_j. \ddlZddlZddlZddlZddlZddlZddlmZee j dZ dZ dZ dddgd d d d dgd d d ddgd d dddgd d dddgdd dddgdd dddgdd gZgdZdZdZdZdZd Zd!Zd"Zd#Zed$kr edSdS)%N)Pathzgma-pronunciation-v1uCác bạn đang nghe truyện được phát từ Gác Mái Audio, chúc các bạn có những giây phút nghe truyện vui vẻ. Hãy ủng hộ chúng tôi bằng cách like video và đăng ký kênh.05 giờ 12 phútunăm giờ mười hai phút narrationz8expand exact clock expression for natural Vietnamese TTS)sourcespokenscopereason 05 giờ 09unăm giờ không chín phút01 giờ 37 phútu!một giờ ba mươi bảy phút 03 giờ 11uba giờ mười một phútAudiou au đi ôintroz2brand pronunciation only; semantic intro unchangedlikelaiz8English CTA pronunciation only; semantic intro unchangedvideou vi đi ôz=English loanword pronunciation only; semantic intro unchanged)rr r r rrrcNtj|S)N)hashlibsha256 hexdigest)datas !script/normalize-pronunciation.py sha_bytesrs >$   ) ) + ++ctj}|d5tfddD]}|| dddn #1swxYwY|S)Nrbc.dS)Ni)read)handlesrzsha_path.. s&++o">">rr)rropeniterupdater)pathdigestblockrs @rsha_pathr's ^  F 4!F>>>>DD ! !E MM%  !!!!!!!!!!!!!!!!     s,A##A'*A'c|jdd||jdz}||t j||dS)NT)parentsexist_ok.part)parentmkdir with_suffixsuffix write_bytesosreplace)r$rparts r atomic_bytesr4%scKdT222   DK'1 2 2DTJtTrc|t|tj|dddzddS)NF ensure_asciiindent utf-8)r4jsondumpsencode)r$values r atomic_jsonr@,s= 5uQGGG$NVVW^__`````rcVt|d|dS)Nr;)lenr>)text char_offsets r byte_offsetrE0s' tL[L!((11 2 22rctjdtj|zdztjtjzS)Nz (?.?s$SVs1vA-?Q$Hr)keyr)r lexicon_indexrrsource_char_spansource_byte_spanoutput_char_spanrWoutput_byte_spanr;z"reverse reconstruction failed for ) enumerateLEXICON_ENTRIESrMfinditerappendstartendgroupsortrBrEjoinr> RuntimeError)rCr candidates entry_indexentrymatchselectedcursoritempieces replacements source_cursor output_charsr^r_ source_valuer unchanged out_startout_endoutputrQabrebuilt reconstructeds r transformrx8s<J'88jj U g & & U8_--66t<< j jE   u{{}}eiikk; TUX]^fXgh i i i i jOOHHOIIIH F 7V   a FLML9A5sKvu,-  i   I&   fF # ("!& !,T5!9!9;tS;Q;Q R!*G 4       MM$}~~&''' WWV__FSS%&1#.vq#9#9;vq;Q;Q"R G F+, svfUl+,,,s8}%%% NN6&''?###GGG$$MG$$ G(<(<<<GGGHHH < rc Dtdz }tj|d}|ddks-|diddkrt d |d }t|d z }t|d z }t|d z }t|dz }t|dz }t|dz }t|dz } ||| fD]W} | s/| | jdzrt d| Xtj|d} | ddks| ddurt d| } | } t| }t| }|| dks|| dkrt dtj|d}|ddks|ddurt d|d|ks|d|krt d|d d!krt d"g}|d#gD]l}t|d$z }| }t||d%krt d&|d$| |mt|d'ksd( || krt d)| d}tjd*||krt d+t#|d,\}}t#t$d-\}}d.t'jd/|D}|rt d0|dd1g}g}t+t,D]\}}d,|d2vr5tt/|d3|nd4}d-|d2vr:tt/|d3t$nd4} d-|d2vp d,|d2v}!|| zd4krd5nd6}"| ||d3|d7|d2|| |"d8|!r!|"d5kr| |d3|rt d9|t3d:|D}#t3d;|D}$t||#kst||$krt d<d=dt4d>t,t6gd?}%tj|%d@dABdCzd}&|d}'t<jt<j j!"}(idddddDt4d|dE|d d|dF|ddGt|'dH|ddIt|&dJ|dK|dLt|dMddNddOd4dPd4t$tt$d|t|d|t|ddQ|(dR})tG||&tG||'tI| |)tK||)dIkstK||)dGkrt dSdT|d<|)dG|dG<|)dI|dU<d|ddV<dd|d||)dGt|d4|(dW|dV<|(|dX<tI||tMtjdt|t||)dGdYd@ZdS)[Nzscript/project-manifest.jsonr;)encoding active_runzrun-002stepsstory completedzDpronunciation blocked until active run-002 Story Promotion Gate PASS active_pathspromotion_receiptspoken_narration_source story_canontts_semantic_assembly_receiptspoken_narrationtts_pronunciation_lexicontts_pronunciation_receiptr+z%pronunciation output must be virgin: statusverifiedTzpromotion not terminal PASSspoken_narration_source_sha256source_canon_sha256z!promotion source/canon hash driftPASSz/TTS semantic assembly receipt not terminal PASSsemantic_source_sha256source_candidate_sha256z#TTS semantic assembly lineage driftseparator_contractz.exactly three blank lines between source partszTTS separator contract drift source_partsr$rzTTS source part hash drift: s z+TTS semantic source reverse assembly failedNFCznarration source must be NFCrrcbg|],}|d|d-S)r)charrD)r`r^).0rgs r zmain..sB Q >>rz[0-9]z.spoken narration has unresolved ASCII digits: r rrresolved unresolvedr)rTrrr narration_occurrencesintro_occurrencesrz$lexicon entries have no occurrence: c3&K|] }|dV dS)rNrPrrQs r zmain..s(OOcS!89OOOOOOrc3&K|] }|dV dS)rNrPrs rrzmain..s(GGc01GGGGGGrz(pronunciation occurrence inventory driftlockedzJexact Unicode token boundary; longest span first; no substring replacement)rrtransform_versionmatchingentriesreviewed_foreign_inventoryrFr6r7r:rspoken_narration_source_pathspoken_narration_pathspoken_narration_sha256 lexicon_pathlexicon_sha256 inventoryrlreplacement_count#outside_replacement_spans_unchangedreverse_reconstruction_verifiedunresolved_countremaining_ascii_digits) source_textsource_text_sha256tts_texttts_text_sha256rlrr)r completed_atz"pronunciation readback hash failedpronunciation_completed tts_pronunciation_lexicon_sha256 pronunciation)rrreceipt source_sha256 output_sha256rrr updated_at)rrlintro_replacementsr)r8)'ROOTr<loads read_textgetrcexistsr.r/ read_bytesrr]rBrbdecode unicodedata normalizerx INTRO_SOURCErHr\rZr[rMfindallsumTRANSFORM_VERSIONREVIEWED_INVENTORYr=r>datetimenowtimezoneutc isoformatr4r@r'print)* manifest_pathmanifestactivepromotion_path source_path canon_pathtts_assembly_path output_pathr receipt_pathrs promotion source_bytes canon_bytes source_sha canon_sha tts_assembly part_bytesrQr$rawrnarration_textnarration_replacements intro_textrrrrindexrfnarration_count intro_countrequiredrexpected_narrationexpected_introlexicon lexicon_bytesnarration_bytesrrs* rmainrrs 99Mz-1171CCDDH||L!!Y..(,,w2K2K2O2OPW2X2X\g2g2gabbb n %FF#677N 9::K}--Jv&EFF 233K&!<==L&!<==L l;QQ ==?? Qf001HIIPPRR QOvOOPP P Q >33W3EEFFI}}X+--z1J1JRV1V1V8999))++L''))K<((J+&&IY]]#CDDDD U^UbUbcxUyUyHyHy>???:/9979KKLLL!!V++|/?/? /K/KSW/W/WLMMM011Z??>S] * *Kc&kKKLL L# :!y~~j99\IIHIII%%g..KUK00K??9:::-6{K-P-P*N*%.|W%E%E"J"[>::kiLbcfdfcfLgiijjjIJ!/22 / / uP[_del_mPmPm#geHo66>>{KKLLLstMTX]^eXfMfMfc'%/22::<HHIIIlm eGn,M uW~0M.NyYfOgOg Y!/0F S!788;`ae *42DQIabc(yI\I\]dIeIe?f?f"yARARSZA[A[7\7\.SI[E\E\/3   #G&}---o... g&&& )9!:::h{>S>SW^_xWy>y>y?@@@2HX*12K*LH &'3:;K3LH /0)4HWo&4FC^<_#g>W6X !788aad!!H_ !H\ x((( $* SAW=X=XpsuGqHqHelmFeGHHW\ ] ] ]^^^^^r__main__)rrr<r1rHrpathlibr__file__resolver)rrrr[rrr'r4r@rErMrxr__name__rPrrrs  tH~~'**` ".LXcWdqkll(GS^R_lfgg".Q]h\ivpqq(FR]Q^keff+ NBCC'GABB+ NMNNxww,,,aaa333hhh7 7 7 tx^x^x^v zDFFFFFr