* BUG: sjoin modifies original frame (#525)
Original pandas table's index names are no
longer changed by sjoin() call. Includes
some unit-tests. Fixes issue #525
* New code to fix issue that is compatible with pandas 0.16.2 - 0.20.3
* applied suggested changes to test_sjoin.py
As of 0.20.0, pandas .ix is deprecated. This switches all such instances
to the label based .loc indexer. (In searching for usage, I also noticed
the CHANGELOG contained an error in describing the introduction of the
.cx indexer.)
All current overlay tests have indices for which loc/iloc give the same
result. Since ix will be deprecated, add a test to distinguish behavior
between the two.
* TST: Set test gdf crs to silence crs warning
The symmetric difference test compared a frame with a crs to one
without. It didn't seem that this was intentional, so set the second
frame crs to the first so that the tests run without warnings.
Warning below:
geopandas/tests/test_geom_methods.py::TestGeomMethods::()::test_symmetric_difference_series
/home/james/python/geopandas/geopandas/base.py:29: UserWarning: GeoSeries crs mismatch: {'init': 'epsg:4326', 'no_defs': True} and None
other.crs))
geopandas/tests/test_geom_methods.py::TestGeomMethods::()::test_symmetric_difference_operator
/home/james/python/geopandas/geopandas/base.py:29: UserWarning: GeoSeries crs mismatch: {'init': 'epsg:4326', 'no_defs': True} and None
other.crs))
* Add test intended to catch CRS warning
`plot_dataframe` now accepts a `legend_kwds` argument, through which a user can pass keyword arguments accepted by the call to matplotlib's legend function (if `legend=True`).
Use geopy's get_geocoder_for_service to map strings to the proper class
to use, so a geopandas user can use any available geopy Geocoder. Also
provide option to use a Geocoder instance, rather than the string
mapping, which is potentially more transparent.
* Fix default cmap for Matplotlib 2.0.0.
* TST: Fix determination of expected dash style.
In Matplotlib 2.0.0, there was a minimum dash length that was eventually
removed because it was problematic. The expected line style does not
apply this minimum.
* Add autodoc to sphinx configuration
* Add support for numpy docstring to sphinx configuration
* Use autodoc for reference documentation, reading documentation from docstrings
* Improve docstrings of properties and methods listed in Reference documentation
* Pass through kwargs for GeoSeries.buffer (#535)
* Fix choropleth typo in doc (#542)
* Remove warnings from numpydoc when generating docs
* Add autodoc to sphinx configuration
* Add support for numpy docstring to sphinx configuration
* Use autodoc for reference documentation, reading documentation from docstrings
* Improve docstrings of properties and methods listed in Reference documentation
* Remove warnings from numpydoc when generating docs
* Add GeoSeries-subsection for consistency
* Consistently only add parameters for functions passing through args/kwargs
* Clarify how contains and within are inverses of each other
* remove plot signature in reference
* Add basic support for remote resources to `read_file`
Tested with GeoJSON files.
* TST: Use GeoJSON file from GeoPandas repo
* Switch to numpydoc style
* Fix syntax in docs
* Formatting and rename test marker
See #509. Polygon cannot be converted to an array, and therefore
fillna works (starting with shapely 1.6, an empty Point can be
converted to an aray -> broke fillna).
However, we should find a better fix for this in the long term.
* unpin shapely, add test builds with older shapely
* adapt test
* change to BaseGeometry
* COMPAT: Alias isnull -> isna, following pandas
Add same alias for notnull -> notna, and add tests of empty geometries
to ensure that the check is working for the case it is supposed to
catch.
* remove hardcoded defaults
* use uniform color by default
* for categorical data use default cmap tab10 instead of Set1
* forgot to remove hardcoded markersize
* remove no longer used gencolor function
* update MultiPolygons test
* fix some tests + mixed geom case with cmap
* enable more tests
* fix for matplotlib 1.4.3
* skip one for 1.4.3
* fix some collection tests
* small updates to docstring
* undo take squaring of the markersize
* Simplify plot_point_collection
Pass values or colors separately to c / color kwargs
Matplotlib automatically handles them well
* Simplify plot_linestring_collection
Handle values or colors separately.
Colors are then handles by the constructor of
LineCollection itself and don't need to be set
afterwards
* Simplify plot_polygon_collection
Handle values or colors separately.
Colors are then handles by the constructor of
PatchCollection itself and don't need to be set
afterwards
Difference is that color sets both face and edgecolor
* Update (and remove legacy support for alpha) plot_series/plot_frame
get plot_series and plot_frame working with the new plot_*_collection functions
removed the workaround to keep old alpha behaviour for polygons
(no longer apply alpha only on fill and not on edge)
changes in behaviour:
- color has now priority over facecolor, not the other way around
MutliPolygons without specifying values don't work yet
* compute s.geometry.type only once
* address review comments
* try to fix tests for older matplotlib versions
* PEP8 clean-up
* skip more tests with older matplotlib
* merge two testing functions
* Remove no longer used baseline images
* Remove dependency on unittest.TestCase (to be able to use pytest functionality) in test_geodataframe.py
* also remove unittest TestCase usage in other test files (+ PEP8 cleaning)
* remove check for old pandas version (not supported anymore)
* automatically close figures after plotting tests
* further cleanup tests/util.py
* ensure recent version of pytest is used
Edward Betts