H&h0ddlZddlZddlZddlmZddlZddlZddlZddl Z ddl m Z ddl d dZ d dZd dZedk(rddlZej$yy) N)etree)Refactor)*cht|dtjjjfd}|;g}t t |D] }|||d|}|j |"|Sg}t t |D] }|d|||}|j |"|S)aI Search Catalogue of Life for for direct children of a particular taxon. :param name: The string to search for. Only exact matches found the name given will be returned, unless one or wildcards are included in the search string. An * (asterisk) character denotes a wildcard; a % (percentage) character may also be used. The name must be at least 3 characters long, not counting wildcard characters. :param id: The record ID of the specific record to return (only for scientific names of species or infraspecific taxa) :param format: format of the results returned. Valid values are format=xml and format=php; if the format parameter is omitted, the results are returned in the default XML format. If format=php then results are returned as a PHP array in serialized string format, which can be converted back to an array in PHP using the unserialize command :param start: The first record to return. If omitted, the results are returned from the first record (start=0). This is useful if the total number of results is larger than the maximum number of results returned by a single Web service query (currently the maximum number of results returned by a single query is 500 for terse queries and 50 for full queries). :param checklist: The year of the checklist to query, if you want a specific year's checklist instead of the lastest as default (numeric). Valid years are 2010 through the previous year from the current date. If none given, the "lastest" checklist is used You must provide one of name or id. The other parameters (format and start) are optional. Returns A list of data.frame's. Usage:: import pytaxize pytaxize.col_children(name=["Apis"]) # An example where there is no classification, results in data.frame with no rows pytaxize.col_children(id=["4fdb38d6220462049eab9e3f285144e0"]) # Use a specific year's checklist pytaxize.col_children(name=["Apis"], checklist="2012") pytaxize.col_children(name=["Apis"], checklist="2009") # Pass in many names or many id's out = pytaxize.col_children(name=["Buteo","Apis","Accipiter"], checklist="2012") # get just one element in list of output out[0] ic d}|nCt|}|dvrtjdd|z|}nd}tjd||}|| d d}|jDcic] \}}| || }}}t ||d j }|j d }t|d k(rtjd g} tt|D]V} || j} | jttgd| ddDcgc]}|jc}X| Scc}}wcc}w)N.https://www.catalogueoflife.org/col/webservice201220112010colannual-checklist/@https://www.catalogueoflife.org/annual-checklist/year/webserviceyearfull)nameidformatresponsestartgetrequest//child_taxa//taxonrz Please enter a valid search namerrrank)strresubitemsrxmlxpathlensysexitrange getchildrenappenddictziptext)xy checklisturlpayloadkvtt childtaxaoutlistitt_rrs [/mounts/lovelace/software/anaconda3/envs/py312/lib/python3.12/site-packages/pytaxize/col.pyfunczcol_children..func=s:>   II22ffU$7)$CSIXffVY4!fQVW$+MMOEDAqq}1a4EE c7E 2 6 6 8HH23 y>Q  HH7 8s9~& AA,**,C NNS-BQ/H1/HIJ   F0Is% D90D9D?N)assert_range_numericdatetimenowrr'r$r)) rrrrr/r:tempr7sss `` r9 col_childrenr@ s^D(*;*;*?*?*A*F*FG6 zs4y! Ad1gtY/B KKO  s2w AdBqE9-B KKO  c d d fd}t|trt}|j||}g}t t |D]"}|||||||} |j| $|S)a :param name: The string to search for. Only exact matches found the name given will be returned, unless one or wildcards are included in the search string. An * (asterisk) character denotes a wildcard; a % (percentage) character may also be used. The name must be at least 3 characters long, not counting wildcard characters. :param downto: The taxonomic level you want to go down to. See examples below. The taxonomic level IS case sensitive, and you do have to spell it correctly. See rank_ref for spelling. :param checklist: The year of the checklist to query, if you want a specific year's checklist instead of the lastest as default (numeric). :param format: The returned format (default = None). If NULL xml is used. Currently only xml is supported. :param start: The first record to return (default = None). If NULL, the results are returned from the first record (start=0). This is useful if the total number of results is larger than the maximum number of results returned by a single Web service query (currently the maximum number of results returned by a single query is 500 for terse queries and 50 for full queries). Returns a list of Pandas DataFrame's. Usage:: import pytaxize pytaxize.col_downstream(name="Apis", downto="Species") pytaxize.col_downstream(name="Insecta", downto="Order") # An example that takes a bit longer pytaxize.col_downstream(name=["Insecta","Animalia"], downto="Class") # Using a checklist from a specific year pytaxize.col_downstream(name="Apis", downto="Species", checklist=2011) rrc.|}nAt|}|dvrtjdd|z}ntjd|}tjdd}t j |}|jDcgc]}|} }g} tt| D]} || | v} | j| |jt j| dg} | | j| jd k(jd | j d }|jDcgc]}|j#d d }}|}d }t jd g}g}d }|d k(r|dz }fd}|||}||dDcgc]}|c}vr$|j|j|d|k(|j|d|k7j d d kDrI|dDcgc]}|}}g}tt|D]} |j|| |v||}n3t%}|j|t j|dg}t'|d|k(rd}n|d}d }|d k(r|Scc}wcc}wcc}wcc}w)Nr r rrpytaxizezdata/rank_ref.csvmatchcolumnsTr,notrankNamec| d d}|jDcic] \}}| || }}}t||dj}|jd}g}t t |D]A}||j } |j| ddDcgc]}|jc}Ctj|gd} | Scc}}wcc}w) Nr)rrrrrrrrrrF) r!rr"r#r'r$r(r)r,pd DataFrame) r-r0r1r2r3r4r5r6r7r8dfrrs r9 searchcolz/col_downstream..func..searchcols#$ &8,3MMOMDAqq}1a4MMc7E:>>@HH%:; s9~.>A#A,224CNNCG#\\'3GH N $=s CCC rfamr)rrr pkg_resourcesresource_filenamerMread_csvranksr'r$r)joinrNlocrEindexshapesplitlistall)rdowntorrr/r0 rank_ref_pathdatr-stuffthingsr7r?dat2subsettoranktogetstop_notoutoutiterrPr4shbbvalscol_urlyear_urls `` r9r:zcol_downstream..funcs  CII22ffU$7)$CWMffVY9%77 DWX kk-(II&q&&s5z" A58#B MM"  xx VgY?@dhhtzzT1288;TZZ]K+1<<8a!''#,q/88zl3un AID 5#&BRZ0!00 266"V*"678vvbjF*+11!4q8!#F,Aa,,s2w-A))BqEVO,-Bv F#dVH=6&>V+,vIunL i'921-s J/J J * J) isinstancerr[r)r'r$) rr]rrr/r:nametmpr>r7r4rmrns @@r9col_downstreamrqesH?GQHAF$&t D 3t9  $q'665) < B KrAcfd}|9g}tt|D]}|j|||d |St|}g}tt|D]}|j|d|| |S)aw Search Catalogue of Life for taxonomic IDs :param name: The string to search for. Only exact matches found the name given will be returned, unless one or wildcards are included in the search string. An * (asterisk) character denotes a wildcard; a % (percentage) character may also be used. The name must be at least 3 characters long, not counting wildcard characters. :param id: The record ID of the specific record to return (only for scientific names of species or infraspecific taxa) :param start: The first record to return. If omitted, the results are returned from the first record (start=0). This is useful if the total number of results is larger than the maximum number of results returned by a single Web service query (currently the maximum number of results returned by a single query is 500 for terse queries and 50 for full queries). :param checklist: The year of the checklist to query, if you want a specific year's checklist instead of the lastest as default (numeric). You must provide one of name or id. The other parameters (format and start) are optional. Usage:: import pytaxize pytaxize.col_search(name=["Apis"]) pytaxize.col_search(id=15669061) # Many names pytaxize.col_search(name=["Apis","Puma concolor"]) # Many ids - DOESNT WORK pytaxize.col_search(id=[15669061,6862841]) # An example where there is no data pytaxize.col_search(id=11935941) # Example with more than 1 result pytaxize.col_search(name=['Poa']) cd} n8 dvrtjdd z|}nd}tjd |}|| d}t||d j}|j d }g}t t |D]r}||}i} t t |D]A} || jD])} | j| j| ji+C|j| t|S) Nrr r rrr)rrrrrz//result) rr rr"r#r'r$riupdatetagr,r))r-r.r0r1r4r`r6r7r8eachger/rs r9r:zcol_search..funcs>   22ffU$7)$CSIXffVY4A6 c7E 2 6 6 8$s5z" !A(CD3s8_ 1Q1AKK01 1 NN4  !rAN)r.)r-r.)r'r$r)r)rrrr/r:r>r7s `` r9 col_searchrysT0 zs4y! /A KKT!W- . / K Ws2w /A KKtr!u- . / KrA__main__)NNNNN)NNNN)r%requestsr<lxmlrpandasrMrjsonrRpytaxize.refactorrpytaxize.utilsr@rqry__name__doctesttestmodrAr9rsV  &WrqfKx zGOOrA