Index TypeScript abstract classes, not just concrete ones

[aala-conga]

Fixes #1031.

Tree-sitter gives an abstract class its own node type, abstract_class_declaration, and
_CLASS_TYPES listed only class_declaration for typescript/tsx. An abstract base therefore
contributed no Class node — and the concrete methods hosted on it lost their parent and were not
emitted either, so shared behaviour on a base class was invisible to the graph while every subclass
was indexed.

The fix adds one string to each of the two lists. abstract_class_declaration is TypeScript-specific;
Java reuses class_declaration and is not affected.

Measured

On an Angular codebase (1696 Class nodes before):

	before	after
abstract classes indexed	0 / 21 (0 %)	21 / 21 (100 %)
concrete classes indexed	634 / 641 (99 %)	unchanged
Class nodes total	1696	1717
the base method's node	<file>::handle	<file>::Base.handle

Java on the same run: 4 abstract classes, 4 indexed, before and after.

Tests

TestTypeScriptAbstractClasses in tests/test_parser.py, a sibling of
TestTypeScriptTypeDeclarations (the #737 class, whose shape it copies). Four cases: the
declaration itself, the methods that disappear with it, the tsx list (the fix touches two lists,
and a .ts-only test would have covered only one), and reachability by qualified name in a built
graph — the form the symptom actually took.

All four fail on staging and pass with the fix:

$ pytest tests/test_parser.py::TestTypeScriptAbstractClasses -q   # staging
4 failed in 0.74s
$ pytest tests/test_parser.py::TestTypeScriptAbstractClasses -q   # with the fix
4 passed in 0.65s

Full suite, fresh clone of staging at f0e4eb7, Python 3.13, pip install -e ".[dev]":

run	result
staging, untouched — baseline	3963 passed, 773 skipped, 2 xfailed, 2 xpassed (349 s)
+ the fix	3963 passed, 773 skipped, 2 xfailed, 2 xpassed (321 s)
+ the fix + these tests	3967 passed, 773 skipped, 2 xfailed, 2 xpassed (333 s)

Nothing in the existing suite depended on abstract classes being absent.

Scope

This does not close #1031's downstream effects on its own. callers_of on a method inherited from an
abstract base is still 0, because this.method() remains bound to the enclosing class rather than
walked up the now-complete INHERITS chain — that is the subject of the follow-up discussion on
#984. What this change does fix is the inheritance chain breaking at an abstract intermediate, and
the node naming of base methods.

I did not check Kotlin, PHP or C#, which also have abstract classes and which I have no codebase to
measure on.

🤖 Generated with Claude Code

[github-actions]

code-review-graph review

Overall risk: 0.25 (LOW) — 7 changed function(s)/class(es), 17 affected flow(s), 1 test gap(s)

Risk-scored changes

Risk	Level	Symbol	Location	Tested
0.25	low	tests/test_parser.py::TestTypeScriptAbstractClasses._project	tests/test_parser.py:2691	yes
0.05	low	tests/test_parser.py::TestTypeScriptAbstractClasses	tests/test_parser.py:2659	no
0.05	low	tests/test_parser.py::TestTypeScriptAbstractClasses.setup_method	tests/test_parser.py:2670	yes
0.05	low	tests/test_parser.py::TestTypeScriptAbstractClasses.test_abstract_class_becomes_a_node	tests/test_parser.py:2698	(test)
0.05	low	tests/test_parser.py::TestTypeScriptAbstractClasses.test_concrete_method_of_an_abstract_class_is_contained_...	tests/test_parser.py:2707	(test)
0.05	low	tests/test_parser.py::TestTypeScriptAbstractClasses.test_abstract_class_in_tsx_becomes_a_node	tests/test_parser.py:2726	(test)
0.05	low	tests/test_parser.py::TestTypeScriptAbstractClasses.test_abstract_base_is_reachable_in_a_built_graph	tests/test_parser.py:2740	(test)

Affected execution flows

• visit_FunctionDef — criticality 0.41, 55 node(s) across 1 file(s)
• visit_AsyncFunctionDef — criticality 0.41, 55 node(s) across 1 file(s)
• visit_ClassDef — criticality 0.41, 54 node(s) across 1 file(s)
• visit_If — criticality 0.36, 3 node(s) across 1 file(s)
• repo_relative_path — criticality 0.36, 9 node(s) across 1 file(s)
• ...and 12 more affected flow(s)

Test gaps

• tests/test_parser.py::TestTypeScriptAbstractClasses (tests/test_parser.py:2659)

Token savings: this graph-backed report used ~202,054 fewer tokens (~91%) than reading every changed file in full (estimated, chars/4 approximation).

---

Powered by code-review-graph — local-first analysis; no code leaves the CI runner.