rhruUddlZddlZddlZddlZddlZddlZddlZddlZddlZddl m Z ddl m Z ddlm Z ddl mZmZmZmZmZmZmZddlmZddlmZdd lmZmZmZdd lmZmZm Z m!Z!dd lm"Z"m#Z#dd l$m%Z%m&Z&m'Z'dd l(m)Z)ddl*m+Z+ddl,m-Z-erddl.m/Z/ej`e1Z2dZ3dZ4gdZ5ee6e7d<dZ8dZ9dZ:dZ;dZddddddddd d!d"e?d#ee?e fd$e?d%ee?d&ee@d'eeee?e?fd(eeee?e?fd)ee6d*e@d+e6fd,ZAGd-d.ejZCeeefZDGd/d0ZEd1eEd d!d"e?d$e?d%e?f d2ZFd1eEdeeeCeeDffd3ZGd4eDddfd5ZHd6eeDd d!d"e?d$e?d%e?ddf d7ZId6eeDd d!d"e?d$e?d%e?ddf d8ZJd6eeDd d!d"e?d$e?d%e?ddf d9ZKGd:d;eZLd4eDdeLfd<ZMd=d>deeDfd?ZNd=d>d@e6deeDfdAZOdBe?ddfdCZPy)DN)datetime)Path)Lock) TYPE_CHECKINGAnyDictListOptionalTupleUnion)quote) constants)CommitOperationAdd UploadInfo_fetch_upload_modes)LocalUploadFileMetadataLocalUploadFilePathsget_local_upload_pathsread_upload_metadata)DEFAULT_REVISION REPO_TYPES)DEFAULT_IGNORE_PATTERNSfilter_repo_objectstqdm) _format_size)is_xet_available) sha_fileobj)HfApi d) 2Kr!}iiXiCOMMIT_SIZE_SCALEii'r#r" paths_listreturnctjdt|tkDr+tj dt|ddtddddlm}|d}|D]}t|j}|j}t|jd k7rt|jnd }||d xxd z cc<t|d d D]7\}}|dk(rd ndj|d |}||dj|9|jD]]\} } | d } t| d} | | z} | t kDs(| d k(rdn| }tj d|d| dd| dd| ddt dd _g}g}|D]y}|j"j%j&}|dz }|t(kDr|j+|j|fS|t,kDs]|j+|j|f{|rkdjd|d dD}t|dkDrdt|dz dnd}tj dt|dt(d ||d!|rkdjd"|d dD}t|dkDrdt|dz dnd}tj dt|d#t,d$||d%tjd&y )'au Validate upload against repository limits and warn about potential issues. Args: paths_list: List of file paths to be uploaded Warns about: - Too many files in the repository (>100k) - Too many entries (files or subdirectories) in a single folder (>10k) - Files exceeding size limits (>20GB recommended, >50GB hard limit) z/Running validation checks on files to upload...zYou are about to upload ,z. files. This exceeds the recommended limit of z files per repository. Consider: - Splitting your data into multiple repositories - Using fewer, larger files (e.g., parquet files) - See: https://huggingface.co/docs/hub/repositories-recommendationsr) defaultdictcdtdS)Nr)filessubdirs)setw/var/www/html/ai-insurance-compliance-backend/venv/lib/python3.12/site-packages/huggingface_hub/_upload_large_folder.pyz)_validate_upload_limits..XsqUXUZ=[r4.r0rN/r1rootzFolder 'z ' contains z entries (z files and z/ subdirectories). This exceeds the recommended z< entries per folder. Consider reorganising into sub-folders.iʚ;z - c34K|]\}}|d|ddywz: z.1fGBNr3.0pathsizes r5 z*_validate_upload_limits..s#!ajdDTF"T#Jb"9!az ... and z more filesFound z files exceeding the zGB hard limit: - zM These files may fail to upload. Consider splitting them into smaller chunks.c34K|]\}}|d|ddywr<r3r>s r5rBz*_validate_upload_limits..s#!\jdDTF"T#Jb"9!\rCz files larger than zGB (recommended limit): - z2 Large files may slow down loading and processing.zValidation checks complete.)loggerinfolenMAX_FILES_PER_REPOwarning collectionsr.r path_in_repopartsstrparent enumeratejoinadditemsMAX_FILES_PER_FOLDER file_pathstatst_sizeMAX_FILE_SIZE_GBappendRECOMMENDED_FILE_SIZE_GB)r*r.entries_per_folderpathsr@rOrQichildfolderdata file_count subdir_count total_entriesfolder_display large_filesvery_large_filesrAsize_gb files_strmore_strs r5_validate_upload_limitsrl;s\ KKAB :++&s:q&9:55G4JKT U ()45[)\ =E&&' &)%5%<T[[!#6"7+q0+"%*- =HAuFSr(;F v &y 1 5 5e < = =+002  '] 4 ?+ "\1 / /'-}V&N NN>*+mA5FGq>\!,<=00DQ/GH::  K>##%--& % %  # #U%7%7$A B / /    2 2G< =>MM!aL\]_^_L`!aa LOP`LadeLe\#&6"7!";!.s1idZ^ZfZfZh  + & / / 1is=$=z**/*)rprqrFz candidate files to uploadzStarting upload...zRecovering from metadata files)desc)statusrurvrxrn)targetkwargsz zIs done: exiting main loopzUpload is complete!), ValueErrorrrr expanduserresolveis_dir isinstancerPros cpu_countmax create_reporHrIrvr repo_info xet_enabledUPLOAD_BATCH_SIZE_XETUPLOAD_BATCH_SIZE_LFSrglobrrJrlrrrNLargeUploadStatusrange threadingThread _worker_jobstartprintcurrent_reporttimesleep_print_overwriteis_doneloggingrS)rurvrwrxrnrorprqrrrsrtnb_coresrepo_urlis_xet_enabledupload_batch_sizefiltered_paths_listrelpathr*r^rUr_threadsthreadlast_report_tss ` r5upload_large_folder_internalrsE& I   "=j\JKK#{#..088:K    +K=8LMNN OS )*+..O<<>&Q(Q,* w)W_cdH KK. +,G   MM  +2@-EZ.ik>N>Nv>Vi%' Obb7(g>bJb KK&Z))CDEJ' KK$% *+KL   $[%2D2DEF E  u&7 8F{#    "&$   G   fv,,../YY[N  1 99; '+= = !6!6!89!YY[N >>  LL5 6    KK%%'( LL&'gc  sMM =(MceZdZejZejZejZejZejZ y) WorkerJobN) __name__ __module__ __qualname__enumautoSHA256GET_UPLOAD_MODE PREUPLOAD_LFSCOMMITWAITr3r4r5rrsC TYY[FdiikODIIKM TYY[F 499;Dr4rcdeZdZdZddeedefdZdefdZde ded e dd fd Z de fd Z de fd Zy )rzBContains information, queues and tasks for a large upload process.rUrc||_tj|_tj|_tj|_tj|_t|_d|_ d|_ d|_ ||_ d|_ d|_d|_t!j"|_d|_t|_|jD]}|\}}|j*|jj-|0|j.|jj-|X|j.dk(r(|j0s|j j-||j2s|j j-|t4j7d|j8dy)NrrlfszSkipping file z! (already uploaded and committed))rUqueueQueue queue_sha256queue_get_upload_modequeue_preupload_lfs queue_commitrlocknb_workers_sha256nb_workers_get_upload_modenb_workers_preupload_lfsrnb_workers_commitnb_workers_waitinglast_commit_attemptrnow _started_at _chunk_idx _chunk_locksha256put upload_mode is_uploaded is_committedrHdebugrN)selfrUritemr^metadatas r5__init__zLargeUploadStatus.__init__%sc 7<{{}@E ">Ckkm 7<{{}F &'/0'-.%&7&''(48 #<<> !%JJ eD"OE8&!!%%d+%%-**..t4%%.x7K7K((,,T2**!!%%d+ ~e.@.@-AAbcd er4r+cl|j5t|jcdddS#1swYyxYwr)rr(rrs r5 target_chunkzLargeUploadStatus.target_chunkGs,    6$T__5 6 6 6s*3successnb_itemsdurationNc |j5|s/tjd|d|xjdzc_nI|t|jk\r3|dkr.tj d|d|xjdz c_t dt|jttdz |_dddy#1swYyxYw)NzFailed to commit z9 files at once. Will retry with less files in next batch.r(zSuccessfully committed z. at once. Increasing the limit for next batch.r) rrHrLrr(rIrminrJ)rrrrs r5 update_chunkzLargeUploadStatus.update_chunkKs    W!28*O:PST:T%UVDO W W Ws B0CCc d}d}d}d}d}d}d}d}d} d} d} |j5|jD]\} } | jr| dz } | | jz } | dz } | j|dz }|| jz }| j dk(r|dz }| j |dz }| j r|dz }|| jz }| js|dz }|| jz }t| }tj}|jd}||jz }t|jdd}d}|d|d |d z }|d z }|d z }|d |d| d t|d|d z }|d|d|d t|d|d z }|dkDr |d|dz }|d|d| d t|d|d z }|d| dz }|dz }|d|jdz }|d|j dz }|d|j"dz }|d|j$dz }|d|j&dz }|dz }|cdddS#1swYyxYw)z >G *4+J+J*K3O OG )F)F(GsK KG d&<&<%=SA AG 4#:#:";2> >G x G_/ / / sB*H4EH44H=c~|j5td|jDcdddS#1swYyxYw)Nc3VK|]!\}}|jxs |j#ywrrrr?rrs r5rBz,LargeUploadStatus.is_done..s(e;1hx,,F0F0FFe'))rallrUrs r5rzLargeUploadStatus.is_dones7 YY feZ^ZdZdee f f fs3<)r)rrr__doc__r JOB_ITEM_TintrrboolfloatrrPrrr3r4r5rr"snL ed:. e3 eD6c6 WD WC W5 WT W==~ffr4rrcP d}t|}|y|\}}|tjk(rX|d} t||jj ||j5|xjdzc_dddnT|tjk(r t!||||||D]}|\} } | j"r| j$dk(r|j&j |@| j$d k(r|j(j |k|jj ||j5|xj*dzc_dddnz|tj,k(r` t/||||||D]}|j(j | |j5|xj0dzc_dddn|tj2k(rt5j4} d} t7|||||t5j4| z }|j9| t;|||j5t5j4|_|xj>dzc_dddnV|tj@k(rCt5jBtD|j5|xjFdzc_#ddd#t $rt$rR} tjd| tj|jj |Yd} ~ d} ~ wwxYw#1swYrxYw#t $rt$r7} tjd| tjYd} ~ d} ~ wwxYw#1swYxYw#t $rt$rY} tjd | tj|D]}|j&j |Yd} ~ fd} ~ wwxYw#1swYExYw#t $rt$r[} tjd | tj|D]}|j(j |d } Yd} ~ }d} ~ wwxYw#1swYxYw#1swYxYw) a Main process for a worker. The worker will perform tasks based on the priority list until all files are uploaded and committed. If no tasks are available, the worker will wait for 10 seconds before checking again. If a task fails for any reason, the item(s) are put back in the queue for another worker to pick up. Read `upload_large_folder` docstring for more information on how tasks are prioritized. TNrzFailed to compute sha256: r)rurvrxrnzFailed to get upload mode: rregularzFailed to preupload LFS: zFailed to commit: F)$_determine_next_jobrr_compute_sha256rrKeyboardInterrupt ExceptionrHerror traceback format_excrrrr_get_upload_moderrrrrr_preupload_lfsrrr_commitrrJrrrrWAITING_TIME_IF_NO_TASKSr)rrurvrxrnnext_jobjobrUrerrstart_tsrrs r5rrs/ AE'v.    U )"" "8D .%,,006 .((A-( . .I-- - ' CI`hi ;" 8))''50..2248))Y6''++D10044T: ; 711Q61 7 7I++ + 9u#w)^fg!2D''++D12 5//14/ 5 5I$$ $yy{HG 39W_`yy{X-H   UX > .-1YY[*((A-( . .INN " JJ/ 0 /))Q.) /u %  . 9!=>$$&##''-- .  . . %  ' :1#>?$$&& '( 7 7%  9 8<=$$&!9D..22489 9  5 5%   1!56$$&!2D''++D12   . .  / /s&K"L?M <N02N#/P9P/R8RL<*AL77L<?M N,NNN #P5AP  PPR 0ARR RR%c.|j5|jdk(r|jjdkDr|jt j |jz dkDrg|xjdz c_t jdtjt|j|jfcdddS|jdk(r|jjdk\rg|xjdz c_t jdtjt|j|jfcdddS|jjtk\re|xjdz c_t jdtdtjt|jtfcdddS|j j|j"k\rr|j$dk(rc|xj$dz c_t jd tj&t|j |j"fcdddS|j(jdkDrg|j*dk(rX|xj*dz c_t jd tj,t/|j(fcdddS|jjdkDrl|jdk(r]|xjdz c_t jd tjt|jtfcdddS|j j|j"k\r|j$dk(st0j2sc|xj$dz c_t jd tj&t|j |j"fcdddS|j(jdkDrX|xj*dz c_t jd tj,t/|j(fcdddS|jjdkDr]|xjdz c_t jdtjt|jtfcdddS|j jdkDrc|xj$dz c_t jd tj&t|j |j"fcdddS|jdk(r|jjdkDr|jt j |jz dkDrg|xjdz c_t jdtjt|j|jfcdddS|jdk(r|jjdkDr|j(jdk(r|jjdk(r|j jdk(r|j*dk(r|jdk(rv|j$dk(rg|xjdz c_t jdtjt|j|jfcdddSt5d|j6Drt j9d dddy|xj:dz c_t jdt<dtj>gfcdddS#1swYyxYw)Nri,rz;Job: commit (more than 5 minutes since last commit attempt)zJob: commit (>100 files ready)zJob: get upload mode (>z files ready)z5Job: preupload LFS (no other worker preuploading LFS)z.Job: sha256 (no other worker computing sha256)z:Job: get upload mode (no other worker getting upload mode)zJob: preupload LFSz Job: sha256zJob: get upload modermz-Job: commit (1 min since last commit attempt)z Job: commitc3VK|]!\}}|jxs |j#ywrrrs r5rBz&_determine_next_job..is(a[Q&&@(*@*@@arz.All files have been processed! Exiting worker.zNo task available, waiting... (zs)) rrrqsizerrrHrrr_get_nrrMAX_NB_FILES_FETCH_UPLOAD_MODErrrrrrrrr_get_onerHF_HUB_ENABLE_HF_TRANSFERrrUrIrrr)rs r5rrs i(  $ $ )##))+a/**6 f8886A  $ $ ) $ LLV W$$fV-@-@&BUBUBW&XYi(i( % % *v/B/B/H/H/Jc/Q  $ $ ) $ LL9 :$$fV-@-@&BUBUBW&XY!i(i(& ) ) / / 15S S  - - 2 - LL23Q2RR_` a--vf6R6RTr/st-i(i(2 ' ' - - /63K3K KPVPoPostPt  + +q 0 + LLP Q++VF4N4NPVPhPh-ij9i(i(> & & (1 ,1I1IQ1N  $ $ ) $ LLI J$$hv/B/B&CDEi(i(J ) ) / / 1A 5&:[:[_`:`  - - 2 - LLU V--vf6R6RTr/stQi(i(X ' ' - - /63K3K K  + +q 0 8[8[  + +q 0 + LL- .++VF4N4NPVPhPh-ijci(i(h & & (1 ,  $ $ ) $ LL '$$hv/B/B&CDoi(i(t ) ) / / 1A 5  - - 2 - LL/ 0--vf6R6RTr/st{i(i(@ ' ' - - /! 3  + +q 0 + LL- .++VF4N4NPVPhPh-ijGi(i(N  $ $ )##))+a/**6 f8886A  $ $ ) $ LLH I$$fV-@-@&BUBUBW&XY]i(i(f  $ $ )##))+a/##))+q0,,2249**002a7((A-11Q6//14  $ $ ) $ LL '$$fV-@-@&BUBUBW&XY}i(i(BaTZT`T`a a KKH IGi(i(N  % % * % LL:;S:TTVW XNNB'Si(i(i(sfB:` B ` $A<` *B` A:` A?` B` 9A+` .A0` (A6` (B9` +D` 2` >A`  `rc|\}}|jB|jjd5}t|j |_ddd|j |y#1swYxYw)z1Compute sha256 of a file and save it in metadata.Nrb)rrWopenrhexsave)rr^rfs r5rrys^OE8 __ ! !$ ' 31)!n002HO 3 MM% 3 3s A&&A/rUc `|Dcgc] }t|}}t||||jt|d|jt ||D]N\}}|\}} |j | _|j| _ |j| _ | j|Pycc}w)zmGet upload mode for each file and update metadata. Also receive info if the file should be ignored. rE)safe) additionsrxrvheadersrnendpointN) _build_hacky_operationr_build_hf_headersr r&zip _upload_moder_should_ignorer _remote_oid remote_oidr ) rUrurvrxrnrr$additionr^rs r5r r s ;@@$'-@I@%%'xb) eY/hx'44!)!8!8&22 e AsB+c|Dcgc] }t|}}|j|||||D]\}}d|_|j|ycc}w)z(Preupload LFS files and update metadata.)rvrxrnr$TN)r'preupload_lfs_filesrr  rUrurvrxrnrr$r^rs r5r r si:?@$'-@I@ !x# eAsAc|Dcgc] }t|}}|j||||d|D]\}}d|_|j|ycc}w)zCommit files to the repo.z(Add files using upload-large-folder tool)rvrxrn operationscommit_messageTN)r' create_commitrr r1s r5rrsl:?@$'-@I@A !x $ eAsAceZdZddZy)HackyCommitOperationAddNcnt|jtrt|j|_yyr)rpath_or_fileobjrrPrs r5 __post_init__z%HackyCommitOperationAdd.__post_init__s* d**D 1#&t';';#AIEIIK >> >sAreportc$|dz }tjjtfd|j D}t |D]@}t jjdt jjdBt jj|t jjdt|j dz zt jjy)zPrint a report, overwriting the previous lines. Since tqdm in using `sys.stderr` to (re-)write progress bars, we need to use `sys.stdout` to print the report. Note: works well only if no other process is writing to `sys.stdout`! rc3@K|]}t|zdzyw)rN)rJ)r?lineterminal_widths r5rBz#_print_overwrite..sSt3t9.2Ssz zrr8N) shutilget_terminal_sizecolumnssum splitlinesrsysstdoutwriterJflush)rGnb_linesrrKs @r5rrs dNF--/77NSv?P?P?RSSH8_# $ "# JJVJJSNS1B1B1DR1H-IIJKJJr4)QrrrrrLrQrrr rpathlibrrtypingrrrr r r r urllib.parser rEr _commit_apirrr _local_folderrrrrrrutilsrrrutils._cache_managerrutils._runtimer utils.sharhf_apir getLoggerrrHrrr(r__annotations__rrrKrVrZr\rlrPrrEnumrrrrrrr r rr7r'rrrr3r4r5rcsZ   IIILLvv3EE.,"   8 $!$O49O]/-A(B]/t]/L#"6:7;!% t( t( t(sDy!t(  t( sm t(d^t(U49c>23t(eDIsN34t(#t(t(t(x ')@@ A ufufpj/ j/ j/j/ j/  j/Zj( 1j(huYPTU_P`E`?a6bj(d*D,7SUXdglp, $z*  3 SV be jn  4 # ' C C [^ cg (=0=  0G &-$z2B?+??Z8H?STr4