| name | qgis-syntax-expressions |
| description | Use when writing QGIS expressions for filtering, labeling, symbology, or field calculations. Prevents expression syntax errors and context misconfiguration. Covers QgsExpression parsing, evaluation contexts, field calculator, data-defined properties, and custom functions. Keywords: QgsExpression, expression, field calculator, label expression, data-defined, @qgsfunction, filter, evaluate, calculate field, formula, conditional label, dynamic value.
|
| license | MIT |
| compatibility | Designed for Claude Code. Requires QGIS 3.44+ / PyQGIS 3.x. |
| metadata | {"author":"OpenAEC-Foundation","version":"1.0"} |
qgis-syntax-expressions
Quick Reference
Expression Evaluation Pipeline
| Step | Class | Purpose |
|---|
| 1. Parse | QgsExpression(string) | Validate syntax, build AST |
| 2. Context | QgsExpressionContext() | Provide variables, fields, scopes |
| 3. Scopes | QgsExpressionContextUtils | Add global/project/layer/feature scopes |
| 4. Evaluate | exp.evaluate(context) | Execute and return result |
| 5. Validate | exp.hasEvalError() | Check for runtime errors |
Expression Syntax Cheatsheet
| Element | Syntax | Example |
|---|
| Field reference | "field_name" (double quotes) | "population" |
| String literal | 'text' (single quotes) | 'hello world' |
| Number literal | 123, 3.14 | 42 |
| NULL check | IS NULL, IS NOT NULL | "name" IS NOT NULL |
| Comparison | =, !=, >, >=, <, <= | "area" > 100 |
| Logical | AND, OR, NOT | "pop" > 1000 AND "type" = 'city' |
| Arithmetic | +, -, *, /, ^, % | "population" * 1.05 |
| Pattern match | LIKE, ILIKE, ~ (regex) | "name" ILIKE '%lake%' |
| Concatenation | || | "first" || ' ' || "last" |
| Conditional | CASE WHEN ... THEN ... END | CASE WHEN "pop" > 1e6 THEN 'large' ELSE 'small' END |
| Geometry vars | $area, $length, $x, $y | $area / 1e6 |
| Current geometry | $geometry | num_points($geometry) |
Common Expression Functions
| Category | Functions |
|---|
| Math | sqrt(), abs(), round(), floor(), ceil(), sin(), cos(), pi() |
| String | upper(), lower(), length(), trim(), replace(), regexp_replace(), substr(), left(), right() |
| Conversion | to_int(), to_real(), to_string(), to_date(), to_datetime() |
| Date/Time | now(), day(), month(), year(), hour(), minute(), age(), day_of_week() |
| Geometry | $area, $length, $x, $y, $perimeter, centroid(), buffer(), area(), length(), num_geometries(), num_points() |
| Aggregates | aggregate(), sum(), count(), mean(), min(), max(), concatenate(), array_agg() |
| Conditionals | if(), coalesce(), nullif(), try() |
| Arrays | array(), array_length(), array_contains(), array_append(), array_to_string() |
| Map/Record |
Critical Warnings
ALWAYS check exp.hasParserError() after creating a QgsExpression. A parser error means the expression string is syntactically invalid and evaluation will fail silently or return NULL.
ALWAYS check exp.hasEvalError() after calling exp.evaluate(). Evaluation errors occur when the expression is syntactically valid but fails at runtime (e.g., field not found, type mismatch).
ALWAYS set up a proper QgsExpressionContext with appropriate scopes when evaluating expressions that reference fields, project variables, or layer properties. Evaluating without context returns NULL for all field references.
NEVER use double quotes for string literals in expressions. Double quotes reference field names: "name" reads the field, 'name' is the string literal.
NEVER forget to call context.setFeature(feature) before evaluating feature-dependent expressions in a loop. Omitting this causes all features to evaluate against stale or empty data.
NEVER use print() inside custom expression functions (@qgsfunction). Use QgsMessageLog instead. Expression functions run during rendering and print() causes thread-safety issues.
ALWAYS specify referenced_columns in @qgsfunction for performance. Use [QgsFeatureRequest.ALL_ATTRIBUTES] if the function reads arbitrary fields; use [] if it reads no fields.
ALWAYS set usesgeometry=True in @qgsfunction if the function accesses feature.geometry(). Omitting this causes the geometry to be unavailable.
Decision Tree: Choosing the Right Expression Approach
Need to use an expression?
|
+-- Filter features? --> QgsFeatureRequest.setFilterExpression()
|
+-- Calculate field values? --> Field Calculator pattern (edit session + evaluate per feature)
|
+-- Drive labeling? --> QgsPalLayerSettings.fieldName + isExpression = True
|
+-- Drive symbology? --> QgsProperty.fromExpression() on symbol/renderer properties
|
+-- Evaluate standalone? --> QgsExpression.evaluate(context)
|
+-- Need custom logic? --> @qgsfunction decorator + QgsExpression.registerFunction()
Essential Patterns
Pattern 1: Parse and Evaluate an Expression
from qgis.core import (
QgsExpression, QgsExpressionContext, QgsExpressionContextUtils
)
exp = QgsExpression('"population" * 1.05')
if exp.hasParserError():
raise ValueError(f"Parse error: {exp.parserErrorString()}")
context = QgsExpressionContext()
context.appendScopes(
QgsExpressionContextUtils.globalProjectLayerScopes(layer)
)
for feature in layer.getFeatures():
context.setFeature(feature)
value = exp.evaluate(context)
if exp.hasEvalError():
raise ValueError(f"Eval error: {exp.evalErrorString()}")
print(f"Feature {feature.id()}: {value}")
Pattern 2: Expression-Based Feature Filtering
from qgis.core import QgsFeatureRequest
request = QgsFeatureRequest().setFilterExpression('"population" >= 50000')
features = list(layer.getFeatures(request))
exp = QgsExpression('"name" ILIKE \'%lake%\'')
request = QgsFeatureRequest(exp)
features = list(layer.getFeatures(request))
Pattern 3: Field Calculator (Update Attributes)
from qgis.core import (
edit, QgsExpression, QgsExpressionContext,
QgsExpressionContextUtils
)
exp = QgsExpression('"population" * 2')
context = QgsExpressionContext()
context.appendScopes(
QgsExpressionContextUtils.globalProjectLayerScopes(layer)
)
with edit(layer):
for f in layer.getFeatures():
context.setFeature(f)
f['doubled_pop'] = exp.evaluate(context)
layer.updateFeature(f)
Pattern 4: Expression-Based Labeling
from qgis.core import QgsPalLayerSettings, QgsVectorLayerSimpleLabeling
settings = QgsPalLayerSettings()
settings.fieldName = '"name" || \' (\' || "population" || \')\''
settings.isExpression = True
settings.enabled = True
text_format = settings.format()
text_format.setSize(12)
settings.setFormat(text_format)
labeling = QgsVectorLayerSimpleLabeling(settings)
layer.setLabeling(labeling)
layer.setLabelsEnabled(True)
layer.triggerRepaint()
Pattern 5: Data-Defined Symbology Properties
from qgis.core import QgsProperty
symbol = layer.renderer().symbol()
symbol.setDataDefinedSize(
QgsProperty.fromExpression('"population" / 1000')
)
symbol.setDataDefinedColor(
QgsProperty.fromExpression(
"CASE WHEN \"type\" = 'city' THEN '#ff0000' ELSE '#0000ff' END"
)
)
layer.triggerRepaint()
Pattern 6: Custom Expression Function
from qgis.core import qgsfunction, QgsExpression
@qgsfunction(args='auto', group='Custom', referenced_columns=[])
def population_density(population, area_km2, feature, parent):
"""
Calculate population density per square kilometer.
<p>Usage: population_density("pop_field", "area_field")</p>
"""
if area_km2 and area_km2 > 0:
return population / area_km2
return None
QgsExpression.registerFunction(population_density)
QgsExpression.unregisterFunction('population_density')
Context Scope Hierarchy
Scopes are stacked from generic to specific. When variable names collide, the most specific scope wins:
Global scope → @qgis_version, @user_full_name, custom global vars
Project scope → @project_title, @project_crs, custom project vars
Layer scope → @layer_name, @layer_id, layer fields
Feature scope → Feature attributes, $geometry, $id, $currentfeature
Adding Custom Variables
from qgis.core import QgsExpressionContextUtils
QgsExpressionContextUtils.setGlobalVariable('my_threshold', 42)
QgsExpressionContextUtils.setProjectVariable(
QgsProject.instance(), 'analysis_year', 2024
)
QgsExpressionContextUtils.setLayerVariable(layer, 'source_date', '2024-01-15')
Aggregate Expressions
Aggregates compute values across features within a layer:
ALWAYS use the filter parameter in aggregates to limit the scope. Aggregating an entire large layer without a filter is a performance bottleneck.
Reference Links
Official Sources