenthetized_expression' }
|
71
|
+
for_symbol('-') { reduce_with 'unpacking_parenthetized_expression' }
|
72
|
+
for_symbol('/') { reduce_with 'unpacking_parenthetized_expression' }
|
73
|
+
for_symbol('_End_') { reduce_with 'unpacking_parenthetized_expression' }
|
74
|
+
for_symbol(')') { reduce_with 'unpacking_parenthetized_expression' }
|
75
|
+
for_symbol('*') { reduce_with 'unpacking_parenthetized_expression' }
|
55
76
|
}
|
56
77
|
|
57
|
-
at_state(
|
58
|
-
for_symbol('
|
78
|
+
at_state(32) {
|
79
|
+
for_symbol('l') { shift_to 16 }
|
80
|
+
for_symbol('FunctionName') { shift_to 23 }
|
81
|
+
for_symbol('n') { shift_to 12 }
|
82
|
+
for_symbol('Function') { shift_to 31 }
|
83
|
+
for_symbol('F') { shift_to 30 }
|
84
|
+
for_symbol('(') { shift_to 17 }
|
85
|
+
for_symbol('h') { shift_to 13 }
|
86
|
+
for_symbol('T') { shift_to 33 }
|
59
87
|
}
|
60
88
|
|
61
|
-
at_state(
|
62
|
-
for_symbol('l') { shift_to
|
63
|
-
for_symbol('
|
64
|
-
for_symbol('
|
65
|
-
for_symbol('Function') { shift_to
|
66
|
-
for_symbol('F') { shift_to
|
67
|
-
for_symbol('
|
68
|
-
for_symbol('
|
69
|
-
for_symbol('T') { shift_to 53 }
|
89
|
+
at_state(15) {
|
90
|
+
for_symbol('l') { shift_to 16 }
|
91
|
+
for_symbol('FunctionName') { shift_to 23 }
|
92
|
+
for_symbol('n') { shift_to 12 }
|
93
|
+
for_symbol('Function') { shift_to 31 }
|
94
|
+
for_symbol('F') { shift_to 35 }
|
95
|
+
for_symbol('(') { shift_to 17 }
|
96
|
+
for_symbol('h') { shift_to 13 }
|
70
97
|
}
|
71
98
|
|
72
99
|
at_state(33) {
|
73
|
-
for_symbol('
|
74
|
-
for_symbol('
|
75
|
-
for_symbol('
|
76
|
-
for_symbol('
|
77
|
-
for_symbol('
|
78
|
-
for_symbol('
|
79
|
-
for_symbol('
|
80
|
-
|
81
|
-
|
82
|
-
|
83
|
-
|
84
|
-
|
85
|
-
|
86
|
-
for_symbol('
|
87
|
-
|
88
|
-
|
89
|
-
|
90
|
-
for_symbol('
|
91
|
-
for_symbol('
|
92
|
-
for_symbol('
|
93
|
-
for_symbol('
|
94
|
-
for_symbol('Function') { shift_to 46 }
|
95
|
-
for_symbol('F') { shift_to 50 }
|
96
|
-
for_symbol('h') { shift_to 41 }
|
97
|
-
for_symbol('(') { shift_to 38 }
|
98
|
-
for_symbol('T') { shift_to 54 }
|
99
|
-
}
|
100
|
-
|
101
|
-
at_state(34) {
|
102
|
-
for_symbol('(') { shift_to 35 }
|
100
|
+
for_symbol('+') { reduce_with 'subtraction' }
|
101
|
+
for_symbol(',') { reduce_with 'subtraction' }
|
102
|
+
for_symbol('-') { reduce_with 'subtraction' }
|
103
|
+
for_symbol('/') { shift_to 15 }
|
104
|
+
for_symbol('_End_') { reduce_with 'subtraction' }
|
105
|
+
for_symbol(')') { reduce_with 'subtraction' }
|
106
|
+
for_symbol('*') { shift_to 21 }
|
107
|
+
}
|
108
|
+
|
109
|
+
at_state(24) {
|
110
|
+
for_symbol('l') { shift_to 16 }
|
111
|
+
for_symbol('FunctionName') { shift_to 23 }
|
112
|
+
for_symbol('n') { shift_to 12 }
|
113
|
+
for_symbol('E') { shift_to 27 }
|
114
|
+
for_symbol('Function') { shift_to 31 }
|
115
|
+
for_symbol('F') { shift_to 30 }
|
116
|
+
for_symbol('Args') { shift_to 25 }
|
117
|
+
for_symbol('(') { shift_to 17 }
|
118
|
+
for_symbol('h') { shift_to 13 }
|
119
|
+
for_symbol(')') { reduce_with 'empty_args' }
|
120
|
+
for_symbol('T') { shift_to 14 }
|
103
121
|
}
|
104
122
|
|
105
|
-
at_state(
|
106
|
-
for_symbol('
|
123
|
+
at_state(11) {
|
124
|
+
for_symbol('l') { shift_to 16 }
|
125
|
+
for_symbol('FunctionName') { shift_to 23 }
|
126
|
+
for_symbol('n') { shift_to 12 }
|
127
|
+
for_symbol('E') { shift_to 36 }
|
128
|
+
for_symbol('Function') { shift_to 31 }
|
129
|
+
for_symbol('F') { shift_to 30 }
|
130
|
+
for_symbol('(') { shift_to 17 }
|
131
|
+
for_symbol('h') { shift_to 13 }
|
132
|
+
for_symbol('T') { shift_to 14 }
|
107
133
|
}
|
108
134
|
|
109
|
-
at_state(
|
110
|
-
for_symbol('
|
135
|
+
at_state(35) {
|
136
|
+
for_symbol('+') { reduce_with 'division' }
|
137
|
+
for_symbol(',') { reduce_with 'division' }
|
138
|
+
for_symbol('-') { reduce_with 'division' }
|
139
|
+
for_symbol('/') { reduce_with 'division' }
|
140
|
+
for_symbol('_End_') { reduce_with 'division' }
|
141
|
+
for_symbol(')') { reduce_with 'division' }
|
142
|
+
for_symbol('*') { reduce_with 'division' }
|
111
143
|
}
|
112
144
|
|
113
|
-
at_state(
|
145
|
+
at_state(31) {
|
114
146
|
for_symbol('+') { reduce_with 'function' }
|
115
147
|
for_symbol(',') { reduce_with 'function' }
|
116
148
|
for_symbol('-') { reduce_with 'function' }
|
117
|
-
for_symbol('#') { reduce_with 'function' }
|
118
149
|
for_symbol('/') { reduce_with 'function' }
|
150
|
+
for_symbol('_End_') { reduce_with 'function' }
|
119
151
|
for_symbol(')') { reduce_with 'function' }
|
120
152
|
for_symbol('*') { reduce_with 'function' }
|
121
153
|
}
|
122
154
|
|
123
|
-
at_state(
|
124
|
-
for_symbol('+') { reduce_with '
|
125
|
-
for_symbol(',') { reduce_with '
|
126
|
-
for_symbol('-') { reduce_with '
|
127
|
-
for_symbol('
|
128
|
-
for_symbol('
|
129
|
-
for_symbol(')') { reduce_with '
|
130
|
-
for_symbol('*') { reduce_with '
|
155
|
+
at_state(26) {
|
156
|
+
for_symbol('+') { reduce_with 'evaluating_function' }
|
157
|
+
for_symbol(',') { reduce_with 'evaluating_function' }
|
158
|
+
for_symbol('-') { reduce_with 'evaluating_function' }
|
159
|
+
for_symbol('/') { reduce_with 'evaluating_function' }
|
160
|
+
for_symbol('_End_') { reduce_with 'evaluating_function' }
|
161
|
+
for_symbol(')') { reduce_with 'evaluating_function' }
|
162
|
+
for_symbol('*') { reduce_with 'evaluating_function' }
|
131
163
|
}
|
132
164
|
|
133
|
-
at_state(
|
134
|
-
for_symbol('
|
135
|
-
for_symbol('
|
136
|
-
for_symbol('
|
137
|
-
for_symbol('E') { shift_to 55 }
|
138
|
-
for_symbol('Function') { shift_to 46 }
|
139
|
-
for_symbol('F') { shift_to 50 }
|
140
|
-
for_symbol('Args') { shift_to 36 }
|
141
|
-
for_symbol('h') { shift_to 41 }
|
142
|
-
for_symbol('(') { shift_to 38 }
|
143
|
-
for_symbol(')') { reduce_with 'empty_args' }
|
144
|
-
for_symbol('T') { shift_to 54 }
|
165
|
+
at_state(18) {
|
166
|
+
for_symbol('+') { shift_to 19 }
|
167
|
+
for_symbol('-') { shift_to 32 }
|
168
|
+
for_symbol(')') { shift_to 34 }
|
145
169
|
}
|
146
170
|
|
147
|
-
at_state(
|
171
|
+
at_state(14) {
|
148
172
|
for_symbol('+') { reduce_with 'term' }
|
149
173
|
for_symbol(',') { reduce_with 'term' }
|
150
174
|
for_symbol('-') { reduce_with 'term' }
|
151
|
-
for_symbol('
|
152
|
-
for_symbol('
|
175
|
+
for_symbol('/') { shift_to 15 }
|
176
|
+
for_symbol('_End_') { reduce_with 'term' }
|
153
177
|
for_symbol(')') { reduce_with 'term' }
|
154
|
-
for_symbol('*') { shift_to
|
155
|
-
}
|
156
|
-
|
157
|
-
at_state(58) {
|
158
|
-
for_symbol('+') { shift_to 52 }
|
159
|
-
for_symbol('-') { shift_to 40 }
|
160
|
-
for_symbol('#') { reduce_with 'expression' }
|
178
|
+
for_symbol('*') { shift_to 21 }
|
161
179
|
}
|
162
180
|
|
163
|
-
at_state(
|
164
|
-
for_symbol('
|
165
|
-
for_symbol('n') { shift_to 42 }
|
166
|
-
for_symbol('FunctionName') { shift_to 34 }
|
167
|
-
for_symbol('E') { shift_to 55 }
|
168
|
-
for_symbol('Function') { shift_to 46 }
|
169
|
-
for_symbol('F') { shift_to 50 }
|
170
|
-
for_symbol('Args') { shift_to 57 }
|
171
|
-
for_symbol('h') { shift_to 41 }
|
172
|
-
for_symbol('(') { shift_to 38 }
|
173
|
-
for_symbol(')') { reduce_with 'empty_args' }
|
174
|
-
for_symbol('T') { shift_to 54 }
|
181
|
+
at_state(25) {
|
182
|
+
for_symbol(')') { shift_to 26 }
|
175
183
|
}
|
176
184
|
|
177
|
-
at_state(
|
178
|
-
for_symbol('+') { reduce_with '
|
179
|
-
for_symbol(',') { reduce_with '
|
180
|
-
for_symbol('-') { reduce_with '
|
181
|
-
for_symbol('
|
182
|
-
for_symbol('
|
183
|
-
for_symbol(')') { reduce_with '
|
184
|
-
for_symbol('*') { reduce_with '
|
185
|
-
}
|
186
|
-
|
187
|
-
at_state(44) {
|
188
|
-
for_symbol('l') { shift_to 47 }
|
189
|
-
for_symbol('n') { shift_to 42 }
|
190
|
-
for_symbol('FunctionName') { shift_to 34 }
|
191
|
-
for_symbol('F') { shift_to 45 }
|
192
|
-
for_symbol('Function') { shift_to 46 }
|
193
|
-
for_symbol('h') { shift_to 41 }
|
194
|
-
for_symbol('(') { shift_to 38 }
|
185
|
+
at_state(22) {
|
186
|
+
for_symbol('+') { reduce_with 'multiplication' }
|
187
|
+
for_symbol(',') { reduce_with 'multiplication' }
|
188
|
+
for_symbol('-') { reduce_with 'multiplication' }
|
189
|
+
for_symbol('/') { reduce_with 'multiplication' }
|
190
|
+
for_symbol('_End_') { reduce_with 'multiplication' }
|
191
|
+
for_symbol(')') { reduce_with 'multiplication' }
|
192
|
+
for_symbol('*') { reduce_with 'multiplication' }
|
195
193
|
}
|
196
194
|
|
197
|
-
at_state(
|
195
|
+
at_state(12) {
|
198
196
|
for_symbol('+') { reduce_with 'getting_literals' }
|
199
197
|
for_symbol(',') { reduce_with 'getting_literals' }
|
200
198
|
for_symbol('-') { reduce_with 'getting_literals' }
|
201
|
-
for_symbol('#') { reduce_with 'getting_literals' }
|
202
199
|
for_symbol('/') { reduce_with 'getting_literals' }
|
200
|
+
for_symbol('_End_') { reduce_with 'getting_literals' }
|
203
201
|
for_symbol(')') { reduce_with 'getting_literals' }
|
204
202
|
for_symbol('*') { reduce_with 'getting_literals' }
|
205
203
|
}
|
206
204
|
|
207
205
|
at_state(36) {
|
208
|
-
for_symbol('
|
206
|
+
for_symbol('+') { shift_to 19 }
|
207
|
+
for_symbol('-') { shift_to 32 }
|
208
|
+
for_symbol('_End_') { reduce_with 'expression' }
|
209
209
|
}
|
210
210
|
|
211
|
-
at_state(
|
212
|
-
for_symbol('+') { reduce_with '
|
213
|
-
for_symbol(',') { reduce_with '
|
214
|
-
for_symbol('-') { reduce_with '
|
215
|
-
for_symbol('
|
216
|
-
for_symbol('
|
217
|
-
for_symbol(')') { reduce_with '
|
218
|
-
for_symbol('*') { reduce_with '
|
211
|
+
at_state(30) {
|
212
|
+
for_symbol('+') { reduce_with 'factor' }
|
213
|
+
for_symbol(',') { reduce_with 'factor' }
|
214
|
+
for_symbol('-') { reduce_with 'factor' }
|
215
|
+
for_symbol('/') { reduce_with 'factor' }
|
216
|
+
for_symbol('_End_') { reduce_with 'factor' }
|
217
|
+
for_symbol(')') { reduce_with 'factor' }
|
218
|
+
for_symbol('*') { reduce_with 'factor' }
|
219
219
|
}
|
220
220
|
|
221
|
-
at_state(
|
221
|
+
at_state(29) {
|
222
222
|
for_symbol(')') { reduce_with 'concatenating_args' }
|
223
223
|
}
|
224
224
|
|
225
|
-
at_state(
|
226
|
-
for_symbol('
|
227
|
-
for_symbol('
|
228
|
-
for_symbol('
|
229
|
-
for_symbol('
|
230
|
-
for_symbol('F') { shift_to 50 }
|
231
|
-
for_symbol('h') { shift_to 41 }
|
232
|
-
for_symbol('(') { shift_to 38 }
|
233
|
-
for_symbol('T') { shift_to 43 }
|
234
|
-
}
|
235
|
-
|
236
|
-
at_state(39) {
|
237
|
-
for_symbol('+') { shift_to 52 }
|
238
|
-
for_symbol('-') { shift_to 40 }
|
239
|
-
for_symbol(')') { shift_to 51 }
|
225
|
+
at_state(27) {
|
226
|
+
for_symbol('+') { shift_to 19 }
|
227
|
+
for_symbol(',') { shift_to 28 }
|
228
|
+
for_symbol('-') { shift_to 32 }
|
229
|
+
for_symbol(')') { reduce_with 'single_args' }
|
240
230
|
}
|
241
231
|
|
242
|
-
at_state(
|
243
|
-
for_symbol('
|
244
|
-
for_symbol('
|
245
|
-
for_symbol('
|
246
|
-
for_symbol('
|
247
|
-
for_symbol('
|
248
|
-
for_symbol('
|
249
|
-
for_symbol('
|
232
|
+
at_state(19) {
|
233
|
+
for_symbol('l') { shift_to 16 }
|
234
|
+
for_symbol('FunctionName') { shift_to 23 }
|
235
|
+
for_symbol('n') { shift_to 12 }
|
236
|
+
for_symbol('Function') { shift_to 31 }
|
237
|
+
for_symbol('F') { shift_to 30 }
|
238
|
+
for_symbol('(') { shift_to 17 }
|
239
|
+
for_symbol('h') { shift_to 13 }
|
240
|
+
for_symbol('T') { shift_to 20 }
|
250
241
|
}
|
251
242
|
|
252
243
|
end
|
@@ -29,7 +29,7 @@ class TestCompiledParser < Test::Unit::TestCase
|
|
29
29
|
end
|
30
30
|
|
31
31
|
def test_compiled_parser_generates_syntax_tree_for_arithmetic_grammar
|
32
|
-
parser_input = ['(','n','-','(','n','/','n','-','n',')','/','n',')'
|
32
|
+
parser_input = ['(','n','-','(','n','/','n','-','n',')','/','n',')']
|
33
33
|
assert_equal \
|
34
34
|
["getting_literals",
|
35
35
|
"factor",
|
@@ -50,16 +50,15 @@ class TestCompiledParser < Test::Unit::TestCase
|
|
50
50
|
"unpacking_parenthetized_expression",
|
51
51
|
"factor",
|
52
52
|
"term",
|
53
|
-
"expression",
|
54
|
-
"start_production"], get_syntax_tree_with_compiled_arithmetic_parser(build_tokens(parser_input, CompiledArithmeticParser.grammar)).linearize
|
53
|
+
"expression"], get_syntax_tree_with_compiled_arithmetic_parser(build_tokens(parser_input, CompiledArithmeticParser.grammar)).linearize
|
55
54
|
end
|
56
55
|
|
57
|
-
def
|
58
|
-
|
56
|
+
def test_parse_result_has_error_if_empty_token_array
|
57
|
+
assert CompiledArithmeticParser.parse([]).has_error?
|
59
58
|
end
|
60
59
|
|
61
60
|
def test_parser_returns_error_result_with_index_of_bad_token_if_parse_error
|
62
|
-
parse_result = CompiledArithmeticParser.parse(build_tokens(['(', '-', ')'
|
61
|
+
parse_result = CompiledArithmeticParser.parse(build_tokens(['(', '-', ')'], CompiledArithmeticParser.grammar))
|
63
62
|
assert parse_result.has_error?
|
64
63
|
assert_equal 1, parse_result.bad_token_index
|
65
64
|
end
|
data/test/nullable_grammar.rb
CHANGED
data/test/parser_test.rb
CHANGED
@@ -76,8 +76,7 @@ class ParserTest < Test::Unit::TestCase
|
|
76
76
|
def test_with_a_different_grammar_with_division
|
77
77
|
grammar = ArithmeticGrammar
|
78
78
|
parser = Dhaka::Parser.new(grammar)
|
79
|
-
|
80
|
-
parser_input = ['(','n','-','(','n','/','n','-','n',')','/','n',')','#']
|
79
|
+
parser_input = ['(','n','-','(','n','/','n','-','n',')','/','n',')']
|
81
80
|
assert_equal \
|
82
81
|
["getting_literals",
|
83
82
|
"factor",
|
@@ -98,10 +97,9 @@ class ParserTest < Test::Unit::TestCase
|
|
98
97
|
"unpacking_parenthetized_expression",
|
99
98
|
"factor",
|
100
99
|
"term",
|
101
|
-
"expression",
|
102
|
-
"start_production"], get_linearized_parse_result(parser_input, parser)
|
100
|
+
"expression"], get_linearized_parse_result(parser_input, parser)
|
103
101
|
|
104
|
-
parser_input = ['h','(','(','n',')','-','n',',','n',')'
|
102
|
+
parser_input = ['h','(','(','n',')','-','n',',','n',')']
|
105
103
|
assert_equal \
|
106
104
|
["max_function",
|
107
105
|
"getting_literals",
|
@@ -122,14 +120,13 @@ class ParserTest < Test::Unit::TestCase
|
|
122
120
|
"function",
|
123
121
|
"factor",
|
124
122
|
"term",
|
125
|
-
"expression",
|
126
|
-
"start_production"], get_linearized_parse_result(parser_input, parser)
|
123
|
+
"expression"], get_linearized_parse_result(parser_input, parser)
|
127
124
|
end
|
128
125
|
|
129
126
|
def test_with_a_grammar_with_nullables_after_terminals
|
130
127
|
grammar = NullableGrammar
|
131
128
|
parser = Dhaka::Parser.new(grammar)
|
132
|
-
parser_input = ['(','a',')'
|
129
|
+
parser_input = ['(','a',')']
|
133
130
|
assert_equal \
|
134
131
|
["literal_a",
|
135
132
|
"empty_element_list",
|
metadata
CHANGED
@@ -3,8 +3,8 @@ rubygems_version: 0.9.0
|
|
3
3
|
specification_version: 1
|
4
4
|
name: dhaka
|
5
5
|
version: !ruby/object:Gem::Version
|
6
|
-
version: 0.0.
|
7
|
-
date: 2006-
|
6
|
+
version: 0.0.3
|
7
|
+
date: 2006-12-04 00:00:00 -05:00
|
8
8
|
summary: An LALR1 parser generator written in Ruby
|
9
9
|
require_paths:
|
10
10
|
- lib
|