Issue — `claude-code-chat-browser` — Refactor `_parse_tool_result` dispatch table (5 pt)

[clean6378-max-it]

Issue — claude-code-chat-browser — Refactor _parse_tool_result dispatch table (5 pt)

Sprint: Wednesday, May 6 — Claude-code Refactor + Exclusion Consolidation (8 pt total; this item = 5 pt, task 1 of 2)
Repo: cppalliance/claude-code-chat-browser
Audit: claude-cursor.md CCC1 (structural half) + claude-code-browser-eval.md §5.1 / §5.4. (Both audit docs are internal; not on GitHub.)
(After opening on GitHub, paste the issue URL here.)

---

Background — what CCC1 actually asks for

CCC1’s Fix column in claude-cursor.md has two prongs:

1. Add direct tests for jsonl_parser.py covering schema variants, malformed entries, and exception paths.
2. Refactor _parse_tool_result’s if/elif chain to a dispatch table so new tool types are a single-site change.

Yesterday closed prong (1). This issue closes prong (2).

Prong	Status	Where
(1) Direct parser tests (14 dispatch arms, malformed inputs, helpers, integration)	Done	Issue #21 · PR #22
(2) Dispatch table refactor (this issue)	Open	This PR

---

Problem

utils/jsonl_parser.py::_parse_tool_result classifies Claude Code toolUseResult blobs through a ~140-line if/elif chain across 14 result families (eval §5.1; arms enumerated in Issue #21 Gap 1 table).

Consequences (eval §5.4):

• High cognitive load for any contributor reading or modifying tool-result classification.
• Schema evolution (new tool, renamed key, new variant of task / web_fetch) requires editing the chain in place — error-prone, merge-conflict-prone.
• The “add a tool type” path is not localized.

The tests landed in PR #22 protect existing behavior; this refactor changes structure only, not semantics.

---

Goal

Replace the pattern-matching chain with an explicit dispatch table / ordered registry of (predicate, handler) (or equivalent), so that:

• Each tool/result family is a small named function.
• New families are added by one registration line + one handler, not by inserting another elif arm.
• Behavior on every input currently exercised by the test suite is unchanged (PR test(jsonl): add jsonl_parser schema coverage and harden null usage #22 stays green).

---

Scope (in)

• Refactor _parse_tool_result only in utils/jsonl_parser.py. Internal helpers may be added in the same module if they aid clarity.
• Preserve the arm ordering / matching semantics documented in Issue #21 Gap 1: bash, file_edit, file_write, glob, grep, file_read, web_search, web_fetch, four task variants, todo_write, user_input, plan, unknown fallback.
• Keep non-dict / null toolUseResult → None behavior intact (Gap 9 in PR test(jsonl): add jsonl_parser schema coverage and harden null usage #22).
• Add a short docstring or README-level note showing how to register a new tool family (one place to add).

Scope (out) — explicit

• CCC2: GitHub Actions pytest workflow.
• CCC3: rewriting other modules’ tests off cross-module _-prefixed imports.
• _process_assistant exception strategy — the “no exception handling” phrase in CCC1’s finding text. PR test(jsonl): add jsonl_parser schema coverage and harden null usage #22 verifies no-crash behavior on bad inputs; adding real try/except + safe error surface is a separate follow-up.
• Any change to what a tool result classifies as (semantics).

---

Acceptance criteria

• _parse_tool_result is implemented via an explicit dispatch table / registry, not a long if/elif ladder.
• All tests added in PR test(jsonl): add jsonl_parser schema coverage and harden null usage #22 (tests/test_jsonl_parser.py::TestParseToolResult + integration) pass without modification.
• Adding a hypothetical new tool family is demonstrably a single primary registration site (verifiable by reading the diff or the new module-level docstring).
• pytest green locally.
• PR description references this issue, CCC1 (structural prong), eval §5.1 / §5.4, and yesterday’s Issue #21 / PR #22.

---

Coverage map (CCC1 closure)

CCC1 piece	Closed by
Tests for 14 dispatch arms, helpers, integration, malformed inputs	PR #22
_parse_tool_result if/elif → dispatch table (eval §5.4)	This PR
_process_assistant exception strategy (CCC1 finding text)	Deferred follow-up
CCC2 CI runs pytest	Deferred follow-up
CCC3 non-parser modules importing _-prefixed parser symbols	Deferred follow-up

After this PR ships, the Fix column of CCC1 is satisfied; the residual three rows above remain as separate tickets.