3 bF @s<ddlmZddlZddlZddlZddlZyddlmZWn ek r\ddl mZYnXy ddl Z Wn ek rddl m Z YnXej ddkrddl Z ddlmZddlmZeejZdd Zd d Zd d ZddZd&ddZddZddZddZddZddZddZd'd d!Zd"d#Z d$d%Z!dS)()with_statementN) OrderedDict)bz2)qconfig) get_loggercCsd}tjj|\}}tj|tjs6tjd|ddd|dkrPtj|dd}n@|dkrrt j |d d}t |}n|dkrPyt j |d d}Wn8tk rtj\}}}tjd t|ddYnX|j}t|dkrtjd|ddt|dkrtjd|dy|j|d}t |}Wn$tk rLtjdddYnXn@y t|}Wn2tk rtj\}}}tj|ddYnX|S)NzPermission denied accessing T) to_stderrexit_with_code.gz.gzipZrt)mode.bz2.bzip2r.zipzCan't open zip file: )r rz Reading %s: zip archive is emptyz3Zip archive must contain exactly one file. Using %sz=Use python 2.6 or newer to work with contigs directly in zip.) exit_code)r r )rr)r)ospathsplitextaccessR_OKloggererrorgzipopenrBZ2File_read_compressed_filezipfileZipFile Exceptionsysexc_infostrZnamelistlenwarningAttributeErrorIOError exception)fpath fasta_file_extZzfileexc_type exc_valuenamesr1@/home/psgendb/BIRCHDEV/pkg/quast-5.2.0/quast_libs/fastaparser.py_get_fasta_file_handler!s>     r3cCs&tjddkr"tjtj|jS|S)Nrr)r" version_infoio TextIOWrapperBytesIOread)Zcompressed_filer1r1r2rMsrc Cs.y|ddjdStk r(dSXdS)zk Extracts name from fasta entry line: ">chr1 length=100500; coverage=15;" ---> "chr1" rNr)split IndexError)liner1r1r2__get_entry_nameSsr=cCst}d}d}t|}xv|D]n}|jddkr:|jd}n|g}xH|D]@}|sPqF|ddkrv|rl|||<d}t|}qF|t|j7}qFWqW|||<|j|S)ze Takes filename of FASTA-file Returns list of lengths of sequences in FASTA-file rN r>)rr3findr:r=r%stripclose)r* chr_lengthslchr_namer+raw_linelinesr<r1r1r2get_chr_lengths_from_fastafile^s(     rIFc Cs~d}i}i}xft|D]Z\}}|jd}t|}||7}tddt|D||<|rh|t||8}|||<qW|||fS)Nrcss"|]\}}|dkr|dVqdS)NrNr1).0xsr1r1r2 sz#get_genome_stats..) read_fastar:r%set enumerate) fasta_fpathZskip_nsZ genome_sizeZreference_chromosomesZns_by_chromosomesnameseqrFchr_lenr1r1r2get_genome_stats|s  rVcCsTd}d}d}d}|d}g}t|}x|D]}|jdd krJ|jd} n|g} x| D]} | s`qV| ddkr|r|j|||t| jt| g||7}d}d}t| }|t| 7}qV|s| } |t| j7}|t| 7}qVWq,WWdQRX|j|||t| jt| gt|d4} x,|D]$} | jdjdd | Dd qWWdQRXdS) Nrz.fair>rr?w cSsg|] }t|qSr1)r$)rKZfsr1r1r2 sz#create_fai_file.. r@) rrAr:appendr%rBr=writejoin)rRrEZ total_offsetZ chr_offsetrFZ fai_fpathZ fai_fieldsZin_frGrHr<Zchr_lineout_fZfieldsr1r1r2create_fai_files<         r_cCs~tjj|stj|d}xRt|D]F}|ddkr\|r@|jttjj|t|dd}|r$|j|q$W|rz|jdS)z Takes filename of FASTA-file and directory to output Creates separate FASTA-files for each sequence in FASTA-file Returns nothing Oops, similar to: pyfasta split --header "%(seqid)s.fasta" original.fasta Nrr?z.farW) rrisdirmkdirrrCr]r=r\)r*output_dirpathZoutFiler<r1r1r2 split_fastas   rcccsd}g}d}t|}xj|D]b}|jd}xR|D]J}|s8q.|ddkrj|sX|dj|fVd}t|}g}q.|j|jq.WqW|s|r|dj|fV|jdS)zH Generator that returns FASTA entries in tuples (name, seq) Tr9r>rr?FN)r3r:r]r=r[rBrC)r*firstrTrSr+rGrHr<r1r1r2rOs&    rOcCs,g}x"t|D]\}}|j||fqW|S)zB Returns list of FASTA entries (in tuples: name, seq) )rOr[)r*list_seqrSrTr1r1r2read_fasta_one_timesrfcCsht|}g}xD|D]<}|jd}x,|D]$}|s0q&|ddkr&|j|jq&WqW|jdj|}|S)z Returns string r>rr?r9)r3r:r[rBrCr])r*r+rerGrHr<Z fasta_strr1r1r2read_fasta_strs     rgcCsPxJ|D]B\}}td|x,tdt|dD]}t|||dq,WqWdS)Nz>%sr<)printranger%)fastarSrTir1r1r2 print_fastas rmrWcCsjt||}xR|D]J\}}|jd|x2tdt|dD]}|j|||ddq8WqW|jdS)Nz>%s rrhrZ)rr\rjr%rC)r*rkr outfilerSrTrlr1r1r2 write_fastas  "rocCsdddddd|jS)NTAGCrJ)rqrprsrrrJ)upper)Zletterr1r1r2compsrucs*ttdddjfddt|DS)NZ ATCGNatcgnZ TAGCNtagcnr9c3s|]}j|dVqdS)r9N)get)rKZ nucleotide)cr1r2rNszrev_comp..)dictzipr]reversed)rTr1)rwr2rev_compsr{)F)rW)" __future__rrr"rr collectionsr ImportErrorZ%quast_libs.site_packages.ordered_dictrZquast_libs.site_packagesr4r5 quast_libsrquast_libs.logrLOGGER_DEFAULT_NAMErr3rr=rIrVr_rcrOrfrgrmrorur{r1r1r1r2s>     ,  #