fix(functemplate): empty ${} inside a call argument truncates the argument list

[chuenchen309]

A path format like %upper{a${}b} evaluates to A${b} instead of A${}B, and %if{1,a${}b,c} collapses to a${}b — the rest of the template leaks out as literal text.

Cause

Parser.parse_symbol handles "no closing brace found" and "identifier is empty" in the same branch:

closer = self.string.find(GROUP_CLOSE, self.pos)
if closer == -1 or closer == self.pos:
    # No closing brace found or identifier is empty.
    self.parts.append(self.string[start_pos : self.pos])

Both emit ${ as literal text, but the two cases need different amounts of input consumed. With an empty identifier the closing brace exists and is left unconsumed, so parse_argument_list reads it as the enclosing call's terminator. The argument list ends early and the remainder becomes raw text:

%foo{a${}b}     -> Call('foo', [Expression(['a', '${'])]), 'b}'
%foo{a${}b,baz} -> Call('foo', [Expression(['a', '${'])]), 'b,baz}'

At the top level } is not a terminator, so a ${} b parses correctly — which is why the existing test_empty_braces_symbol passes and this went unnoticed. Only nesting inside a function-call argument triggers it.

Fix

Split the branch so an empty identifier consumes through the closing brace. ${} stays literal text in both contexts; the closer == -1 path is unchanged.

Verification

Both new tests fail on master (assert 2 == 1 — the leaked 'b,baz}' part) and pass with the fix. Full test/util/test_functemplate.py passes (49/49), as does test/test_library.py (170 passed). ruff check, ruff format --check, mypy, and docstrfmt --check on the changelog are all clean.

---

AI disclosure: drafted with Claude Code (Opus 4.8), including the root-cause trace and tests. I ran the repro and the full test suite locally and reviewed the diff before opening this.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 75.60%. Comparing base (2cf20aa) to head (e71110a).
⚠️ Report is 14 commits behind head on master.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #6846   +/-   ##
=======================================
  Coverage   75.60%   75.60%           
=======================================
  Files         163      163           
  Lines       21299    21302    +3     
  Branches     3358     3359    +1     
=======================================
+ Hits        16103    16106    +3     
  Misses       4406     4406           
  Partials      790      790           

Files with missing lines	Coverage Δ
beets/util/functemplate.py	88.76% <100.00%> (+0.12%)	⬆️

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.