|
| 1 | +# Expression DSL |
| 2 | + |
| 3 | +The PyIceberg library provides a powerful expression DSL (Domain Specific Language) for building complex row filter expressions. This guide will help you understand how to use the expression DSL effectively. This DSL allows you to build type-safe expressions for use in the `row_filter` scan argument. |
| 4 | + |
| 5 | +They are composed of terms, predicates, and logical operators. |
| 6 | + |
| 7 | +## Basic Concepts |
| 8 | + |
| 9 | +### Terms |
| 10 | + |
| 11 | +Terms are the basic building blocks of expressions. They represent references to fields in your data: |
| 12 | + |
| 13 | +```python |
| 14 | +from pyiceberg.expressions import Reference |
| 15 | + |
| 16 | +# Create a reference to a field named "age" |
| 17 | +age_field = Reference("age") |
| 18 | +``` |
| 19 | + |
| 20 | +### Predicates |
| 21 | + |
| 22 | +Predicates are expressions that evaluate to a boolean value. They can be combined using logical operators. |
| 23 | + |
| 24 | +#### Comparison Predicates |
| 25 | + |
| 26 | +```python |
| 27 | +from pyiceberg.expressions import EqualTo, NotEqualTo, LessThan, LessThanOrEqual, GreaterThan, GreaterThanOrEqual |
| 28 | + |
| 29 | +# age equals 18 |
| 30 | +age_equals_18 = EqualTo("age", 18) |
| 31 | + |
| 32 | +# age is not equal to 18 |
| 33 | +age_not_equals_18 = NotEqualTo("age", 18) |
| 34 | + |
| 35 | +# age is less than 18 |
| 36 | +age_less_than_18 = LessThan("age", 18) |
| 37 | + |
| 38 | +# Less than or equal to |
| 39 | +age_less_than_or_equal_18 = LessThanOrEqual("age", 18) |
| 40 | + |
| 41 | +# Greater than |
| 42 | +age_greater_than_18 = GreaterThan("age", 18) |
| 43 | + |
| 44 | +# Greater than or equal to |
| 45 | +age_greater_than_or_equal_18 = GreaterThanOrEqual("age", 18) |
| 46 | +``` |
| 47 | + |
| 48 | +#### Set Predicates |
| 49 | + |
| 50 | +```python |
| 51 | +from pyiceberg.expressions import In, NotIn |
| 52 | + |
| 53 | +# age is one of 18, 19, 20 |
| 54 | +age_in_set = In("age", [18, 19, 20]) |
| 55 | + |
| 56 | +# age is not 18, 19, oer 20 |
| 57 | +age_not_in_set = NotIn("age", [18, 19, 20]) |
| 58 | +``` |
| 59 | + |
| 60 | +#### Null Predicates |
| 61 | + |
| 62 | +```python |
| 63 | +from pyiceberg.expressions import IsNull, NotNull |
| 64 | + |
| 65 | +# Is null |
| 66 | +name_is_null = IsNull("name") |
| 67 | + |
| 68 | +# Is not null |
| 69 | +name_is_not_null = NotNull("name") |
| 70 | +``` |
| 71 | + |
| 72 | +#### String Predicates |
| 73 | + |
| 74 | +```python |
| 75 | +from pyiceberg.expressions import StartsWith, NotStartsWith |
| 76 | + |
| 77 | +# TRUE for 'Johnathan', FALSE for 'Johan' |
| 78 | +name_starts_with = StartsWith("name", "John") |
| 79 | + |
| 80 | +# FALSE for 'Johnathan', TRUE for 'Johan' |
| 81 | +name_not_starts_with = NotStartsWith("name", "John") |
| 82 | +``` |
| 83 | + |
| 84 | +### Logical Operators |
| 85 | + |
| 86 | +You can combine predicates using logical operators: |
| 87 | + |
| 88 | +```python |
| 89 | +from pyiceberg.expressions import And, Or, Not |
| 90 | + |
| 91 | +# TRUE for 25, FALSE for 67 and 15 |
| 92 | +age_between = And( |
| 93 | + GreaterThanOrEqual("age", 18), |
| 94 | + LessThanOrEqual("age", 65) |
| 95 | +) |
| 96 | + |
| 97 | +# FALSE for 25, TRUE for 67 and 15 |
| 98 | +age_outside = Or( |
| 99 | + LessThan("age", 18), |
| 100 | + GreaterThan("age", 65) |
| 101 | +) |
| 102 | + |
| 103 | +# NOT operator |
| 104 | +not_adult = Not(GreaterThanOrEqual("age", 18)) |
| 105 | +``` |
| 106 | + |
| 107 | +## Advanced Usage |
| 108 | + |
| 109 | +### Complex Expressions |
| 110 | + |
| 111 | +You can build complex expressions by combining multiple predicates and operators: |
| 112 | + |
| 113 | +```python |
| 114 | +from pyiceberg.expressions import And, Or, Not, EqualTo, GreaterThan, LessThan, In |
| 115 | + |
| 116 | +# (age >= 18 AND age <= 65) AND (status = 'active' OR status = 'pending') |
| 117 | +complex_filter = And( |
| 118 | + And( |
| 119 | + GreaterThanOrEqual("age", 18), |
| 120 | + LessThanOrEqual("age", 65) |
| 121 | + ), |
| 122 | + Or( |
| 123 | + EqualTo("status", "active"), |
| 124 | + EqualTo("status", "pending") |
| 125 | + ) |
| 126 | +) |
| 127 | + |
| 128 | +# NOT (age < 18 OR age > 65) |
| 129 | +age_in_range = Not( |
| 130 | + Or( |
| 131 | + LessThan("age", 18), |
| 132 | + GreaterThan("age", 65) |
| 133 | + ) |
| 134 | +) |
| 135 | +``` |
| 136 | + |
| 137 | +### Type Safety |
| 138 | + |
| 139 | +The expression DSL provides type safety through Python's type system. When you create expressions, the types are checked at runtime: |
| 140 | + |
| 141 | +```python |
| 142 | +from pyiceberg.expressions import EqualTo |
| 143 | + |
| 144 | +# This will work |
| 145 | +age_equals_18 = EqualTo("age", 18) |
| 146 | + |
| 147 | +# This will raise a TypeError if the field type doesn't match |
| 148 | +age_equals_18 = EqualTo("age", "18") # Will fail if age is an integer field |
| 149 | +``` |
| 150 | + |
| 151 | +## Best Practices |
| 152 | + |
| 153 | +1. **Use Type Hints**: Always use type hints when working with expressions to catch type-related errors early. |
| 154 | + |
| 155 | +2. **Break Down Complex Expressions**: For complex expressions, break them down into smaller, more manageable parts: |
| 156 | + |
| 157 | +```python |
| 158 | +# Instead of this: |
| 159 | +complex_filter = And( |
| 160 | + And( |
| 161 | + GreaterThanOrEqual("age", 18), |
| 162 | + LessThanOrEqual("age", 65) |
| 163 | + ), |
| 164 | + Or( |
| 165 | + EqualTo("status", "active"), |
| 166 | + EqualTo("status", "pending") |
| 167 | + ) |
| 168 | +) |
| 169 | + |
| 170 | +# Do this: |
| 171 | +age_range = And( |
| 172 | + GreaterThanOrEqual("age", 18), |
| 173 | + LessThanOrEqual("age", 65) |
| 174 | +) |
| 175 | + |
| 176 | +status_filter = Or( |
| 177 | + EqualTo("status", "active"), |
| 178 | + EqualTo("status", "pending") |
| 179 | +) |
| 180 | + |
| 181 | +complex_filter = And(age_range, status_filter) |
| 182 | +``` |
| 183 | + |
| 184 | +## Common Pitfalls |
| 185 | + |
| 186 | +1. **Type Mismatches**: Always ensure that the types of your literals match the field types in your schema. |
| 187 | + |
| 188 | +2. **Null Handling**: Be careful when using `IsNull` and `NotNull` predicates with required fields. The expression DSL will automatically optimize these cases: |
| 189 | + - `IsNull` on a required field will always return `False` |
| 190 | + - `NotNull` on a required field will always return `True` |
| 191 | + |
| 192 | +3. **String Comparisons**: When using string predicates like `StartsWith`, ensure that the field type is actually a string type. |
| 193 | + |
| 194 | +## Examples |
| 195 | + |
| 196 | +Here are some practical examples of using the expression DSL: |
| 197 | + |
| 198 | +### Basic Filtering |
| 199 | + |
| 200 | +```python |
| 201 | + |
| 202 | +from datetime import datetime |
| 203 | +from pyiceberg.expressions import ( |
| 204 | + And, |
| 205 | + EqualTo, |
| 206 | + GreaterThanOrEqual, |
| 207 | + LessThanOrEqual, |
| 208 | + GreaterThan, |
| 209 | + In |
| 210 | +) |
| 211 | + |
| 212 | +active_adult_users_filter = And( |
| 213 | + EqualTo("status", "active"), |
| 214 | + GreaterThanOrEqual("age", 18) |
| 215 | +) |
| 216 | + |
| 217 | + |
| 218 | +high_value_customers = And( |
| 219 | + GreaterThan("total_spent", 1000), |
| 220 | + In("membership_level", ["gold", "platinum"]) |
| 221 | +) |
| 222 | + |
| 223 | +date_range_filter = And( |
| 224 | + GreaterThanOrEqual("created_at", datetime(2024, 1, 1)), |
| 225 | + LessThanOrEqual("created_at", datetime(2024, 12, 31)) |
| 226 | +) |
| 227 | +``` |
| 228 | + |
| 229 | +### Multi-Condition Filter |
| 230 | + |
| 231 | +```python |
| 232 | +from pyiceberg.expressions import And, Or, Not, EqualTo, GreaterThan |
| 233 | + |
| 234 | +complex_filter = And( |
| 235 | + Not(EqualTo("status", "deleted")), |
| 236 | + Or( |
| 237 | + And( |
| 238 | + EqualTo("type", "premium"), |
| 239 | + GreaterThan("subscription_months", 12) |
| 240 | + ), |
| 241 | + EqualTo("type", "enterprise") |
| 242 | + ) |
| 243 | +) |
| 244 | +``` |
0 commit comments