The two methods were ~90% duplicate, differing only in how they obtain
settings/cache/bouquets and in the divergent GeoIP + signal backends
(GeoIPService/file-signals vs GeoIP/Redis) that stay inline. Extract the
identical blocks into private helpers, called by both:
loadUserRow - credential resolution (token/cache file or DB),
resolving $rUserID by reference so the cached
re-verification keeps its exact behaviour
verifyCachedCredentials - cached access-token / username+password re-check
decodeUserFields - JSON line fields -> arrays
resolveOutputFormats - allowed output-format keys
aggregateBouquetIds - bouquet -> channel/series/vod/live/radio id lists
resolveCategoryIds - bouquet -> category ids
Also de-obfuscate the remaining `if (cond) {} else { body }` blocks in both
methods. Net -74 lines despite adding the shared helpers (~240 lines of
duplication removed). Behaviour preserved; the file-vs-Redis signal divergence
is intentionally left as-is (separate decision).
Tests: UserRepositoryTest covers the pure helpers (aggregate/category/decode/
verify), 12 tests total. Validated live: player_api/testxc returns auth=1,
Active, allowed_output_formats=[m3u8,ts,rtmp]; on-demand /live start still works.
Verified: php -l, phpunit (430 tests, 962 assertions), make gates.
- getE2Info: collapse the obfuscated `if (cond) {} else { body }` chain into
straight positive-form ifs; drop the redundant re-init of pair_line_info.
- getUserInfo: same de-obfuscation of the ISP block, and fix the same
"Undefined array key isp_asn" bug already fixed in getStreamingUserInfo — the
ISP-persist step ran even on a GeoIP miss (con_isp_name null), reading the
undefined isp_asn key and writing a null isp_desc/as_number to the line.
- Extract the persist predicate both methods shared into a pure, tested
UserRepository::ispChanged() helper (+ tests/Unit/UserRepositoryTest.php).
Verified: php -l, phpunit (423 tests, 944 assertions), make gates.
Divarion_D