bpo-46784: Add newly exported expat symbols to the namespace.

[yilei]

The libexpat 2.4.1 upgrade from https://bugs.python.org/issue44394 introduced the following new exported symbols:

• testingAccountingGetCountBytesDirect
• testingAccountingGetCountBytesIndirect
• unsignedCharToPrintable
• XML_SetBillionLaughsAttackProtectionActivationThreshold
• XML_SetBillionLaughsAttackProtectionMaximumAmplification

We need to adjust Modules/expat/pyexpatns.h

(The newer libexpat upgrade https://bugs.python.org/issue46400 has no new symbols).

https://bugs.python.org/issue46784

Automerge-Triggered-By: GH:gpshead

[yilei]

Should I create a news entry in the "Core and Builtins" category?

[gpshead]

Should I create a news entry in the "Core and Builtins" category?

I don't remember the categories, extension modules or library would be my first choices as this only impacts the xml modules which should be dynamically loaded in most people's builds. if those don't exist core and builtins is fine. the main thing is having a news entry. :)

[corona10]

@hartwork

I'm not sure about the case of dynamically using an older version.
I want to ask for advice.

[gpshead]

This PR is just continuing what CPython pyexpatns.h already does to avoid symbol conflicts between its own embedded C expat and anyone else linking in their own expat.

It comes up in embedding contexts where CPython is linked into a C/C++ application that also uses upstream expat itself. It'd presumably also be possible when a C/C++ extension module uses expat itself.

Dynamically, without this, code either uses the first version loaded or there's a duplicate symbol error upon loading the extension module using the second version. Which problem, if any, probably depends on exactly how things are linked.

[corona10]

@gpshead Thank you for explain :)

[hartwork]

The libexpat 2.4.1 upgrade from https://bugs.python.org/issue44394 introduced the following new exported symbols:

• testingAccountingGetCountBytesDirect

• testingAccountingGetCountBytesIndirect

• unsignedCharToPrintable

I would like to note that these three^^ are not considered public API and that Expat's own two build systems (CMake and GNU Autotools) tell the linker which symbols are public and hide everything else. Maybe you'll decide that doesn't matter, but I wanted to be sure it's clear these are not public symbols of Expat.

• XML_SetBillionLaughsAttackProtectionActivationThreshold

• XML_SetBillionLaughsAttackProtectionMaximumAmplification

I still hope that someone can make those two^^ accessible (with additional glue code) to the user on pyexpat level in CPython.

[gpshead]

• XML_SetBillionLaughsAttackProtectionActivationThreshold
• XML_SetBillionLaughsAttackProtectionMaximumAmplification

I still hope that someone can make those two^^ accessible

I filed https://bugs.python.org/issue46793 to at least track that.

[gpshead]

regarding the symbol exports, as we're renaming them to not conflict I doubt it matters much - though it is effectively dead code in the shared library. Avoiding export requires modifying how the module gets linked, probably via flags in setup.py and possibly the PCbuild windows stuff. It would be nice, but if we were doing that we could've done renames via the linker or avoided exporting things at all? Though we might have _elementtree depend on getting them from pyexpat? I'm not going to let untangling our embedded expat copy mess hold this up. :)

[hartwork]

it is effectively dead code in the shared library.

@gpshead not unsignedCharToPrintable but the other two, yes, interesting point.

I'm not going to let untangling our embedded expat copy mess hold this up. :)

That sounds healthy 👍

[yilei]

News entry added.

[miss-islington]

@yilei: Status check is done, and it's a success ✅ .

[miss-islington]

Thanks @yilei for the PR 🌮🎉.. I'm working now to backport this PR to: 3.9, 3.10.
🐍🍒⛏🤖

[bedevere-bot]

GH-31417 is a backport of this pull request to the 3.9 branch.

[miss-islington]

Thanks @yilei for the PR 🌮🎉.. I'm working now to backport this PR to: 3.7.
🐍🍒⛏🤖

[miss-islington]

Thanks @yilei for the PR 🌮🎉.. I'm working now to backport this PR to: 3.8.
🐍🍒⛏🤖

[bedevere-bot]

GH-31418 is a backport of this pull request to the 3.7 branch.

[bedevere-bot]

GH-31419 is a backport of this pull request to the 3.8 branch.

[miss-islington]

Thanks @yilei for the PR 🌮🎉.. I'm working now to backport this PR to: 3.10.
🐍🍒⛏🤖

[miss-islington]

Thanks @yilei for the PR 🌮🎉.. I'm working now to backport this PR to: 3.10.
🐍🍒⛏🤖

[bedevere-bot]

GH-31420 is a backport of this pull request to the 3.10 branch.