Step 1: extract RuTrackerParser interface + TDD on real fixtures

[andrikeev]

Step 1 of decoupling parsing from app releases

Goal of the whole effort: be able to change rutracker parsing rules without shipping a new release, by eventually swapping in a config-driven parser. This PR builds the seam that makes that swap a one-line change, and pins current behaviour with tests on real pages.

What this PR does

• New RuTrackerParser interface — a pure String html → DTO / Boolean contract, no network, no IO. It's the single place that holds all markup/text knowledge.
• RuTrackerHtmlParser (ksoup) implements it, keeping the focused Parse*UseCase objects as the internal building blocks it delegates to (ParseTopicPageUseCase, ParseCommentsPageUseCase, ParseTorrentUseCase, ParsePostUseCase, ParseTorrentStatusUseCase).
• Use cases become "fetch + delegate": the page-parsing use cases (search, category, forum, topic, comments, torrent, profile, current-user, favorites) and the page-state checks now call the parser instead of parsing inline. Parsing logic that was inlined in Get*UseCase / Utils moved into the parser.
• RuTrackerApiFactory constructs the parser and injects it — this is the line Step 3 will swap.
• Public NetworkApi contract is unchanged → blast radius stays inside :core:network:rutracker.

Tests (TDD on real HTML)

Characterization tests run against real saved rutracker pages in src/test/resources/fixtures/ — :core:network:rutracker:test is green (9/9). Covered: parseForum, parseCategoryPage, parseTorrent, parseTopicPage, parseCommentsPage (incl. quote/spoiler/image post content), parseProfile, and the topic-exists / forum-exists / is-authorized / is-torrent-topic states.

Known gap — authenticated fixtures needed

The search / favorites / main_authorized captures came back as login stubs (the session cookie wasn't authenticated — no logged-in-username anywhere), so these parsers aren't test-covered yet:

• parseSearchPage, parseFavorites (+ page count), parseCurrentUserId, isAuthorized == true
• the authenticated topic branch (#tor-status-resp status + #tor-size-humn size)

The interface methods exist and are implemented; tests will be added once authenticated fixtures are re-captured. The auth/login/form-token/captcha/action-result callers (WithFormTokenUseCase, LoginUseCase, add/remove favorite, add comment) are also still inline — they'll be routed through the parser in a follow-up increment.

Next

• Step 2: design the versioned parsing config and a config-driven RuTrackerParser (also TDD against these same fixtures).
• Step 3: swap the implementation in RuTrackerApiFactory.

🤖 Generated with Claude Code

https://claude.ai/code/session_01EPxhiZbmURTJvhRzAHcpGu

---

Generated by Claude Code