| 1 | #include <Common/typeid_cast.h> |
| 2 | |
| 3 | #include <Interpreters/IdentifierSemantic.h> |
| 4 | |
| 5 | namespace DB |
| 6 | { |
| 7 | |
| 8 | namespace ErrorCodes |
| 9 | { |
| 10 | extern const int AMBIGUOUS_COLUMN_NAME; |
| 11 | } |
| 12 | |
| 13 | namespace |
| 14 | { |
| 15 | |
| 16 | const DatabaseAndTableWithAlias & extractTable(const DatabaseAndTableWithAlias & table) |
| 17 | { |
| 18 | return table; |
| 19 | } |
| 20 | |
| 21 | const DatabaseAndTableWithAlias & extractTable(const TableWithColumnNames & table) |
| 22 | { |
| 23 | return table.table; |
| 24 | } |
| 25 | |
| 26 | template <typename T> |
| 27 | IdentifierSemantic::ColumnMatch tryChooseTable(const ASTIdentifier & identifier, const std::vector<T> & tables, |
| 28 | size_t & best_table_pos, bool allow_ambiguous) |
| 29 | { |
| 30 | using ColumnMatch = IdentifierSemantic::ColumnMatch; |
| 31 | |
| 32 | best_table_pos = 0; |
| 33 | auto best_match = ColumnMatch::NoMatch; |
| 34 | size_t same_match = 0; |
| 35 | |
| 36 | for (size_t i = 0; i < tables.size(); ++i) |
| 37 | { |
| 38 | auto match = IdentifierSemantic::canReferColumnToTable(identifier, extractTable(tables[i])); |
| 39 | if (match != ColumnMatch::NoMatch) |
| 40 | { |
| 41 | if (match > best_match) |
| 42 | { |
| 43 | best_match = match; |
| 44 | best_table_pos = i; |
| 45 | same_match = 0; |
| 46 | } |
| 47 | else if (match == best_match) |
| 48 | ++same_match; |
| 49 | } |
| 50 | } |
| 51 | |
| 52 | if ((best_match != ColumnMatch::NoMatch) && same_match) |
| 53 | { |
| 54 | if (!allow_ambiguous) |
| 55 | throw Exception("Ambiguous column '" + identifier.name + "'" , ErrorCodes::AMBIGUOUS_COLUMN_NAME); |
| 56 | return ColumnMatch::Ambiguous; |
| 57 | } |
| 58 | return best_match; |
| 59 | } |
| 60 | |
| 61 | } |
| 62 | |
| 63 | std::optional<String> IdentifierSemantic::getColumnName(const ASTIdentifier & node) |
| 64 | { |
| 65 | if (!node.semantic->special) |
| 66 | return node.name; |
| 67 | return {}; |
| 68 | } |
| 69 | |
| 70 | std::optional<String> IdentifierSemantic::getColumnName(const ASTPtr & ast) |
| 71 | { |
| 72 | if (ast) |
| 73 | if (const auto * id = ast->as<ASTIdentifier>()) |
| 74 | if (!id->semantic->special) |
| 75 | return id->name; |
| 76 | return {}; |
| 77 | } |
| 78 | |
| 79 | std::optional<String> IdentifierSemantic::getTableName(const ASTIdentifier & node) |
| 80 | { |
| 81 | if (node.semantic->special) |
| 82 | return node.name; |
| 83 | return {}; |
| 84 | } |
| 85 | |
| 86 | std::optional<String> IdentifierSemantic::getTableName(const ASTPtr & ast) |
| 87 | { |
| 88 | if (ast) |
| 89 | if (const auto * id = ast->as<ASTIdentifier>()) |
| 90 | if (id->semantic->special) |
| 91 | return id->name; |
| 92 | return {}; |
| 93 | } |
| 94 | |
| 95 | std::optional<ASTIdentifier> IdentifierSemantic::uncover(const ASTIdentifier & identifier) |
| 96 | { |
| 97 | if (identifier.semantic->covered) |
| 98 | { |
| 99 | std::vector<String> name_parts = identifier.name_parts; |
| 100 | return ASTIdentifier(std::move(name_parts)); |
| 101 | } |
| 102 | return {}; |
| 103 | } |
| 104 | |
| 105 | void IdentifierSemantic::coverName(ASTIdentifier & identifier, const String & alias) |
| 106 | { |
| 107 | identifier.setShortName(alias); |
| 108 | identifier.semantic->covered = true; |
| 109 | } |
| 110 | |
| 111 | bool IdentifierSemantic::canBeAlias(const ASTIdentifier & identifier) |
| 112 | { |
| 113 | return identifier.semantic->can_be_alias; |
| 114 | } |
| 115 | |
| 116 | void IdentifierSemantic::setMembership(ASTIdentifier & identifier, size_t table_pos) |
| 117 | { |
| 118 | identifier.semantic->membership = table_pos; |
| 119 | identifier.semantic->can_be_alias = false; |
| 120 | } |
| 121 | |
| 122 | std::optional<size_t> IdentifierSemantic::getMembership(const ASTIdentifier & identifier) |
| 123 | { |
| 124 | return identifier.semantic->membership; |
| 125 | } |
| 126 | |
| 127 | bool IdentifierSemantic::chooseTable(const ASTIdentifier & identifier, const std::vector<DatabaseAndTableWithAlias> & tables, |
| 128 | size_t & best_table_pos, bool ambiguous) |
| 129 | { |
| 130 | static constexpr auto no_match = IdentifierSemantic::ColumnMatch::NoMatch; |
| 131 | return tryChooseTable<DatabaseAndTableWithAlias>(identifier, tables, best_table_pos, ambiguous) != no_match; |
| 132 | } |
| 133 | |
| 134 | bool IdentifierSemantic::chooseTable(const ASTIdentifier & identifier, const std::vector<TableWithColumnNames> & tables, |
| 135 | size_t & best_table_pos, bool ambiguous) |
| 136 | { |
| 137 | static constexpr auto no_match = IdentifierSemantic::ColumnMatch::NoMatch; |
| 138 | return tryChooseTable<TableWithColumnNames>(identifier, tables, best_table_pos, ambiguous) != no_match; |
| 139 | } |
| 140 | |
| 141 | std::pair<String, String> IdentifierSemantic::extractDatabaseAndTable(const ASTIdentifier & identifier) |
| 142 | { |
| 143 | if (identifier.name_parts.size() > 2) |
| 144 | throw Exception("Logical error: more than two components in table expression" , ErrorCodes::LOGICAL_ERROR); |
| 145 | |
| 146 | if (identifier.name_parts.size() == 2) |
| 147 | return { identifier.name_parts[0], identifier.name_parts[1] }; |
| 148 | return { "" , identifier.name }; |
| 149 | } |
| 150 | |
| 151 | std::optional<String> IdentifierSemantic::(const ASTIdentifier & identifier, const String & table_name) |
| 152 | { |
| 153 | if (identifier.name_parts.size() == 3 && table_name == identifier.name_parts[0]) |
| 154 | return identifier.name_parts[1] + '.' + identifier.name_parts[2]; |
| 155 | else if (identifier.name_parts.size() == 2) |
| 156 | return identifier.name_parts[0] + '.' + identifier.name_parts[1]; |
| 157 | return {}; |
| 158 | } |
| 159 | |
| 160 | bool IdentifierSemantic::doesIdentifierBelongTo(const ASTIdentifier & identifier, const String & database, const String & table) |
| 161 | { |
| 162 | size_t num_components = identifier.name_parts.size(); |
| 163 | if (num_components >= 3) |
| 164 | return identifier.name_parts[0] == database && |
| 165 | identifier.name_parts[1] == table; |
| 166 | return false; |
| 167 | } |
| 168 | |
| 169 | bool IdentifierSemantic::doesIdentifierBelongTo(const ASTIdentifier & identifier, const String & table) |
| 170 | { |
| 171 | size_t num_components = identifier.name_parts.size(); |
| 172 | if (num_components >= 2) |
| 173 | return identifier.name_parts[0] == table; |
| 174 | return false; |
| 175 | } |
| 176 | |
| 177 | IdentifierSemantic::ColumnMatch IdentifierSemantic::canReferColumnToTable(const ASTIdentifier & identifier, |
| 178 | const DatabaseAndTableWithAlias & db_and_table) |
| 179 | { |
| 180 | /// database.table.column |
| 181 | if (doesIdentifierBelongTo(identifier, db_and_table.database, db_and_table.table)) |
| 182 | return ColumnMatch::DbAndTable; |
| 183 | |
| 184 | /// alias.column |
| 185 | if (doesIdentifierBelongTo(identifier, db_and_table.alias)) |
| 186 | return ColumnMatch::TableAlias; |
| 187 | |
| 188 | /// table.column |
| 189 | if (doesIdentifierBelongTo(identifier, db_and_table.table)) |
| 190 | { |
| 191 | if (!db_and_table.alias.empty()) |
| 192 | return ColumnMatch::AliasedTableName; |
| 193 | else |
| 194 | return ColumnMatch::TableName; |
| 195 | } |
| 196 | |
| 197 | return ColumnMatch::NoMatch; |
| 198 | } |
| 199 | |
| 200 | /// Strip qualificators from left side of column name. |
| 201 | /// Example: 'database.table.name' -> 'name'. |
| 202 | void IdentifierSemantic::setColumnShortName(ASTIdentifier & identifier, const DatabaseAndTableWithAlias & db_and_table) |
| 203 | { |
| 204 | auto match = IdentifierSemantic::canReferColumnToTable(identifier, db_and_table); |
| 205 | size_t to_strip = 0; |
| 206 | switch (match) |
| 207 | { |
| 208 | case ColumnMatch::TableName: |
| 209 | case ColumnMatch::AliasedTableName: |
| 210 | case ColumnMatch::TableAlias: |
| 211 | to_strip = 1; |
| 212 | break; |
| 213 | case ColumnMatch::DbAndTable: |
| 214 | to_strip = 2; |
| 215 | break; |
| 216 | default: |
| 217 | break; |
| 218 | } |
| 219 | |
| 220 | if (!to_strip) |
| 221 | return; |
| 222 | |
| 223 | std::vector<String> stripped(identifier.name_parts.begin() + to_strip, identifier.name_parts.end()); |
| 224 | |
| 225 | DB::String new_name; |
| 226 | for (const auto & part : stripped) |
| 227 | { |
| 228 | if (!new_name.empty()) |
| 229 | new_name += '.'; |
| 230 | new_name += part; |
| 231 | } |
| 232 | identifier.name.swap(new_name); |
| 233 | } |
| 234 | |
| 235 | void IdentifierSemantic::setColumnLongName(ASTIdentifier & identifier, const DatabaseAndTableWithAlias & db_and_table) |
| 236 | { |
| 237 | String prefix = db_and_table.getQualifiedNamePrefix(); |
| 238 | if (!prefix.empty()) |
| 239 | { |
| 240 | String short_name = identifier.shortName(); |
| 241 | identifier.name = prefix + short_name; |
| 242 | prefix.resize(prefix.size() - 1); /// crop dot |
| 243 | identifier.name_parts = {prefix, short_name}; |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | } |
| 248 | |