Xf*dZddlmZddlmZddlmZddlmZddlm Z ddl m Z ddl m Z dd l mZd d lmZd Zd ZdZdZddZdZdZeeeedZdZddZddZGddeZdS)zBio.SeqIO support for the SnapGene file format. The SnapGene binary format is the native format used by the SnapGene program from GSL Biotech LLC. )datetimesub)unpack) parseString)Seq) SeqFeature)SimpleLocation) SeqRecord)SequenceIteratorc#K |d}t|dkrdStd|d}|d}t|dkrtdtd|d}||}t||krtd|||fV) a?Iterate over the packets of a SnapGene file. A SnapGene file is made of packets, each packet being a TLV-like structure comprising: - 1 single byte indicating the packet's type; - 1 big-endian long integer (4 bytes) indicating the length of the packet's data; - the actual data. Tr Nz>BrzUnexpected end of packetz>I)readlenr ValueError)handle packet_typelengthdatas 4lib/python3.11/site-packages/Bio/SeqIO/SnapGeneIO.py_iteraters*kk!nn {  a   FT;//2 Q v;;??788 8f%%a({{6"" t99v  788 8FD))))*c|jrtdtd|dz z|\}}t|d|_d|jd<|dzr d|jd<d Sd |jd<d S) z}Parse a DNA sequence packet. A DNA sequence packet contains a single byte flag followed by the sequence itself. z*The file contains more than one DNA packetz>B%dsr ASCIIDNA molecule_typecirculartopologylinearN)seqrrrdecode annotations)rrrecordflagssequences r_parse_dna_packetr'7s zGEFFFW 3T::OE8X__W--..FJ*/F' t|2)3:&&&)1:&&&rct|d}t|d}|dkr d|jd<n d|jd<t|d}|rt j|d|jd <t|d }|r||_t|d }|r8|d d d|_||_ |s|j|_dSdSdS)zParse a 'Notes' packet. This type of packet contains some metadata about the sequence. They are stored as a XML string with a 'Notes' root node. UTF-8Type SyntheticSYNdata_file_divisionUNC LastModifiedz%Y.%m.%ddateAccessionNumberComments r rN) rr"_get_child_valuer#rstrptimeidsplitname description)rrr$xmltyper0acccomments r_parse_notes_packetr>Is  dkk'** + +C C ( (D {38/0038/0 C 0 0D I%-%6tZ%H%H6" 3 1 2 2C  sJ//G$mmC++A. $ $ FIII $$ $ $rctd|\}}}}|ddkrtddS)zParse a SnapGene cookie packet. Every SnapGene file starts with a packet of this type. It acts as a magic cookie identifying the file as a SnapGene file. z>8sHHHrSnapGenez%The file is not a valid SnapGene fileN)rr"r)rrr$cookieseq_type exp_version imp_versions r_parse_cookie_packetrEfsJ 28$1G1G.FHk; }}W++@AAA,+rFc d|dD\}}|dz }|r |dz }|dz }||kr7t|t||}td||}||z}nt|||}|S)Nc34K|]}t|VdSN)int).0xs r z"_parse_location..rs(77Q#a&&777777r-r )strandr)r7r r) rangespecrNr$ is_primerstartendl1l2locations r_parse_locationrVqs77)//#"6"6777JE3 AIE   q s{{ E3v;;v > > > As6 2 2 27!%V<<< Orc t|d}|dD]}i}t|dd}d}t t|dd}|d krd }d } g} d |d D]}} t| dddkrt| d} dz t | ||} | s| } n|d kr| | z} n| | z} t| d}|r| |g~t| d krD|d krtfd| D} d d| Dg|d<| std|dD] }t|dd}g}|dD]}| dr3|t|j djJ| dr3|t|j dj| dr2|t |j dj|||< t|d}|rd|vr|g|d<n||dvr|g|d<t| ||}|j|d S) zParse a sequence features packet. This packet stores sequence features (except primer binding sites, which are in a dedicated Primers packet). The data is a XML string starting with a 'Features' root node. r)Featurer; misc_featuredefaultr directionality1NrSegmentstandardgapranger8c*g|]\}}|z dz|gS)r )rJir8n_partss r z*_parse_features_packet..s*$U$U$UDgkAot%<$U$U$Ur;c3*K|]\}}|d|VdS):Nre)rJrfr8s rrLz)_parse_features_packet..s0&M&MD!}}d}}&M&M&M&M&M&MrpartszMissing feature locationQzMissing qualifier nameerrorVtextpredefrIlabelr; qualifiers)rr"getElementsByTagName_get_attribute_valuerIrVappendrreversedjoinr hasAttribute_decode attributesvaluer features)rrr$r:featurequalsr;rNr\rUsubpartssegmentrng next_locationr8 qualifierqnameqvaluesr~rgs @r_parse_features_packetrs~ dkk'** + +C++I66D(D(#GV^LLL *:C H H H   Q  F33I>> 1 1G#GVZ@@EII&w88C qLG+C@@M 4(2(83#m3'88D 1$000 x==1  ||#$U$U$U$UH$U$U$UVV!hh&M&MH&M&M&MMMNE'N 9788 8 55c:: # #I(6)AEG"77<< G G%%f--GNN75+;F+C+I#J#JKKKK''11GNN75+;H+E+K#L#LMMMM''..GNN3u'7'>'D#E#EFFF"E%LL#GV44  'e##"&gU7^++"&f XDUCCCw''''ID(D(rc Vt|d}|dD]}i}t|d}|r|g|d<g}|dD]}t|dd} t t|d d } | d krd } nd } t | | |d} t t|dd d k} | r| |vr{|| t| d|} |j| dS)zParse a Primers packet. A Primers packet is similar to a Features packet but specifically stores primer binding features. The data is a XML string starting with a 'Primers' root node. r)Primerr8rs BindingSiterUzMissing binding site locationrn boundStrand0rZr r_T)rP simplified primer_bindrtN) rr"rvrwrIrVrxr r)rrr$r:primerrr8 locationssiterrNrUrrs r_parse_primers_packetrs| dkk'** + +C**844,,#FF33  $"VE'N // >> , ,D&j(GC-dM3OOOPPF{{&sFFdKKKH1$ cRRRSSWXXJ h)33   X & & & " G O " "7 + + + +- ,,,r)r c$tdd|S)Nz<[^>]+>r)rqs rr|r|s y"d # ##rNc||rt|j|jS|rt ||SrH)r{r|r}r~r)noder8r[ros rrwrw sK tt,2333 rc||}|rG|djr:|djj|jkrt |djjS|rt||S)Nr)rv childNodes firstChildnodeType TEXT_NODEr|rr)rr8r[rochildrens rr4r4sx((..H  QK "  QK " +t~ = =x{-2333 rc.eZdZdZfdZdZdZxZS)SnapGeneIteratorzParser for SnapGene files.cPt|dddS)aParse a SnapGene file and return a SeqRecord object. Argument source is a file-like object or a path to a file. Note that a SnapGene file can only contain one sequence, so this iterator will always return a single record. br@)modefmtN)super__init__)selfsource __class__s rrzSnapGeneIterator.__init__&s* cz:::::rc0||}|S)z9Start parsing the file, and return a SeqRecord generator.)iterate)rrrecordss rparsezSnapGeneIterator.parse0s,,v&&rc#Ktd}t|} t|\}}}n#t$rt ddwxYw|dkrt dt ||||D]/\}}}t |}| ||||0|jst d|VdS)z.Iterate over the records in the SnapGene file.Nz Empty file. z5The file does not start with a SnapGene cookie packetzNo DNA packet in file) r rnext StopIterationrrE_packet_handlersgetr!)rrr$packetsrrrhandlers rrzSnapGeneIterator.iterate5s46"" 6(,W %K 6 6 6]++ 5 6 $  TUU UVT6222)0 . . %K&**;77G"f---z 6455 5 s 6A)__name__ __module__ __qualname____doc__rrr __classcell__)rs@rrr#s\$$;;;;; rr)F)NN)rrrerstructrxml.dom.minidomrBio.SeqrBio.SeqFeaturer r Bio.SeqRecordr Interfacesr rr'r>rErVrrrr|rwr4rrerrrs ''''''%%%%%%))))))######((((((***:222$$$$:BBB&L(L(L(^&,&,&,T    $$$     '''''''''''r