o Rŀg0@sdZddlZddlmZddlmZddlmZddlmZddl m Z ddl m Z dd l m Z dd l mZdd lmZd d lmZd dlmZddddddZdddZddZddZddZddZddZd d!ZGd"d#d#eZGd$d%d%eZdS)&zBio.SeqIO support for the "xdna" file format. The Xdna binary format is generated by Christian Marck's DNA Strider program and also used by Serial Cloner. N)match)pack)unpack)BiopythonWarning)Seq) ExactPosition) SeqFeature)SimpleLocation) SeqRecord)SequenceIterator)SequenceWriterDNARNAprotein)rr linearcircular)rr cCs&||}t||krtd||S)z9Read the specified number of bytes from the given handle.z Cannot read %d bytes from handle)readlen ValueError)handlelengthdatarD/var/www/html/myenv/lib/python3.10/site-packages/Bio/SeqIO/XdnaIO.py_read's   rcCs2tdt|dd}td|t||ddS)zRead a Pascal string. A Pascal string comprises a single byte giving the length of the string followed by as many bytes. >Br rz%dsASCII)rrdecode)rrrrr _read_pstring/sr"cCs tt|S)N)intr")rrrr_read_pstring_as_integer9s r$cCs*t|}|dkrt|t|}||fSdS)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)NN)r$rabs)rroverhangrrr_read_overhang=s r'cCsVdd|dDD]}td|}|r|\}}|g||<q d|vr(|g|d<q dS)zParse the description field of a Xdna feature. The 'description' field of a feature sometimes contains several GenBank-like qualifiers, separated by carriage returns (CR, 0x0D). cSsg|] }t|dkr|qS)r)r).0xrrr _sz._parse_feature_description.. z^([^=]+)="([^"]+)"?$"noteN)splitrgroups)desc qualifierslinemqualvaluerrr_parse_feature_descriptionXs    r6cCst|}t|}t|p d}t|}t|}tdt|d\}}} |r&d} nd} ||}}t|t|d|| d} i} |rC|g| d<t|| t| || d} |j| d S) zRead a single sequence feature. misc_featurez>BBxBrr )strandlabel)typer1N) r"r$rrr r6rfeaturesappend)rrecordnamer0r;startendforwarddisplayarrowr9locationr1featurerrr _read_featureks$    rGc0eZdZdZfddZddZddZZS) 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. bXdna)modefmtNsuper__init__)selfsource __class__rrrPs zXdnaIterator.__init__cCs:|d}|s tdt|dkrtd|||}|S)z9Start parsing the file, and return a SeqRecord generator.pz Empty file.z2Improper header, cannot read 112 bytes from handle)rrriterate)rQrheaderrecordsrrrparses   zXdnaIterator.parseccstd|\}}}}}}|dkrtd|tvrtdt||d} t||d} | dd} tt| | | | d} t|rHt|| jd<|t vrSt || jd <t | d d krt |t |td t|d d} | dkrt || | d 8} | dksr| Vd S) z.Parse the file and generate SeqRecord objects.z>BBB25xII60xI12xrzUnsupported XDNA versionzUnknown sequence typer  ) descriptionr?id molecule_typetopologyr rN)rr _seq_typesrr!r.r r annotations_seq_topologiesrrr'rG)rQrrWversionseq_typer^r neg_length com_lengthsequencecommentr?r> num_featuresrrrrVs2  zXdnaIterator.iterate)__name__ __module__ __qualname____doc__rPrYrV __classcell__rrrSrrIs   rIcrH) 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)rLNrN)rQtargetrSrrrPszXdnaWriter.__init__c Cst|}zt|}Wn tytddwzt|tdty(Ynwd|_|jd}|dur9d}nd|vr@d}nd |vrGd }n d |vrNd }nd}|jd ddkr\d}nd}|j|j ri|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 ||D]} || jddgdd} | jD](} | dvrq| j| D]} t | dkr | d} | | d| d } qq|| || jt| jjd} t| jj}d}| jjd!krW|| } }d}|t| |t||j t d"|ddd|d#q|jrtd$tdS)%aWrite the specified record to a Xdna file. Note that the function expects a list (or iterable) of records as per the SequenceWriter interface, but the list should contain only one record as the Xdna format is a mono-record format. zMust have one sequenceNzMore than one sequence foundFr]rrr rrrrr^rrrZz>BBB25xII60xI11xBr r0cSs,g|]}t|jjtrt|jjtr|qSr) isinstancerEr@rrA)r(frrrr*.s  z)XdnaWriter.write_file..z Dropping z features with fuzzy locationsz%Too many features, dropping the last r:)r: translationr+z="r,r8z>BBBBz 127,127,127z1Some annotations were truncated to 255 characters)iternext StopIterationr_has_truncated_stringsr`getr[ startswithr\rwriterrbytesseqencode_write_pstringr<warningswarnrr1r;r#rEr@rAr9str)rQrXr>r]seqtyper^rgr<droprFr[qnamevalr@rAr9rrr write_files                   zXdnaWriter.write_filecCsJt|dkrd|_|dd}|jtdt||j|ddS)z*Write the given string as a Pascal string.rqTNrr )rrzrr}rr)rQsrrrrgs  zXdnaWriter._write_pstring)rirjrkrlrPrrrmrrrSrrns   rn)rlrrerstructrrBiorBio.SeqrBio.SeqFeaturerrr Bio.SeqRecordr Interfacesr r r_rarr"r$r'r6rGrIrnrrrrs8             &I