KeyError: '__file__' #912
Comments
|
I too am getting this problem on Windows 10 with python 3.10. Is there something I can do to work around this or is this confirmed to be a bug with mypyc on Windows at the moment? For example I tried importing mypyc inside a python script to try and call the entrypoint that way but no luck. |
|
mypyc appears to be broken on 0.930 (see python/mypy#11700 (comment) for more info). You can work around this by using the |
|
@pranavrajpal Here is the fix I'm testing right now for Debian's package of mypy 0.930; do you think anything else would be needed? diff --git a/setup.py b/setup.py
index 1004227..001df39 100644
--- a/setup.py
+++ b/setup.py
@@ -107,6 +107,8 @@ if USE_MYPYC:
)) + (
# Don't want to grab this accidentally
os.path.join('mypyc', 'lib-rt', 'setup.py'),
+ ) + (
+ os.path.join('mypyc', '__main__.py'),
)
everything = ( |
|
Thanks @pranavrajpal , downgrading to 0.921 got me back up and running for now. I am using pip-compile for my build and this worked nicely:
|
|
No, I think that should be enough to fix this. |
mr-c
added a commit
to mr-c/mypy
that referenced
this issue
Dec 29, 2021
hauntsaninja
pushed a commit
to hauntsaninja/mypy
that referenced
this issue
Dec 29, 2021
This should catch mypyc/mypyc#912
hauntsaninja
added a commit
to python/mypy
that referenced
this issue
Dec 29, 2021
This should catch mypyc/mypyc#912 We never ran mypyc tests against wheels, dating back to https://github.com/python/mypy/tree/783297c9b4ed3b2af9000c3536d6e1be34fed9ea/misc Co-authored-by: hauntsaninja <>
hauntsaninja
added a commit
to python/mypy
that referenced
this issue
Dec 29, 2021
Fixes mypyc/mypyc#912 Co-authored-by: Shantanu <12621235+hauntsaninja@users.noreply.github.com>
|
python/mypy#11865 runs mypyc tests against installed wheels, preventing this kind of regression going forward (confirmed failing on current master) |
JukkaL
pushed a commit
to python/mypy
that referenced
this issue
Jan 5, 2022
Fixes mypyc/mypyc#912 Co-authored-by: Shantanu <12621235+hauntsaninja@users.noreply.github.com>
JukkaL
pushed a commit
to python/mypy
that referenced
this issue
Jan 5, 2022
This should catch mypyc/mypyc#912 We never ran mypyc tests against wheels, dating back to https://github.com/python/mypy/tree/783297c9b4ed3b2af9000c3536d6e1be34fed9ea/misc Co-authored-by: hauntsaninja <>
|
(I think this issue can be closed) |
danparisd
added a commit
to WycliffeAssociates/DOC
that referenced
this issue
Jan 13, 2022
* Improve local var name * Remove unused _content instance var on Resource Previous refactoring made obivated the _content instance var * Make a method a function (static method) and pass in state Prefer function over method when it makes sense. Prefer properties for book_number and book_title because they are derivative values that are only retrieved and don't need to be set. The overarching theme of this change is to lean up __init__ and state in general. * Better organize property methods Group related property methods together * Remove unused _verses_html instance variable * Rename _content instance var to _html_content Better naming * More accurate typing Make mypy and pyre happy with more precise types. This actually caught a potential bug in assembly_strategies.py too in which tn_verses and tq_verses could be accessed before being defined due to conditional logic scope blocks. In practice this never happened, but it was overlooked without stricter typing via pyre and mypy --strict. * Sort imports isort sorted * More accurate typing Continue work to make mypy --strict work * Remove unnecessary classmethod * Replace use of __getattr__ with explicit delegation * Reverse experiment of having *HtmlInitializer classes * Ignore some flake8 lint rules * Prefer short expressions over function call encapsulating them * Use generators to simplify Use of generators in this case achieves the following goals: - More functionalish code: fewer local variables. - Better use of memory by not using so many list appends. * Function renaming Leading underscore has been used to indicate protected variables and methods in an OO context. Refactor of document_generator toward a more functional scheme obviates the need for this, thus remove leading underscore. * Simplify Resource state Remove a couple instance variables whose state can be supplied in another way, i.e., prefer small objects with little state. Prefer derived properties over object state. * Update Makefile targets Add a new smoke test that is useful during development without having to run longer test suites. Update some target preconditions. Fix a .PHONY name. * Improve source comment * Fix docker-compose gunicorn stanza Make docker-compose up work. gunicorn stanza in docker-compose.yml was missing pythonpath cli switch and value. * Add missing types Make mypy --strict happy. * Refactor to functional (in the mathematical sense) paradigm Use more immutable data types like Sequence where warranted. Prefer immutability over mutability where possible. Use generators in a few more places where it cleans up the code. Make the top level exception handling code more robust by moving initialization of success message into else block of try/except to prevent the possibility (detected during this refactor) of a success message getting set when in fact failure had occurred. Add a test wherein at least one resource request is unloadable and another unfindable while other resource requests in the document request are fine. This exercises the case where we still want to render the PDF document with what resources were available while reporting (on the cover of the PDF) those that were not. Handle more USFM resources. Specifically, handle USFM resources whose resource asset files are organized with one directory per chapter and one file per chapter. Compile most modules with mypyc for at least 2x speedup. mypyc compiled modules types are checked at runtime so we also don't need typeguard. Also not using icontract because mypyc doesn't like its use of lambdas since lambdas can't be typed by mypy. Change source layout organization: src -> backend. * Small change to directory layout Rename api -> backend. Dockerfile -> backend/Dockerfile. Reason for this change is to prepare for also having frontend in the same repo each with their own Dockerfiles but composed with docker-compose into one container. * Update a few source code comments * Add frontend submodule and associated config Docker container for backend and frontend composed into one container using docker-compose. * Return lang codes and names sorted alphabetically by name Also add the language code to the displayed value at least for now because then it makes it easier to discover lang codes for the purpose of creating tests that are first discovered manually. * Use exception handler approach for top level exception scope Follow fastapi's example for this: https://fastapi.tiangolo.com/tutorial/handling-errors/ * mypyc Makefile target doesn't rely on mypy target * Pull latest commits from frontend submodule * Pull latest submodule frontend commits * Pull latest commits from frontend submodule * Update docker-compose to make back and front ends communicate You can now do: make build make frontend-server to get a locally accessible frontend http://localhost:8000 (which talks to the backend). * Make optional RUN_TESTS functionality work in docker build * Better docs and comment out docker volume for PDFs Since we have a frontend that serves up results, we no longer need to map PDFs from the container to the host system to be able to view the generated PDFs. * Update docs * Pull latest commits on frontend submodule * Handle case where English tn and tn-wa are requested translations.json offers English translation notes in two locations: git repo and downloadable zip. If both were requested through the UI then filter out the tn format and use the repo only. We don't want them both in the same document as they are actually the same material. Update a few source comments and docstrings. Make isort happy by removing imports that are no longer used. * Add clean-mypyc-artifacts Makefile target mypyc creates C binaries. When making changes and verifying them in a quick edit-test-run development workflow having to recompile using mypyc in between test runs slows down the flow of work. This Makefile target can be used to quickly remove the generated binaries so that we are back to running interpreted Python temporarily. Then when we have our tests passing and changes we can go back to compiling source with mypyc via make mypyc. * Clean up Jinja2 templates Remove extraneous html elements, comment out unused css classes, remove unused template.html, add css classes that will allow two column layout. * Update log message Make log message less misleading * Upgraded Python packages * Remove some vars from .env file I don't want a couple key directories /working/temp and /working/output to be easily changed so I have removed them from being initialized in .env. They are now just initialized in config.py. Also removed some unused and commented out env vars from .env. * Fix a Makefile target dependency relationship * Update docstrings and a few source comments Docstrings needed updating to match current functionality. * Better organize and simplify some code Move some methods out of Settings class in config.py into resource_lookup and document_generator modules where it belongs. * Two column layout for scripture verses and their helps Puts USFM resources on left 50% of layout and their associated helps on the right 50% of the layout. The rest of the layout is not subdivided. * Update design.org design doc Remove some obselete sections, update a few others. More to do still. * Add pygout package Used for generating project stats in design doc. * Update Python packages pip-compile --upgrade automatic upgrades to packages. * Update Docker image to Python 3.10.1 * Update some docstrings and source comments * Avoid use of magic string * Go with Python 3.10 union | use * Workaround for mypyc regression upstream I reported this here: mypyc/mypyc#912 * Add generate-class-diagrams Makefile target Uses pyreverse which comes with pylint package to generate UML class diagrams from source code. * Workaround for mypyc regression upstream part 2 Fix to make pip-compile generate the requirements-prod.txt with mypy (and thus mypyc) pinned at an older (working) version. * Copy generated PDFs to own directory... ...for mapping into host file system. * Update a few source comments * Tiny bit of automatic code formatting via black * Only start translation word definition section if words present Fix issue where a language that did not provide translation words would at least print a header for its section despite there not being any content. * Rename a few attributes Done for consistency and simplicity * Show the max number of chapters and verses available Prior to this commit you could get an undesirable state of affairs in book then language assembly wherein if the first usfm language/book resource was missing a verse, then other usfm language/book resources in that same document request would skip that verse because he chapter and verse loops were arbitrarily based off the first usfm resource. Now the usfm resource with the most chapters for a particular book and subsequently the usfm resource with the most verses within each respective chapter are used as the outer loops pumps for chapters and verses repsectively. This way the max available chapters and verses are displayed in the resulting document. * Use final type decorator to indicate design Mypy then enforces that classes so decorated are not subclass-able (by design). * Add a log message for a possible exception point Easier diagnosis using logs. * Fix footnotes regression Footnotes were not being shown with their associated verse and thus the footnotes section at the end of the chapter showed footnotes but they were not clickable. This is fixed now. * Move custom exceptions to own module Better organization and now we can use the module in other modules where its exceptions are needed. * Better naming Follow Pythonic convention and conventions of good style to name functions after what they return rather than names which imply how they do that. Also avoiding nebulous names like initialize...blah. * Tighter control of acceptable resource types requested Make sure we don't accept resource types that we do not support yet or which are not valid. * Stricter adherence to functional style Don't use "globals" even if they are just a module reference when it can be passed as a defaulted function argument instead. Pass in what we need as parameters to keep things more pure. I say more because we aren't yet to a place with the code that doesn't have side-effects, that is coming though. * Skip tests that test functionality not yet supported Skip tests that deal with TA resource type as we do not support it yet. * Minor tweak to docs * Update many source code comments * Change magic number to constant * Greater referential transparency Use more dependency injection at the function parameter level via defaulted arguments so as to increase the referential transparency of the code and make it more easily testable. * Improve a log message slightly * Remove unused imports * Prune unused exception variable * Make e2e-tests suite use all language codes when run outside Docker So that more more random language combinations will be discovered by randomized test fixtures. Also use Sequence instead of list type in order to communicate immutable use of collection. * Skip two tests that fail in Docker Investigate further later. * Tell jinja2 to autoescape HTML To help prevent potential XSS attacks. * Fix mypy type error and preserve immutability Instead of extending a collection, unpack the two collections into one new collection. * Update submodule pointer ...to not point to old submodule (https://github.com/linearcombination/InterleavedResourcesGeneratorUI). * New submodule pointer ...pointing to https://github.com/WycliffeAssociates/DOC-UI Co-authored-by: linearcombination <4829djaskdfj@gmail.com>
tushar-deepsource
pushed a commit
to DeepSourceCorp/mypy
that referenced
this issue
Jan 20, 2022
This should catch mypyc/mypyc#912 We never ran mypyc tests against wheels, dating back to https://github.com/python/mypy/tree/783297c9b4ed3b2af9000c3536d6e1be34fed9ea/misc Co-authored-by: hauntsaninja <>
tushar-deepsource
pushed a commit
to DeepSourceCorp/mypy
that referenced
this issue
Jan 20, 2022
Fixes mypyc/mypyc#912 Co-authored-by: Shantanu <12621235+hauntsaninja@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I successfully use mypyc in my project and it has performed well until just a couple days ago. I now get the following error:
Line 18 above, i.e., the line that is failing, is just
which I wouldn't expect to fail. I am in my venv virtualenv when I execute mypyc using the same command as has always worked before.
I thought perhaps a regression was introduced in mypyc so I used git to go back in time to see if that line had changed in any recent version of mypy, but it hadn't.
I also tried downgrading mypy to an older version that worked before but that version also failed with the same error. To be sure it wasn't being experienced by others I also checked the issues at the mypy repo on github and did a search for
__file__to see if that part of the error message showed up and it didn't. Perhaps it is some weird issue with my environment?I experience the issue with venv virtualenvs created with Python 3.10, 3.10.1 but also 3.9.9 too. It worked fine on Python 3.10 before. Any ideas on what to investigate next?
The text was updated successfully, but these errors were encountered: