:>"^- @smdZddlmZddlmZmZddlZddlmZm Z ddl m Z ddl m Z ddlmZdd lmZmZmZdd lmZdd lmZdejd ejd ejdejdejiZddd diZddZddZddZ ddZ!ddZ"ddZ#ddZ$Gd d!d!eZ%dS)"zBio.SeqIO support for the "xdna" file format. The Xdna binary format is generated by Christian Marck's DNA Strider program and also used by Serial Cloner. )match)packunpackN)AlphabetBiopythonWarning) _read_header)Seq)SequenceWriter) SeqFeatureFeatureLocation ExactPosition) SeqRecord) as_handlelinearcircularcCs5|j|}t||kr1td||S)z9Read the specified number of bytes from the given handle.z Cannot read %d bytes from handle)readlen ValueError)handlelengthdatar5/tmp/pip-build-ww9dw3qa/biopython/Bio/SeqIO/XdnaIO.py_read&srcCsCtdt|dd}td|t||djdS)zRead a Pascal string. A Pascal string comprises a single byte giving the length of the string followed by as many bytes. z>Brrz%dsASCII)rrdecode)rrrrr _read_pstring.sr cCstt|S)N)intr )rrrr_read_pstring_as_integer8sr"cCs?t|}|dkr7t|t|}||fSdSdS)aRead an overhang specification. An overhang is represented in a XDNA file as: - a Pascal string containing the text representation of the overhang length, which also indicates the nature of the overhang: - a length of zero means no overhang, - a negative length means a 3' overhang, - a positive length means a 5' overhang; - the actual overhang sequence. Examples: - 0x01 0x30: no overhang ("0", as a P-string) - 0x01 0x32 0x41 0x41: 5' AA overhang (P-string "2", then "AA") - 0x02 0x2D 0x31 0x43: 3' C overhang (P-string "-1", then "C") Returns a tuple (length, sequence). rN)NN)r"rabs)rrZoverhangrrr_read_overhang<s    r$cCs{xtdd|jdDD]V}td|}|rZ|j\}}|g||^s z._parse_feature_description.. z^([^=]+)="([^"]+)"?$"ZnoteN)splitrgroups)desc qualifierslinemZqualvaluerrr_parse_feature_descriptionWs# r1cCst|}t|}t|p'd}t|}t|}tdt|d\}}} |rrd} nd } ||}}t|t|d|d| } i} |r|g| dBBxBrrstrandlabeltyper-N) r r"rrr r1r featuresappend)rrecordnamer,r4startendZforwardZdisplayZarrowr2locationr-featurerrr _read_featurejs$    !     r>c cst|dl}t|d}td|\}}}}}}|dkrZtd|tkrrtdt||jd}t||jd} | jdd} tt |t|d | d | d | } |t krt || j d BBB25xII60xI12xrzUnsupported XDNA versionzUnknown sequence typer descriptionr9idtopologyrz>BN)rrrr _seq_typesrrr*r r_seq_topologies annotationsrrr$r>) rheaderversionr4rDrZ neg_lengthZ com_lengthsequencecommentr9r8Z num_featuresrrr XdnaIterators.    (    rLc@s.eZdZdZddZddZdS) XdnaWriterzWrite files in the Xdna format.c Cs|stdt|dkr0td|d}d|_tj|jj}t|tjrsd}n<t|tj rd}n!t|tj rd}nd}|j j dd d krd}nd}|j j|jr|j }nd j|j|j }|jjtd d||t|dt|d |jjt|jjd|jj|jd|jjtdd|jd|jddd|jD}t|jt|}|dkr tjdj|tt|d krVt|d }tjdj|t|dd }|jjtdt|xa|D]Y} |j| jj ddgdd} xh| jD]]} | dkrqxE| j| D]6} t| dkr| d} | d| | f} qWqW|j| |j| j| jjjd} | jj j}d}| jj!dkr|| } }d}|jt| |jt||jjtd|ddd|jdq|W|jrtjdtdS) zWrite the specified record to a Xdna file. Note that the function expects a list of records as per the SequenceWriter interface, but the list should contain only one record as the Xdna format is a mono-record format. zMust have one sequencerzMore than one sequence foundrFrrrDrrz{} {}z>BBB25xII60xI11xBrz>B0cSsFg|]<}t|jjtkrt|jjtkr|qSr)r4r<r:r r;)r%frrrr' s z)XdnaWriter.write_file..z)Dropping {} features with fuzzy locationsz'Too many features, dropping the last {}Nr3 translationr(z%s="%s"z>BBBBz 127,127,127z1Some annotations were truncated to 255 characters)zlabelz translationr5)"rr_has_truncated_stringsrZ_get_base_alphabetseqZalphabet isinstanceZ DNAAlphabetZ RNAAlphabetZProteinAlphabetrGgetrB startswithrCformatrwriterstrencode_write_pstringr6warningswarnrr-r4r<r:positionr;r2)selfrecordsr8ZalptypeZseqtyperDrKr6Zdropr=rBqnamevalr:r;r2rrr write_files             "       #    "  zXdnaWriter.write_filecCsgt|dkr+d|_|dd}|jjtdt||jj|jddS)z*Write the given string as a Pascal string.rNTNz>Br)rrSrrYrr[)r`srrrr\Es  zXdnaWriter._write_pstringN)__name__ __module__ __qualname____doc__rdr\rrrrrMs  yrM)&rirerstructrrr]ZBiorrZ Bio._utilsrZBio.SeqrZBio.SeqIO.Interfacesr ZBio.SeqFeaturer r r Z Bio.SeqRecordr ZBio.FilerZgeneric_alphabetZ generic_dnaZ generic_rnaZgeneric_proteinrErFrr r"r$r1r>rLrMrrrr s0          & 9