limsq.py

#!/usr/bin/env python2

import argparse
import os
import commands

""" Var definition """
options_to_url = {"project" : "project__name=",
                  "subproject" : "subprojects__subproject_name=",
                  "pedigree" : "pedigree=",
                  "cohort" : "cohort__name=",
                  "sample" : "barcode=",
                  "SampleName" : "name=",
                  "aliquot" : "library__aliquot__aliquot_barcode=",
                  "library" : "library__name=",
                  "material" : "subprojects__sample_type__name=",
                  "flowcell" : "library__loadedwith__lane__flowcell__name=",
                  "lane" : "library__loadedwith__lane__lane=",
                  "index" : "library__multiplex_index__multiplex_index_name=",
                  "GenomeRef" : "library__loadedwith__mapping__reference_filename=",
                  "subprojectApplication" : "subprojects__application__name=",
                  "motherId" : "mother_id=",
                  "fatherId" : "father_id=",
                  "libraryGBS" : "library__gbs_index__multiplex_index_name=",
                  "mappingVersion" : "library__loadedwith__mapping__version=",
                  "mapperName" : "mapper=",
                  "mapperVersion" : "mapperversion="
}

options_to_url_QC={"libraryPassFail" : "library__passfail__status=",
                   "LanePassFail" : "library__loadedwith__passfail__status="}

lims_header = "project;subproject;id;SampleName;sample;library;SubprojectId;material;aliquot;flowcell;lane;index;libraryPassFail;LanePassFail;libApplication;Kit;Organism;Sex;pedigree;cohort;GenomeRef;SplitIndexcohort;SubprojectCov;motherId;fatherId;library__gbs_index__multiplex_index_name;readLength;sampleStatus;mappingVersion;mapperName;mapperVersion"




def main():
    lims_url = "https://lims.cnag.cat/lims/lims_q_2/?"
    options = get_options()
    lims_url = generate_lims_query(lims_url,options,options_to_url)
    lims_output = commands.getstatusoutput("wget -q -O- \""+ lims_url + "\" |sort|uniq|grep -v \"^#\"")[1].split("\n")  
    #For some reason lims doesn't allow me to query everything at once, with the filters so they have to be added one by one
    lims_filtering_list = generate_filtering_list(lims_url, options)
    print(lims_header)
    for lims_out_line in lims_output:
        if lims_out_line not in lims_filtering_list:
            print(lims_out_line)
         

#Some strange characters are permitted which causes conflicts in the html ("#" for instance)
#Convert them to hex so all the options are now codified as hex ASCII
def text_fixer(character):
     return("%"+str(hex(ord(character))).split("x")[1])
 
def generate_lims_query(lims_url, options, options_to_url):
    for url_key in options_to_url.keys():
        url_to_query = options_to_url[url_key]
        value = getattr(options, url_key)
        if value != None:
            for query_value in value.split(","):
                query_value = "".join(map(str,map(text_fixer,query_value)))
                lims_url = add_to_lims_query(lims_url, url_to_query, query_value)
    return(lims_url)

def generate_filtering_list(lims_url,options):
    lims_filtering_list = list()
    for url_key in options_to_url_QC.keys():
         lims_url_QC = lims_url
         url_to_query = options_to_url_QC[url_key]
         value = getattr(options, url_key)
         if value != None:
            for query_value in value.split(","):
                lims_url_QC = add_to_lims_query(lims_url_QC, url_to_query, query_value)
                lims_filtering_list = lims_filtering_list + commands.getstatusoutput("wget -q -O- \""+ lims_url_QC + "\"")[1].split("\n")
    return(lims_filtering_list)

def add_to_lims_query(lims_url, url_to_query, query_value):
    lims_url = lims_url+"&"+url_to_query+query_value
    return(lims_url)

def get_options():
    parser = argparse.ArgumentParser()
    parser.add_argument('-project',
			'-p', 
			dest="project",
                        help="OPT CSV",
                        )
    parser.add_argument('-subproject',
                        '-sp',
                        help="OPT CSV", 
                        dest="subproject",
                        )
    parser.add_argument('-pedigree',
                        '-ped', 
                        help="OPT CSV",
                        dest="pedigree",
                        )
    parser.add_argument('-cohort',
                        help="OPT CSV",
                        dest="cohort",
                        )
    parser.add_argument('-sample',
                        '-s',
                        help="OPT CSV",
                        dest="sample",
                        )
    parser.add_argument('-sampleName',
                        '-sn', 
                        help="OPT CSV",
                        dest="SampleName",
                        )
    parser.add_argument('-aliquot',
                        '-a', 
                        help="OPT CSV",
                        dest="aliquot",
                        )
    parser.add_argument('-library',
                        '-lib', 
                        help="OPT CSV",
                        dest="library",
                        )
    parser.add_argument('-material',
                        '-m', 
                        help="OPT CSV",
                        dest="material",
                        )
    parser.add_argument('-flowcell',
                        '-f', 
                        help="OPT CSV",
                        dest="flowcell",
                        )
    parser.add_argument('-lane',
                        '-l', 
                        help="OPT CSV",
                        dest="lane",
                        )
    parser.add_argument('-index',
                        '-i', 
                        help="OPT CSV",
                        dest="index",
                        )
    parser.add_argument('-genome',
                        '-g', 
                        help="OPT CSV",
                        dest="GenomeRef",
                        )
    parser.add_argument('-subprojectapplication',
                        '-spap',
                        help="OPT CSV",
                        dest="subprojectApplication",
                        )
    parser.add_argument('-libpf',
                        '-libraryPassFail',
                        help="OPT CSV of library PassFail status to EXCLUDE",
                        dest="libraryPassFail",
                        )
    parser.add_argument('-lanepf',
                        '-lanePassFail',
                        help="OPT CSV of lane PassFail status to EXCLUDE",
                        dest="LanePassFail",
                        )
    parser.add_argument('-mother_id',
                        '-mid',
                        help="OPT CSV",
                        dest="motherId",
                        )
    parser.add_argument('-father_id',
                        '-fid',
                        help="OPT CSV",
                        dest="fatherId",
                        )
    parser.add_argument('-gbs_index',
                        '-gbs',
                        help="OPT CSV",
                        dest="libraryGBS",
			            )
    parser.add_argument('-mappingVersion',
                        '-mV',
                        help="OPT CSV",
                        dest="mappingVersion",
                        )
    parser.add_argument('-mapperName',
                        '-MN',
                        help="OPT CSV",
                        dest="mapperName",
            			)
    parser.add_argument('-mapperVersion',
                        '-MV',
                        help="OPT CSV",
                        dest="mapperVersion",
            			)
                       
    options = parser.parse_args()
    return options


if __name__ == "__main__":
    exit(main())