[PATCH] D142989: [extract_symbols.py] Better handling of templates

[John Brawn via Phabricator via llvm-commits]

john.brawn marked an inline comment as done.
john.brawn added a subscriber: glandium.
john.brawn added a comment.

In D142989#4095488 <https://reviews.llvm.org/D142989#4095488>, @glandium wrote:

> Could you add before/after figures of the numbers of symbols in, say, clang.symbols in the summary?

Will do.



================
Comment at: llvm/utils/extract_symbols.py:183
     # <throw-spec> ::= exceptions are not allowed
     elif re.search('(llvm|clang)@@[A-Z][A-Z0-9_]*[A-JQ].+(X|.+@|.*Z)$', symbol):
         return symbol
----------------
glandium wrote:
> would it be worth changing this to use `parse_microsoft_mangling`?
The important thing here is detecting that we have a function type mangling, which parse_microsoft_mangling doesn't do.


================
Comment at: llvm/utils/extract_symbols.py:334
+        # Check for a simple name
+        match = re.match('(\w+)@(.+)', arg)
+        if match:
----------------
glandium wrote:
> you probably want to pass the `re.ASCII` flag to these match invocations.
Names in symbols aren't necessarily ASCII, e.g.
```
int あ() { return 0; }
```
results in the symbol `?あ@@YAHXZ`. Though we probably don't use non-ASCII identifiers in LLVM, so it doesn't really matter.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D142989/new/

https://reviews.llvm.org/D142989