| 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 | # Edge Case: Tree doesn't start with ';' |
| 61 | if not content.startswith(';'): |
| 62 | raise ValueError("tree with no nodes") |
| 63 | |
| 64 | return _parse_tree(content) |
| 65 | |
| 66 | |
| 67 | def _parse_tree(content: str) -> SgfTree: |
| 68 | """ |
| 69 | Parse a tree from its content string. |
| 70 | |
| 71 | Args: |
| 72 | content: The content of the tree without outer parentheses |
| 73 | |
| 74 | Returns: |
| 75 | SgfTree: The parsed tree |
| 76 | """ |
| 77 | # Edge Case: Content is empty |
| 78 | if not content: |
| 79 | raise ValueError("tree with no nodes") |
| 80 | |
| 81 | # Edge Case: Content doesn't start with ';' |
| 82 | if not content.startswith(';'): |
| 83 | raise ValueError("tree with no nodes") |
| 84 | |
| 85 | i = 1 # Skip the initial ';' |
| 86 | properties = {} |
| 87 | children = [] |
| 88 | |
| 89 | # Parse properties |
| 90 | while i < len(content) and content[i] != '(' and content[i] != ';': |
| 91 | # Parse key |
| 92 | key_start = i |
| 93 | while i < len(content) and content[i].isupper(): |
| 94 | i += 1 |
| 95 | |
| 96 | # Edge Case: Property key is not in uppercase |
| 97 | if i == key_start: |
| 98 | raise ValueError("property must be in uppercase") |
| 99 | |
| 100 | key = content[key_start:i] |
| 101 | |
| 102 | # Edge Case: Property has no values |
| 103 | if i >= len(content) or content[i] != '[': |
| 104 | raise ValueError("properties without delimiter") |
| 105 | |
| 106 | # Parse values |
| 107 | values = [] |
| 108 | while i < len(content) and content[i] == '[': |
| 109 | i += 1 # Skip '[' |
| 110 | value_start = i |
| 111 | |
| 112 | # Parse value with escape sequences |
| 113 | value_parts = [] |
| 114 | while i < len(content): |
| 115 | if content[i] == ']': |
| 116 | # Check if this bracket is escaped |
| 117 | # Count consecutive backslashes immediately before this bracket |
| 118 | backslash_count = 0 |
| 119 | j = i - 1 |
| 120 | while j >= 0 and content[j] == '\\': |
| 121 | backslash_count += 1 |
| 122 | j -= 1 |
| 123 | # If odd number of backslashes, the bracket is escaped |
| 124 | is_escaped = backslash_count % 2 == 1 |
| 125 | |
| 126 | if not is_escaped: |
| 127 | # This is a real closing bracket, stop parsing this value |
| 128 | break |
| 129 | else: |
| 130 | # This bracket is escaped, add it to the value |
| 131 | value_parts.append(content[i]) |
| 132 | i += 1 |
| 133 | else: |
| 134 | # For any other character, add it to the value |
| 135 | value_parts.append(content[i]) |
| 136 | i += 1 |
| 137 | |
| 138 | # Edge Case: Unclosed value bracket |
| 139 | if i >= len(content) or content[i] != ']': |
| 140 | raise ValueError("properties without delimiter") |
| 141 | |
| 142 | value = ''.join(value_parts) |
| 143 | # Process SGF text type |
| 144 | value = _process_sgf_text(value) |
| 145 | values.append(value) |
| 146 | i += 1 # Skip ']' |
| 147 | |
| 148 | properties[key] = values |
| 149 | |
| 150 | # Parse children |
| 151 | while i < len(content) and content[i] == '(': |
| 152 | # Find matching parenthesis |
| 153 | paren_count = 1 |
| 154 | start = i + 1 |
| 155 | i += 1 |
| 156 | |
| 157 | while i < len(content) and paren_count > 0: |
| 158 | if content[i] == '(': |
| 159 | paren_count += 1 |
| 160 | elif content[i] == ')': |
| 161 | paren_count -= 1 |
| 162 | i += 1 |
| 163 | |
| 164 | # Edge Case: Unmatched parenthesis |
| 165 | if paren_count != 0: |
| 166 | raise ValueError("tree missing") |
| 167 | |
| 168 | child_content = content[start:i-1] |
| 169 | children.append(_parse_tree(child_content)) |
| 170 | |
| 171 | # Handle shorthand notation (next node without parentheses) |
| 172 | if i < len(content) and content[i] == ';': |
| 173 | child_content = content[i:] |
| 174 | children.append(_parse_tree(child_content)) |
| 175 | |
| 176 | return SgfTree(properties, children) |
| 177 | |
| 178 | |
| 179 | def _process_sgf_text(text: str) -> str: |
| 180 | """ |
| 181 | Process SGF Text type according to SGF specification. |
| 182 | |
| 183 | Args: |
| 184 | text: Raw text value from SGF |
| 185 | |
| 186 | Returns: |
| 187 | str: Processed text |
| 188 | """ |
| 189 | result = [] |
| 190 | i = 0 |
| 191 | while i < len(text): |
| 192 | if text[i] == '\\': |
| 193 | i += 1 |
| 194 | # Edge Case: Escape at end of string |
| 195 | if i >= len(text): |
| 196 | break |
| 197 | |
| 198 | # If next character is newline, remove both backslash and newline |
| 199 | if text[i] == '\n': |
| 200 | i += 1 |
| 201 | continue |
| 202 | else: |
| 203 | # Any non-whitespace character after \ is inserted as-is |
| 204 | # Any whitespace character after \ is converted according to rules |
| 205 | char = text[i] |
| 206 | if char in ' \t\r': |
| 207 | # Convert whitespace to space |
| 208 | result.append(' ') |
| 209 | else: |
| 210 | result.append(char) |
| 211 | i += 1 |
| 212 | else: |
| 213 | char = text[i] |
| 214 | if char == '\n': |
| 215 | result.append('\n') |
| 216 | elif char in ' \t\r': |
| 217 | # All whitespace other than newline converted to space |
| 218 | result.append(' ') |
| 219 | else: |
| 220 | result.append(char) |
| 221 | i += 1 |
| 222 | |
| 223 | return ''.join(result) |
| 224 | # Handled Edge Cases: Empty input, Input doesn't start with '(', Input doesn't end with ')', Empty tree with no nodes, Tree doesn't start with ';', Content is empty, Content doesn't start with ';', Property key is not in uppercase, Property has no values, Unclosed value bracket, Escape at end of string, Unmatched parenthesis |
| 225 | # Handled Edge Cases: Empty input, Input doesn't start with '(', Input doesn't end with ')', Empty tree with no nodes, Tree doesn't start with ';' |
| 226 | # Handled Edge Cases: Content is empty, Content doesn't start with ';', Property key is not in uppercase, Property has no values, Unclosed value bracket, Escape at end of string, Unmatched parenthesis |
| 227 | # Handled Edge Cases: Escape at end of string |