U Ug* @s2ddlZddlZddlZddlmZddlZddlZddlZddlZddl Z ddl Z ddl Z ddl m Z ddlZddlmZddlmmZddejDZe jdejddd d gd d d ggdddgdejddd d gddge jddddgddZe jdejddd d gd d d ggdddgdejddd d gddge jdddgddZe jddddiejfdiej fdiej!fgddZ"e jd ddd!d"ejd#d!ifdiej d$d%ifdiej!ifgd&d'Z#d(d)Z$d*d+Z%d,d-Z&d.d/Z'd0d1Z(d2d3Z)e jdejddd d gd d d ggdddgdejddd d gddge jddddgd4d5Z*e jdejddd d gd d d ggdddgdejddd d gddge jddddgd6d7Z+e jd8d9e j,fd:ej-fgd;d<Z.d=d>Z/d?d@Z0dAdBZ1dCdDZ2dS)EN)Path)is_platform_windowscCsi|]\}}||qSr).0keyvaluerrD/tmp/pip-unpacked-wheel-u1j4u1ka/pandas/tests/io/test_compression.py sr objd翶~Ϛ?9DܜJ?>D)?.4i(@33333@̤AXYZcolumnsnamemethodZ to_pickleZto_jsonto_csvc Csr|dkrddd}tN}t||||dtj|}t|||ddtj|}||ksdtW5QRXdS)Ntarw:gzrmode compressiontm ensure_cleangetattrospathgetsizeAssertionError)r rcompression_onlyr&compressed_sizeuncompressed_sizerrrtest_compression_sizes     r,c CstV}tj||dkrdnd|d"}t|||j|jjrDtW5QRXtj |}W5QRXtV}tj|ddd"}t|||j|jjrtW5QRXtj |}||kstW5QRXdS)Nrrwr) r"r#icom get_handler$handleclosedr(r%r&r')r rr)r&handlesr*r+rrrtest_compression_size_fh1s   r3z'write_method, write_kwargs, read_methodindexFc Csttjdddgdddggddd gd }t|}td |$}t|||f||||d }W5QRXt||dS) Ng?rg333333 @rrrr compressedr)pd DataFrame_compression_to_extensionr"r#r$assert_frame_equal) write_method write_kwargs read_methodr)input extensionr&outputrrr,test_dataframe_compression_defaults_to_inferNs $rCz1write_method,write_kwargs,read_method,read_kwargsT)r4headersqueezetypZseriesc Cstjddddgdd}t|}td|^}t|||f|d|krr|}|d=||fd |i|d } n||fd |i|} W5QRXtj| |d d dS) Nrr6 rrr8rEr rF)Z check_names) r9Seriesr;r"r#r$copyrEZassert_series_equal) r=r>r?Z read_kwargsr)r@rAr&kwargsrBrrr)test_series_compression_defaults_to_inferbs rLc Cstjddddgdddggdd d gd }tH}tj|d |d ,}tt|j|j |d W5QRXW5QRXW5QRXdS)Nr r r rrrrrrrrr-r) r9r:r"r#r.r/Zassert_produces_warningRuntimeWarningrr0)r)dfr&r2rrrtest_compression_warning}s  rOc Cst}tP}t|dd }|j|d|d|dW5QRXt|tj|d|dW5QRXt }|j|d|d|dt|tj|d|ddS)z? Binary file handles support compression. GH22555 wb)r)rr r)Z index_colr N) r" makeDataFramer#openrseekr<r9read_csvioBytesIO)r)rNr&filerrrtest_compression_binarys   rXc Cspt}ddd}tL}t|}|j||dtd|}|j||d||ksbtW5QRXdS)z Gzip should create reproducible archives with mtime. Note: Archives created with different filenames will still be different! GH 28103 gziprmtimerr7N) r"rQr#rrtimesleep read_bytesr()rNcompression_optionsr&rBrrr#test_gzip_reproducibility_file_names   racCsht}ddd}t}|j||dd|}tdt}|j||dd||ksdtdS)zL Gzip should create reproducible archives with mtime. GH 28103 rYrZr[rP)r rr7N) r"rQrUrVrgetvaluer]r^r()rNr`bufferrBrrr%test_gzip_reproducibility_file_objects  rdcCs&td}tjtjd|gtjddS)z6Tests if import pandas works when lzma is not present.zT import sys sys.modules['lzma'] = None import pandas -cstderrNtextwrapdedent subprocess check_outputsys executablePIPEcoderrrtest_with_missing_lzmasrrcCs&td}tjtjd|gtjddS)z\Tests if RuntimeError is hit when calling lzma without having the module available. z import sys import pytest sys.modules['lzma'] = None import pandas as pd df = pd.DataFrame() with pytest.raises(RuntimeError, match='lzma module'): df.to_csv('foo.csv', compression='xz') rerfNrhrprrrtest_with_missing_lzma_runtimes rsc CsftT}t|||ddtj|}t|||ddddtj|}||ksXtW5QRXdS)NrYrrZr compresslevelr!)r rr&Zcompressed_size_defaultZcompressed_size_fastrrrtest_gzip_compression_levels   rvc Cs0t}t|||ddddW5QRXdS)zGH33196 bzip needs file size > 100k to show a size difference between compression levels, so here we just check if the call works when compression is passed as a dict. bz2rZrtrN)r"r#r$)r rr&rrrtest_bzip_compression_levels rxzsuffix,archive.zipz.tarc CsVtj|d@}||dW5QRXtjtddt|W5QRXW5QRXdS)Nfilenamer-zZero files foundmatch)r"r#pytestraises ValueErrorr9rT)suffixarchiver&rrrtest_empty_archive_zips   rc CsptjddZ}t|d}|dd|ddW5QRXtjtddt |W5QRXW5QRXdS) Nryrzr-a.csvzfoo,barb.csvz Multiple files found in ZIP filer|) r"r#zipfileZipFilewritestrr~rrr9rT)r&rWrrrtest_ambiguous_archive_zip#s  rc Cs|d}t|d}|dW5QRX|d}t|d}|dW5QRX|d}t|d}||d||dW5QRXtjtddt |W5QRXdS)Nrr-zfoo,bar rz archive.tarz#Multiple files found in TAR archiver|) rRwritetarfileTarFileaddr~rrr9rT)Ztmp_pathZcsvAPathaZcsvBPathbZtarpathrrrrtest_ambiguous_archive_tar,s   rc Cstjdd}tjddggddgdj|dd d d d t|j}tj|d R}| }t |dksjt | |d d}trd}nd}||kst W5QRXW5QRXW5QRXdS)Nz.foorz12ZfoobarrrrrF)r r4)fileobjrZrutf8zfoo,bar 1,2 z foo,bar 1,2 )r"r#r9r:rrYrRrr getmemberslenr( extractfilereaddecoder)rWZ uncompressedrmemberscontentexpectedrrr!test_tar_gz_to_different_filename=s$ rc Cs2t }tj|ddW5QRXW5QRXdS)Nr-)rr)rUrVr.Z _BytesTarFile)rcrrrtest_tar_no_error_on_closeQs r)3rYrUr%pathlibrrkrmrrir]rr~Z pandas.compatrZpandasr9Zpandas._testingZ_testingr"Zpandas.io.commoncommonr.Zextension_to_compressionitemsr;markZ parametrizer:rIr,r3rT read_jsonZ read_picklerCrLrOrXrardrrrsrvrxrrrrrrrrrrrs