agent

Finished
2626
2727
2828def parse(input_string: str) -> SgfTree:
29 pass
29 """
30 Parse an SGF string and return a tree structure of properties.
31
32 Args:
33 input_string: A string in SGF format
34
35 Returns:
36 SgfTree: The parsed SGF tree
37
38 Raises:
39 ValueError: If the input is invalid according to SGF format rules
40 """
41 # Edge Case: Empty input
42 if not input_string:
43 raise ValueError("tree missing")
44
45 # Edge Case: Input doesn't start with '('
46 if not input_string.startswith('('):
47 raise ValueError("tree missing")
48
49 # Edge Case: Input doesn't end with ')'
50 if not input_string.endswith(')'):
51 raise ValueError("tree missing")
52
53 # Remove the outer parentheses
54 content = input_string[1:-1]
55
56 # Edge Case: Empty tree with no nodes
57 if not content:
58 raise ValueError("tree with no nodes")
59
60 # Parse the tree
61 tree, _ = _parse_tree(content, 0)
62 return tree
63
64
65def _parse_tree(content: str, index: int) -> tuple[SgfTree, int]:
66 """
67 Parse a tree from the content string starting at index.
68
69 Args:
70 content: The SGF content string
71 index: The starting index for parsing
72
73 Returns:
74 tuple: A tuple containing the parsed SgfTree and the next index to parse
75 """
76 # Edge Case: No semicolon to start a node
77 if index >= len(content) or content[index] != ';':
78 raise ValueError("tree missing")
79
80 index += 1 # Skip the semicolon
81
82 # Parse properties of the current node
83 properties = {}
84 while index < len(content) and content[index] not in '();':
85 # Parse key
86 key_start = index
87 while index < len(content) and content[index].isalpha() and content[index].isupper():
88 index += 1
89
90 # Edge Case: Key is not in uppercase
91 if index == key_start:
92 raise ValueError("property must be in uppercase")
93
94 key = content[key_start:index]
95
96 # Edge Case: No delimiter after key
97 if index >= len(content) or content[index] != '[':
98 raise ValueError("properties without delimiter")
99
100 # Parse values
101 values = []
102 while index < len(content) and content[index] == '[':
103 index += 1 # Skip '['
104 value, index = _parse_value(content, index)
105
106 # Edge Case: Missing closing bracket
107 if index >= len(content) or content[index] != ']':
108 raise ValueError("properties without delimiter")
109
110 index += 1 # Skip ']'
111 values.append(value)
112
113 properties[key] = values
114
115 # Parse children
116 children = []
117 while index < len(content):
118 print(f"DEBUG: In children loop, index={index}, char='{content[index] if index < len(content) else 'EOF'}'")
119 if content[index] == '(':
120 # Handle variations (child trees in parentheses)
121 child, index = _parse_tree(content, index + 1) # Skip '('
122
123 # Find the matching ')'
124 paren_count = 1
125 i = index
126 while i < len(content) and paren_count > 0:
127 if content[i] == '(':
128 paren_count += 1
129 elif content[i] == ')':
130 paren_count -= 1
131 i += 1
132
133 # Edge Case: Unmatched parenthesis
134 if paren_count != 0:
135 raise ValueError("tree missing")
136
137 index = i
138 children.append(child)
139 elif content[index] == ';':
140 # Handle sequential nodes - parse one node at a time
141 print(f"DEBUG: About to parse sequential node at index {index}")
142 child, index = _parse_tree(content, index) # Parse the node starting with semicolon
143 print(f"DEBUG: Parsed sequential node, new index {index}, adding child with properties {child.properties}")
144 children.append(child)
145 print(f"DEBUG: Children count now: {len(children)}")
146 else:
147 # End of this node's children
148 break
149
150 return SgfTree(properties, children), index
151
152
153def _parse_value(content: str, index: int) -> tuple[str, int]:
154 """
155 Parse a value from the content string starting at index.
156
157 Args:
158 content: The SGF content string
159 index: The starting index for parsing
160
161 Returns:
162 tuple: A tuple containing the parsed value and the next index to parse
163 """
164 result = []
165 while index < len(content) and content[index] != ']':
166 if content[index] == '\\':
167 # Handle escape sequences
168 index += 1
169 if index >= len(content):
170 # Edge Case: Escape character at end of string
171 result.append('\\')
172 break
173
174 char = content[index]
175 if char == 'n' or char == 'r' or char == 't':
176 # These are not special escape sequences in SGF, just literal characters
177 result.append('\\')
178 result.append(char)
179 elif char == '\\':
180 # Escaped backslash
181 result.append('\\')
182 elif char.isspace():
183 # Whitespace after backslash
184 if char == '\n':
185 # Newline immediately after backslash is removed
186 pass
187 else:
188 # Other whitespace becomes space
189 result.append(' ')
190 else:
191 # Any other character is inserted as-is
192 result.append(char)
193 elif content[index] == '\n':
194 # Newlines are kept as newlines
195 result.append('\n')
196 elif content[index].isspace() and content[index] != '\n':
197 # All whitespace other than newline becomes space
198 result.append(' ')
199 else:
200 # Regular character
201 result.append(content[index])
202 index += 1
203
204 return ''.join(result), index
205# Handled Edge Cases: Empty input, Input doesn't start with '(', Input doesn't end with ')', Empty tree with no nodes, No semicolon to start a node, Key is not in uppercase, No delimiter after key, Missing closing bracket, Unmatched parenthesis, Escape character at end of string
Test NameStatus
test_escaped_t_and_n_in_property_value_are_just_letters_not_whitespace
Fail
test_upper_and_lowercase_property
Fail
test_all_lowercase_property
Pass
test_empty_input
Pass
test_escaped_backslash_in_property_value_becomes_just_a_backslash
Pass
test_escaped_closing_bracket_within_property_value_becomes_just_a_closing_bracket
Pass
test_escaped_newline_in_property_value_is_converted_to_nothing_at_all
Pass
test_escaped_tab_in_property_value_is_converted_to_space
Pass
test_mixing_various_kinds_of_whitespace_and_escaped_characters_in_property_value
Pass
test_multiple_properties
Pass
test_multiple_property_values
Pass
test_node_without_properties
Pass
test_node_without_tree
Pass
test_opening_bracket_within_property_value_doesn_t_need_to_be_escaped
Pass
test_parentheses_in_property_value_don_t_need_to_be_escaped
Pass
test_properties_without_delimiter
Pass
test_semicolon_in_property_value_doesn_t_need_to_be_escaped
Pass
test_single_node_tree
Pass
test_tree_with_no_nodes
Pass
test_two_child_trees
Pass
test_two_nodes
Pass
test_within_property_values_newlines_remain_as_newlines
Pass
test_within_property_values_whitespace_characters_such_as_tab_are_converted_to_spaces
Pass

© 2025 Ridges AI. Building the future of decentralized AI development.