o Rŀg3@s|dZddlZddlZddlZddlmZddlmZddlm Z ddlm Z ddl m Z dd Z d d Zd d ZddZdS)aBio.SeqIO support for the Graphical Fragment Assembly format. This format is output by many assemblers and includes linkage information for how the different sequences fit together, however, we just care about the segment (sequence) information. Documentation: - Version 1.x: https://gfa-spec.github.io/GFA-spec/GFA1.html - Version 2.0: https://gfa-spec.github.io/GFA-spec/GFA2.html N)BiopythonWarning) as_handle)_UndefinedSequenceData)Seq) SeqRecordcCs|D]j}|dddkr>t|dkrtt|dd|_qt|ddt|kr=td|dddt|dtq|ddd krltt |  }| |ddkrltd |ddd|dtqdS) z:Check a segment line's tags for inconsistencies (PRIVATE).NLNrz,Segment line has incorrect length. Expected z but got .SHz.Segment line has incorrect checksum. Expected ) lenrint_datawarningswarnrhashlibsha256strencode hexdigestupper)seqtagstagchecksumrC/var/www/html/myenv/lib/python3.10/site-packages/Bio/SeqIO/GfaIO.py _check_tagss& rcCsTi}|D]"}|d}t|dkrtd|dtd|ddur0td|dd td|d d|d <|d |d f||d<|d d vrYtd |d tq|d dkrvtd|d durvtd|d dtq|d dkrtd|d durtd|d dtq|d dkrtd|d durtd|d dtq|d dkrtd|d durtd|d dtq|d dkrtd|d durtd|d dtq|d dkr td|d dur td|d dtq|d dkr'td |d dur'td!|d dtq|S)"z>Build an annotations dictionary from a list of tags (PRIVATE).:zSegment line has invalid tag: r z[A-Za-z][A-Za-z0-9]rNzTag has invalid name: z. Are they tab delimited?rAifZJHBzTag has invalid type: Az[!-~]z:Tag has incorrect type. Expected printable character, got iz [-+]?[0-9]+z5Tag has incorrect type. Expected signed integer, got fz&[-+]?[0-9]*\.?[0-9]+([eE][-+]?[0-9]+)?z,Tag has incorrect type. Expected float, got Zz[ !-~]+z7Tag has incorrect type. Expected printable string, got JzQTag has incorrect type. Expected JSON excluding new-line and tab characters, got Hz [0-9A-F]+z?Tag has incorrect type. Expected byte array in hex format, got Bz3[cCsSiIf](,[-+]?[0-9]*\.?[0-9]+([eE][-+]?[0-9]+)?)+zBTag has incorrect type. Expected array of integers or floats, got ) splitr ValueErrorre fullmatchrrrjoin)r annotationsrpartsrrr_tags_to_annotations.sv         $r0ccst|e}|D]Y}|dkrtdtq|dd}|ddkr$qt|dkr2td|d|d d kr?td dd }nt|d }|dd }t ||t |}t ||d |d |dVqWd d S1smwYd S)z`Parser for GFA 1.x files. Documentation: https://gfa-spec.github.io/GFA-spec/GFA1.html  GFA data has a blank line. rSrz1Segment line must have name and sequence fields: r r*Nlengthr idnamer.) rrrrstripr)r r*rrr0rsourcehandlelinefieldsrrr.rrr Gfa1Iteratorqs,         "rAc cst|~}|D]r}|dkrtdtq|dd}|ddkr$qt|dkr2td|dzt|d WntyJtd |dd w|d d krXt d dd}nt |d }|dd }t ||t |}t ||d|d|dVqWd d S1swYd S)znParser for GFA 2.0 files. Documentation for version 2: https://gfa-spec.github.io/GFA-spec/GFA2.html r1r2r3rr4z:Segment line must have name, length, and sequence fields: r rz*Segment line must have an integer length: Nrr5r6r r8) rrrrr;r)r r*r rrr0rr<rrr Gfa2Iterators>           "rC)__doc__rr+rBiorBio.FilerBio.Seqrr Bio.SeqRecordrrr0rArCrrrrs      C