The block added in #504 only runs when `_out_scalar is _in_scalar`, which
is true for basic_int, numpy_int, and memview_int, but not for basic_str.
Coverage is only measured on the basic_str copy of the shared API file,
so the block showed up as uncovered and failed the 100% coverage gate on
master.
Co-Authored-By: Claude Code <noreply@anthropic.com>
In h3.api.basic_int, numpy_int, and memview_int, _out_scalar is the identity
function, incurring unnecessary Python function call frames and wrapper
overhead on every scalar call.
When _out_scalar is _in_scalar (integer-returning flavours), bind scalar
functions directly to their _cy Cython implementations while preserving
original docstrings and signatures.
Eliminates ~18% (and up to 50%+ on ARM) latency on scalar calls.
Fixes#501.
* bump h3lib to unreleased a9c0a2e (cellsToMultiPolygon)
* initial changes
* move around dependencies
* update CI
* ci separation
* h3.pc.in
* clean up .gitignore
* changelog
* better changelog
* add cellsToMultiPolygon to h3-py in separate release
* sdist exclude and coverage for version test
* warn that changing some justfile rules will affect CI
* try to silence some CI warnings
* one more warning
* bump h3lib to v4.4.0
* bump h3lib to v4.4.1
* tests pass with new version
* add new error codes and some helper functions around error codes
* jupyter-book<2
* is_valid_index
* tests for is_valid_index and get_index_digit
* linting
* good set of tests
* lint
* thank you, coverage tests for tests!
* int api test
* function docstrings
* maybe not
* changelog. don't expose error code helper functions (yet)
* ticks
* don't even mention 'em
* `list` rather than `tuple`, since users are likely to mutate result
* remove justfile
* failing tests
* tests should be passing
* more tests
* swap docs order
* polygon to cells just takes string arguments
* don't need enum
* docstrings
* have tests use h3shape_to_cells_experimental
* clean up tests
* don't repeat tests
* experimental warning
* Being experimental, try `contain` instead of `containment` in params
* try typing with literal
* add polygon_to_cells_experimental.
Contains coverage and lint errors -- just checking that they raise in CI
* ruff fix
* docs and changelog
* add test_polygon_to_cells_experimental for full coverage
* test notes
* simplify coverage and lint actions
* move around the testing for coverage
* try 3.11
* clean up a bit
* separate tests
* run tests twice
* no coverage target for cython tests
* full tests for polygon_to_cells_experimental
* i think that gets the actions how we want
* cell_to_child_pos and child_pos_to_cell
* start exposing cell_to_children_size
* checking resolutions
* cell_to_children_size python function
* adding tests from https://github.com/uber/h3-py/pull/404
* fiddling
* fix test_cell_to_vertex; helps coverage by removing untouched branch
* more tests
* lint
* exhaustive test for a few resolution pairs
* lint
* doc strings
* error messages
* stricter test
* playing with argument order. example code in tests looks nicer
* use parent_res instead of res
* Use `res_*` instead of `*_res`
* fix up changelog
* failing test
* better test
* lint
* dah! lat/lng order *still* trips me up!
* 3d poly test
* names and notes
* fix accidental replace from when we changed names to LatLngPoly
* introduce LatLngPoly
* LatLngPoly in files other than *.py
* drop 3.7 tests
* intro LatLngMultiPoly to .py files
* ...and to non .py files
* h3shape to just shape in .py files
* Class names
* function names
* H3Shape -> Shape
* filename
* Revert "H3Shape -> Shape"
This reverts commit 389725312f.
* Revert "filename"
This reverts commit b263a3e9dc.
* revert
* fix docs
* fix what David found
* rename params from `shape` to `h3shape`
* ...and a few more
* one more
* def cells_to_h3shape(cells, *, tight=True)
* initial stab at documenting polygon interface
* Separate out verbose descriptions
* Nicer flow
* improvements
* avoid using "set"
* move compact functions
* move geo functions
* "grid distance" is better than "H3 distance"
* lint
* units
* clean up contributing
* warning about function name changes
* don't need to specify title
* update copyright
* add a warning about lat/lng order
* try this
* give the geo_interface link
* sphinx 7.3 introduced an issue
* document bug
* better organization of functions
* green is a friendlier color
* don't need this
* sphinx bug is fixed with 7.3.3
* docs building correctly again
* drop beta announcement
* cleaning up docs a bit
* i think this is where it was at...
* bump beta in readme announcement
* update action
* pssssst. it's a secret
* separate upload docs step
* setps taken
* want --warningiserror
* need requirements for notebook
* no need to run linter here
* does this default thing work?
* alas, chatgpt led me astray
* gimme an artifact to check
* bump to actions/upload-artifact@v4.3.1
* slightly better names
* artifacts need names now
* set python version for sdist
* try this one
* all unique names
* try cibw-wheels-${{ matrix.os }}-${{ strategy.job-index }}
* try merge action
* try combining sdist and wheels into single result
* upload should work
* bump to pypa/gh-action-pypi-publish@v1.8.14
* try pypa/cibuildwheel@v2.17.0
* Revert "try pypa/cibuildwheel@v2.17.0"
This reverts commit fa8f7c9ab3.
* SDist doesn't need a strategy matrix
* shorten names for cibw
* bump actions/download-artifact@v4.1.4
* bump pypa/cibuildwheel@v2.17.0
* .
* shouldn't need CIBW_SKIP
* keep intermediate results for comparison for now
* see if that gets us building the correct subset
* formatting
* split up musllinux because it currently takes 20 minutes
* can still include cp36-manylinux_aarch64
* delete-merged
* test dropping CIBW_ARCHS_MACOS
* docs building correctly again
* drop beta announcement
* cleaning up docs a bit
* i think this is where it was at...
* bump beta in readme announcement
* update action
* pssssst. it's a secret
* separate upload docs step
* setps taken
* want --warningiserror
* need requirements for notebook
* no need to run linter here
* does this default thing work?
* alas, chatgpt led me astray
* gimme an artifact to check
* bump to actions/upload-artifact@v4.3.1
* slightly better names
* artifacts need names now
* set python version for sdist
* try this one
* all unique names
* try cibw-wheels-${{ matrix.os }}-${{ strategy.job-index }}
* try merge action
* try combining sdist and wheels into single result
* upload should work
* bump to pypa/gh-action-pypi-publish@v1.8.14
* try pypa/cibuildwheel@v2.17.0
* Revert "try pypa/cibuildwheel@v2.17.0"
This reverts commit fa8f7c9ab3.
* SDist doesn't need a strategy matrix
* shorten names for cibw
* bump actions/download-artifact@v4.1.4
* fix: moved cython_example.pyx and test_cython.py tests to their own folder
* updated test.yml and makefile
* fix: updated coverage-lint.yml to match new path
* fix: all *.c files in tests/ folder and subfolders of tests/ will now be ignored
---------
Co-authored-by: Tim S <racer4308@gmail.com>
* Replace _out_unordered and _out_ordered with _out_collection
* update batch of tests
* lint
* few more tests
* group tests around APIs
* few more tests
* few more tests
* couple few
* final tests
* test_icosahedron_faces
* lint
* remove redundant tests
* consolidate same_set
* same_set
* update makefile
* remove the .so files
* Add note indicating that there is no guaranteed ordering of cells for many functions
* bump to pypa/cibuildwheel@v2.16.5
* starting refactor
* remove stuff
* unindent
* remove self from function signatures
* remove self.
* remove staticmethod
* set up symlinks
* some linting
* test for bad H3Shape
* symlink note
* just put everything in __init__.py
* reorganize conversion functions
* rename _api_funcs.py to _convert.py
* use lazy import of numpy so we don't have to special case h3.api.numpy_int
* DRY
* polygons_to_cells
* Helpers for cells_to_geojson and geojson_to_cells
* ValueError
* Remove gj (#316)
* Remove geojson logic from cell_to_boundary and directed_edge_to_boundary
* fix a warning about return values in tests from pytest
* Remove geo_json logic from cells_to_multi_polygon
* remove some unused tests
* add back in polygons_to_cells
* actions/setup-python@v4.5.0
* Solution copied from https://github.com/actions/setup-python/issues/544
* playing around with polygons
* make a directory
* bah
* start migrating `h3.Polygon` to `H3Poly`
* migrate most
* finish migrating
* H3MultiPoly repr
* shape_to_cells and cells_to_shape
* tests: polygon_to_cells to shape_to_cells
* finish migrating tests
* update util.py and example
* notes
* bah
* fix lint
* fix
* remove X_to_shape
* run tests
* run other CIs too
* don't use f-string
* test update
* change errors per review
* remove unused per review
* fix lint
* test run
* fix exclude logic in tests
* " to ' in yaml
* move conversion logic to geo_to_cells function
* linting
* few more tests
* lint
* clean up
* complete tests
* group geojson/polygon tests into a folder
* move MockGeoInterface into the file that uses it
* move test_input_format to its own file, since it has a lot of helper functions not used elsewhere
* test module description
* rename _polygon.py to _h3shape.py
* make H3Shape an ABC
* shape -> h3shape
* geos to polys
* rename geo files to latlng
* docstrings
* __init__.py
* lint
* notebooks
* more interface tests
* notebooks
* merp
* clean up geo_to_h3shape()
* simplify sydney coords
* simplify far east test
* simplify sf coords
* tuples
* geo interface tests
* more interface tests
* clean up maine lat/lngs
* messed up lat/lng order in tests. this mistake gets made all the time!
* expand geo_to_h3shape test
* plotting with contextily and geoviews
* reprs for H3Shape objects should work better with geopandas
* different representations of the same information. which is best?
* us "loopcode" in H3Poly and H3MultiPoly reprs
* set tight=True as default
* can't use f-strings in python 3.5
* notebooks
* basic plotting
* plotting
* plotting
* tutorial notebook
* cleaning up
* clean up geoviews example
* remove geoviews example
* todos
* reworking tutorial order
* notebook dump
* cleaner
* move tutorial
* Update src/h3/api/_api_template.py
Co-authored-by: Jongbin Jung <jungjongbin86@gmail.com>
* Update src/h3/api/_api_template.py
Co-authored-by: Isaac Brodsky <isaac@isaacbrodsky.com>
* Update src/h3/api/_api_template.py
Co-authored-by: Jongbin Jung <jungjongbin86@gmail.com>
* setting loops
* bug and lint
---------
Co-authored-by: Isaac Brodsky <isaac@isaacbrodsky.com>
Co-authored-by: Jongbin Jung <jungjongbin86@gmail.com>
* Drop cibuildwheel v1 and 3.5 tests
* try cibuildwheel@v2.15.0
* drop down to cibuildwheel@v2.14.1
* bump ubuntu from 20.04 to 22.04
* Revert "bump ubuntu from 20.04 to 22.04"
This reverts commit f94bea6d87.
* bump Python version from 3.8 to 3.11 for coverage/lint and docs
* bump cibuildwheel
* beta announce
* try this instead
* double the warning!
* meh, let's see how github renders it
* try the quote block again
* try br
* try some hrules
* bah
* move around
* merp
* headers
* revert because it adds a horizontal line in github
* ooofffffff
* WIP: Starting work on v4, The Sequel (#238)
* cimport for h3fake2 doesn't seem to work
* back to where we were
* meh, close to what i was hoping for
* h3fake2 in github actions
* comment out current bindings that consume the h3c v3 library
* don't need the coveragerc file at the moment; clear out c files in /tests
* sdist needs h3fake2
* try CIBW_BEFORE_BUILD to get h3fake2
* need h3fake2 for cibuildwheel tests
* update h3 C library
* comment out most of h3lib.pxd
* calling from v4 c lib works! (tests dont' work yet tho)
* tests shall pass
* starting to move over first few functions
* little steps
* util submodule fully moved over; some hackery with h3fake2 errors
* getting even trickier with h3fake2 errors
* few more simple functions
* geo_to_h3 and h3_to_geo
* grid_distance function
* grid_disk
* parents and kids
* compact/uncompact
* getNumCells
* comment out sdist test temporarily
* commenting out `make_sdist` seems to have turned off the other CI jobs
* no, it was that we're pushing to the dev branch
* keep the sdist job in there, but commented out
* ajfriend/more v4 funcs (#241)
* get_pentagon_indexes and get_res0_indexes
* line
* cell area functions
* get_faces (#248)
* initial pass
* thoughts
* the results of thinking
* this commit is associated with the previous commits
* nogil
* i'm a bill
* remove the typical nonsense
* v4: local ij functions to v4 cython (#240)
* move local ij functions to v4 implementation
* revert some changes and add gridPathCells
* swap implementation
* revert gridPathCells
* ring (#251)
* Uncomment cython example, and exclude sdist from wheels test temporarilty (#252)
* uncomment cython test
* exclude make_sdist temporarily from the wheels tests
* v4: polygonToCells (#254)
* v4: geo boundary, set to multi polygon (#255)
* v4: cell boundary
* v4: h3_set_to_multi_polygon
* v4: convert edge from h3fake2 (#253)
* v4: convert edge from h3fake2
* on error not neighbors
* except *
* check for neighbor error
* Fix sdist build (#239)
* make sure `make_sdist` CI job still broken
* reproduce locally
* update h3lib
* still seems broken
* try sdist on mac and ubuntu
* pyest!
* debugging
* ensurepath
* Disable building docs
* just build sdist once
* colon
Co-authored-by: Isaac Brodsky <isaac@unfolded.ai>
* remove h3fake2 transitional library (#257)
* Remove more sdist files (#259)
* remove scripts/examples/apps
* BUILD_FUZZERS OFF
* ENABLE_FORMAT off
* turn off all the things (and sort options)
* mac attack
* see if there's a build time difference
* back to the future
* New error system (#260)
* starting to play around with new error system
* move enum to h3lib.pxd
* this structure is weird :(
* error system gets its own file
* go back to using H3ErrorCodes
* extern the H3ErrorCodes
* remove H3Index in favor of just H3int
* " to '
* import ordering
* something
* exception hierarchy
* formatting
* errors list
* remove H3ResolutionError
* remove H3EdgeError
* remove H3CellError
* remove H3DistanceError
* H3FailedError
* domain error
* H3UnrecognizedException
* most of em done
* remove H3ValueError
* remove some checks
* ideas ideas
* document what the function raises?
* thoughts
* lint
* one half idea for DRYing (kinda)
* potentially make the E_SUCCESS path faster
* typos
* maple syrup
* cimport H3ErrorCodes as ec
* clean up
* whitespace
* clarify cases
* trying new error idiom
* thanks comment
* notes
* applied idiom a few times now
* alternative idiom?
* handling some more potential errors
* idioms and ideas
* clean
* removing check
* interesting allocation example
* lint
* just trying to beat 2.7 for sport right now
* works?
* bow to lint
* using enum values directly!
* now that's a hierarchy
* remove raise_with_msg
* ij funcs
* lint
* remove some checks
* remove err
* comment
* cleaned up `ring`, and i think i fixed a memory leak bug
* think i fixed another memory leak bug
* another memory leak bug
* memory leaks!
* clean
* removing some comments
* meh
* H3Exception shouldn't be appearing as a concrete object
* renaming some base exceptions
* clean up
* don't have a test for this anyway
* documented
* puttering around
* words
* backing up this nonsense with an airtight argument
* air toight
* todos
* more words
* skepticism
* rewording
* editing
* ordering
* tests on the error codes
* more error code tests
* note
* address some comments
* we raise!
* lint
* comments
* Squash #261: Use new H3MemoryManager system (#268)
* Squash #261: Use new H3MemoryManager system
Co-authored-by: AJ Friend <ajfriend@gmail.com>
* add some error checks
* comments
* comments
* clearer logical flow
* arrays are different from memoryviews!
* little cleaner
* Revert "comments"
This reverts commit 80c3f4d382.
* Revert "Revert "comments""
This reverts commit a9d6590b04.
* Revert "comments"
This reverts commit 24904bdcea.
* comments
* arrays are different from memoryviews!
* try _copy_to_mv
* i don't get enough opportunities to consider using the word "penultimate"
* ugh, and there i go using it incorrectly...
* hmmm. was this really all it was?
* That's what it was!
This reverts commit 7e72adb590.
* clarify branches
* remove simple_mv
* moving around
* clean up
* comments
* comments
* comments
Co-authored-by: Isaac Brodsky <isaac@unfolded.ai>
* remove: from h3 import h3 (#270)
* is_valid_cell (#269)
* Remove alias functions (#271)
* remove alias function hex_range
* note test_k_ring and test_hex_range are the same
* remove test_hex_range
* duplicate tests
* remove test_hex_range2
* same test
* remove test_hex_range_pentagon
* remove alias: k_ring_distances
* tests are the same!
* same test
* combine tests
* remove alias: h3_is_res_class_iii
* lint
* Update most functions to v4.0 names (#272)
* grid_distance
* cell_to_parent
* get_resolution
* is_pentagon
* is_res_class_III
* get_pentagons
* get_res0_cells
* get_base_cell_number
* compact and uncompact
* get_faces
* cell_to_boundary
* cell_to_latlng
* latlng_to_cell
* cell_to_children
* int_to_string and string_to_int
* cell_to_center_child
* moving things around
* is_valid_directed_edge
* grid_path_cells
* moving around
* cells_to_multi_polygon
* are_neighbor_cells
* cell_to_local_ij and local_ij_to_cell
* get_icosahedron_faces
* great_circle_distance
* get_num_cells
* note
* remove hex_range_distances and hex_ranges (easily implementable by user)
* grid_ring
* grid_disk
* moving around
* origin_to_directed_edges
* directed_edge_to_boundary
* cells_to_directed_edge
* directed_edge_to_cells
* todo: directedEdgeToCells
* get_directed_edge_origin
* get_directed_edge_destination
* grouping functions
* more grouping
* _binding to _b
* todos
* average_hexagon_area
* average_edge_length
* average_edge_length and edge_length
* average_hexagon_edge_length
* ordering
* more ordering
* Get the makefile working on Linux (#274)
* bump core lib to 4.0 rc5 (#275)
* bump core lib to 4.0 rc5
* more pythonic names for string<->int conversion functions
* cython hex2int to str_to_int to match the python functions
* better argument names for great_circle_distance
* great_circle_distance comments
* revert great_circle_distance tuple unpacking since it breaks python2. drop python 2 before release
* polygon_to_cells (#276)
* start converting tests over
* converted main tests
* linting
* better null island test
* test_compact_and_uncompact_cells
* convert a few more tests
* add h3.Polygon class
* lint
* migrating some tests
* this shift_circular_list thing makes these tests too hard to understand...
* remove shift_circular_list and convert a few more tests
* last of the tests in test_h3.py converted
* last of the tests converted
* note for polygons_to_cells in the future
* Try out a __repr__ for h3.Polygon
* test_polygon_class.py
* use `cells` instead of `hexes` throughout the library and tests
* docstring for Polygon
* docstrings for functions
* use `res` convention for resolution parameters
Co-authored-by: Isaac Brodsky <isaac@unfolded.ai>
Co-authored-by: David Ellis <isv.damocles@gmail.com>
2022-08-22 23:31:19 -07:00
109 changed files with 7904 additions and 5032 deletions