o Rŀgu@sdZddlmZddlmZddlmZddlmZddl m Z ddl m Z dd l mZdd l mZGd d d ejZGd ddeZGdddeZdS)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) SeqRecord)SequenceIterator)SequenceWritercseZdZdZfddZddZddZdd Zd d Zd d Z ddZ ddZ ddZ ddZ ddZddZddZddZddZd d!Zd"d#Zd$d%Zd&d'ZZS)(ContentHandlerz5Handles XML events generated by the parser (PRIVATE).cs>td|_d|_d|_d|_d|_d|_d|_g|_ dS)z&Create a handler to handle XML events.N) super__init__source sourceVersion seqXMLversion ncbiTaxID speciesNamestartElementNSdatarecordsself __class__F/var/www/html/myenv/lib/python3.10/site-packages/Bio/SeqIO/SeqXmlIO.pyr s  zContentHandler.__init__cCs |j|_dS)z2Set XML handlers when an XML declaration is found.N)startSeqXMLElementrrrrr startDocument,s zContentHandler.startDocumentc Csj|dkrtd|durtdd}|D]R\}}|\}}|durR|dkr*||_q|dkr2||_q|dkr:||_q|dkrFt|} ||_q|d krN||_qtd |d kra|d kr]|}qtd td|d|jdurrtd|jdvr{tdd|jd} |dur|| krtd||jf|jr|jdkrtd|jr|jdkrtd|j |_ |j |_ dS)z!Handle start of a seqXML element.)NseqXMLz*Failed to find the start of seqXML elementNz#Unexpected qname for seqXML elementrrrrrz#Unexpected attribute for XML Schema)http://www.w3.org/2001/XMLSchema-instancenoNamespaceSchemaLocationz0Unexpected attribute for XML Schema in namespaceUnexpected namespace 'z' for seqXML attributezFailed to find seqXMLversion)0.10.20.30.4zUnsupported seqXMLversionzhttp://www.seqxml.org/z /seqxml.xsdzDXML Schema '%s' found not consistent with reported seqXML version %sr%z@Attribute 'speciesName' on root is only supported in version 0.4z>Attribute 'ncbiTaxID' on root is only supported in version 0.4) ValueError RuntimeErroritemsrrrintrrendSeqXMLElement endElementNSstartEntryElementr) rnameqnameattrsschemakeyvalue namespace localnamenumberurlrrrr0s`    z!ContentHandler.startSeqXMLElementcCsX|\}}|durtd|d|durtd|d|dkr$tdd|_d|_dS)z!Handle end of the seqXML element.Nr!z' for seqXML endUnexpected qname 'rz$Failed to find end of seqXML element)r'rr+rr-r.r3r4rrrr*hs zContentHandler.endSeqXMLElementc Cs"|dkrtd|durtdtddd}|jdur!|j|jd<|jdur,|j|jd<|j|jd<|D]7\}}|\}}|durf|d krJ||_q6|dkr^|j d ksX|j d kr^||jd<q6td |d td|d|jdurwtd|j ||j dkr|j |_ n|j|_ |j|_dS)z>Set new entry with id and the optional entry source (PRIVATE).Nentryz.Expected to find the start of an entry elementN"Unexpected qname for entry element)idorganism ncbi_taxidrr<r$r%zUnexpected attribute z in entry elementr!z' for entry attributezFailed to find entry IDr")r&r'rr annotationsrrr(r<rrappendstartEntryFieldElementVersion01rstartEntryFieldElementendEntryElementr+) rr-r.r/recordr1r2r3r4rrrr,ts>              z ContentHandler.startEntryElementcCsL|dkrtd|durtd|jdjdurtd|j|_|j|_dS)zHandle end of an entry element.r9z,Expected to find the end of an entry elementNr;z+Failed to find a sequence for entry element)r&r'rseqr,rr*r+)rr-r.rrrrCs zContentHandler.endEntryElementcC|\}}|durtd|d|d|dur"td|d|d|dkr+||S|dkr4||S|dvr=||S|d krF||S|d krO||Std |d ) zCReceive a field of an entry element and forward it for version 0.1.Nr!' for  elementr7species description)dnaSeqrnaSeqaaSeq alternativeIDpropertyUnexpected field in entryr&r'startSpeciesElementstartDescriptionElementstartSequenceElementstartDBRefElementstartPropertyElementrr-r.r/r3r4rrrrA$     z.ContentHandler.startEntryFieldElementVersion01cCrG) zFReceive a field of an entry element and forward it for versions >=0.2.Nr!rHrIr7rJrK)DNAseqRNAseqAAseqDBRefrPrQrRrSrYrrrrBrZz%ContentHandler.startEntryFieldElementc Csd}d}|D]-\}}|\}}|dur.|dkr|}q|dkr&t|}|}qtd|dtd|d|dur>td|durFtd |jd } || jd <|| jd <|j|_dS) zParse the species information.Nr-rUnexpected attribute 'z' found in species tagr!z' for species attributezFailed to find species namezFailed to find ncbiTaxIdrEr=r>)r(r)r&rr?endSpeciesElementr+) rr/r-rr1r2r3r4r5rDrrrrTs0      z"ContentHandler.startSpeciesElementcCsT|\}}|durtd|d|durtd|d|dkr$td|j|_dS)z Handle end of a species element.Nr!z' for species endr7rJz%Failed to find end of species elementr'rCr+r8rrrr`s z ContentHandler.endSpeciesElementcC:|rtd|jdurtd|jdd|_|j|_dS)zParse the description.z2Unexpected attributes found in description elementNUnexpected data found: '')r&rr'endDescriptionElementr+rr/rrrrU   z&ContentHandler.startDescriptionElementcCst|\}}|durtd|d|durtd|d|dkr$td|jd}|j}|r1||_d|_|j|_dS)z(Handle the end of a description element.Nr!z' for description endr7rKz)Failed to find end of description elementrE)r'rrrKrCr+)rr-r.r3r4rDrKrrrrfs   z$ContentHandler.endDescriptionElementcCrb)z$Parse DNA, RNA, or protein sequence.z/Unexpected attributes found in sequence elementNrcrdre)r&rr'endSequenceElementr+rgrrrrVrhz#ContentHandler.startSequenceElementcCs|\}}|durtd|d|durtd|d|jd}|dkr*|jdks3|dkr9|jdkr9d |jd <n8|d krB|jdksK|d krQ|jdkrQd |jd <n |dkrZ|jdksc|dkri|jdkrid|jd <ntd|dt|j|_d|_|j|_dS)z%Handle the end of a sequence element.Nr!z' for sequence endr7rEr[r"rLDNA molecule_typer\rMRNAr]rNproteinz,Failed to find end of sequence (localname = )) r'rrr?rrrFrCr+)rr-r.r3r4rDrrrris*       z!ContentHandler.endSequenceElementc Csd}d}d}|D]0\}}|\}}|dur3|dkr|}q |dkr$|}q |dkr+|}q td|dtd|d|durCtd |durKtd |dur]|jd ksY|jd kr]td |jdurktd|jdd|_|jd} |d|} | | jvr| j| |j|_ dS)z!Parse a database cross reference.Ntyperr<r_z' found for DBRef elementr!z' for DBRef attributez'Failed to find source for DBRef elementz#Failed to find id for DBRef elementr#r$z%Failed to find type for DBRef elementrcrdrerE:) r(r&rrr'rdbxrefsr@endDBRefElementr+) rr/TYPErIDr1r2r3r4rDdbxrefrrrrW;sB       z ContentHandler.startDBRefElementcCs|\}}|durtd|d|durtd|d|dkr%|jdks.|dkr6|jdkr6td|d|jrBtd |jd d|_|j|_dS) z"Handle the end of a DBRef element.Nr!z' for DBRef elementr7r^r"rOUnexpected localname 'z-Unexpected data received for DBRef element: 'rd)r'rrrCr+r8rrrrrds  zContentHandler.endDBRefElementc Csd}d}|D]&\}}|\}}|dur'|dkr|}q|dkr"|}qtd|td|d|dur7td|jd}|d krO|j||vsIJ||j|<n||jvrYg|j|<|j|||j|_dS) z'Handle the start of a property element.Nr-r2z4Unexpected attribute '%s' found for property elementr!z' for property attributez(Failed to find name for property elementrErk)r(r&rr?r@endPropertyElementr+) rr/ property_nameproperty_valuer1r2r3r4rDrrrrXvs2      z#ContentHandler.startPropertyElementcCs\|\}}|durtd|d|durtd|d|dkr(td|d|j|_dS)z%Handle the end of a property element.Nr!z' for property elementr7rPrvrar8rrrrws   z!ContentHandler.endPropertyElementcCs |jdur|j|7_dSdS)zHandle character data.N)r)rrrrr characterss zContentHandler.characters)__name__ __module__ __qualname____doc__r rrr*r,rCrArBrTr`rUrfrVrirWrrrXrwrz __classcell__rrrrr s*  8 & !   )#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.TbSeqXML)modefmtN) r make_parserparserr setContentHandler setFeaturerfeature_namespacesr r )rstream_or_pathr3content_handlerrrrr s  zSeqXmlIterator.__init__cCs|j}|}|j} ||}|s|jdurtdtd|||j}|dur,nq ||_|j|_|j |_ |j |_ |j |_ | |}|S)z9Start parsing the file, and return a SeqRecord generator.TNz Empty file.zXML file contains no data.) rgetContentHandlerBLOCKreadrr&feedrrrrriterate)rhandlerrrtextrrrrrparses*     zSeqXmlIterator.parseccst|j}|}|j}|j} t|dkr|d}|V||}|s%n||q||EdH| dS)z)Iterate over the records in the XML file.TrrN) rrrrlenpoprrcloseclear)rrrrrrrDrrrrrs"      zSeqXmlIterator.iterate)N) r{r|r}r~rr rrrrrrrrs  rcsdeZdZdZ dfdd 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 r rr xml_generatorrrsource_versionrJ ncbiTaxId)rtargetrrrJrrrrrr s   zSeqXmlWriter.__init__cCsdddd}|jdur|j|d<|jdur|j|d<|jdur.t|jts)td|j|d <|jdurDt|jttfs?td |j|d <|j d t |dS) z'Write root node with document metadata.rz$http://www.seqxml.org/0.4/seqxml.xsdr%)z xmlns:xsizxsi:noNamespaceSchemaLocationrNrrz species should be of type stringr)ncbiTaxID should be of type string or intrr) rrrJ isinstancestr TypeErrorrr)r startElementrrgrrr write_headers"         zSeqXmlWriter.write_headercCs|jr|jdkr tdt|jtstdd|ji}d|jvr;|j|jdkr;t|jdts4td|jd|d<|jdt || || || || ||||jddS) zWrite one record.z zSeqXML requires identifierz#Identifier should be of type stringr<rzsource should be of type stringr:N)r<r&rrrr?rrrr_write_species_write_description _write_seq_write_dbxrefs_write_properties endElement)rrDattrbrrr write_record5s"        zSeqXmlWriter.write_recordcCs|jd|jdS)z0Close the root node and finish the XML document.rN)rr endDocumentrrrr write_footerOs zSeqXmlWriter.write_footercCsd}d|jvr+|jd}t|tr+t|dkr|d}nt|dkr%d}ntd|d|jvrm|ro|jd}t|ts@tdt|ttfsKtd||jksU||j krq|t|d }|j d t ||j d dSdSdSdS) z%Write the species if given (PRIVATE).Nr>rrz9Multiple entries for record.annotations['ncbi_taxid'], %rr=z!organism should be of type stringr)r-rrJ)r?rlistrr&rrr)rJrrrrr)rrDlocal_ncbi_taxid local_orgattrrrrrTs2         zSeqXmlWriter._write_speciescCsp|jr4t|jts td|j}|dkrd}t|jdkr6|jdti|j||j ddSdSdS)z)Write the description if given (PRIVATE).z$Description should be of type stringzrerrKN) rKrrrrrrrrzr)rrDrKrrrrss  zSeqXmlWriter._write_descriptioncCst|j}t|dkstd|jd}|durtdd|vr$d}nd|vr+d }nd |vr2d }ntd |d |j|ti|j ||j |dS)zWrite the sequence (PRIVATE). Note that SeqXML requires the molecule type to contain the term "DNA", "RNA", or "protein". rz,The sequence length should be greater than 0rkNzmolecule_type is not definedrjr[rlr\rmr]zunknown molecule_type 'rd) bytesrFrr&r?getrrrrzr)rrDrFrkseqElemrrrrs    zSeqXmlWriter._write_seqcCs~|jdur;|jD]4}t|tstd|ddkrtd|dd\}}||d}|jdt ||j dqdSdS)z.Write all database cross references (PRIVATE).Nz(dbxrefs should be of type list of stringrprz9dbxrefs should be in the form ['source:id', 'source:id' ])rr<r^) rqrrrfindr&splitrrrr)rrDrudbsourcedbidrrrrrs    zSeqXmlWriter._write_dbxrefscCs|jD]h\}}|dvrm|dur%d|i}|jdt||jdqt|trO|D]!}|dur7d|i}n|t|d}|jdt||jdq,qt|t t tfrm|t|d}|jdt||jdqdS)ztWrite all annotations that are key value pairs with values of a primitive type or list of primitive types (PRIVATE).)r=r>rNr-rP)r-r2) r?r(rrrrrrrr)float)rrDr1r2rvrrrrs,    zSeqXmlWriter._write_properties)NNNN)r{r|r}r~r rrrrrrrrrrrrrrsrN)r~xmlrxml.saxrxml.sax.saxutilsrxml.sax.xmlreaderrBio.Seqr Bio.SeqRecordr Interfacesr r r rrrrrrs       O