Python package for the threedigrid administration
Threedigrid: The 3Di grid admin framework
The Python package for the threedigrid administration.
- Free software: BSD license
- Documentation: https://threedigrid.readthedocs.io.
- access to the threedicore administration by a single instance of the GridH5Admin object
- query the model data by pre-defined subsets and django style filters
- export model data to gis formats like shapefile, geopackage
- serialize model data as geojson
The standard threedigrid distribution is pretty lightweight, installing as little dependencies as possible. If you want to make use of all capabilities threedigrid has to ofter (e.g. spatial operations and command line tools) install like this:
$ pip install threedigrid[geo,results]
Using the 3digrid_explore shortcut, simply run:
$ 3digrid_explore --grid-file=<path to grid file> --ipy
This will invoke an ipython session with a GridH5Admin instance already loaded.
To get a quick overview of the threedimodels meta data omit the --ipy option or explicitly run:
$ 3digrid_explore --grid-file=<the to grid file> --no-ipy
This will give you output like this:
Overview of model specifics: model slug: v2_bergermeer-v2_bergermeer_bres_maalstop-58-b1f8179f1f3c2333adb08c9e6933fa7b9a8cd163 threedicore version: 0-20180315-3578e9b-1 threedi version: 1.63.dev0 has 1d: True has 2d: True has groundwater: True has levees: True has breaches: True has pumpstations: True
Get a grid admin instance:
from threedigrid.admin.gridadmin import GridH5Admin f = 'gridadmin.h5' ga = GridH5Admin(f)
The grid admin directly holds some model specific attributes like whether the model has a 1D or 2D or groundwater section:
In : ga.has_groundwater Out: False In : ga.has_1d Out: True
There are different types of filters but a filter, generally speaking, acts on field. That means you can filter by value. If you have a line model instance you can filter the data by the kcu field:
or by the lik value:
The filtering is lazy, that is, to retrieve data you have to call data explicitly:
ga.lines.filter(lik__eq=4).data # will return an ordered dict
Remote procedure calls
Currently only the client-side is included. The server-side might be added in a later stage. Note: this is an advanced feature used inside the 3Di stack, probably you don’t need this. Note2: you need Python 3.7 or higher for this to work.
$ pip install threedigrid[rpc]
ga = GridH5ResultAdmin('rpc://REDIS_HOST/SIMULATION_ID', 'rpc://REDIS_HOST/SIMULATION_ID') # Replace REDIS_HOST and SIMULATION_ID with actual values. future_result = ga.nodes.filter(lik__eq=4).data data = await future_result.resolve()
subscription = await future_result.subscribe() async for item in subscription.enumerate(): # do something with item
- Bugfix: crest_level is also inverted by Inpy. Use the raw value in the prepare step to include the correct (non inverted) value.
- Removed numba as dependency, since it did not really give any performance gain on Linux.
- Creating fresh release after upload failed.
- Added extra field ‘discharge_coefficient’ to channels and pipes. These fields default to 1.0.
- Bugfix: don’t use the z-coordinate when making line_geometries during the prepare step
- Invert_level_start_point and end point where inverted by Inpy. After this change the values will be correct again, however present gridadmin files will still have the incorrect value.
- Added Breaches model under lines with specific breach fields mapped from the ‘breaches’ h5py datagroup.
- Added MappedSubsetArrayField allowing to map arrays from other h5py datagroups to a model on another datagroup subset. Breaches uses this to map the array’s under ‘breaches’ to ‘lines’
- Fixed problem with previous release
- Nodgrid generation bugfixes
- Added missing numba requirement in setup.py
- Fixed RPC breaches/pumps bug
- Added fast nod_grid generation on cells
- Add groundwater_cells to exporter for frontend.
- Added content_pk to the export_constants of all structures which have a content_pk.
- An empy array  is returned now instead of None if there is no dataset.
- Line geojson items need to use line_geometry values
- Bugfix for Model._get_subset_idx not instantiating new subsets with their parent’s mixins.
- Use ‘ga.xxx.id.size’ to check if certain submodels (like channels/weirs/manholes) are available for geojson exports.
- Allow older pyrpoj versions. (pre 2.2.0)
- Bugfix for timeseries start_time=0 selection and allow indexes=slice(x,x,x) in combination with sample() method.
- Added try-except surrounding all imports of the package geojson. This package is only available when threedigrid is installed with the extra [geo] extension.
- Add ORIFICES_EXPORT_FIELDS to export_constants.
- Add extra field pixel_width to cells
- Bugfix for GeometryIntersectionFilter: filter was only checking on intersecting bounding boxes
- The sample method needs to skip the last timestamp for SWMR to work correctly. (time dataset can have one item more than datasets with timeseries)
- Bugfix: reprojection with no coordinates (empty array’s)
Added general GeoJsonSerializer which allows you to specify the field names you want to serialize and extract to geojson. The GeoJsonSerializer allows you to specify nested fields.
Added a set of standard export fields for each model.
Automatically pick the correct serializer based on file extention
- .json/.geojson –> to_geojson
- .gpgk –> to_geopackage
- .shp –> to_shape
- Minor bugfix, need to check if h5py filepath is a str or bytes string during initialization
- First release with RPC integration.
- Bumped version of asyncio-rpc
- Fixed incorrect version number
- Added RPC datasource which enables to use the majority of threedigrid in a RPC setting. Uses asyncio-rpc for sending/handling RPC calls.
- RPC datasource allows both one time executing (resolve()) and pub/sub (subscribe()) functionialty.
- Refactored to allow using RPC datasource
- Only use pyproj Transformer if it is present else revert to old transform method
- Added content_pk to the pumps model.
- Bumped package versions
- Reduced reprojection overhead of line_geometries.
- Removed max capacity from Orifice model/serializer.
- Fixed group update for default null values.
- Do not use 0 has a default when converting database objects to numpy arrays in the prepare phase.
- Fixed _field_model_dict being a class variable.
- Added sumax to nodes
- Bug fix in h5py_file method mapping.
- Added sources and sinks (q_sss) to threedigrid.
- Manholes preparation fixed mapping in connection_node_pk.
- Added to_structured_array method for retrieving (filtered) results as Numpy structured array instead of an OrderedDict
- Set fixed type to the fields code, display_name and shape. These fields now have a fixed lenght of 32, 64 and 4 characters respectively.
- Bug fix: dict.values() and dict.keys() in python 3 are causing some unintended behaviour.
- New release due to failing uploads.
- Add aggregation option ‘current’ to volume and intercepted_volume.
- Using a non-tuple sequence for multidimensional indexing is deprecated; use arr[tuple(seq)] instead of arr[seq].
- Properties should be strings so we can use string methods on them.
- Do not prepare levees if there aren’t any.
- Split requirements file to allow for finer grained builds (for instance to generate the documentation).
- Add ‘intercepted_volume’ to NodesAggregateResultsMixin.
- Split requirements file to allow for finer grained builds (for instance to generate the documentation).
- Added BooleanArrayField for boolean values and use it for is_manhole filter. NO_DATA_VALUE is interpreted as False.
- Do not throw exception on cftime ImportError
- Add boolean filter for manholes.
- Patch for converting numpy strings/bytes to float for both python2/3.
- Dropped NetCDF library and replaced opening NetCDF files with h5py
- Bumped h5py to 2.8.0
- Made threedigrid >= Python 3.5 compatible.
- Bug fix for issue #44: use the method get_filtered_field_value() instead of get_field_value() for the count property.
- Properly closes netcdf-file in GridH5ResultAdmin.
- Add export functions for 2D to the export_all() collection.
- Do not use pkg_resources to determine the current version but use zest_releaser to update the version string in threedigrid/init.py
- Use the custom NumpyEncoder to convert specific numpy types to native python types when calling (geo-)json.dumps().
- Introducing subset fields that can be used to query results that are collected only for subsets of the model, like the 2D section.
- Fix lookup_index functionality for composite fields.
- Make model name property optional. That is, ‘unknown’ will be returned if the name cannot be derived.
- Changed Depth/width fields on breach-timeseries to breach_depth and breach_width.
- _get_composite_meta() does not raise an AssertionError anymore if composite field attributes differ. Instead a warning is issued.
- Bug fix: threedicore_result_version must be a property.
Added additional exporters for
Added method get_model_instance_by_field_name to the GridH5ResultAdmin class. This makes it possible to do reverse lookups in situations where you have a field name but do not know which model it belongs to. N.B the field must be unique otherwise an IndexError will be raised.
Added property dt_timestamps to the timeseries_mixin module.
The version number is added to the __init__ file dynamically using the pkg_resources API.
Timestamps of all timeseries fields are shown for aggregation results.
Timestamps in the aggregation results are filtered when retrieving subsets of timeseries.
Introducing the ModelMeta class. Its main purpose at this moment is to compute all possible combinations of composite_fields and aggregation variables.
Fixed return statement of method slice (in class Model) which now takes **new_class_kwargs.
Empty or missing datasets are now displayed as np.array(None) instead of raising an error.
- New release using twine 1.11.
- Added support for composite fields which can be used to fetch data from multiple source variables as a single field. Like this result_3di netcdfs can be queried the same way as gridadmin files.
Changed ResultMixin to dynamically add attributes based on the netcdf variables.
Added basic result proccesing for line/node data.
The filter mask is computed only for array’s affected and before applying it to all array’s
The ‘only’ filter works much faster because the filter mask is only applied on fields that are affected.
The filter mask is cached on the line/node instance after getting the first value. You can thus do something like:
queryset = gridadmin.lines.filter(kcu=2) ids = queryset.id line_coords = queryset.line_coords
and the filter mask will only be computed once.
Add click console scripts 3digrid_explore and 3digrid_export for quick overviews and data exports.
Make ogr/gdal imports optional to avoid breaking parts of the documentation.
Added documentation and setup for sphinx documentation pipeline.
Use linear referencing for embedded channels to keep the original geometry intact when preparing line geometries for visualisation.
Define extra’s to make the standard threedigrid distribution as lightweight as possible.
Convert strings in attrs to numpy.string_ to fix crashes under Windows.
- Remove property has_groundwater from GridH5Admin. Should always be provided by the threedicore itself. Gives a warning for backwards compatibility.
- Get model extent now always returns a bbox (minX, minY, maxX, maxY)
- All imports are absolute.
- Added install info using pip.
- First release with fullrelease.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size threedigrid-1.0.25-py2.py3-none-any.whl (107.6 kB)||File type Wheel||Python version py2.py3||Upload date||Hashes View|
|Filename, size threedigrid-1.0.25.tar.gz (10.1 MB)||File type Source||Python version None||Upload date||Hashes View|
Hashes for threedigrid-1.0.25-py2.py3-none-any.whl