Winnow is made for applying sql to geojson in memory. It is useful for working against geojson objects but also has built-in primitives for piping streams.
Build and apply a query to a feature collection object or an array of features
const features = Object
const options = {
where: String // A sql where statement
geometry: Object // GeoJSON or Esri geometry Object
spatialPredicate: String // ST_Within || ST_Contains || ST_Intersects
fields: Array // Set of fields to select from feature properties
aggregates: Object // Describes the set of aggregations to perform on fields
groupBy: Array // Set of fields for grouping statistics
limit: Number // number of results to return
offset: Number // number of return features to offset
order: Array // Set of fields or aggregates by which to order results
outputCrs: Number || String // An EPSG code, an OGC WKT or an ESRI WKT used to convert geometry
inputCrs: Number || String // An EPSG code, an OGC WKT or an ESRI WKT defining the coordinate system of the input data. Defaults to 4326 (WGS84)
toEsri: Boolean // return Esri feature collection
geometryPrecision: Number // number of digits to appear after decimal point for geometry
classification: Object // GeoJSON or geoservices classification Object
}
winnow.query(features, options)
// Returns the set of features that match the queryA sql where statement.
'Trunk_Diameter > 10''Trunk_Diameter > 10 AND Genus like 'Quercus%'(Genus like '%Quercus%' OR Common_Name like '%Live Oak%') AND Street_Type like '%AVE%'
A GeoJSON or Esri Geometry Object
// GeoJSON Polygon
{
type: 'Polygon',
coordinates: [[[-118.163, 34.162], [-118.108, 34.162], [-118.108, 34.173], [-118.163, 34.173], [-118.163, 34.162]]],
}
// Esri Envelope (aka Bounding box)
{
xmin: -13155799.066536672,
ymin: 4047806.77771083,
xmax: -13143569.142011061,
ymax: 4050673.16627152,
spatialReference: {
wkid: 102100
}
}Specifies the relationship between the passed-in geometry and the features in the data
- ST_Within: Features in the data must be completely within the passed-in geometry
- ST_Contains: Features in the data must completely contain the passed-in geometry
- ST_Intersects: Features in the data must intersect the passed-in geometry
Can also specify the esri-style predicates esriSpatialRelWithin, esriSpatialRelContains, esriSpatialRelIntersects
An array that specifies fields should be returned from each features properties or attributes. Will also accept a comma-delimited string.
e.g. [Trunk_Diameter, Common_Name, Genus]
An array that specifies aggregations to apply across all features or properties. Must specify at least a type and a field. Providing a name for the aggregation is optional
Types: [sum, avg, count, max, min, first, last]
e.g:
[
{
type: 'sum',
field: 'Trunk_Diameter',
name: 'Total_Trunk_Diameter'
},
{
type: 'avg',
field: 'Trunk_Diameter'
}
]An array of fields used to group the results. Must be used with aggregates. Note this will not return any geometry
The total number of results to return
The amount to offset returned features (e.g., 10 will skip the first 10 returned features). limit is required to used offset.
An array of fields use to sort the output
Can be an EPSG code, an OGC wkt or an Esri wkt. This parameter controls how the geometry will be projected to another coordinate system.
Can be an EPSG code, an OGC wkt or an Esri wkt.. This parameter defines the coordinate system of input data. If the incoming data is a GeoJSON feature collection with a "named" crs property defined according to specification, winnow will use it's value if inputCrs is undefined. If neither is defined, Winnow assumes the input data has coordinate system WGS84.
If true, the object returned will be an esri feature collection.
Winnow will automatically determine field types from the first feature passed in. If a given attribute is null, Winnow will assume it is a string.
You can also pass in a metadata object that describes the fields in the feature collection. This is recommended if you know your schema ahead of time
e.g.
{
type: 'FeatureCollection',
features: [],
metadata: {
fields: [
{
name: 'SomeDateField',
type: 'Date'
},
{
name: 'SomeDoubleField',
type: 'Double'
},
{
name: 'SomeIntegerField',
type: 'Integer'
},
{
name: 'SomeStringField',
type: 'String'
}
]
}
}A number for geometry precision. Geometry values will be truncated to the supplied decimal place.
An object for classification aggregation. Classification ouputs an array of breaks on features. There are two supported classification types: Class Breaks and Unique Value. Classification supports input from FeatureServer's generateRenderer classificationDef.
Class Breaks is used to classify numeric data based on a number of breaks and a statistical method. Features can also be normalized before being classified.
{
*type: 'classes',
*field: '<field1>',
*method: 'equalInterval' | 'naturalBreaks' | 'quantile' | 'std',
*breakCount: 7,
normType: 'field' | 'log' | 'percent',
normField: '<field2>' // mandatory if normType === 'field'
}
*requirede.g. An example feature collection has a field called field1 ranging in value from 0 - 29.
Input:
{
type: 'classes',
field: 'field1',
method: 'equalInterval',
breakCount: 5,
}Output (array of class intervals):
[ [0-5],
[6-11],
[12-17],
[18-23],
[24-29] ]Unique Value is used to classify data based on a unique field(s). The output is an array of objects for each unique value combination. Each object contains an instance count, and the classifying unqiue field names and values.
{
*type: 'unique',
*fields: ['<field1>', '<field2>', '<field3>'] // up to three fields
}
*requirede.g. An example feature collection has unique fields called employeeID and customerID.
Input:
{
type: 'unique',
fields: ['employeeID', 'customerID']
}Output (array of instance objects):
[
{count: 3, employeeID: 'A', customerID: 'M'},
{count: 1, employeeID: 'A', customerID: 'N'},
{count: 1, employeeID: 'B', customerID: 'M'},
{count: 2, employeeID: 'B', customerID: 'N'},
{count: 2, employeeID: 'B', customerID: 'O'},
{count: 1, employeeID: 'C', customerID: 'O'},
]Returns a function that can be applied directly to a feature collection object, an array of features, or a single feature. Useful when you want to pass a stream of features through a filter.
const options = {
where: String,
geometry: Object,
spatialPredicate: String,
fields: Array,
aggregates: Array
}
const filter = winnow.prepareQuery(options)
filter(geojson)
// returns the set of feature that match the queryExecute sql directly against the query engine.
- Replace any variables with ?
- Table name should always be replaced by ?
- Non-string values always be replaced by ?
const statement = 'Select * from ? where Genus in ?'
const data = geojson
const genus = ['Quercus']
winnow.querySql(statement, [geojson, genus])
// returns all features that match the queryPass in a statement and return a filter than can be applied to a feature collection object, an array of features or a single feature. Variables work in the same way as winnow.sql
const statement = 'Select Trunk_Diameter from ? where Trunk_Diameter > 100'
const filter = winnow.prepareSql(statement)
filter(geojson)
// returns all the features that match the queryWhen option toEsri is true, winnow will check that features have a unique-identifier field. This is field is flagged with the idField option. If not present, winnow will look for a field named OBJECTID and use it as the unique-identifier by default. If no OBJECTID is present, winnow creates one by generating a numeric hash of the feature. By default, winnow uses FarmHash for hashing. Some cloud deployments currently have trouble executing Farmhash (Heroku, 2020-09-30), so you can use native Javascript hashing as an alternative. Simply set the environment variable OBJECTID_FEATURE_HASH=javascript.
Find a bug or want to request a new feature? Please let us know by submitting an issue.
Esri welcomes contributions from anyone and everyone. Please see our guidelines for contributing and the RELEASE.md for a description of our release process.