:>"^g@sdZddlZddlmZddlmZddlmZddlm Z ddl m Z ddl m Z dd lmZdd lmZdd lmZd d lmZGdddejZGdddeZGdddeZdS)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 N) XMLGenerator)AttributesImpl)handler)sax) basestring)Alphabet)Seq) UnknownSeq) SeqRecord)SequentialSequenceWriterc@seZdZdZddZddZddZdd Zd d Zd d Z ddZ ddZ ddZ ddZ ddZddZddZddZddZd d!Zd"d#Zd$d%Zd&S)'ContentHandlerz5Handles XML events generated by the parser (PRIVATE).cCs\tjj|d|_d|_d|_d|_d|_d|_d|_ g|_ dS)z&Create a handler to handle XML events.N) rr __init__source sourceVersion seqXMLversion ncbiTaxID speciesNamestartElementNSdatarecords)selfr7/tmp/pip-build-ww9dw3qa/biopython/Bio/SeqIO/SeqXmlIO.pyr%s       zContentHandler.__init__cCs|j|_dS)z2Set XML handlers when an XML declaration is found.N)startSeqXMLElementr)rrrr startDocument1szContentHandler.startDocumentc Cs|dkrtd|dk r0tdd}x|jD]\}}|\}}|dkr|dkr||_q:|dkr||_q:|dkr||_q:|dkrt|} ||_q:|d kr||_q:td qC|d kr*|d kr|}q:td qCtd|qCW|jdkrYtdd|j} || krtd||jf|j |_ |j |_ dS)z!Handle start of a seqXML element.NseqXMLz*Failed to find the start of seqXML elementz#Unexpected qname for seqXML elementrrrrrz#Unexpected attribute for XML Schemaz)http://www.w3.org/2001/XMLSchema-instanceZnoNamespaceSchemaLocationz0Unexpected attribute for XML Schema in namespacez.Unexpected namespace '%s' for seqXML attributezFailed to find seqXMLversionz#http://www.seqxml.org/%s/seqxml.xsdzDXML Schema '%s' found not consistent with reported seqXML version %s)Nr) ValueError RuntimeErroritemsrrrintrrendSeqXMLElement endElementNSstartEntryElementr) rnameqnameattrsZschemakeyvalue namespace localnamenumberurlrrrr5s@                        z!ContentHandler.startSeqXMLElementcCsr|\}}|dk r(td||dk rDtd||dkr\tdd|_d|_dS)z!Handle end of the seqXML element.Nz(Unexpected namespace '%s' for seqXML endz$Unexpected qname '%s' for seqXML endrz$Failed to find end of seqXML element)rrr")rr$r%r)r*rrrr!\s      zContentHandler.endSeqXMLElementc Csc|d krtd|dk r0tdtddd}|jdk ra|j|jd<|jdk r|j|jd<|j|jd Set new entry with id and the optional entry source (PRIVATE).Nentryz.Expected to find the start of an entry elementz"Unexpected qname for entry elementidorganism ncbi_taxidrz(Unexpected attribute %s in entry elementz-Unexpected namespace '%s' for entry attributezFailed to find entry ID)Nzentry)rrr r annotationsrrrr/rappendstartEntryFieldElementrendEntryElementr") rr$r%r&recordr'r(r)r*rrrr#hs0           z ContentHandler.startEntryElementcCsL|dkrtd|dk r0td|j|_|j|_dS)zHandle end of an entry element.Nr-z,Expected to find the end of an entry elementz"Unexpected qname for entry element)Nzentry)rrr#rr!r")rr$r%rrrr5s      zContentHandler.endEntryElementcCs|\}}|dk r.td||f|dk rPtd||f|dkri|j|S|dkr|j|S|d kr|j|S|d kr|j|S|d kr|j|Std |dS) z3Receive a field of an entry element and forward it.Nz(Unexpected namespace '%s' for %s elementz$Unexpected qname '%s' for %s elementspecies descriptionDNAseqRNAseqAAseqDBRefpropertyzUnexpected field %s in entry)r9r:r;)rrstartSpeciesElementstartDescriptionElementstartSequenceElementstartDBRefElementstartPropertyElement)rr$r%r&r)r*rrrr4s              z%ContentHandler.startEntryFieldElementc Csd}d}x|jD]|\}}|\}}|dkr|dkrR|}q|dkrst|}|}qtd|qtd|qW|dkrtd|dkrtd|jd } || jd <|| jd <|j|_dS) zParse the species information.Nr$rz.Unexpected attribute '%s' found in species tagz/Unexpected namespace '%s' for species attributezFailed to find species namezFailed to find ncbiTaxIdr r0r1)rr rrr2endSpeciesElementr") rr&r$rr'r(r)r*r+r6rrrr>s(              z"ContentHandler.startSpeciesElementcCsl|\}}|dk r(td||dk rDtd||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$r%r)r*rrrrDs     z ContentHandler.endSpeciesElementcCsM|rtd|jdk r4td|jd|_|j|_dS)zParse the description.z2Unexpected attributes found in description elementNzUnexpected data found: '%s'r.)rrrendDescriptionElementr")rr&rrrr?s   z&ContentHandler.startDescriptionElementcCs|\}}|dk r(td||dk rDtd||dkr\td|jd}|j}|r||_d|_|j|_dS)z(Handle the end of a description element.Nz-Unexpected namespace '%s' for description endz)Unexpected qname '%s' for description endr8z)Failed to find end of description elementr rC)rrrr8r5r")rr$r%r)r*r6r8rrrrEs         z$ContentHandler.endDescriptionElementcCsM|rtd|jdk r4td|jd|_|j|_dS)z$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 r(td||dk rDtd||dkr\tj}n@|dkrttj}n(|dkrtj}ntd||jd }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 endr9r:r;z/Failed to find end of sequence (localname = %s)r rC) rrZ generic_dnaZ generic_rnaZgeneric_proteinrrrseqr5r")rr$r%r)r*alphabetr6rrrrFs            z!ContentHandler.endSequenceElementc Cs4d}d}x~|jD]p\}}|\}}|dkry|dkrR|}q|dkrg|}qtd|qtd|qW|dkrtd|dkrtd|jdk rtd|jd |_|jd }d ||f} | |jkr$|jj| |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.r z%s:%srC) rrrrrdbxrefsr3endDBRefElementr") rr&rZIDr'r(r)r*r6dbxrefrrrrAs.            z ContentHandler.startDBRefElementcCs|\}}|dk r(td||dk rDtd||dkr`td||jr|td|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$r%r)r*rrrrJs      zContentHandler.endDBRefElementc Csd}d}x~|jD]p\}}|\}}|dkry|dkrR|}q|dkrg|}qtd|qtd|qW|dkrtd|jd}||jkrg|j|<|j|j||j|_dS) z'Handle the start of a property element.Nr$r(z4Unexpected attribute '%s' found for property elementz0Unexpected namespace '%s' for property attributez(Failed to find name for property elementr rC)rrrr2r3endPropertyElementr") rr&Z property_nameZproperty_valuer'r(r)r*r6rrrrB*s$          z#ContentHandler.startPropertyElementcCsp|\}}|dk r(td||dk rDtd||dkr`td||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$r%r)r*rrrrLBs    z!ContentHandler.endPropertyElementcCs"|jdk r|j|7_dS)zHandle character data.N)r)rrrrr charactersMszContentHandler.charactersN)__name__ __module__ __qualname____doc__rrrr!r#r5r4r>rDr?rEr@rFrArJrBrLrMrrrrr "s&   '         r c@sbeZdZdZdZdddZddZdd Zej d d kr^d d Z dS)SeqXmlIteratora)Breaks seqXML file into SeqRecords. Assumes valid seqXML please validate beforehand. It is assumed that all information for one record can be found within a record element or above. Two types of methods are called when the start tag of an element is reached. To receive only the attributes of an element before its end tag is reached implement _attr_TAGNAME. To get an element and its children as a DOM tree implement _elem_TAGNAME. Everything that is part of the DOM tree will not trigger any further method calls. iNcCsntj|_t}|jj||jjtjdyt|}Wn$t k rt||_ d|_ YnX||_ d|_ |j }ysxl|j j |}|s|jdkrtdn td|jj||j}|dk rPqWWn+tk r0|j r)|j jYnX||_|j|_|j|_|j|_|j|_dS)z0Create the object and initialize the XML parser.TFNz Empty file.zXML file contains no data.)r make_parserparserr setContentHandler setFeaturerfeature_namespacesopen TypeErrorhandleshould_close_handleBLOCKreadrrfeedr Exceptioncloserrrr)rZstream_or_pathr)content_handlerrZr\textrrrrrbs@                 zSeqXmlIterator.__init__cCs|S)Nr)rrrr__iter__szSeqXmlIterator.__iter__cCs|jj}|j}|j}yYxR|jj|}|s@P|jj|t|dkr'|jd}|Sq'WWn+t k r|j r|jj YnXy|jd}Wnt k rYnX|S|jj |j r|jj t dS)z)Iterate over the records in the XML file.r rN)rTgetContentHandlerrr\rZr]r^lenpopr_r[r` IndexError StopIteration)rrarr\rbr6rrr__next__s2          zSeqXmlIterator.__next__rcCs |jS)z8Python 2 style alias for Python 3 style __next__ method.)ri)rrrrnextszSeqXmlIterator.next) rNrOrPrQr\rrcrisys version_inforkrrrrrRSs '  !rRc@seZdZdZddddddZddZddZd d Zd d Zd dZ ddZ ddZ ddZ dS) SeqXmlWriterzWrites SeqRecords into seqXML file. SeqXML requires the sequence alphabet be explicitly RNA, DNA or protein, i.e. an instance or subclass of Bio.Alphapet.RNAAlphabet, Bio.Alphapet.DNAAlphabet or Bio.Alphapet.ProteinAlphabet. NcCsWtj||t|d|_|jj||_||_||_||_dS)z*Create Object and start the xml generator.zutf-8N) r rr xml_generatorrrsource_versionr7 ncbiTaxId)rrZrrpr7rqrrrrs    zSeqXmlWriter.__init__cCstj|ddddddi}|jdk rA|j|d<|jdk r]|j|d <|jdk rt|jtstd |j|d <|jdk rt|jtt fstd |j|d <|j j dt |dS)z'Write root node with document metadata.z xmlns:xsiz)http://www.w3.org/2001/XMLSchema-instancezxsi:noNamespaceSchemaLocationz$http://www.seqxml.org/0.4/seqxml.xsdrz0.4Nrrz species should be of type stringrz)ncbiTaxID should be of type string or intrr) r write_headerrrpr7 isinstancerrYrqr ro startElementr)rr&rrrrrs"        zSeqXmlWriter.write_headercCs|j s|jdkr%tdt|jtsCtdd|ji}d|jkr|j|jdkrt|jdtstd|jd|d<|jjdt ||j ||j ||j ||j ||j||jjddS) zWrite one record.z zSeqXML requires identifierz#Identifier should be of type stringr/rzsource should be of type stringr-N)r/rrsrrYr2rrortr_write_species_write_description _write_seq_write_dbxrefs_write_properties endElement)rr6Zattrbrrr write_records"        zSeqXmlWriter.write_recordcCs.tj||jjd|jjdS)z0Close the root node and finish the XML document.rN)r write_footerrorz endDocument)rrrrr|s zSeqXmlWriter.write_footercCs:d}d|jkry|jd}t|tryt|dkrP|d}n)t|dkrkd}ntd|d|jkr6|r6|jd}t|tstdt|ttfstd||jks||j kr6d |d t |i}|j j d t ||j jd dS) z%Write the species if given (PRIVATE).Nr1r rz9Multiple entries for record.annotations['ncbi_taxid'], %rr0z!organism should be of type stringz)ncbiTaxID should be of type string or intr$rr7)r2rslistrerrrYr r7rqstrrortrrz)rr6Zlocal_ncbi_taxidZ local_orgattrrrrrus*      zSeqXmlWriter._write_speciescCs|jrt|jts'td|j}|dkrBd}t|jdkr|jjdti|jj||jj ddS)z)Write the description if given (PRIVATE).z$Description should be of type stringzr.rr8N) r8rsrrYrerortrrMrz)rr6r8rrrrv#s    zSeqXmlWriter._write_descriptioncCst|jtrtdt|j}t|dksKtdtj|jj }t|tj r{d}nBt|tj rd}n't|tj rd}n td|j j|ti|j j||j j|dS) ziWrite the sequence (PRIVATE). Note that SeqXML requires a DNA, RNA or protein alphabet. z8Sequence type is UnknownSeq but SeqXML requires sequencerz,The sequence length should be greater than 0r:r9r;z#Need a DNA, RNA or Protein alphabetN)rsrGr rYrrerrZ_get_base_alphabetrHZ RNAAlphabetZ DNAAlphabetZProteinAlphabetrortrrMrz)rr6rGalphaZseqElemrrrrw3s       zSeqXmlWriter._write_seqcCs|jdk rx|jD]}t|ts:td|jddkr[td|jdd\}}d|d|i}|jjdt ||jj dqWdS) z.Write all database cross references (PRIVATE).Nz(dbxrefs should be of type list of string:r z9dbxrefs should be in the form ['source:id', 'source:id' ]rr/r<) rIrsrrYfindrsplitrortrrz)rr6rKZdbsourceZdbidrrrrrxOs  zSeqXmlWriter._write_dbxrefscCsCx<|jjD]+\}}|dkr|dkrld|i}|jjdt||jjdqt|trx|D]Y}t|tt t frd|d|i}|jjdt||jjdqWqt|tt t frd|dt |i}|jjdt||jjdqWdS) ztWrite all annotations that are key value pairs with values of a primitive type or list of primitive types (PRIVATE).r0r1rNr$r=r()zorganismz ncbi_taxidzsource) r2rrortrrzrsr~r floatrr)rr6r'r(rvrrrrybs"     zSeqXmlWriter._write_properties) rNrOrPrQrrrr{r|rurvrwrxryrrrrrns       rn)rQrlxml.sax.saxutilsrZxml.sax.xmlreaderrZxml.saxrxmlrZ Bio._py3krZBiorZBio.Seqrr Z Bio.SeqRecordr Z Interfacesr r objectrRrnrrrrs 2a