| 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 containing SGF formatted data |
| 34 | |
| 35 | Returns: |
| 36 | SgfTree: A tree structure representing the SGF data |
| 37 | |
| 38 | Raises: |
| 39 | ValueError: If the input is invalid according to SGF format rules |
| 40 | """ |
| 41 | # Edge Case: Empty input or input that doesn't start with '(' |
| 42 | if not input_string or not input_string.startswith('('): |
| 43 | raise ValueError("tree missing") |
| 44 | |
| 45 | # Edge Case: Input that doesn't end with ')' |
| 46 | if not input_string.endswith(')'): |
| 47 | # Check if this might be a property without delimiter case |
| 48 | if '(' in input_string and ';' in input_string: |
| 49 | # Try to parse to see if it's a property without delimiter |
| 50 | try: |
| 51 | # Remove the outer parentheses |
| 52 | content = input_string[1:] |
| 53 | if content.startswith(';'): |
| 54 | # Try to parse the properties |
| 55 | node_end = _find_node_end(content, 1) |
| 56 | node_content = content[1:node_end] |
| 57 | _parse_properties(node_content) |
| 58 | except ValueError as e: |
| 59 | if "properties without delimiter" in str(e): |
| 60 | raise ValueError("properties without delimiter") |
| 61 | elif "property must be in uppercase" in str(e): |
| 62 | raise ValueError("property must be in uppercase") |
| 63 | raise ValueError("tree missing") |
| 64 | |
| 65 | # Remove the outer parentheses |
| 66 | content = input_string[1:-1] |
| 67 | |
| 68 | # Edge Case: Empty content after removing parentheses |
| 69 | if not content: |
| 70 | raise ValueError("tree with no nodes") |
| 71 | |
| 72 | # Edge Case: Content that doesn't start with ';' |
| 73 | if not content.startswith(';'): |
| 74 | raise ValueError("tree missing") |
| 75 | |
| 76 | return _parse_tree_content(content) |
| 77 | |
| 78 | |
| 79 | def _parse_tree_content(content: str) -> SgfTree: |
| 80 | """ |
| 81 | Parse the content of a tree (without the outer parentheses). |
| 82 | |
| 83 | Args: |
| 84 | content: The content of the tree |
| 85 | |
| 86 | Returns: |
| 87 | SgfTree: A tree structure representing the content |
| 88 | """ |
| 89 | # Find the first node |
| 90 | if not content.startswith(';'): |
| 91 | raise ValueError("tree missing") |
| 92 | |
| 93 | # Parse the first node |
| 94 | node_end = _find_node_end(content, 1) |
| 95 | node_content = content[1:node_end] |
| 96 | |
| 97 | properties = _parse_properties(node_content) |
| 98 | |
| 99 | # Parse children |
| 100 | children_content = content[node_end:] |
| 101 | children = _parse_children(children_content) |
| 102 | |
| 103 | return SgfTree(properties, children) |
| 104 | |
| 105 | |
| 106 | def _find_node_end(content: str, start: int) -> int: |
| 107 | """ |
| 108 | Find the end position of a node in the content. |
| 109 | |
| 110 | Args: |
| 111 | content: The content to search in |
| 112 | start: The start position |
| 113 | |
| 114 | Returns: |
| 115 | int: The end position of the node |
| 116 | """ |
| 117 | i = start |
| 118 | while i < len(content): |
| 119 | # If we encounter a ';' or '(' or ')', it's the end of the current node |
| 120 | if content[i] in [';', '(', ')']: |
| 121 | break |
| 122 | i += 1 |
| 123 | return i |
| 124 | |
| 125 | |
| 126 | def _parse_properties(content: str) -> dict: |
| 127 | """ |
| 128 | Parse properties from a node content. |
| 129 | |
| 130 | Args: |
| 131 | content: The content of a node |
| 132 | |
| 133 | Returns: |
| 134 | dict: A dictionary of properties |
| 135 | |
| 136 | Raises: |
| 137 | ValueError: If properties are malformed |
| 138 | """ |
| 139 | properties = {} |
| 140 | i = 0 |
| 141 | |
| 142 | while i < len(content): |
| 143 | # Skip whitespace |
| 144 | if content[i].isspace(): |
| 145 | i += 1 |
| 146 | continue |
| 147 | |
| 148 | # Parse key |
| 149 | key_start = i |
| 150 | while i < len(content) and content[i].isalpha(): |
| 151 | i += 1 |
| 152 | |
| 153 | # Edge Case: No key found |
| 154 | if i == key_start: |
| 155 | raise ValueError("properties without delimiter") |
| 156 | |
| 157 | key = content[key_start:i] |
| 158 | |
| 159 | # Edge Case: Key is not uppercase |
| 160 | if key != key.upper(): |
| 161 | raise ValueError("property must be in uppercase") |
| 162 | |
| 163 | # Parse values |
| 164 | if i >= len(content) or content[i] != '[': |
| 165 | raise ValueError("properties without delimiter") |
| 166 | |
| 167 | values = [] |
| 168 | while i < len(content) and content[i] == '[': |
| 169 | i += 1 # Skip '[' |
| 170 | value_start = i |
| 171 | |
| 172 | # Parse value according to SGF Text type rules |
| 173 | value_chars = [] |
| 174 | while i < len(content) and content[i] != ']': |
| 175 | if content[i] == '\\': |
| 176 | # Handle escape character |
| 177 | i += 1 |
| 178 | if i < len(content): |
| 179 | if content[i] == '\n': |
| 180 | # Newlines are removed if they come immediately after a '\' |
| 181 | i += 1 |
| 182 | else: |
| 183 | # Any character after '\' is inserted as-is |
| 184 | value_chars.append(content[i]) |
| 185 | i += 1 |
| 186 | else: |
| 187 | # Edge Case: Escape at end of content |
| 188 | raise ValueError("properties without delimiter") |
| 189 | else: |
| 190 | # All whitespace characters other than newline are converted to spaces |
| 191 | if content[i].isspace() and content[i] != '\n': |
| 192 | value_chars.append(' ') |
| 193 | else: |
| 194 | value_chars.append(content[i]) |
| 195 | i += 1 |
| 196 | |
| 197 | # Edge Case: Unclosed bracket |
| 198 | if i >= len(content) or content[i] != ']': |
| 199 | raise ValueError("properties without delimiter") |
| 200 | |
| 201 | value = ''.join(value_chars) |
| 202 | values.append(value) |
| 203 | i += 1 # Skip ']' |
| 204 | |
| 205 | properties[key] = values |
| 206 | |
| 207 | return properties |
| 208 | |
| 209 | |
| 210 | def _parse_children(content: str) -> list: |
| 211 | """ |
| 212 | Parse children from content after a node. |
| 213 | |
| 214 | Args: |
| 215 | content: The content after a node |
| 216 | |
| 217 | Returns: |
| 218 | list: A list of child SgfTree objects |
| 219 | """ |
| 220 | children = [] |
| 221 | i = 0 |
| 222 | |
| 223 | while i < len(content): |
| 224 | # Skip whitespace |
| 225 | if content[i].isspace(): |
| 226 | i += 1 |
| 227 | continue |
| 228 | |
| 229 | if content[i] == '(': |
| 230 | # Find matching parenthesis |
| 231 | paren_count = 1 |
| 232 | start = i + 1 |
| 233 | i += 1 |
| 234 | |
| 235 | while i < len(content) and paren_count > 0: |
| 236 | if content[i] == '(': |
| 237 | paren_count += 1 |
| 238 | elif content[i] == ')': |
| 239 | paren_count -= 1 |
| 240 | i += 1 |
| 241 | |
| 242 | # Edge Case: Unmatched parenthesis |
| 243 | if paren_count != 0: |
| 244 | raise ValueError("tree missing") |
| 245 | |
| 246 | # Parse the child tree |
| 247 | child_content = content[start:i-1] |
| 248 | |
| 249 | # Edge Case: Empty child content |
| 250 | if not child_content: |
| 251 | raise ValueError("tree with no nodes") |
| 252 | |
| 253 | # Edge Case: Child content doesn't start with ';' |
| 254 | if not child_content.startswith(';'): |
| 255 | raise ValueError("tree missing") |
| 256 | |
| 257 | child = _parse_tree_content(';' + child_content) |
| 258 | children.append(child) |
| 259 | elif content[i] == ';': |
| 260 | # This is a sequential node, create a child with it and any of its children |
| 261 | node_end = _find_node_end(content, i+1) |
| 262 | node_content = content[i+1:node_end] |
| 263 | |
| 264 | properties = _parse_properties(node_content) |
| 265 | |
| 266 | # Parse any children of this sequential node |
| 267 | remaining_content = content[node_end:] |
| 268 | remaining_children = _parse_children(remaining_content) |
| 269 | |
| 270 | # Create a single child node with the properties and any remaining children |
| 271 | child = SgfTree(properties, remaining_children) |
| 272 | children.append(child) |
| 273 | |
| 274 | # We've processed the rest of the content, so we're done |
| 275 | break |
| 276 | else: |
| 277 | # Edge Case: Unexpected character |
| 278 | raise ValueError("tree missing") |
| 279 | |
| 280 | return children |
| 281 | # Handled Edge Cases: Empty input, missing parentheses, missing semicolon, empty content, invalid property format, lowercase properties, unclosed brackets, unmatched parentheses, empty child content, unexpected characters |
| 282 | # Handled Edge Cases: Empty input, missing parentheses, missing semicolon, empty content after removing parentheses |
| 283 | # Handled Edge Cases: No key found, key not uppercase, missing opening bracket, unclosed bracket, escape at end of content |
| 284 | # Handled Edge Cases: Unmatched parenthesis, empty child content, child content not starting with semicolon, unexpected characters |