From 3b6a01d3a4b5276a717bd6bed8fa9cb302d2e166 Mon Sep 17 00:00:00 2001
From: jinye_huang <jinye_huang@foxmail.com>
Date: Sun, 18 May 2025 23:36:43 +0800
Subject: [PATCH] =?UTF-8?q?=E4=BC=98=E5=8C=96=E4=BA=86=E5=AD=98=E5=82=A8?=
 =?UTF-8?q?=E5=86=85=E5=AE=B9=EF=BC=8C=E6=89=80=E6=9C=89=E7=BB=93=E6=9E=9C?=
 =?UTF-8?q?=E7=9A=84json=E4=B8=AD=E5=AD=98=E5=82=A8=E4=B8=BAbase64?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../content_judger.cpython-312.pyc            | Bin 41493 -> 41504 bytes
 .../output_handler.cpython-312.pyc            | Bin 22584 -> 25276 bytes
 utils/content_judger.py                       |   6 +-
 utils/output_handler.py                       | 167 +++++++++++-------
 4 files changed, 105 insertions(+), 68 deletions(-)

diff --git a/utils/__pycache__/content_judger.cpython-312.pyc b/utils/__pycache__/content_judger.cpython-312.pyc
index 5f5b05caca5c4f836b6d81f579f8012012129f01..6776532b903477593c30803cd86e2e42b0aa53c0 100644
GIT binary patch
delta 88
zcmV-e0H^<z!~&qi0u9Rz4GI7N002JrDQGaU4S)s$05P+V2DAwzAY5+myo}_yjp(L<
u<iD=wl)2`JmgR_`>6xVFjiTtGjOVPTvtAHy9|0`0%`f@}0V1=<Hco4GIVj-(

delta 55
zcmZ2*glXy#Cf?J$yj%=Gu<4j)hU!M%1~#Uz2AliYws0~laVbsa5|G}UC6FV(`MUlu
Kw#}E!{IUV$rxIEK

diff --git a/utils/__pycache__/output_handler.cpython-312.pyc b/utils/__pycache__/output_handler.cpython-312.pyc
index 6896c913beafba709918f6b76c50f3ddccf184cc..85bb16dc6ce4f410e52689918d12ba3551c792bb 100644
GIT binary patch
delta 6909
zcma)BeNY=mmY>mRB=jwT1o{F5;!A)8=0g}TU_OOqFkiL-+r+^P!C)J(5V6e)likf;
zaMqi^buQpbawxZzic)blB(A&F-llSFw(g49cyGp%BU$U(b#9HF{1qJg%6CcK)pd^$
z79`GERCT}U_j|v7{ko-jukkM8AN?CHy&{naG5Gn-|Co+;w|yX8RSU&@44cDT*bolz
zszh)iz8JJ<4kcX_yh|b^hT#~+{=LhIJUe9|$kBtZG&azx;pK_mAR@ZS7^fhN!Z241
zMNr&}<hw+8q!^pzyZA$Kpf@hgD0bnmVUz?~;t^ICrw*IJd&V(rQV_urQkMXntWg>$
z86}TQ0u2?L4A)7C*a$f83Bbw-SoH+p_y{=R3Bc+ISn~v6Z3L`)0<b;;PJ9AzQUsj*
z1Yko1obm+V)G*je88?B`qBms*KYyBw2_AEHxKz3;2He-wmSnL9A^Hf31X$D|5<UJ9
z$xKm)2c3ZwZJdgr(k9@E+M<C+PL{o845A4xr2t=PlX3PybEzDc2;}Ju+!@aKw*Z|T
zpK<=`X`&lDg&Q$veB@lGVGiS@&?$>N52@J`o#EoagP9o}gKoI`N#R-H`9Ftyu#&CK
zI6w5HlzF)>(c>u3tW%%DPhnv`lNp%xMK}SD=u^QteZ~~m;ympfS3@_AhA*Ex_vJs%
zuTIZ>@utU+<rp76JV=!pCJ2*IIF5c0j1h<-L=Fy9{ii30J$cpN;9;%VaA)qVJFhR4
z8A4pD*V)gSm<_A%|K^Ji-h_j^kt0L>ea`ja+yC<RD)@2Plp(|)b&OEsP$7p?m<?gY
z-S=L-`*&|aE6KX{@bQs>!M=6lFMs*c9rs62OO83u9PeMZ43BPDidfycU6}jsYrneV
zeqr_Hx!WJjL2H?_cgiu)?>yW)P7U@C_eF@pfDIy*vv(6}vn$X2_KS~akt?i&-*8~8
z3~(XgSl)T{g*&fbynAjMwyVGX@b<sHvHBhi@FI+=q2A#bae#l)TXN&gX0xHAyJK&C
zv!S)UrMY#l0gebKlxTckM}2$LD9|U4tM<VO9Q$DtB2PI6o&6yKt^zD1#&b+!@Nr^Z
zW(?fa!=t^<qao5cGCJ57!uvxa7|9c(;o&hcG&UFjo@C>65x|~QJf#-2;WrhknPcw4
z#RR{ibPeN3x^ZvYZMkwzFe{kPai3fqUuvT}I<L#S9+F7;fQLzxFE^cQqIH$tA9+vE
znkK)v8IG#8bEjrc&ADb>ZkxyDZSZbiDx%GezJy)VjcAUeZo)I*S6J6DvBZvhQ-X5E
z)Es}7PiqSnt&3USbXvXBFRu<JX5Ghx3OhbiA5^O5Y_qocA&=^P%{!XK5^ufFR6}Rg
zE)~%F2EWp_6=3xi`%KkzR?Si>t*`eh8@2+rd-wTFwRBe9QVp%Q`KOePTM?z+exIq1
z&Z=K(r}d3~<*s!Aon-NhzklYPGv4IobYEU8ZE2(1I%xe~zjEJJj>q}=h$C^C^yTfL
zrz|bY4qD&tS9WZjhAy=)*ZcC?X-fy)v5(g8_bWTUi)p3XJAHXww56NwK1k~i`ISA7
zccus~=riT6dny#&ha}tu4gNs8v76=IMsFFdY4D3}55(X~(o6bU9BueL_Dp>9P>QxK
zym{X~NCE@N15@bd^w-iRJ@z|N*F+G%GgQaIhffL`6issCQ?3C%Toy>16R^t()uLt*
zaV6CNA65vmNy1-=iGk`BIoYJ)uc-J))(e_rh?R6rGeKM>NGM+wk<Ci}RXHD8T}=?|
zf#U{Eb0u-Ll7zAY6dIC+XC+ZdiV}<(s&XY1*A??D?hv3nLj!$;J$DD}hXoV?HJF2B
z5tsuPfVLDh_+Nu~pqS!87>QsMp9P7j7?uXM6n~v2hSfu2NFt_$8+yUIUdZav^hht-
z&~w-IBCsb#hx5Tuicun<#F4yZQV4dYCP=tZBO*AG%9#`+1$0mZ+EWr@r4%n>Ldjm|
zOiBdcx2bz@Ie5vKo~Vet#at39Hp<UR=)V9o88hK(*HSg#gN}>x8lkx?n)?VXiRM;<
z9mclrVNtOxC?=Q#u?Sff;Ukt!9>%k16rnI$#G*xyK_{?i$p%`&q9u>en><v5C8H7i
zC{3iLbgT*5LLIvW5=w`toNC~1(Q}3pQD#ueqx6xmPRd}8kqi;+$0eg+K6NoEDp8d?
zQzF`-RAeoYxa1U%N~R1KQ{E+kKD{I^)g^lkdxI0%4}9}<#XKr~V>-pA=@j3c&Lwat
zUGkw!G?hjGU-}u*#4e#j5z)p8A5G{Ity{tLMw5Y(g-urCM_NIqOB~j+u4lOvo3WcB
zwT(w*Z@3=2$@SRpaUHo09Z1M=g&mJh>DP6-T$hktRJ2ayP_c`GwmLJL>10GFPcHEz
zeS!tK4}%hIzs@boiwcNj$OKN-Y4vqj&$AeI2v#hUag$1CPGmD(N@p(EnW@6rNb}<$
zX}&WLjAp7jw!OiPB7)=*n@7EKTSRs#DGpVj-AFSbPhoWI&v!=li$rszf#J*qya>c(
zRmT;(Vt+XHB^$Ai-xT}!$HYEGtAvLT?JK?=#F~t_6#QA2Y}C4>o9BVqC}&c7Rtl5b
zyI~3DqdJ0f{x#61Q;`{yQl|ps)FqmX=wMT6tbJ5WqU;wp*%w3mg!NfXKQOCp!zvNC
zARB<^7!s{6)%W<RI;(;e#7<C|V?;mSIu(&V$*!xg)D{0j-ksruXNgLT%fL!b`xNOM
zbRO>yNk<%m1B1i8#}9{}`F$h9&i-L%NWP(nj7T;{oV^2M4n5rbkaTFA8t6aV3klAQ
z4UUB*;pegOzP|pkF%!q}GDJCEMa>Yxlf7eogM%Tl6Z|AsPl^x`cr~{qBc$9!e)vf5
zSiiL>q}ogm{yn#ebci4km@<#(CZRAM3mc(u^;d7+aeG&%|K`p=%zX9H-wVOl=95M#
zRIR?Wu=@7TVJ>|4y^qTb<Lpor7CddS;$vXW^0zvz;~9t#ar;I_&xB&A{{B%UIerQU
zL0w*s<#{Nub2|<psRuQwsCf=G=<DJ5F=~E>nxBA=^V*KQ3B{0baCmgwc^DQ7y^dL=
z1E_fgsS-98y?LD>(a5-S!{|i>6|e6Co9jAWKorxk0SEI_@t=dU`Q7~U@NtabgFodb
zO{p|<J+nRT9X?h5v~Z1wi4`+;MwH~XE@~NL`K^>3Pd$@TFm0cC_L?{;D2bgJVk8Fl
ze$NDxUV2L`hZb_}%V*D>oj>X!{W43?m<6$?8L{D(EPg(lk)_e5(j_fp+8xX*^=2}8
zH8Vn5m&3?&gEIBJotctl&^Z-L2bi3e;I?w_E@oQ;0+<=OB`8zPYZ;l*BlAu$S-XOk
z?cR9CvU5f_U&Y9C;KbUrg`^8f9{ysiPg6844eHVtDlb%eOpEP4-F7-g9aN_-NH0j;
z&w3m_b-}bGn40^x<xLB{t$L}zm)bzbB(Djlu!NMk;n`t#muK7;UmQ%#TzKljQ=S%>
z2q)Iw<8u>))1ozLqM$0W9FKO<cFnZorllxgsbnmb-ihT*rm8(q)y-6O(+7HeRY!c5
zzJO(ju?*4296rm~jA(ujbods#URj2xY;o{f&d!_p)<FJFW-5Q@l6H9)Q@uY>-NRJ(
z&`<UIs-N-Y4+QdunfzgT<XK<-sTmQi&t>Fh6zuqLOoWTpYguJC&4mGT1!J!84lTzs
zm92rwPNuSx?mFzNeA;L34VaHI=A-oBsLy;7deO|tvk`^CUB<}Fi`jHV^R=mhJvU1#
z0wp%4#I`J?JNlSLD$sb0X*@<BAM-UjeI?_85*JhATEnojcs=JnhHY=)AcqPWdC@Id
ziu(W~%UiV5Rc+Ub+HaOt1xj}@rMs51>3su?{b;~G!q`XX(NjMAX<zA?hZt5?&AG?6
zkg$fcYYH)z*;&S7Ms7v^CA;Cd$z9ZXTdr+ux@p}Jur@H(hNTI*y_c~a3D^c1+aNu3
z(r0t{tYZP|1Y@0mWG=iGah24uTm_80aO+yXZ>eohyw=;|IJT_Ql={VGYeGyM2fqp;
zL(pjU3@}FP^zL~+JoOKFU{_&w5-bvXA73PbR|@l{Vg;4Ax0HP?u<W}Lml7#a&Z2^d
zf|2$si|+9-Wm=@b(dYV==Fj5{5g_HRaOctb0>85G^SJB?Dyfhz+D#WW(fVe;a?eAJ
zSki(&&`0VSz$r?PQwZwV`gX`|T;%%|+t)A&C@b1laN`m8u%`!p8(?j;)32<)vB5q+
zOq=WI-1?=xwBGJl?gqz-e!){nB;^nE;QxxuRivPOOXZA9(!Fi4s6{pHyvQ$4dq}{c
zNK|!2iw75*X-&;_aqWLUs0AmBGsNFKRBjYl;7W0c1g+p6ERS1oRAI@iSt2;$Jq$MJ
zxyG731sU&Qu=y4i+;Yv=7>?Y48`pfUnbuc$k9m8RQ<vM8YiVt3K-|WN+rC{R5aye)
zKf>R{-*2}W_v^5g7-=Ug!{T+FLSki`0m@ekq+RjYAF%59E&=f$@do%nCrG<vF*?1r
zu$v_O8UuW|uAPz|kYd-fVh`ZN^)dsLKhp_1`NU^w>P`+3;E+%b2+2+*KOl>Nw1EUc
z7mo-SbX_<R#7QUz1!PwoKPZiXGX&!W-5esAtnR8Hf)yl`Z;)iSjDJHI11Gx?E9kBy
zZs^qArNoU=63QW*)E*G=L%bL`aY!mSu!9IC=nj++p%U`Ij!W*6b2vwfgPhXWIq^97
zv~;&{8r~pBU0|1&&2GND7E<spl2gGfJ$!jBFE7@iS1J2}S9)2K{HQBECCX?ojg{x(
z#S0K&5;>lR-CubTZrt5Co(DH}{DP;4j*Wc=j#m_y@&whKtyeG<?mY;<J9C;@&3w5>
zw3JS_w$V==xvuC#cQ91=4hH<OLMP>;M|5jL!6Op`xRv%CIjY8MxUd>G*2u9>xMYol
z|A~N)WVxUgCqB_6)$AZX*+JIg;7nx({?a9P<%b0R2asHqe92LhiE}S%VIA{!?HFzV
zx;hP>idGC7>ykw2Y)D`;lmT<;a;Sf=P6HCMl_2yTZOWyz`oG{x6r*u;@xS`o$E!bm
z8D4{#R6u7tB!3Q$MWqVQgY&jz{Ab{ztq^|)+_H5G(ThFAInnQU5!5#B=VhUgsV*%v
z&f&NUl-Qpu;VZIlY7*y9csjf#eoZZmg6b)J>Evy-enD_S;Lh??d+V1|=&nQ8)jcq7
zPzk>@D)`R{d$P`m$<=GA@GAsgqYpJm7dA&hH0_>m-Cc=&s+QV|u}?E}b`x<KH$wTc
ziL@8+FI)IXE*9)ACoWeO?6wkD^0^Ry#Y)1FE9C@|D+SG5VufGOR7I>*k<DD7X!;yK
zdg*@CD_odkO}G7t_aR#UC2+j`Aie;W+pEFpyhI@DNKbhSDcEcZB~l@h8b2{Q=3rAU
z_89#h?CY@OF9LUmniqhGi2~q`O8^(zb@=g1_d4{Pl7B`F4X9~CO$%xe(ouw(T+}F0
z(;jK;P#Cj-y8Q>IOypkNQ@Dnq-n%#4)9w4#1!~|vt6AHRH*&~!+|#j!v4VGOU7$Pm
zN2F(lBGMVx{V?q7y<9F?wI&vl)f^ApG1Pne!ad!(m+ovl$h01W_kVEwp1P9Mpb21y
z=wZi_qn;XIpW$24#<6oE@_x<%_=be~wVfQ$-`T+X6fN+V;DgR>_<w@Gbk-@lNL)RC
L;=eG|vw`@3Xmr3x

delta 4759
zcma)9dr(`~nZH+8SJFjqkdTA~67Pp3Ug9M%=51_)E%UAy2*zB3AqH$o#KsY(oi-_%
zlG@H`YuqGb+-YXiiQCY6cC03yUE8r+zdF6w5`kqrjXPtSO=owDYck%%|Lk`pgh@8p
z&Hdw?@1FDfzTbDgd+zzZ^Sc{lV1ZQJk;`KUcz*UD*@t%S{Iz0LA5%mT#96{ejF70n
z8i!PdHxyn4#^C27^Nt9;YcQZU3zN`a4LeHZKJpEMRlG&LBZwfT7>4jN5t$%-3`?=X
zi}X8S6hFnpl-MU8F`)jqIHmF(GBQdM;W41oQmpC(41Q_mLbC=<=SW1)fRvDpMZZik
zV@IesAA|lg$s(A&f~p)^Ds@VNJ}NY)N_`^mO$NSg^6V#AVoGvSGJ7x5qhv>dV77Ow
zTAs$oe4X0lhWat`7;&81PaGq&wTQM9kwR2%NugJ$fx{luW640DB^S}cT<>u&vZd%v
zp9^AQPH60QyQQP8q1Cdl)wO3|y9M53J}1%slnJ{CmhA4pO(!%fOvCYk!6V+`@v#+V
za8sm@xre=j*)iy8N}h%sT#+7ekB=Vl_8sZ>9`fL?h+0w`1>YgjPg9GVX{Lp|uhyP^
zVcxNz3#pxJ1SuaQ11<MenzPIqX3jc)bYbF33)j{a>OH)u8ht|H`G>N&Q_Zh7bH?p~
zmja`lp*bYmvmS#UrvFfpi030#p010~4>N|Q4*=(7MSVVTmA2HUQCHI~AYaQ-G+2pi
zWi`?UGj*M^z{~YCy526no-Kyy>-kKBiMsAIH0Y=sIvV5~X1XCud?QT^$QxE>2aFdP
zcEnP@kEKC&R|AQ=2ATmJr09b;m=f=)$6o0dw(CM>b*YLK`Xn!sZkbQQQU(A)FWeUC
zsY+r@%2F^-xW!={;f6OB#?ha$bR9G+itc1j8Y_Ys8f&rwki?skWK&|o66Pb^GBzg6
zE5UplyfJtlTPCIGgRE4Pn=PYa7y@0$u^>K6E0MCYFdwSU(US_)Z%vnO{uWcxC=Ar3
zs4hnvK_opx$b3>DQ(+PxUalG>^h);L2vEKWgsE7hw6=W3cpNg@Q{|dx{|I;o{;K1j
zMPf<;$AJNk15I8OHYM>Xz-WAU`MU7(aTs-#1gqbC(^sOO7Fu+Ms1iwd6Ts8(Gh#wV
z?Qx_TA2N9&))<ZYloHO5<S5FgWKFE%V&azbL*-3H`yD1=@u^r5o5Y&E8BzE}3mi|6
z?4^mxKII$4TY~?OxIo0FZ4#@$K&<}X#6H!C6<x^HiPL>CV4XtWDRd@g_|)H!DQlA{
z_6tn0U&HhlhgOSEqV%l?Sl9<RJaswPhp(PfTA!`kCmFG$`TYr|dZPD5g6M%$dMa*8
z<IRhP%%}0@qrc~CNh|&mbQ~D9q4Nd0Hv3z`EoNZpQH^K4w>T>MG^~KlP1roQ!AjoN
z@XvOI9R~(6W=Qpxq6-CDG9O(os7WmF#eTyOIyOTX|Ai37e@zH6AI;|DjDkNFS%ls%
zOeN(wJ5XAm9FdMVwfv;~v|#ocE0_WaAM6{+0Hcxt3mAC>^L7LSIU|ug04N<L_o`8z
zD=|9-Z#Op8Cx>-xi5g^^uvM@>u|?Opg~3c^M*P^|XxhXXDRLIGXwmMX6spjt4SS+J
zsXa~3x&Z|ec!E)77)7T~ZFu7-_Qn5S-b@N299xoD3DOm}b|}W(!$ZSk{fGO!!`{P#
zE2_=+f$=f#;FxzsF*3mp4fgc|X3{h4S&?rdy!}I-6*{cvHbJ;o#2Y&3--@kBU6O}J
zOB$wE<Pig&iGhJZkH-yXHyoSLOxU4WedXoVU;TFV@^3zV|J1+#>-SfG_3G-&{=WuZ
z{`mAOfB)U=>Zx}hzW?^>d#6@s&aTe>WFm1>BLV@9kNrP*82IJt_uu>EN9W2dCuIF@
z@9@Ck!Q2tg_?UYb_Q0)@Uf~4jp_<a8Xt~ruzKCRHuahQpzDx_*duY&$ep_aBO#&qB
z+e$*8r+;kNJA7iWZ(}`u&yTzN!UlUVsuwpYxH*QKOxzsDO*p0Fjpu#|#X4J>Uk7nT
zIzHh&GU4rG`@Q|{@8QXpaq|P*%;M%GZcah7B8&3j^zNR*LlUSb!r30(aHcTRFyzre
zgC05Kk*K^vjNSI!qS4Osr0MvCv!iE5=bMA^1vAn$5g}8bZsMip`5g;ZKCSXWirqiR
zrxeXJ&53xKd0AzgYv5HW%ktRMFYxl@`R5m;d`86sJj|(5KwV`#HT~-J+;P7)q|94R
z%LYXWFH3%4%b#iD^jW+tdpWldi#fc^x@^mzjt)OiYUhl+GR<$kBIUE2mh;L3C464(
zY3ZDcSJ}XFLi&00IkR84kP<W$&nT9S8Rx6cRr#F@2ZKf@C(*Bo;+U$3#^fcVl{Z@b
zP4BvX;R-ZfX${(TF4<hX&Bg8O3fjIEH11zA9^j1!xc);y<M8xM^Zmrsr9?ZQX!pPH
z?)Wdp1H(6CgSpL1xov!I8`s_)%<Ty#_AVu|d?L#Y9u6js&g=zutfjCt1ux57wv|WM
zfY)!Cui%xr3l&_I>z-rZ{nE;%(nh|taam*LEcv{qU`bQPYsvz;pr+<AMV3w1kssCV
zx>|BG^Mk52icp(4lY>_kJy0gkck{~pg>J5<?Ot*FeP`8@a~JR26}D5zYaC0Oa$Zv&
zC=F`rz)pDs`B6jD)y|uZAM|aQDdtrr>mov~2^VSxXGm7Q78A=<JuqdScb#+jU4f3E
zsjg*a4|r`fFC{qn1ZTh+OxQ8g1S2*lEKkYHvhbR8b5>rN4%G8vPLs8)G|s)mD|7I>
zjMLbowV+{VNVW@#z+YG9Btr$+|M2T7NLnBtL!VXVm5Z3|525rtG*9`5Lh2G+G}e)U
z<YkrmtoV$0ZabHd<4@!As<_;0PG1vJ)#Bo_4i}&3XWMh)<V=0IbRC&bT@Z)VWorb3
zKH6S@YO7uoF^s%oU4s^@>^nqE*Ow|>f!r2od9cAEoo9J{PDo{a0)E+^Ma)>V(7mvS
zGt@51>i&DZ5!F;@P*2x$P;GT)j0{#lJ|gfsJk_Og?9`*M^SI)@T%9Ea_D2LXPuJto
z->P$_!7_&1o~}y>mEj)*iG7l5mgGsiJc&!Ly;5<-e$&mhws9@(oVjC3-pR{5pRQ3D
z_qpdW{IAzi>$6)ji9g6|q^<G9O{bzQmbfLhw9(Y9Obf`jvlQ)W;&zFyT}a)LTR^_6
zP;|(NyQVt2W167;lwyH5E>+RV5L|(_V;jYbEFkllOlus)=j&S))I9|a^1XPbO+ei<
z>su?SdzCcEL7Hw;ii0r{;0eYu?Iaa68r!x}!EH3iAtBwa5Qi8EaE8=OdnFZ0(6^US
zp)wleMS|{#DG@Ky5};p{GaXK9F<#$ML@gH4Sl&i=io_q15^&{1hUqM&K8!VX7E&J;
z(w(K3Y_+eFf+qyJT=%BnF@ZYjo5<JEh5E~1zG(I|L@t`Bp$P4C$f&F-F;3l=K5ej(
z<)<Nf*)sP5=-l7nDBiqXxhGL)<7=YNK+m&;K4~l-pqU!Mm+wlGWS|?y&Kk}b<|_Qs
zE9u<67Vf!&p+m<O)yMIEoqTp#+I_R)rk(5R;d&2oJ;R(~WKs6Q6B<MC)}JKt?)mo4
z-BvBWDAh7>QA(}V64w~IHdcI1A;z+nsS{Dxl9K93>I0Il6QNIcW{`82OuK$VkwH|@
zY`*kg&6%XaYXPsr&&33&bPanwWD0t1uK^u&XwZ9mlgLacM@;C?T@rN4B}PJ50?eCT
z1~La<xKXof3*yo#*WXADj^YHqO|8E1FRR~$+`5n*k6vr((SB7rn?#?rm_<0LWoyt?
zsTMg}4dk0>PwW1eaDozy4!TdGh1O1y5N_O__)D_3vm~iOKX3nbnMj>;-;g*r>Te5_
zh75HO8LfxBa`e95be=iK%xC#)0u48lxo`D^4jx_9yCK9d|4?H@-5q9QGNID1S>P9g
zgz!X%XCnGIvOn&a+t*AHxAcnL<;1NF%kEt2wkQ?k+qv}aQt@qv7|Z2MGeO;<3Y#jZ
zJH<l4->INs<Ss#he3xSOFw|XH;T|FNM<Kn3K~tT7A?q&nb^S;PCBl05pG7~1MZ4cd
z%RSxX2~^ZuL%Zi-1fA*4NIr|F!X*IC{wp*)F?z(~4%ZOjP4~}exp#L0&g|h^x@Gfm
zYp~`V+V)&2*?7tOoJm0X(U}8r<Q#hYKr(4Ze>l)R9S{Ga-?2vEKF}WNxz>&ifgO4r
zo>|L3PSRbZzio{O3jxoDz_oQorRPVY(xmTk1ojUvC8Vp?WDH#+@WVI3ePAHcbNkx4
zuI@&DU-#qi_@nqr+JFh5!{(t28^=966n<YLh(^I~K~(M({1g1%Fu!(^M4$BU5WSDr
d`(spmumCpo(7}4OM@Z`DMn5HR@3z70{{RV~G=u;E

diff --git a/utils/content_judger.py b/utils/content_judger.py
index acb1f69..e9d16db 100644
--- a/utils/content_judger.py
+++ b/utils/content_judger.py
@@ -62,7 +62,7 @@ class ContentJudger:
 关键字词：价、元、r、人民币、rmb、优惠、活动、福利、赠、免费、折、DIY、跟拍、送、摄影、兑、服务、￥、包、课、提供、选、专业、补、差
 4. 字数控制：每个文案的标题字数都必须少于20个字（计数包括文字、符号、数字和emoji）。如果标题超过20个字，请在符合文案风格的前提下修改标题到20个字以内，尽量保留emoji，必须保证标题流畅通顺。
 5. 敏感字词替换：请删去标题中的数字后面的"元"和"r"，并将正文中数字后面的"元"字修改为"r"。例如：标题中的399元修改为399，正文中的399元修改为399r                
-6. 特征语句保留：请保留文案中原本的引流语句，不要修改或删除。请保留文案中的换行符"\n"，不要修改或删除。
+6. 特征语句保留：请保留文案中原本的引流语句，不要修改或删除。请保留文案中的换行符 \\n，不要修改或删除换行符。
 7. 面向人群保留：请尽量保留文案原本的面向人群和风格，这是同一产品面向多种人群营销的策略。例如产品资料中写明亲子游时，文案写"为情侣定制的山水秘境"是可以接受的。
 8. 案例如下，请参考案例评判真假信息的尺度，逐行逐句仔细分析不符点和修改思路，并按照分析思路落实对每一处不符的修改措施，严格审查每一篇文案：
 {  
@@ -118,9 +118,9 @@ class ContentJudger:
     8、产品材料中未提及房内配有加厚床垫/卡通洗漱杯/尿布台（无需额外购买），因此属于不符内容。应回顾产品资料中关于房内配置的内容，修改为：房内配置：55英寸超大纯平电视+独立的浴缸+超大的落地玻璃窗，尽览蕉门河风景，尽享亲子度假时光。
     9、产品材料中未提及五一专属加码，但文案中提到5月1-5日期间入住，凭房卡可免费领取儿童防晒冰袖+湿巾礼包，因此属于不符内容。应回顾产品资料，找到现有文案未提及的产品特色，修改为：套餐专属福利：1、豪华客房一间一晚(周一至四只开放双床房) 2、2大1小自助早晚餐 3、赠送2大1小水鸟世界门票（酒店前台领取），无需额外购买             
     10、产品资料中未提及水鸟世界门票领取有时间限制，但文案中提到水鸟世界门票需提前1小时至前台领取纸质票，因此属于不符内容。应修改为：酒店前台领取水鸟世界纸质门票
-    综合以上分析结果，将修改应用到原文案中，得到修改后的文案。"
+    综合以上分析结果，将修改应用到原文案中，得到修改后的文案。",
     "title": "五一遛娃👶必囤！喜来登1088景观房",
-    "content": "五一不想挤人潮？南沙这家酒店直接承包遛娃+度假双重快乐‼️\n地铁直达！2大1小1088r住景观房，含双早+自助晚餐+水鸟世界门票，儿童乐园/泳池/健身房全开放！\n🌟【遛娃刚需全配齐】\n✅ 儿童乐园：酒店设有免费儿童乐园，提供丰富的游乐设施，让孩子们尽情玩耍\n✅ 户外泳池：酒店配有户外无边泳池，供大人小孩一同享受清凉时光  \n✅ 健身房：酒店提供免费健身中心，适合家庭成员共同锻炼。\n\n📍【1小时玩转南沙】\n① 南沙天后宫（车程20分钟）：穿汉服拍大片，听妈祖传说涨知识\n② 南沙湿地公园（40分钟）：5月芦苇摇曳，带娃认鸟类+乘船探秘\n③ 十九涌海鲜街（45分钟）：现捞现煮生猛海鲜，人均50r吃到撑  \n\n🍽️【家长友好细节】  \n• 自助餐厅：供应鲜美海鲜、精美甜品等任君选择，大人小孩都爱吃  \n• 房内配置：55英寸超大纯平电视+独立的浴缸+超大的落地玻璃窗，尽览蕉门河风景，尽享亲子度假时光  \n• 安全保障：酒店设有完善的监控系统和安保措施，全力保障您与家人的安全  \n\n🎁【套餐专属福利】\n1、豪华客房一间一晚(周一至四只开放双床房) \n2、2大1小自助早晚餐 \n3、赠送2大1小水鸟世界门票（酒店前台领取），无需额外购买  \n\n📌Tips：  \n1. 周一至周四仅限双床房型，周五起可选大床房  \n2. 酒店前台领取水鸟世界纸质门票  \n3. 地铁四号线金洲站下车，打车15分钟直达酒店  \n\n这个五一，南沙喜来登让你躺着遛娃！不用长途跋涉，家门口就能玩出仪式感～\n
+    "content": "五一不想挤人潮？南沙这家酒店直接承包遛娃+度假双重快乐‼️\n地铁直达！2大1小1088r住景观房，含双早+自助晚餐+水鸟世界门票，儿童乐园/泳池/健身房全开放！\n🌟【遛娃刚需全配齐】\n✅ 儿童乐园：酒店设有免费儿童乐园，提供丰富的游乐设施，让孩子们尽情玩耍\n✅ 户外泳池：酒店配有户外无边泳池，供大人小孩一同享受清凉时光  \n✅ 健身房：酒店提供免费健身中心，适合家庭成员共同锻炼。\n\n📍【1小时玩转南沙】\n① 南沙天后宫（车程20分钟）：穿汉服拍大片，听妈祖传说涨知识\n② 南沙湿地公园（40分钟）：5月芦苇摇曳，带娃认鸟类+乘船探秘\n③ 十九涌海鲜街（45分钟）：现捞现煮生猛海鲜，人均50r吃到撑  \n\n🍽️【家长友好细节】  \n• 自助餐厅：供应鲜美海鲜、精美甜品等任君选择，大人小孩都爱吃  \n• 房内配置：55英寸超大纯平电视+独立的浴缸+超大的落地玻璃窗，尽览蕉门河风景，尽享亲子度假时光  \n• 安全保障：酒店设有完善的监控系统和安保措施，全力保障您与家人的安全  \n\n🎁【套餐专属福利】\n1、豪华客房一间一晚(周一至四只开放双床房) \n2、2大1小自助早晚餐 \n3、赠送2大1小水鸟世界门票（酒店前台领取），无需额外购买  \n\n📌Tips：  \n1. 周一至周四仅限双床房型，周五起可选大床房  \n2. 酒店前台领取水鸟世界纸质门票  \n3. 地铁四号线金洲站下车，打车15分钟直达酒店  \n\n这个五一，南沙喜来登让你躺着遛娃！不用长途跋涉，家门口就能玩出仪式感～\n"
 }
 
 8. 必须按照以下格式输出修改后内容，不需要输出无关内容
diff --git a/utils/output_handler.py b/utils/output_handler.py
index 7855812..18353f6 100644
--- a/utils/output_handler.py
+++ b/utils/output_handler.py
@@ -101,8 +101,27 @@ class FileSystemOutputHandler(OutputHandler):
         topics_path = os.path.join(run_dir, f"tweet_topic_{run_id}.json")
         try:
             with open(topics_path, "w", encoding="utf-8") as f:
-                json.dump(topics_list, f, ensure_ascii=False, indent=4, ignore_nan=True, cls=SafeJSONEncoder)
+                # 不使用自定义编码器，直接使用标准json
+                json.dump(topics_list, f, ensure_ascii=False, indent=4, ignore_nan=True)
             logging.info(f"Topics list saved successfully to: {topics_path}")
+            
+            # 额外创建txt格式输出
+            txt_path = os.path.join(run_dir, f"tweet_topic_{run_id}.txt")
+            with open(txt_path, "w", encoding="utf-8") as f:
+                f.write(f"# 选题列表 (run_id: {run_id})\n\n")
+                for topic in topics_list:
+                    f.write(f"## 选题 {topic.get('index', 'N/A')}\n")
+                    f.write(f"- 日期: {topic.get('date', 'N/A')}\n")
+                    f.write(f"- 对象: {topic.get('object', 'N/A')}\n")
+                    f.write(f"- 产品: {topic.get('product', 'N/A')}\n")
+                    f.write(f"- 产品策略: {topic.get('product_logic', 'N/A')}\n")
+                    f.write(f"- 风格: {topic.get('style', 'N/A')}\n")
+                    f.write(f"- 风格策略: {topic.get('style_logic', 'N/A')}\n")
+                    f.write(f"- 目标受众: {topic.get('target_audience', 'N/A')}\n")
+                    f.write(f"- 受众策略: {topic.get('target_audience_logic', 'N/A')}\n")
+                    f.write(f"- 逻辑: {topic.get('logic', 'N/A')}\n\n")
+            logging.info(f"选题文本版本已保存到: {txt_path}")
+            
         except Exception as e:
             logging.exception(f"Error saving topic JSON file to {topics_path}:")
             
@@ -124,81 +143,82 @@ class FileSystemOutputHandler(OutputHandler):
         
         # 创建输出数据的副本，避免修改原始数据
         import copy
-        output_data = copy.deepcopy(content_data)
+        input_data = copy.deepcopy(content_data)
         
         # 统一使用tags字段，避免tag和tags重复
-        if "tag" in output_data and "tags" not in output_data:
+        if "tag" in input_data and "tags" not in input_data:
             # 只有tag字段存在，复制到tags
-            output_data["tags"] = output_data["tag"]
-            del output_data["tag"]
-        elif "tag" in output_data and "tags" in output_data:
-            # 两个字段都存在，保留tags并删除tag
-            del output_data["tag"]
+            input_data["tags"] = input_data["tag"]
+        elif "tag" in input_data and "tags" in input_data:
+            # 两个字段都存在，保留tags
+            pass
         
         # 确保即使在未启用审核的情况下，字段也保持一致
-        if not output_data.get("judged", False):
-            output_data["judged"] = False
-            # 添加original_title、original_content和judge_analysis字段，值为null
-            output_data["original_title"] = None
-            output_data["original_content"] = None
-            output_data["judge_analysis"] = None
-            # 添加original_tags字段
-            if "tags" in output_data and "original_tags" not in output_data:
-                output_data["original_tags"] = output_data["tags"]
+        if not input_data.get("judged", False):
+            input_data["judged"] = False
+            # 添加original字段（临时），值为当前值
+            if "title" in input_data and "original_title" not in input_data:
+                input_data["original_title"] = input_data["title"]
+            if "content" in input_data and "original_content" not in input_data:
+                input_data["original_content"] = input_data["content"]
+            if "tags" in input_data and "original_tags" not in input_data:
+                input_data["original_tags"] = input_data["tags"]
         
-        # 保存原始值用于调试
-        original_title = output_data.get("title", "")
-        original_content = output_data.get("content", "")
+        # 保存原始值用于txt文件生成和调试
+        original_title = input_data.get("title", "")
+        original_content = input_data.get("content", "")
+        original_tags = input_data.get("tags", "")
+        original_judge_analysis = input_data.get("judge_analysis", "")
         
-        # 添加Base64编码内容
+        # 创建一个只包含元数据和base64编码的输出数据对象
+        output_data = {
+            # 保留元数据字段
+            "judged": input_data.get("judged", False),
+            "judge_success": input_data.get("judge_success", False)
+        }
+        
+        # 为所有内容字段创建base64编码版本
         try:
-            # 编码标题和内容
-            title_base64 = base64.b64encode(output_data.get("title", "").encode('utf-8')).decode('ascii')
-            content_base64 = base64.b64encode(output_data.get("content", "").encode('utf-8')).decode('ascii')
+            # 1. 标题和内容
+            if "title" in input_data and input_data["title"]:
+                output_data["title_base64"] = base64.b64encode(input_data["title"].encode('utf-8')).decode('ascii')
             
-            # 添加到输出数据
-            output_data["title_base64"] = title_base64
-            output_data["content_base64"] = content_base64
+            if "content" in input_data and input_data["content"]:
+                output_data["content_base64"] = base64.b64encode(input_data["content"].encode('utf-8')).decode('ascii')
             
-            # 如果有原始内容，也编码
-            if "original_title" in output_data and output_data["original_title"]:
-                output_data["original_title_base64"] = base64.b64encode(
-                    output_data["original_title"].encode('utf-8')).decode('ascii')
-            if "original_content" in output_data and output_data["original_content"]:
-                output_data["original_content_base64"] = base64.b64encode(
-                    output_data["original_content"].encode('utf-8')).decode('ascii')
+            # 2. 标签
+            if "tags" in input_data and input_data["tags"]:
+                output_data["tags_base64"] = base64.b64encode(input_data["tags"].encode('utf-8')).decode('ascii')
+            
+            # 3. 原始内容
+            if "original_title" in input_data and input_data["original_title"]:
+                output_data["original_title_base64"] = base64.b64encode(input_data["original_title"].encode('utf-8')).decode('ascii')
+            
+            if "original_content" in input_data and input_data["original_content"]:
+                output_data["original_content_base64"] = base64.b64encode(input_data["original_content"].encode('utf-8')).decode('ascii')
+            
+            # 4. 原始标签
+            if "original_tags" in input_data and input_data["original_tags"]:
+                output_data["original_tags_base64"] = base64.b64encode(input_data["original_tags"].encode('utf-8')).decode('ascii')
+            
+            # 5. 审核分析
+            if "judge_analysis" in input_data and input_data["judge_analysis"]:
+                output_data["judge_analysis_base64"] = base64.b64encode(input_data["judge_analysis"].encode('utf-8')).decode('ascii')
                 
             logging.info("成功添加Base64编码内容")
         except Exception as e:
             logging.error(f"Base64编码内容时出错: {e}")
         
-        # 对内容进行深度清理，确保安全序列化
-        try:
-            # 暂存judge_success状态
-            judge_success = output_data.get("judge_success", False)
+        # 保存可能有用的额外字段
+        if "error" in input_data:
+            output_data["error"] = input_data["error"]
             
-            # 深度清理
-            output_data = self._sanitize_content_for_json(output_data)
-            
-            # 恢复judge_success状态
-            output_data["judge_success"] = judge_success
-            
-            # 移除可能的错误标志 - 我们通过尝试序列化来决定是否设置它
-            if "error" in output_data:
-                del output_data["error"]
-            if "raw_result" in output_data:
-                del output_data["raw_result"]
-                
-            logging.info("内容已经过安全清理，可以序列化")
-        except Exception as e:
-            logging.error(f"内容清理过程中出错: {e}")
-        
-        # 保存统一格式的article.json
+        # 保存统一格式的article.json (只包含base64编码和元数据)
         content_path = os.path.join(variant_dir, "article.json")
         try:
             with open(content_path, "w", encoding="utf-8") as f:
-                # 使用自定义的SafeJSONEncoder
-                json.dump(output_data, f, ensure_ascii=False, indent=4, ignore_nan=True, cls=SafeJSONEncoder)
+                # 使用标准json
+                json.dump(output_data, f, ensure_ascii=False, indent=4, ignore_nan=True)
             logging.info(f"Content JSON saved to: {content_path}")
         except Exception as e:
             logging.exception(f"Failed to save content JSON to {content_path}: {e}")
@@ -206,19 +226,35 @@ class FileSystemOutputHandler(OutputHandler):
         # 创建一份article.txt文件以便直接查看
         txt_path = os.path.join(variant_dir, "article.txt")
         try:
-            # 使用原始内容
+            # 使用原始内容，保留所有换行符
             with open(txt_path, "w", encoding="utf-8") as f:
-                f.write(f"{original_title}\n\n{original_content}")
+                if original_title:
+                    f.write(f"{original_title}\n\n")
+                
+                # 保持原始内容的所有换行符
+                if original_content:
+                    f.write(original_content)
+                
+                if original_tags:
+                    f.write(f"\n\n{original_tags}")
+                
+                if original_judge_analysis:
+                    f.write(f"\n\n审核分析：\n{original_judge_analysis}")
+            
             logging.info(f"Article text saved to: {txt_path}")
         except Exception as e:
             logging.error(f"Failed to save article.txt: {e}")
             
-        # 记录调试信息，无论是否成功
+        # 记录调试信息，无论是否成功 (包含原始数据的完整副本以便调试)
         debug_path = os.path.join(variant_dir, "debug_content.txt")
         try:
             with open(debug_path, "w", encoding="utf-8") as f:
                 f.write(f"原始标题: {original_title}\n\n")
                 f.write(f"原始内容: {original_content}\n\n")
+                if original_tags:
+                    f.write(f"原始标签: {original_tags}\n\n")
+                if original_judge_analysis:
+                    f.write(f"审核分析: {original_judge_analysis}\n\n")
                 f.write("---处理后---\n\n")
                 for key, value in output_data.items():
                     if isinstance(value, str):
@@ -234,11 +270,10 @@ class FileSystemOutputHandler(OutputHandler):
         prompt_path = os.path.join(variant_dir, "tweet_prompt.txt")
         try:
             with open(prompt_path, "w", encoding="utf-8") as f:
-                 # Assuming prompt_data is the user prompt used for this variant
-                f.write(prompt_data + "\n")
+                f.write(prompt_data)
             logging.info(f"Content prompt saved to: {prompt_path}")
         except Exception as e:
-             logging.exception(f"Failed to save content prompt to {prompt_path}: {e}")
+            logging.error(f"Failed to save content prompt to {prompt_path}: {e}")
              
     def _ultra_safe_clean(self, text):
         """执行最严格的字符清理，确保100%可序列化"""
@@ -252,7 +287,8 @@ class FileSystemOutputHandler(OutputHandler):
         config_path = os.path.join(run_dir, f"topic_{topic_index}_poster_configs.json")
         try:
             with open(config_path, 'w', encoding='utf-8') as f_cfg_topic:
-                 json.dump(config_data, f_cfg_topic, ensure_ascii=False, indent=4, ignore_nan=True, cls=SafeJSONEncoder)
+                 # 不使用自定义编码器，使用标准json
+                 json.dump(config_data, f_cfg_topic, ensure_ascii=False, indent=4, ignore_nan=True)
             logging.info(f"Saved complete poster configurations for topic {topic_index} to: {config_path}")
         except Exception as save_err:
             logging.error(f"Failed to save complete poster configurations for topic {topic_index} to {config_path}: {save_err}")
@@ -308,7 +344,8 @@ class FileSystemOutputHandler(OutputHandler):
                 metadata_path = os.path.join(os.path.dirname(save_path), metadata_filename)
                 try:
                     with open(metadata_path, 'w', encoding='utf-8') as f:
-                        json.dump(metadata, f, ensure_ascii=False, indent=4, ignore_nan=True, cls=SafeJSONEncoder)
+                        # 不使用自定义编码器，使用标准json
+                        json.dump(metadata, f, ensure_ascii=False, indent=4, ignore_nan=True)
                     logging.info(f"保存{image_type}元数据到: {metadata_path}")
                 except Exception as me:
                     logging.error(f"无法保存{image_type}元数据到{metadata_path}: {me}")