| 26 | 26 | |
| 27 | 27 | |
| 28 | 28 | def 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 | |
| 65 | def _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 to parse from |
| 72 | |
| 73 | Returns: |
| 74 | tuple: (SgfTree, next_index) |
| 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].isupper(): |
| 85 | # Parse key |
| 86 | key_start = index |
| 87 | while index < len(content) 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 values for the property |
| 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_chars = [] |
| 105 | |
| 106 | # Parse value, handling escapes |
| 107 | while index < len(content): |
| 108 | if content[index] == ']': |
| 109 | # Check if this bracket is escaped |
| 110 | if value_chars and value_chars[-1] == '\\': |
| 111 | # This bracket is escaped, add it and continue |
| 112 | value_chars.append(content[index]) |
| 113 | index += 1 |
| 114 | else: |
| 115 | # This is the real closing bracket |
| 116 | break |
| 117 | elif content[index] == '\\': |
| 118 | # Add the escape character and the next character |
| 119 | value_chars.append(content[index]) |
| 120 | index += 1 |
| 121 | if index < len(content): |
| 122 | value_chars.append(content[index]) |
| 123 | index += 1 |
| 124 | else: |
| 125 | value_chars.append(content[index]) |
| 126 | index += 1 |
| 127 | |
| 128 | # Edge Case: Unclosed value bracket |
| 129 | if index >= len(content) or content[index] != ']': |
| 130 | raise ValueError("properties without delimiter") |
| 131 | |
| 132 | value = ''.join(value_chars) |
| 133 | # Process the value according to SGF Text type rules |
| 134 | value = _process_text_value(value) |
| 135 | values.append(value) |
| 136 | index += 1 # Skip ']' |
| 137 | |
| 138 | properties[key] = values |
| 139 | |
| 140 | # Check if we have a non-uppercase character where we expect a property |
| 141 | if index < len(content) and not content[index].isupper() and content[index] not in ';()': |
| 142 | # This might be a lowercase property key |
| 143 | if content[index].isalpha(): |
| 144 | raise ValueError("property must be in uppercase") |
| 145 | |
| 146 | # Parse children |
| 147 | children = [] |
| 148 | |
| 149 | # Handle sequential nodes (separated by ;) |
| 150 | if index < len(content) and content[index] == ';': |
| 151 | child, index = _parse_tree(content, index) |
| 152 | children.append(child) |
| 153 | |
| 154 | # Handle variations (nodes in parentheses) |
| 155 | while index < len(content) and content[index] == '(': |
| 156 | index += 1 # Skip '(' |
| 157 | child, index = _parse_tree(content, index) |
| 158 | children.append(child) |
| 159 | # Skip ')' |
| 160 | if index < len(content) and content[index] == ')': |
| 161 | index += 1 |
| 162 | |
| 163 | return SgfTree(properties, children), index |
| 164 | |
| 165 | |
| 166 | def _process_text_value(value: str) -> str: |
| 167 | """ |
| 168 | Process a text value according to SGF Text type rules. |
| 169 | |
| 170 | Args: |
| 171 | value: The raw text value |
| 172 | |
| 173 | Returns: |
| 174 | str: The processed text value |
| 175 | """ |
| 176 | result = [] |
| 177 | i = 0 |
| 178 | while i < len(value): |
| 179 | if value[i] == '\\': |
| 180 | # Edge Case: Escape at end of string |
| 181 | if i + 1 >= len(value): |
| 182 | result.append('\\') |
| 183 | break |
| 184 | |
| 185 | next_char = value[i+1] |
| 186 | # If next character is newline, remove the backslash and newline |
| 187 | if next_char == 'n': |
| 188 | i += 2 # Skip backslash and newline |
| 189 | continue |
| 190 | else: |
| 191 | # Any non-whitespace character after \ is inserted as-is |
| 192 | # Any whitespace character after \ follows the above rules |
| 193 | if next_char.isspace() and next_char != 'n': |
| 194 | result.append(' ') |
| 195 | else: |
| 196 | result.append(next_char) |
| 197 | i += 2 |
| 198 | else: |
| 199 | # All whitespace characters other than newline are converted to spaces |
| 200 | if value[i].isspace() and value[i] != '\n': |
| 201 | result.append(' ') |
| 202 | else: |
| 203 | result.append(value[i]) |
| 204 | i += 1 |
| 205 | |
| 206 | return ''.join(result) |
| 207 | # 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 values for the property, Unclosed value bracket, Escape at end of string |