Mi SSKrSSKrSSKrSSKrSSKJr SSKJrJrJ r SSK r SSK r SSK JrJr \R "S5rS\S\S\4S jrS \S\\4S jrS rS rSrSS\S\S\ R0S\S\4 SjjrSrS\4SjrS\4SjrS S\ \ R<\\4S\\ S\ \S44Sjjr!S S\\\S\\ S\ \S44Sjjr"S S\\\S\\ S\ \S44Sjjr#S\\\S\4Sjr$Sr%g)!N)Path)ListOptionalUnion)elements_from_jsonelements_to_textzunstructured.evaldocpath output_typereturncUS:Xa[[U55nU$US:Xa [U5nU$[SUS35e![an[R SU35 UeSnAff=f)zu Convert given input document (path) into cct-ready. The function only support conversion from `json` or `txt` file. jsontxtzZFile type not supported. Expects one of `json` or `txt`, but received z instead.zCould not read the file N)rr_read_text_file ValueErrorloggererror)r r output_cctes t/var/www/eduai.edurigo.com/storigo/production/storigo_env/lib/python3.13/site-packages/unstructured/metrics/utils.py_prepare_output_cctrs  & )*_display..UsC ST^^ s!# c3P># UHupURTU5v M g7fr?ljust)rBiheader col_widthss rrDrEWs%XEW  Z]33EWs#&-r7z.3fc3R># UHnTURTU5v M g7fr?rH)rBrJrL formatted_rows rrDrEas*^D]q]1%++JqM::D]s$')r@r1tolistmaxclickechor enumeratesumiterrows isinstancefloatrrArange)r3headersrKr"rowrCrLrOs @@r_displayr\Ms" 2w!|jj!GSZSZCKC CCDSZJ JJsxxXYwEWXXY JJsS_$sc'lQ.>'??@++- D$&&$$Sz3$$SY/   HH^E#mJ\D]^ ^  s1E< directoryr$r3mode overwritecUS;a [S5eU(a[U5RSS9 UR5nSUR;aUSR [ 5US'SUR;a!SUR;aURSS/S9nU(d [X5nUR[RRX5S US US :HS 9 g )z Save the metrics report to tsv file. The function allows an option 1) to choose `mode` as `w` (write) or `a` (append) and 2) to `overwrite` the file if filename existed or not. )waz/Mode not supported. Mode must be one of [w, a].T)exist_okr,r$ connector)by Fra)sepr^indexrKN) rrmkdircopyr1astypeint sort_values_get_non_duplicated_filenameto_csvrrr)r]r$r3r^r_s r_write_to_filerpes :JKK Yt, B"**k((-7 RZZK2::$= ^^ Z8^ 9 / DII  Y)t$eUY]`U`r'c\[R"SU5nU(a[US5$g)z A function that defines the sorting method for duplicated file names. For example, with filename.ext filename (1).ext filename (2).ext filename (10).ext - this function extracts the integer in the bracket and sort those numbers ascendingly. z(\d+)r)refindallrl)r$numberss r _sorting_keyrv}s+jj8,G72;r'c 2URSS5up#S[R"U5S[R"U5S3n[UVs/sH"n[R"XE5(dM UPM$ sn[ S9n/nUHLn[R "SU5n U (dM#UR[U RS555 MN UR5 Sn UHn X:XaU S- n M O US-[U 5-S -U-$s snf) z Checks the duplicity of the file name from the list and run the numerical check of the minimum number needed as extension to not overwrite the exising file. Returns a string of file name in the format of `filename ().ext`. rr7^z(?: \((\d+)\))?\.$)keyz \((\d+)\)z (z).) rsplitrsescapesortedmatchrvsearchrrlgroupsortrA) file_listtarget_filenameoriginal_filename extensionpatternfduplicated_filesrufiler~counternumbers r_uniquity_filers $3#9#9#q#A 299./00A"))IBVAWWXYG)L)Qrxx7Kq)LR^_G  ,- 5 NN3u{{1~. /!  LLNG   qLG   t #c'l 2T 9I EE#Ms D2DcF[[R"U5U5nU$)zg Helper function to calls the `_uniquity_file` function. Takes in directory and file name to check on. )rrr )rr$s rrnrns bjjox8H Or'scoresroundingcv[U5S:Xag[R"U5nU(dU$[X!5$)z Find mean from the list. Returns None if no element in the list. Args: rounding (int): optional argument that allows user to define decimal points. Default at 3. rN)r@ statisticsr)round)rrr)s rr-r-s4 6{a ??6 "D    r'cUVs/sH o"cMUPM nn[U5S::agU(d[R"U5$[[R"U5U5$s snf)z Find standard deviation from the list. Returns None if only 0 or 1 element in the list. Args: rounding (int): optional argument that allows user to define decimal points. Default at 3. Nr7)r@rr*rrrscores rr.r.sX"( =eF = 6{a '' !!&)8 44 > A$A$cUVs/sH o"cMUPM nn[U5S::agU(d[R"U5$[[R"U5U5$s snf)z Find population standard deviation from the list. Returns None if only 0 or 1 element in the list. Args: rounding (int): optional argument that allows user to define decimal points. Default at 3. Nr7)r@rr+rrs rr/r/sX"( =eF = 6{a   (( ""6*H 55 >rc[U5$)z$ Returns the row count of the list. )r@)rs rr0r0s v;r'c[RRU5(d[SUS35e[ USS9nUR 5nSSS5 U$!,(df  W$=f![ an[SUSU35eSnAff=f)z? Reads the contents of a text file and returns it as a string. z The file at z does not exist.ignore)errorsNz+An error occurred when reading the file at z: )rrexistsFileNotFoundErroropenreadOSErrorIOError)rrtextrs rrrs 77>>$  ,tf4D EFFQ $x (A668D) ) ( QCD6A3OPPQs4 A,A A, A)$A,)A,, B 6BB )raT))&loggingrrsrpathlibrtypingrrrrRpandasr9unstructured.staging.baserr getLoggerrrArr&r5r<r\ DataFrameboolrprvrrnSeriesrXrlr-r.r/r0rr'rrs (( J   . /33*CDI$ )/ 2Y]!')||;>QU0  F#F83 !% 4;./ !8C= !QVW\^bWbQc !54(5HSM5%PUW[P[J\5$ 6D%) 6Xc] 65QVX\Q\K] 6 4(UQr'