o RŀgT~@s<dZddlmZddlmZddiZejZeD] \Z Z ee e qdZ Gddde Z d d Zd d Ze d fddZddZddZedfddZddZefddZddZefddZd:ddZd d!Zd"d#Zd$d%Zd&d'Zd(d)Zd*d+ZGd,d-d-Z d.d/Z!d0d1Z"d;d3d4Z#d5d6Z$Gd7d8d8Z%d9S).fix_single__iter__c3s|]}|VqdSNr ).0trr r jszwrite..) phylogenieszXFirst argument must be a Phyloxml, Phylogeny, Tree, or iterable of Trees or Phylogenies.)encodingindent) rrPhyloxmlrrr to_phyloxmlhasattrrWriterwrite)objrr'r(r r$r r-Ls  r-cCs&|ddkr||dddS|S)z;Extract the local tag from a namespaced tag name (PRIVATE).r{}N)indextagr r r _localws r5cCs2z |ddddWStyd|fYSw)z;Split a tag into namespace and local tag strings (PRIVATE).r1Nr0)splitrr3r r r _split_namespace~s   r8cCsd|d|S)z5Format an XML tag with the given namespace (PRIVATE).r/r0r )r4 namespacer r r _nssr:cCs"|t|}|dur||SdS)z|Find a child node by tag, and pass it through a constructor (PRIVATE). Returns None if no matching child is found. N)findr:parentr4 constructchildr r r _get_child_assr@cCs.|t|}|dur|jr||jSdSdS)z~Find a child node by tag; pass its text through a constructor (PRIVATE). Returns None if no matching child is found. N)r;r:textr<r r r _get_child_texts rBcfdd|t|DS)zFind child nodes by tag; pass each through a constructor (PRIVATE). Returns an empty list if no matching child is found. csg|]}|qSr r r"r?r>r r sz$_get_children_as..findallr:r=r4r>r rEr _get_children_asrJcrC)zFind child nodes by tag; pass each node's text through a constructor (PRIVATE). Returns an empty list if no matching child is found. csg|] }|jr|jqSr rArDrEr r rFsz&_get_children_text..rGrIr rEr _get_children_textrKrMcCsd|d}t|r?|jr|js|d|_|D]}t||d|jr*|js/|d|_q|jr8|js=||_dSdS|rN|jrI|jsP||_dSdSdS)zAdd line breaks and indentation to ElementTree in-place (PRIVATE). Sources: - http://effbot.org/zone/element-lib.htm#prettyprint - http://infix.se/2007/02/06/gentlemen-indent-your-xml  z r1N)lenrAstrip_indenttail)elemlevelier r r rQs     rQcCs4|dks|dkr dS|dks|dkrdStd|)z$Convert string to boolean (PRIVATE).true1Tfalse0Fz*String could not be converted to boolean: )rrLr r r _str2bools  r[cCs.|}|D]}||vrt||||<q|S)zQReturn a new dictionary where string values are replaced with booleans (PRIVATE).)copyr[)dctkeysoutkeyr r r _dict_str2bools racC,|durzt|WStyYdSwdS)z$Return text as an integer (PRIVATE).N)int ExceptionrLr r r _int  recCrb)z!Return text as a float (PRIVATE).N)floatrdrLr r r _floatrfrhcCs|dur d|SdS)aReplace all spans of whitespace with a single space character (PRIVATE). Also remove leading and trailing whitespace. See "Collapse Whitespace Policy" in the phyloXML spec glossary: http://phyloxml.org/documentation/version_100/phyloxml.xsd.html#Glossary N )joinr7rLr r r _collapse_wspacesrkcCs"dD] }||vr||d}q|S)zReplace tab, LF and CR characters with spaces, but don't collapse (PRIVATE). See "Replace Whitespace Policy" in the phyloXML spec glossary: http://phyloxml.org/documentation/version_100/phyloxml.xsd.html#Glossary ) rN ri)replace)rAcharr r r _replace_wspaces  rpc@seZdZdZddZddZddZdd Zgd Zd d d ddZ e e e  gdZ ddZddZddZddZddZddZddZdd Zd!d"Zd#d$Zd%d&Zd'd(Zd)d*Zd+d,Zd-d.Zd/d0Zd1d2Zd3d4Zd5d6Z d7d8Z!d9d:Z"d;d<Z#d=d>Z$d?S)@raMethods for parsing all phyloXML nodes from an XML stream. To minimize memory use, the tree of ElementTree parsing events is cleared after completing each phylogeny, clade, and top-level 'other' element. Elements below the clade level are kept in memory until parsing of the current clade is finished -- this shouldn't be a problem because clade is the only recursive element, and non-clade nodes below this level are of bounded size. cCs.ttj|dd}t|\}}||_||_dS)zInitialize the class.)startend)eventsN)iterr iterparsenextrcontext)selfrrweventrr r r __init__s  zParser.__init__c Cstdd|jD}d}|jD]M\}}t|j\}}|dkr;|tdkr,|d7}q|dkr;||}|j ||dkr_|tdkr_|d8}|dkr_| |||}|j ||j q|S) zszParser.read..rrqrr1 phylogenyrr) rr)ritemsrwr8r4 NAMESPACES_parse_phylogenyr&appendotherclear) rxphyloxml other_depthryrSr9localtagr}otrr r r rs&     z Parser.readccs<td}|jD]\}}|dkr|j|kr||VqdS)z@Parse the phyloXML file incrementally and return each phylogeny.r}rqN)r:rwr4r)rxphytagryrSr r r r1s z Parser.parsecCs6tjdit|jddg}ddddd}|jD]\}}t|j\}}|dkr;|d kr;|jd ur4td | ||_q|d kr|d krJ| |S||vr^t ||| t |||q|dvrnt ||t |||q|dvr|t ||t|jq|tdkr|j ||||| qtd|q|S)aParse a single phylogeny within the phyloXML tree (PRIVATE). Recursively builds a phylogenetic tree with help from parse_clade, then clears the XML event history for the phylogeny element and returns control to the top-level parsing function. rooted rerootable confidences propertiesclade_relationssequence_relations) confidencepropertyclade_relationsequence_relationrqcladeNz)Phylogeny object should only have 1 claderrr})dateid)name descriptionrMisidentified tag: r )rrraattribrwr8r4rr _parse_cladergetattrrsetattrrkrArrr)rxr=r} list_typesryrSr9r4r r r r:s@      zParser._parse_phylogeny)colorrsbinary_charactersrr distributions referencesr)r distribution referencer) branch_lengthrnode_idwidthcCstjdi|j}|jdurt|j|_g}|jD]\}}t|j\}}|dkr\|dkr6|j | |q|dkrD|j | |q|dkrR|j ||q||jvr\| ||dkr|dkrk||S||dkrrq|||jvrt||j| t|||q||jvrt||t|||q|dkr|jdurtd t|j|_q|d krt|j|_q|d krt|j|_q|d krt|j|jd |_q|t dkr|j! |!||||qtd|q|S)z;Parse a Clade node and its children, recursively (PRIVATE).Nrqrtaxonomysequencerrrz7Attribute branch_length was already set for this Clade.rrrproviderrrr )"rrrrrgrwr8r4cladesrr taxonomies_parse_taxonomy sequences_parse_sequence_clade_tracked_tagsrpop_clade_list_typesr_clade_complex_typesrrrhrArrkrIdrPgetrrr)rxr=r tag_stackryrSr9r4r r r rush            "zParser._parse_cladecCstjd i|j}|jD]d\}}t|j\}}|dkrp|dkr&||S|dvr6t||t|||q |dkrD|j | |q |dkrOt |j |_q |dvr[t|||j q |tdkrp|j |||||q |S) z%Parse a molecular sequence (PRIVATE).rrr) accessionmol_sequridomain_architecture annotationr)symbollocationrNr )rSequencerrwr8r4rrr annotationsrrrkrArrr)rxr=rryrSr9r4r r r rs*  zParser._parse_sequencecCstjd i|j}|jD]e\}}t|j\}}|dkrq|dkr&||S|dvr6t||t|||q |dkrD|j t |j q |dkrP|j |j q |dvr\t|||j q |tdkrq|j |||||q |S) z2Parse taxonomic information for a clade (PRIVATE).rrr)rr common_namesynonym)codescientific_name authorityrankrNr )rTaxonomyrrwr8r4rrr common_namesrrkrAsynonymsrr)rxr=rryrSr9r4r r r rs*  zParser._parse_taxonomycs4tj|||j|jr|jpdfdd|DdS)z/Create an Other object, a non-phyloXML element.Ncs$g|]}j|gt|jRqSr )rr8r4rDrxr r rFsz Parser.other..)valuechildren)rOtherrrArP)rxrSr9rr rr rs z Parser.othercCst|j|dS)zCreate accession object.source)r AccessionrArPrrxrSr r r rzParser.accessionc CsDtjdtt|dt|d|jt|d|jt|d|jd|j S)zCreate annotation object.descrrr)rrrrNr ) r AnnotationrkrBr@rrJrrrrr r r rs    zParser.annotationc Csrdd}tj|dt|dt|dt|dt|dt|d|t|d |t|d |t|d |d S) z Create binary characters object.cSs t|dS)z$Get binary characters from subnodes.bc)rMrSr r r bc_getters z+Parser.binary_characters..bc_gettertype gained_count lost_count present_count absent_countgainedlostpresentabsent) rrrrrrrrr)rBinaryCharactersrrer@)rxrSrr r r rs        zParser.binary_charactersc Cs6tj|d|d|d|dt|d|jdS)z!Create clade relationship object.rid_ref_0id_ref_1distancerrr)r CladeRelationrr@rrr r r r s zParser.clade_relationcs&fdddD\}}}t|||S)zCreate branch color object.c3s|] }t|tVqdSr!)rBrc)r"rrr r r%s  zParser.color..redgreenblue)r BranchColor)rxrSrrrr rr rs  z Parser.colorcCstt|j|dS)zCreate confidence object.r)r ConfidencerhrArrr r r rrzParser.confidencec Cs<tj|dtt|dt|dtt|dtt|dtdS)zCreate date object.unitrrminimummaximum)rrrrr)rDaterrkrBrgrr r r r s    z Parser.datecCs.tjtt|dt|d|jt|d|jdS)z&Create geographic distribution object.rpointpolygon)rpointspolygons)r DistributionrkrBrJrrrr r r r*s    zParser.distributioncCsBtj|jt|ddt|dt|d|ddS)zCreate protein domain object.fromr1torr)rr)r ProteinDomainrArPrcrrhrr r r domain2s  z Parser.domaincCs"tjt|dt|d|jdS)z"Create domain architecture object.lengthr)rdomains)rDomainArchitecturercrrJrrr r r r<s  zParser.domain_architecturec Cs<tjt|dt|dtt|dtt|dtt|d|jdS)zCreate events object.r duplications speciationslossesrrrrrr)rEventsrBrcr@rrr r r rsCs    z Parser.eventscCs&|dp |d}t|j|S)zCreate identifier object.rr)rrrrArP)rxrSrr r r rMsz Parser.idcCs.|d}|dur t|}tj|j|dS)z!Create molecular sequence object. is_alignedNr)rr[rMolSeqrArP)rxrSrr r r rRs zParser.mol_seqcCs8tj|dt|dtt|dtt|dt|ddS)z,Create point object, coordinates of a point.geodetic_datumlatlongaltalt_unit)rr)rPointrrBrgrr r r rYs   z Parser.pointcCstjt|d|jdS)z&Create polygon object, list of points.r)r)rPolygonrJrrr r r rcszParser.polygonc Cs:tj|j|d|d|d|d|ddS)z*Create properties from external resources.ref applies_todatatyperid_ref)rr )rPropertyrArPrrr r r rgszParser.propertycCstj|dt|ddS)z#Create literature reference object.doir)rr)r ReferencerrBrr r r rrszParser.referencec Cs:tj|d|d|dt|dt|d|jdS)zHCreate sequence relationship object, relationship between two sequences.rrrrrr)rSequenceRelationrrhr@rrr r r rvs  zParser.sequence_relationcCs&tj|jt|d|ddS)z(Create uri object, expected to be a url.rrrr)rUrirArPrkrrr r r rs  z Parser.uriN)%rrr r rzrrrrrsetunionr^rrrrrrrrrrrrrrrrsrrrrrrrrr r r r rsN  .  9        rcCs4t|tr t|St|trt|St|S)zEConvert a Python primitive to a phyloXML-compatible string (PRIVATE).)rrgstrupperboollower)rr r r _serializes    rcCs0i}|D]}t||}|durt|||<q|S)zNCreate a dictionary from an object's specified, non-None attributes (PRIVATE).N)rr)r.attrsr_r`r{r r r _clean_attribs  rFcs$fdd}dd|_|S)z2Handle to serialize nodes with subnodes (PRIVATE).cstt|}D]2}t|tr't||dur&|t||t||q |\}}t||D] }|t|||q0q rFt|j|_ |S)z$Wrap nodes and subnodes as elements.N) rElementrrrrrrrrA)rxr.rSsubnmethodpluralitemattribshas_textsubnodesr4r r wrappeds  z _handle_complex..wrappedz Serialize a z and its subnodes, in order.r )r4r#r%r$r&r r"r _handle_complexsr(csfdd}dd|_|S)z+Handle to serialize simple nodes (PRIVATE).cst}t||_|S)zWrap node as element.)rrrrArxr.rSr3r r r&s  z_handle_simple..wrappedzSerialize a simple z node.r')r4r&r r3r _handle_simples r*c@s:eZdZdZddZedfddZddZd d Ze d d d Z e dddZ e dddddZ e dddZ ddZe dddZe dddZe dd dddZe d!d"d#Ze d$dd%Zd&d'Ze d(d)d*Ze d+dd,Ze d-d.dddZe d/d0dddZe d1d.dddZe d2d3d4Ze d5dd6Ze d7d8dddZe d9d:d;Ze dZe d?ddZe d@ddAZ e dBdCdddZ!e"dDZ#e"dEZ$e"dFZ%e"dGZ&e"dHZ'e"dIZ(e"dJZ)e"dKZ*e"dLZ+e"dMZ,e"dNZ-e"dOZ.e"dPZ/e"dQZ0e"dRZ1e"dSZ2e"dTZ3e"dUZ4e"dVZ5e"dWZ6e"dXZ7e"dYZ8e"dZZ9e"d[Z:e"d\Z;e"d]ZsF      +