##// END OF EJS Templates
Improve typing and MIME hook API for inspector (#14342)...
Improve typing and MIME hook API for inspector (#14342) Fixes https://github.com/ipython/ipython/issues/14339 ### Additions Adds `InfoDict` type to improve the typing of `info()` result. Adds missing `"subclasses"` to `info_fields` list (these were added to the field list in https://github.com/ipython/ipython/pull/11486 but we forgot to update `info_fields` variable at the time) - the newly added `InfoDict` type will ensure that this won't happen again. Adds `InspectorHookData` dataclass which is passed to the MIME hooks which now should expect a single argument. Having a single dataclass argument enables us to deprecate individual fields, or add new fields without breaking the existing hooks. The old hooks will still work (if any are out there since this mechanism got just added in the previous point version). ### Deletions A comment over `info_fields` gets deleted: - Contrarily to the comment (which is getting deleted in this PR), `info_fields` were not defining the order of display since at least 2015 (https://github.com/ipython/ipython/pull/7903 - I did not feel the need to go further in the history to find when exactly it happened). - Also contrarily to this comment, current Jupyter messaging spec does not define the contents of `info_fields` (I guess this was lost during IPython/Jupyter split), but the newly added `InfoDict` at least properly annotates their type (if you know where I can find the old IPython messaging spec with the descriptions I can add these as doc comments). Unused `cast_unicode` import gets deleted. If someone imported it from here... well they really should not have. ### Deprecations - mime hooks taking two arguments (`obj, info`)

File last commit:

r28480:87ab1c59
r28661:2084e7f3 merge
Show More
py3compat.py
68 lines | 1.6 KiB | text/x-python | PythonLexer
# coding: utf-8
"""Compatibility tricks for Python 3. Mainly to do with unicode.
This file is deprecated and will be removed in a future version.
"""
import platform
import builtins as builtin_mod
from .encoding import DEFAULT_ENCODING
def decode(s, encoding=None):
encoding = encoding or DEFAULT_ENCODING
return s.decode(encoding, "replace")
def encode(u, encoding=None):
encoding = encoding or DEFAULT_ENCODING
return u.encode(encoding, "replace")
def cast_unicode(s, encoding=None):
if isinstance(s, bytes):
return decode(s, encoding)
return s
def safe_unicode(e):
"""unicode(e) with various fallbacks. Used for exceptions, which may not be
safe to call unicode() on.
"""
try:
return str(e)
except UnicodeError:
pass
try:
return repr(e)
except UnicodeError:
pass
return "Unrecoverably corrupt evalue"
# keep reference to builtin_mod because the kernel overrides that value
# to forward requests to a frontend.
def input(prompt=""):
return builtin_mod.input(prompt)
def execfile(fname, glob, loc=None, compiler=None):
loc = loc if (loc is not None) else glob
with open(fname, "rb") as f:
compiler = compiler or compile
exec(compiler(f.read(), fname, "exec"), glob, loc)
PYPY = platform.python_implementation() == "PyPy"
# Cython still rely on that as a Dec 28 2019
# See https://github.com/cython/cython/pull/3291 and
# https://github.com/ipython/ipython/issues/12068
def no_code(x, encoding=None):
return x
unicode_to_str = cast_bytes_py2 = no_code