B b1@sdZddlZddlmZddlmZddlmZddlmZddl m Z ddl m Z dd l m Z dd l mZdd lmZd d lmZd dlmZddddddZdddZddZddZddZddZddZd d!ZGd"d#d#eZGd$d%d%eZdS)&zBio.SeqIO support for the "xdna" file format. The Xdna binary format is generated by Christian Marck's DNA Strider program and also used by Serial Cloner. N)match)pack)unpack)BiopythonWarning)Seq) ExactPosition)FeatureLocation) SeqFeature) SeqRecord)SequenceIterator)SequenceWriterDNARNAprotein)rr linearcircular)rr cCs&||}t||kr"td||S)z9Read the specified number of bytes from the given handle.z Cannot read %d bytes from handle)readlen ValueError)handlelengthdatar/lib/python3.7/site-packages/Bio/SeqIO/XdnaIO.py_read(s   rcCs2tdt|dd}td|t||ddS)zRead a Pascal string. A Pascal string comprises a single byte giving the length of the string followed by as many bytes. z>Br rz%dsASCII)rrdecode)rrrrr _read_pstring0sr!cCs tt|S)N)intr!)rrrr_read_pstring_as_integer:sr#cCs.t|}|dkr&t|t|}||fSdSdS)aRead an overhang specification. An overhang is represented in a XDNA file as: - a Pascal string containing the text representation of the overhang length, which also indicates the nature of the overhang: - a length of zero means no overhang, - a negative length means a 3' overhang, - a positive length means a 5' overhang; - the actual overhang sequence. Examples: - 0x01 0x30: no overhang ("0", as a P-string) - 0x01 0x32 0x41 0x41: 5' AA overhang (P-string "2", then "AA") - 0x02 0x2D 0x31 0x43: 3' C overhang (P-string "-1", then "C") Returns a tuple (length, sequence). r)NNN)r#rabs)rrZoverhangrrr_read_overhang>s r%cCsZxTdd|dDD]<}td|}|r@|\}}|g||<qd|kr|g|d<qWdS)zParse the description field of a Xdna feature. The 'description' field of a feature sometimes contains several GenBank-like qualifiers, separated by carriage returns (CR, 0x0D). cSsg|]}t|dkr|qS)r)r).0xrrr `sz._parse_feature_description.. z^([^=]+)="([^"]+)"?$"ZnoteN)splitrgroups)desc qualifierslinemZqualvaluerrr_parse_feature_descriptionYs   r2cCst|}t|}t|pd}t|}t|}tdt|d\}}} |rLd} nd} ||}}t|t|d|| d} i} |r|g| d<t|| t| || d} |j| d S) zRead a single sequence feature.Z misc_featurez>BBxBrr )strandlabel)typer.N) r!r#rrrr2r featuresappend)rrecordnamer-r6startendZforwardZdisplayZarrowr4locationr.featurerrr _read_featurels$    r?cs0eZdZdZfddZddZddZZS) XdnaIteratorzParser for Xdna files.cstj|ddddS)aParse a Xdna file and return a SeqRecord object. Argument source is a file-like object in binary mode or a path to a file. Note that this is an "iterator" in name only since an Xdna file always contain a single sequence. bZXdna)modeZfmtN)super__init__)selfsource) __class__rrrDs zXdnaIterator.__init__cCs:|d}|stdt|dkr*td|||}|S)z9Start parsing the file, and return a SeqRecord generator.pz Empty file.z2Improper header, cannot read 112 bytes from handle)rrriterate)rErheaderrecordsrrrparses   zXdnaIterator.parseccstd|\}}}}}}|dkr&td|tkr6tdt||d} t||d} | dd} tt| | | | d} t|rt|| jd<|t krt || jd <t | d d krt |t |td t|d d} x| dkrt || | d 8} qW| Vd S) z.Parse the file and generate SeqRecord objects.z>BBB25xII60xI12xrzUnsupported XDNA versionzUnknown sequence typer ) descriptionr:id molecule_typetopologyr z>BN)rr _seq_typesrr r+r r annotations_seq_topologiesrrr%r?)rErrJversionZseq_typerQrZ neg_lengthZ com_lengthZsequencecommentr:r9Z num_featuresrrrrIs,   zXdnaIterator.iterate)__name__ __module__ __qualname____doc__rDrLrI __classcell__rr)rGrr@s r@cs0eZdZdZfddZddZddZZS) XdnaWriterzWrite files in the Xdna format.cstj|dddS)zInitialize an Xdna writer object. Arguments: - target - Output stream opened in binary mode, or a path to a file. wb)rBN)rCrD)rEtarget)rGrrrDszXdnaWriter.__init__c Cs t|}y t|}Wntk r2tddYnXyt|tdWntk r\YnXd|_|jd}|dkr~d}n.d|krd}n d |krd }nd |krd }nd}|jd ddkrd}nd}|j|j r|j}n|j d|j}|j t dd||t |dt |d|j t|j|j |d|j t dd|d|ddd|jD}t |jt |}|dkrtd|dtt |dkrt |d}td|t|dd}|j t dt |x|D] } || jddgdd} xZ| jD]P} | dkr4q"x:| j| D],} t | dkrZ| d} | d| | f} q@Wq"W|| || j| jjjd} | jjj}d}| jjd kr|| } }d}|t| |t||j t d!|ddd|d"qW|jrtd#tdS)$aWrite the specified record to a Xdna file. Note that the function expects a list (or iterable) of records as per the SequenceWriter interface, but the list should contain only one record as the Xdna format is a mono-record format. zMust have one sequenceNzMore than one sequence foundFrPrrr rrrrrQrrrMz>BBB25xII60xI11xBrz>B0cSs0g|](}t|jjtkrt|jjtkr|qSr)r6r=r;rr<)r&frrrr(0sz)XdnaWriter.write_file..z Dropping z features with fuzzy locationsz%Too many features, dropping the last r5)r5Z translationr)z%s="%s"r3z>BBBBz 127,127,127z1Some annotations were truncated to 255 characters)iternext StopIterationr_has_truncated_stringsrSgetrN startswithrOrwriterrbytesseqencode_write_pstringr7warningswarnrr.r6r=r;Zpositionr<r4str)rErKr9rPZseqtyperQrVr7Zdropr>rNZqnamevalr;r<r4rrr write_files               zXdnaWriter.write_filecCsJt|dkrd|_|dd}|jtdt||j|ddS)z*Write the given string as a Pascal string.r_TNz>Br)rrfrrirrl)rEsrrrrmhs   zXdnaWriter._write_pstring)rWrXrYrZrDrrrmr[rr)rGrr\s  r\)rZrnrerZstructrrZBiorZBio.SeqrZBio.SeqFeaturerrr Z Bio.SeqRecordr Z Interfacesr r rRrTrr!r#r%r2r?r@r\rrrr s4             &I