B bk@sdZddlmZddlmZddlmZddlmZddl m Z ddl m Z ddl m Z d d lmZd d lmZGd d d ejZGdddeZGdddeZdS)aBio.SeqIO support for the "seqxml" file format, SeqXML. This module is for reading and writing SeqXML format files as SeqRecord objects, and is expected to be used via the Bio.SeqIO API. SeqXML is a lightweight XML format which is supposed be an alternative for FASTA files. For more Information see http://www.seqXML.org and Schmitt et al (2011), https://doi.org/10.1093/bib/bbr025 )sax)handler) XMLGenerator)AttributesImpl)Seq) UnknownSeq) SeqRecord)SequenceIterator)SequenceWritercseZdZdZfddZddZddZdd Zd d Zd d Z ddZ ddZ ddZ ddZ ddZddZddZddZddZd d!Zd"d#Zd$d%ZZS)&ContentHandlerz5Handles XML events generated by the parser (PRIVATE).cs>td|_d|_d|_d|_d|_d|_d|_g|_ dS)z&Create a handler to handle XML events.N) super__init__source sourceVersion seqXMLversion ncbiTaxID speciesNamestartElementNSdatarecords)self) __class__1lib/python3.7/site-packages/Bio/SeqIO/SeqXmlIO.pyr s zContentHandler.__init__cCs |j|_dS)z2Set XML handlers when an XML declaration is found.N)startSeqXMLElementr)rrrr startDocument,szContentHandler.startDocumentc Cs$|dkrtd|dk r tdd}x|D]\}}|\}}|dkr|dkrV||_q|dkrf||_q|dkrv||_q|dkrt|} ||_q|d kr||_qtd q.|d kr|d kr|}qtd q.td|q.W|jdkrtdd|j} || krtd||jf|j |_ |j |_ dS)z!Handle start of a seqXML element.)NseqXMLz*Failed to find the start of seqXML elementNz#Unexpected qname for seqXML elementrrrrrz#Unexpected attribute for XML Schemaz)http://www.w3.org/2001/XMLSchema-instanceZnoNamespaceSchemaLocationz0Unexpected attribute for XML Schema in namespacez.Unexpected namespace '%s' for seqXML attributezFailed to find seqXMLversionz#http://www.seqxml.org/%s/seqxml.xsdzDXML Schema '%s' found not consistent with reported seqXML version %s) ValueError RuntimeErroritemsrrrintrrendSeqXMLElement endElementNSstartEntryElementr) rnameqnameattrsZschemakeyvalue namespace localnamenumberZurlrrrr0sF     z!ContentHandler.startSeqXMLElementcCsP|\}}|dk rtd||dk r0td||dkr@tdd|_d|_dS)z!Handle end of the seqXML element.Nz(Unexpected namespace '%s' for seqXML endz$Unexpected qname '%s' for seqXML endrz$Failed to find end of seqXML element)rrr#)rr%r&r*r+rrrr"\s  zContentHandler.endSeqXMLElementc Cs|dkrtd|dk r tdtddd}|jdk rB|j|jd<|jdk rX|j|jd<|j|jd <xb|D]V\}}|\}}|dkr|d kr||_q|d kr||jd <qtd |qntd |qnW|jdkrtd |j ||j |_ |j |_dS)z>Set new entry with id and the optional entry source (PRIVATE).)Nentryz.Expected to find the start of an entry elementNz"Unexpected qname for entry element)idorganism ncbi_taxidrr/z(Unexpected attribute %s in entry elementz-Unexpected namespace '%s' for entry attributezFailed to find entry ID)rrrr annotationsrrr r/rappendstartEntryFieldElementrendEntryElementr#) rr%r&r'recordr(r)r*r+rrrr$hs4          z ContentHandler.startEntryElementcCs4|dkrtd|dk r td|j|_|j|_dS)zHandle end of an entry element.)Nr-z,Expected to find the end of an entry elementNz"Unexpected qname for entry element)rrr$rr"r#)rr%r&rrrr5s zContentHandler.endEntryElementcCs|\}}|dk r td||f|dk r8td||f|dkrJ||S|dkr\||S|dkrn||S|dkr||S|dkr||Std |dS) z3Receive a field of an entry element and forward it.Nz(Unexpected namespace '%s' for %s elementz$Unexpected qname '%s' for %s elementspecies description)DNAseqRNAseqAAseqDBRefpropertyzUnexpected field %s in entry)rrstartSpeciesElementstartDescriptionElementstartSequenceElementstartDBRefElementstartPropertyElement)rr%r&r'r*r+rrrr4s$     z%ContentHandler.startEntryFieldElementc Csd}d}xb|D]V\}}|\}}|dkr\|dkr8|}qh|dkrNt|}|}qhtd|qtd|qW|dkr|td|dkrtd|jd} || jd <|| jd <|j|_dS) zParse the species information.Nr%rz.Unexpected attribute '%s' found in species tagz/Unexpected namespace '%s' for species attributezFailed to find species namezFailed to find ncbiTaxIdr0r1)r r!rrr2endSpeciesElementr#) rr'r%rr(r)r*r+r,r6rrrr>s,    z"ContentHandler.startSpeciesElementcCsL|\}}|dk rtd||dk r0td||dkr@td|j|_dS)z Handle end of a species element.Nz)Unexpected namespace '%s' for species endz%Unexpected qname '%s' for species endr7z%Failed to find end of species element)rr5r#)rr%r&r*r+rrrrDs  z ContentHandler.endSpeciesElementcCs6|r td|jdk r$td|jd|_|j|_dS)zParse the description.z2Unexpected attributes found in description elementNzUnexpected data found: '%s'r.)rrrendDescriptionElementr#)rr'rrrr?s  z&ContentHandler.startDescriptionElementcCsl|\}}|dk rtd||dk r0td||dkr@td|jd}|j}|rZ||_d|_|j|_dS)z(Handle the end of a description element.Nz-Unexpected namespace '%s' for description endz)Unexpected qname '%s' for description endr8z)Failed to find end of description elementrC)rrrr8r5r#)rr%r&r*r+r6r8rrrrEs   z$ContentHandler.endDescriptionElementcCs6|r td|jdk r$td|jd|_|j|_dS)z$Parse DNA, RNA, or protein sequence.z/Unexpected attributes found in sequence elementNzUnexpected data found: '%s'r.)rrrendSequenceElementr#)rr'rrrr@s  z#ContentHandler.startSequenceElementcCs|\}}|dk rtd||dk r0td||jd}|dkrNd|jd<n4|dkrbd |jd<n |d krvd |jd<n td |t|j|_d|_|j|_dS) z%Handle the end of a sequence element.Nz*Unexpected namespace '%s' for sequence endz&Unexpected qname '%s' for sequence endrCr9DNA molecule_typer:RNAr;proteinz/Failed to find end of sequence (localname = %s))rrr2rrseqr5r#)rr%r&r*r+r6rrrrFs"        z!ContentHandler.endSequenceElementc Csd}d}xZ|D]N\}}|\}}|dkrT|dkr8|}q`|dkrF|}q`td|qtd|qW|dkrttd|dkrtd|jdk rtd|jd |_|jd }d ||f} | |jkr|j| |j|_dS) z!Parse a database cross reference.Nrr/z1Unexpected attribute '%s' found for DBRef elementz-Unexpected namespace '%s' for DBRef attributez'Failed to find source for DBRef elementz#Failed to find id for DBRef elementzUnexpected data found: '%s'r.rCz%s:%s) r rrrrdbxrefsr3endDBRefElementr#) rr'rZIDr(r)r*r+r6dbxrefrrrrAs2      z ContentHandler.startDBRefElementcCsj|\}}|dk rtd||dk r0td||dkrDtd||jrXtd|jd|_|j|_dS)z"Handle the end of a DBRef element.Nz+Unexpected namespace '%s' for DBRef elementz'Unexpected qname '%s' for DBRef elementr<z+Unexpected localname '%s' for DBRef elementz0Unexpected data received for DBRef element: '%s')rrr5r#)rr%r&r*r+rrrrM0s    zContentHandler.endDBRefElementc Csd}d}xX|D]L\}}|\}}|dkrR|dkr8|}q^|dkrF|}q^td|qtd|qW|dkrrtd|jd}|dkr|j||kst||j|<n$||jkrg|j|<|j|||j|_dS) z'Handle the start of a property element.Nr%r)z4Unexpected attribute '%s' found for property elementz0Unexpected namespace '%s' for property attributez(Failed to find name for property elementrCrH)r rrr2AssertionErrorr3endPropertyElementr#) rr'Z property_nameZproperty_valuer(r)r*r+r6rrrrBDs.     z#ContentHandler.startPropertyElementcCsP|\}}|dk rtd||dk r0td||dkrDtd||j|_dS)z%Handle the end of a property element.Nz.Unexpected namespace '%s' for property elementz*Unexpected qname '%s' for property elementr=z.Unexpected localname '%s' for property element)rr5r#)rr%r&r*r+rrrrPgs   z!ContentHandler.endPropertyElementcCs|jdk r|j|7_dS)zHandle character data.N)r)rrrrr charactersvs zContentHandler.characters)__name__ __module__ __qualname____doc__rrrr"r$r5r4r>rDr?rEr@rFrArMrBrPrQ __classcell__rr)rrr s& , ! !   !#r cs6eZdZdZdZd fdd ZddZdd ZZS) SeqXmlIteratoraNParser for seqXML files. Parses seqXML files and creates SeqRecords. Assumes valid seqXML please validate beforehand. It is assumed that all information for one record can be found within a record element or above. Two types of methods are called when the start tag of an element is reached. To receive only the attributes of an element before its end tag is reached implement _attr_TAGNAME. To get an element and its children as a DOM tree implement _elem_TAGNAME. Everything that is part of the DOM tree will not trigger any further method calls. iNcsBt|_t}|j||jtjdtj |ddddS)z0Create the object and initialize the XML parser.TbZSeqXML)modeZfmtN) rZ make_parserparserr ZsetContentHandlerZ setFeaturerZfeature_namespacesr r)rZstream_or_pathr*content_handler)rrrrs   zSeqXmlIterator.__init__cCs|j}|}|j}xH||}|s@|jdkr8tdntd|||j}|dk rPqW||_|j|_|j |_ |j |_ |j |_ | |}|S)z9Start parsing the file, and return a SeqRecord generator.Nz Empty file.zXML file contains no data.) rZgetContentHandlerBLOCKreadrrfeedrrrrriterate)rhandlerZr[r]textrrrrrparses(     zSeqXmlIterator.parseccst|j}|}|j}|j}x:t|dkr8|d}|V||}|sHP||qW|EdH|| dS)z)Iterate over the records in the XML file.r rN) rZr\rr]lenpopr^r_clearclose)rrarZr[rr]r6rbrrrr`s    zSeqXmlIterator.iterate)N) rRrSrTrUr]rrcr`rVrr)rrrW|s   rWcsbeZdZdZdfdd ZddZddZd d Zd d Zd dZ ddZ ddZ ddZ Z S) SeqXmlWriterzWrites SeqRecords into seqXML file. SeqXML requires the SeqRecord annotations to specify the molecule_type; the molecule type is required to contain the term "DNA", "RNA", or "protein". NcsFt|d|j}t|d|_|j||_||_||_||_ dS)a1Create Object and start the xml generator. Arguments: - target - Output stream opened in binary mode, or a path to a file. - source - The source program/database of the file, for example UniProt. - source_version - The version or release number of the source program or database from which the data originated. - species - The scientific name of the species of origin of all entries in the file. - ncbiTaxId - The NCBI taxonomy identifier of the species of origin. wbzutf-8N) r rrar xml_generatorrrsource_versionr7 ncbiTaxId)rtargetrrkr7rlra)rrrrs  zSeqXmlWriter.__init__cCsdddd}|jdk r |j|d<|jdk r4|j|d<|jdk r\t|jtsRtd|j|d <|jdk rt|jttfs~td |j|d <|j d t |dS) z'Write root node with document metadata.z)http://www.w3.org/2001/XMLSchema-instancez$http://www.seqxml.org/0.4/seqxml.xsdz0.4)z xmlns:xsizxsi:noNamespaceSchemaLocationrNrrz species should be of type stringrz)ncbiTaxID should be of type string or intrr) rrkr7 isinstancestr TypeErrorrlr!rj startElementr)rr'rrr write_headers          zSeqXmlWriter.write_headercCs|jr|jdkrtdt|jts,tdd|ji}d|jkrv|j|jdkrvt|jdtshtd|jd|d<|jdt || || || || ||||jddS) zWrite one record.z zSeqXML requires identifierz#Identifier should be of type stringr/rzsource should be of type stringr-N)r/rrnrorpr2rrjrqr_write_species_write_description _write_seq_write_dbxrefs_write_properties endElement)rr6Zattrbrrr write_records"        zSeqXmlWriter.write_recordcCs|jd|jdS)z0Close the root node and finish the XML document.rN)rjrxZ endDocument)rrrr write_footers zSeqXmlWriter.write_footercCsd}d|jkrV|jd}t|trVt|dkr8|d}nt|dkrJd}n td|d|jkr|r|jd}t|tstdt|ttfstd||jks||j kr|t|d }|j d t ||j d dS) z%Write the species if given (PRIVATE).Nr1r rz9Multiple entries for record.annotations['ncbi_taxid'], %rr0z!organism should be of type stringz)ncbiTaxID should be of type string or int)r%rr7)r2rnlistrdrrorpr!r7rlrjrqrrx)rr6Zlocal_ncbi_taxidZ local_orgattrrrrrs s*        zSeqXmlWriter._write_speciescCsh|jrdt|jtstd|j}|dkr,d}t|jdkrd|jdti|j||j ddS)z)Write the description if given (PRIVATE).z$Description should be of type stringzr.rr8N) r8rnrorprdrjrqrrQrx)rr6r8rrrrt@s  zSeqXmlWriter._write_descriptioncCst|jtrtdt|j}t|dks2td|jd}|dkrPtdn6d|kr^d}n(d |krld }nd |krzd }n td ||j |t i|j ||j |dS)zWrite the sequence (PRIVATE). Note that SeqXML requires the molecule type to contain the term "DNA", "RNA", or "protein". z8Sequence type is UnknownSeq but SeqXML requires sequencerz,The sequence length should be greater than 0rHNzmolecule_type is not definedrGr9rIr:rJr;zunknown molecule_type '%s')rnrKrrpbytesrdrr2getrjrqrrQrx)rr6rKrHZseqElemrrrruPs$       zSeqXmlWriter._write_seqcCs~|jdk rzxn|jD]d}t|ts(td|ddkr>td|dd\}}||d}|jdt ||j dqWdS)z.Write all database cross references (PRIVATE).Nz(dbxrefs should be of type list of string:r z9dbxrefs should be in the form ['source:id', 'source:id' ])rr/r<) rLrnrorpfindrsplitrjrqrrx)rr6rNZdbsourceZdbidr|rrrrvns    zSeqXmlWriter._write_dbxrefscCsx|jD]\}}|dkr |dkrLd|i}|jdt||jdq t|trx|D]B}|dkrrd|i}n|t|d}|jdt||jdq\Wq t|t t tfr |t|d}|jdt||jdq WdS)ztWrite all annotations that are key value pairs with values of a primitive type or list of primitive types (PRIVATE).)r0r1rNr%r=)r%r)) r2r rjrqrrxrnr{ror!float)rr6r(r)r|vrrrrws$    zSeqXmlWriter._write_properties)NNNN)rRrSrTrUrrrryrzrsrtrurvrwrVrr)rrrhs rhN)rUZxmlrZxml.saxrZxml.sax.saxutilsrZxml.sax.xmlreaderrZBio.SeqrrZ Bio.SeqRecordrZ Interfacesr r r rWrhrrrrs         aM