jRdZddlmZddlZddlmcmZddl Z ddl m Z ddl Z e e jjjjZeee j$vr!e j$j'eee j$j)deeddlmZmZmZmZmZmZmZddlmZmZm Z m!Z!m"Z"ddl#m$Z$m%Z%m&Z&m'Z'm(Z(dZ)d Z*dd Z+e jXdd Z-d Z.d Z/ddZ0e jbjedgdddZ3e jbjedgd ddZ4ddZ5dZ6dZ7dZ8dZ9y)utests/regression/test_auto_finalize_chain_default_2529.py 회귀 테스트 — task-2529 auto finalize chain default (시스템 결함 fix). 회장 §명시 (2026-05-10): 봇이 본질 작업과 자체 검증을 완료하면, 회장이 별도로 말하지 않아도 자동으로 commit → push → PR → CI/Gemini → bot identity merge → smoke → reconcile까지 진입한다. 명시적 opt-out이 없는 모든 code task는 자동 finalize chain을 기본값으로 가진다. 회장 §명시 5 회귀 (정확히 5건): 1. task-2524+1 사례 박제 — 자체 검증 PASS 후 PR 미생성 → lifecycle stuck SELF_VERIFIED_BUT_NOT_FINALIZED 자동 분류 1-b. task-2528 사례 박제 (task-2529+1 보강, 2026-05-10) — 자체 검증 PASS 후 commit/push/PR 미진입 → lifecycle stuck CODE_DONE_BUT_NO_COMMIT / SELF_VERIFIED_BUT_NOT_FINALIZED 자동 분류 2. task md에 12단계 누락 → wrapper가 footer 자동 삽입 3. read_only task → finalize 생략 4. report_only task → finalize 생략 5. code task → finalize 자동 진입 (footer 삽입 + lifecycle 정상) 본 사건 fixture (회장 §1 / §6 사례): - task-2524+1 (dev5 사라스와티, 2026-05-10) — 회귀/구현 PASS 후 PR 미생성 - task-2528 (dev1 헤르메스, 2026-05-10) — 자체 검증 PASS 후 커밋/푸시 미진입 ("커밋/푸시는 별도 회장 명령 시 진행하겠습니다") 테스트 카운트 메모 (task-2529+1 정정, 2026-05-10): - def test_ 함수: 10개 (task-2528 fixture 추가 후) - parametrize 전개 후 실제 pytest collect 카운트: 15개 - 계산: def 10 + test_regression_3(3 cases - 1) + test_regression_4(4 cases - 1) = 15 - 봇 보고 기준 task-2529 시점 14는 def 9 + (3-1) + (4-1) = 14 — parametrize collapse가 아니라 실제 실행 카운트로 일관됐음. 봇 보고 9건은 def test_ 함수 수, 14건은 collect 카운트. ) annotationsN)Path)AUTO_FINALIZE_FOOTER_MARKERAUTO_FINALIZE_OPT_OUT_TOKENSDispatchStatusauto_inject_finalize_footeris_finalize_opt_outsafe_cron_dispatch"should_auto_inject_finalize_footer) TASK_KIND_BOTTASK_KIND_FOLLOWUP_ROTASK_KIND_HUMAN_RESPONSETASK_KIND_INDEPENDENTTASK_KIND_MERGE)LifecycleEvidence StuckReasondetect_stuck_casesdetermine_stateLifecycleState 6937032012f3e244a7f4f0d036c tdiddddddddddd dd dd dd dd ddddddddddddddddddddddddddddddddd}|j|tdi|S)uWLifecycleEvidence 기본값 (FINALIZED 직전 상태) → 회귀에서 필드 override.task_idz task-9999 pr_numberNpr_state merge_commitmerged_into_mainF ci_status smoke_status timer_statustimer_end_timehas_donehas_done_ackedhas_merge_done has_qc_result has_followuphas_escalate_markerescalate_marker_age_minutestelegram_reply_truncatedbot_session_statusworktree_existsbranch_pushed_to_remoteworktree_mtime_seconds_agohas_pushed_commitsreport_artifact_present process_alivepr_open_age_secondsreport_artifact_age_seconds)dictupdater) overridesbases M/home/jay/workspace/tests/regression/test_auto_finalize_chain_default_2529.py_make_evidencer9Rs(                     "  %)! ""'# $ % &' (!&) *$(+ ,!- .!&/ 01 2!3 4%)5 D8 KK   $t $$c |dz S)u<production audit-jsonl 오염 방지 — tmp_path에 격리.zcron-targeting-audit.jsonlr3)tmp_paths r8 audit_pathr=ts 2 22r:c  tddddddddddd }t|}|Dchc]}|j}}tj}||v}|st j d|fd||fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd d z}t jd |Dcgc]}|jjc}d zd|iz}tt j|dx}}tj}||v}|st j d|fd||fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd d z}dd|iz}tt j|dx}}t!|\}} t"j$}||k(}|st j d|fd||fdtjvst j|rt j|nddtjvst jt"rt jt"ndt j|d z}dd|iz}tt j|dx}}ycc}wcc}w)u본 사건 fixture: dev5 사라스와티가 회귀/구현 PASS 후 PR 미생성한 사례. 재발 시 lifecycle_reconciliation_manager가 stuck으로 자동 분류해야 한다. z task-2524+1T@NF) rr/r2rrrr+r-r.r,r inzG%(py2)s {%(py2)s = %(py0)s.SELF_VERIFIED_BUT_NOT_FINALIZED } in %(py4)srreasonspy0py2py4u;task-2524+1 본 사건이 stuck 분류되지 않음. cases= >assert %(py6)spy6z?%(py2)s {%(py2)s = %(py0)s.CODE_DONE_BUT_NO_COMMIT } in %(py4)sassert %(py6)s==z=%(py0)s == %(py4)s {%(py4)s = %(py2)s.STUCK_NEEDS_RECONCILE }stater)r9rreasonrSELF_VERIFIED_BUT_NOT_FINALIZED @pytest_ar_call_reprcompare @py_builtinslocals_should_repr_global_name _saferepr_format_assertmsgvalueAssertionError_format_explanationCODE_DONE_BUT_NO_COMMITrrSTUCK_NEEDS_RECONCILE) evidencecasescrC @py_assert1 @py_assert3 @py_format5 @py_format7rO_s r8Ktest_regression_1_task_2524_plus_1_self_verified_but_no_pr_finalize_missingrfsm  $$)#( %H" x (E!&'Aqxx'G'  6 6 6' A 6'   7;B;B F_dFeZ[qxx~~FeEfg  . .9 .' 9999 .'999999;999;999 .999999'999'9999999x(HE1"88858 88885888888858885888888N888N88888888888(Gfs M;#N c 6 tddddddddddd }t|}|Dchc]}|j}}g}tj}||v}|}|stj }||v} | }|s%t jd|fd||fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd d z} d d | iz} |j| |st jd fd|fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd dz} dd| iz} |j| t j|diz}t jd|Dcgc]}|jjc}dzd|iz}t!t j"|dx}x}x}x}x}} tj }||v}|st jd|fd||fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd dz}t jd|Dcgc]}|jc}dzd|iz} t!t j"| dx}}tj}||v}|st jd|fd||fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd dz}t jd|Dcgc]}|jc}dzd|iz} t!t j"| dx}}t%|\}}t&j(}||k(}|st jd|fd||fdtjvst j|rt j|ndd tjvst jt&rt jt&nd t j|dz}d!d|iz} t!t j"| dx}}ycc}wcc}wcc}wcc}w)"u^task-2528 사례 박제 — 자체 검증 PASS 후 commit/push/PR 미진입. 회장 §6 (2026-05-10) 명시 충족: task-2524 / task-2528 사례 모두 회귀 fixture에 박제. 회장 §명시 (2026-05-10): "lifecycle stuck enum 또는 audit log만으로 task-2528 fixture 충족을 대체하지 않는다" → 명시 def test_ 함수로 박제. Source: - dev1 헤르메스, 2026-05-10 — 자체 검증 보고서(memory/reports/task-2528.md) 작성 후 commit/push 단계 미진입. - 봇 명시 발언: "커밋/푸시는 별도 회장 명령 시 진행하겠습니다" (auto-finalize-chain-missing-260510.jsonl). - 결과: AUTO_FINALIZE_CHAIN_MISSING audit log 생성, 회장 §결정으로 task-2528+1 발행. 재발 시 lifecycle_reconciliation_manager가 자동으로 stuck 분류해야 한다. z task-2528Tr?FN) rr/r2r.r,rrrr+r-r r@)zG%(py4)s {%(py4)s = %(py2)s.SELF_VERIFIED_BUT_NOT_FINALIZED } in %(py6)srrC)rFrGrIz%(py8)spy8)zC%(py12)s {%(py12)s = %(py10)s.CODE_DONE_BUT_NO_COMMIT } in %(py14)s)py10py12py14z%(py16)spy16u9task-2528 본 사건이 stuck 분류되지 않음. cases=z >assert %(py19)spy19rJrDu[CODE_DONE_BUT_NO_COMMIT 미분류 — task-2528은 commit 단계 미진입 사례. reasons=rHrIrBukSELF_VERIFIED_BUT_NOT_FINALIZED 미분류 — task-2528은 자체 검증 PASS 후 finalize 미완. reasons=rLrNrOrrK)r9rrPrrQr\rRrSrTrUrVrWappend_format_booloprXrYrZr[rrr])r^r_r`rCrarb @py_assert5 @py_assert0 @py_assert11 @py_assert13rd @py_format9 @py_format15 @py_format17 @py_format18 @py_format20rcrrOres r8Qtest_regression_1b_task_2528_dev1_hermes_self_verified_but_finalize_chain_missingr{s  $$)! %#(#H( x (E!&'Aqxx'G'333w>  . . .' 9 3w    4  8? 8?    .'    /  3: 3:     */0Q!((..01 3    . . .' 9 .'   /3:3: %,-AGG-. 0  6 6 6' A 6'   7;B;B %,-AGG-. 0 x(HE1"88858 88885888888858885888888N888N888888888885(1..sXX X 5X c  d}t|v}|stjd|fdt|fdtjvstj trtj tnddtjvstj |rtj |nddz}dd|iz}ttj|d }t|}| }|sd d tjvstj trtj tnd dtjvstj |rtj |ndtj |d z}ttj|d x}}t|d tttd |}|j}tj }||k(} | stjd| fd||fdtjvstj |rtj |ndtj |dtjvstj trtj tndtj |dz} tj"d|j$dzd| iz} ttj| d x}x} }t'|j(} d} | | v}|stjd|fd| | ftj | dtjvstj | rtj | nddz}dd|iz}ttj|d x} }| j+d}| |dz}t|v}|stjd|fdt|fdtjvstj trtj tndd tjvstj |rtj |nd dz}tj"d!d"zd|iz}ttj|d }d#D]}||v}|stjd|fd||fd$tjvstj |rtj |nd$d tjvstj |rtj |nd dz}tj"d%|d&d"zd|iz}ttj|d }y )'utask md (cron prompt)에 finalize chain 명시가 빠져 있으면 wrapper가 footer를 자동 삽입. safe_cron_dispatch가 ALLOWED 상태로 진입하고 audit/preview에 footer가 포함된다. uM[task-9001] dev2 indra: utils/feature_x.py에 새 helper 추가 + 회귀 3건not inz%(py0)s not in %(py2)sr bare_promptrErFassert %(py4)srGNz0assert not %(py3)s {%(py3)s = %(py0)s(%(py1)s) }r rEpy1py32mpromptschedulechattarget_bot_key task_kind session_idr=rLzK%(py2)s {%(py2)s = %(py0)s.status } == %(py6)s {%(py6)s = %(py4)s.ALLOWED }resultrrErFrGrIuCfooter 자동 삽입은 preflight 통과해야 함. blocked_reason=z >assert %(py8)srh--cronr@)z%(py1)s in %(py3)sargv)rrassert %(py5)spy5rmz%(py0)s in %(py2)sinjected_promptu3wrapper가 footer를 prompt 끝에 삽입해야 함 >assert %(py4)s)commitpushPRmergesmoke reconcilekeywordu footer에 u 누락)rrRrSrTrUrVrWrZr[r r CHAIR_CHATDEV2_INDRA_KEYr statusrALLOWEDrXblocked_reasonlist command_argvindex)r=rra @py_format3rc @py_assert2 @py_assert4rrqrbrdrurrr @py_format4 @py_format6cron_idxrrs r8Atest_regression_2_task_md_missing_12_steps_wrapper_injects_footerrs bK 'k 9999 &k999999 &999 &999999k999k9999999";/// // //////"///"//////;///;//////////  %F ==N22=2 2=2   ++3 NfNcNcMde ## $D 8t 8t8ttzz(#H8a<(O &/ 9 &/ ' '+:+: >KK/)JJJw/JJJJJJwJJJwJJJJJJ/JJJ/JJJJZ{'+JJJJJJJKr:opt_out_phrase)zread_only: truezread_only:truezREAD_ONLY: TRUEc  d|d}t|}|stjd|dzdtjvstj trtj tnddtjvstj |rtj |ndtj |dz}ttj|d}tt|}| }|s d d tjvstj trtj tnd d tjvstj trtj tnd dtjvstj |rtj |ndtj |d z}ttj|dx}}tt|}||k(} | stjd | fd||fdtjvstj |rtj |nddtjvstj |rtj |nddz} tjddzd| iz} ttj| d} t|v} | stjd| fdt|fdtjvstj trtj tnddtjvstj |rtj |nddz} dd| iz} ttj| d} t|dtt td|} | j"} t$j&}| |k(}|stjd |fd| |fdtjvstj | rtj | ndtj | dtjvstj t$rtj t$ndtj |dz} dd| iz}ttj|dx} x}}t)| j*j-d }| j*|d!z}t|v} | stjd| fd"t|fdtjvstj trtj tndtj |d#z}d$d%|iz}ttj|dx} }y)&uXread_only opt-out이 명시된 task는 자동 finalize footer가 삽입되지 않는다.uask-9002] audit only — u status scan 후 보고.opt-out 토큰 인식 실패: . >assert %(py3)s {%(py3)s = %(py0)s(%(py1)s) }r rrNz9assert not %(py4)s {%(py4)s = %(py0)s(%(py1)s, %(py2)s) }r r )rErrFrGrLz%(py0)s == %(py2)soutru8read_only task에 footer가 삽입됨 — opt-out 위반rrGr}rrrrrrrrrassert %(py8)srhrrmz%(py0)s not in %(py3)srErrr)r rRrXrTrUrVrWrZr[r r rrSrr rrrrrrrr)rr=rrrrbrqrrrarrcrrdrurs r8/test_regression_3_read_only_task_skips_finalizersR+>*::U VF v &[ &[[*HHZ([[[[[[[ [[[ [[[[[[v[[[v[[[ &[[[[[[1-HHH HH HHHHHH1HHH1HHHHHH-HHH-HHHHHHHHHHHHHHHHHHH &mV 2BBW XF v &[ &[[*HHZ([[[[[[[ [[[ [[[[[[v[[[v[[[ &[[[[[[ %&;V DC &=3&33&& &c 1111 &c111111 &111 &111111c111c1111111  %'F ==2N222=2 2222=222222262226222=222222N222N22222222222F''(..x8H.4.A.A(Q,.OO &.O OOOO &.OOOOOOO &OOO &OOO.OOOOOOOOr:cd}tttfD]}t||}||k7}|st j d|fd||fdt jvst j|rt j|nddt jvst j|rt j|nddz}t j|ddzd |iz}tt j|d }t|v}|st j d |fd t|fd t jvst jtrt jtnd dt jvst j|rt j|nddz}dd |iz}tt j|d }t||}||k(}|st j d|fd||fdt jvst j|rt j|nddt jvst j|rt j|nddz}t j|ddzd |iz}tt j|d }|j}|t}d} || k(} | s t j d| fd|| fdt jvst j|rt j|ndt j|d t jvst jtrt jtnd t j|t j| dz} dd| iz} tt j| d x}x}x} } tt fD]}t||}||k(}|st j d|fd||fdt jvst j|rt j|nddt jvst j|rt j|nddz}t j|ddzd |iz}tt j|d }t#|dt$t&td |} | j(}t*j,}||k(}|st j d|fd||fdt jvst j| rt j| ndt j|dt jvst jt*rt jt*ndt j|dz}dd |iz} tt j| d x}x}}t/| j0j3d!}| j0|dz}t|v}|st j d |fd t|fd t jvst jtrt jtnd d"t jvst j|rt j|nd"dz}dd |iz}tt j|d }t5d#d$d%d&d'd(d)d*d+d'd'd'd'd',}t7|\}}t8j:}||k(}|st j d|fd-||fd.t jvst j|rt j|nd.d/t jvst jt8rt jt8nd/t j|d0z}d1d2|iz}tt j|d x}}g}||k(}|st j d|fd3||fd4t jvst j|rt j|nd4t j|d5z}t jd6|Dcgc]}|j<j>c}d7zd8|iz}tt j|d x}}y cc}w)9u[code task (independent / merge / bot) 모두 footer 자동 삽입 + finalize 박제 정합.u1[task-9004] dev2: utils/foo.py 변경 + 회귀 +1)!=)z%(py0)s != %(py2)sr code_promptru에서 footer 미삽입rrGNr@rrrrLrtwiceu footer 멱등성 위반rm)zK%(py5)s {%(py5)s = %(py2)s {%(py2)s = %(py0)s.count }(%(py3)s) } == %(py8)s)rErFrrrhzassert %(py10)sriu)는 footer 자동 삽입 대상이 아님rrrrrrrrhr final_promptz task-9004cMERGED abc1234567TSUCCESSPASS completedz2026-05-10T12:00:00Z)rrrrrrrr r!r"r#r$r%r/)z1%(py0)s == %(py4)s {%(py4)s = %(py2)s.FINALIZED }rOrrDrKrI)z%(py0)s == %(py3)sstuckruFINALIZED 상태에서 stuck 분류는 0건이어야 함 — AUTO_FINALIZE_CHAIN_MISSING 4종이 false positive로 잡히면 안 됨. got: z >assert %(py5)sr) rrr rrRrSrTrUrVrWrXrZr[rcountr rr rrrrrrrrr9rr FINALIZEDrPrY)r=rkindrrarrcrr @py_assert7 @py_assert6ru @py_format11rrqrbrdrrfinalized_evidencerOrrrsrs r86test_regression_5_code_task_auto_enters_finalize_chainrjsEK'G=)$ <k!CCCskCCCCCCsCCCsCCCCCCkCCCkCCCCdV+B#CCCCCCC*c1111*c111111*111*111111c111c1111111+D#6|>>>u>>>>>>u>>>u>>>>>>>>>>>>>v%=>>>>>>>{{<{67<1<71<<<<71<<<<< read_only report_only analysis_onlyfinalize_policy:ru$opt-out 토큰 카탈로그 누락:  >assert not %(py0)srEmissingN) rsplitstriprRrXrTrUrVrWrZr[)expected_logicaltokfoundrra @py_format2s r8(test_opt_out_tokens_catalog_completenessrsW2N O3SYYs^A  $ $ & OE O&G;H;HH>wiHHHHHHHwHHHwHHHHHH Ps)CcVtdiddddddddd d d d d ddd dd dd dddd dddd dd dd}t|}tjtjtj tj h}|Dchc]}|jc}|z}| }|s~tjd|dzddtjvstj|rtj|ndiz}ttj|d}ycc}w) ui이미 PR이 MERGED인 task evidence는 AUTO_FINALIZE_CHAIN_MISSING 4종 어느 것도 잡히면 안 됨.rz task-9005rdrrr def4567890rTrrrrr"r#r$r rr/r2g~.Ar.r,r-uPPR_MERGED 상태에서 AUTO_FINALIZE_CHAIN_MISSING이 false positive로 잡힘: rrE triggeredNr3)r9rrr\COMMIT_DONE_BUT_NO_PRPR_OPEN_BUT_NO_MERGE_ATTEMPTrQrPrRrXrTrUrVrWrZr[)merged_evidencer_auto_finalize_reasonsr`rrars r89test_auto_finalize_does_not_inject_for_merged_pr_evidencersv$"     !!%%- !% $,!O$  /E++))0033  $))a),AAI== [[dZef*sD&c tddddddddd }t|}|Dchc]}|j}}tj}||v}|st j d |fd ||fd tjvst jtrt jtnd t j|d tjvst j|rt j|nd d z}t jd|dzd|iz}tt j|dx}}ycc}w)uQPR_OPEN_BUT_NO_MERGE_ATTEMPT — pr_state=OPEN + ci=SUCCESS + 5분+ 경과한 PR.z task-9006OPENNrr?T) rrrrrr1r+r,r.r@)zD%(py2)s {%(py2)s = %(py0)s.PR_OPEN_BUT_NO_MERGE_ATTEMPT } in %(py4)srrCrDu@PR_OPEN + CI SUCCESS + 10분 경과인데 stuck 분류 안 됨: rHrI)r9rrPrrrRrSrTrUrVrWrXrZr[stuck_evidencer_r`rCrarbrcrds r8+test_pr_open_but_no_merge_attempt_detectionrs+#! $ N ~ .E!&'Aqxx'G'  3 3 3w > 3w   48?8? K7)T(sEc tdddddddd}t|}|Dchc]}|j}}tj}||v}|st j d|fd||fdtjvst jtrt jtndt j|d tjvst j|rt j|nd d z}t jd |d zd |iz}tt j|dx}}ycc}w)uICOMMIT_DONE_BUT_NO_PR — branch push되었지만 PR 없고 5분+ 경과.z task-9007NTr?)rrrrr+r,r.r-r@)z=%(py2)s {%(py2)s = %(py0)s.COMMIT_DONE_BUT_NO_PR } in %(py4)srrCrDuDbranch_push + PR missing + 10분 경과인데 stuck 분류 안 됨: rHrI)r9rrPrrrRrSrTrUrVrWrXrZr[rs r8$test_commit_done_but_no_pr_detectionrs(# $#( N ~ .E!&'Aqxx'G'  , , , 7 ,   -1818 OwiX(sE)returnr)r<rrr)r=r)rstrr=r):__doc__ __future__rbuiltinsrT_pytest.assertion.rewrite assertionrewriterRsyspathlibrpytest__file__resolveparent_WORKTREE_ROOTrpathremoveinsertscripts.safe_cron_dispatchrrrrr r r utils.cron_targeting_auditr r rrr&utils.lifecycle_reconciliation_managerrrrrrrrr9fixturer=rfr{rmark parametrizerrrrrrrr3r:r8rsh B#  h'')0077>>~#(("HHOOC'(3~&' #%D33"9V?9L%KXPP>PP%)PP<:BIB(r: