def on_pmid_button_clicked(self, pmid,*args):
"""
Given a PMID, resquest, download the abstract from NCBI,
save it in the current folder - as pmid_number.xml -
and load it on the interface for annotation
"""
#TODO: this method could be better organized
pmid = self._pmid_entry.get_text()
pre_process = self._preprocess_checkbutton.get_active()
parallel = self._parallel_checkbutton.get_active()
#Check if the input has a valid PMID
try:
int(pmid)
except:
self.show_message("Ops! Check if you provided a valid PMID")
return 0
self.current_open_file = pmid + ".xml"
raw_ncbi_xml= entrez.get_abstract_from_ncbi(pmid, self.current_open_file)
data = xml_tools.get_data_from_ncbi_xml(self.current_open_file)
#just display the raw abstract, withou any preprocessing
if data and not pre_process:
self._clear_interface()
self._update_interface(data)
#automaticaly recognize MESH terms and display the annotated abstract
elif data and pre_process:
self.show_message("This can take a few seconds. Do not close the program")
#get mesh terms from database
abstract_text = xml_tools.get_abstract_text_from_etree(data)
abstract_text = tools.translate_greek2latin(abstract_text)
#recognize MESH terms automatically
annotated_filename = "annotated_" + self.current_open_file
self.current_open_file = annotated_filename
self.create_annotated_abstract_xml(raw_ncbi_xml,
abstract_text,
annotated_filename,
parallel)
#load annotated xml
data,tags = xml_tools.load_annotated_xml(annotated_filename)
if not data:
self.show_message("PubMed Tagger was not able to parse %s" %annotated_filename)
return 0
self._clear_interface()
self._update_interface(data, textbuffer=True)
print tags.keys()
self.tag_text(tags)
self.update_annotation_liststore(tags)
def on_filechooser_ok_button_clicked(self, *args):
"""
Get the active filename in the filechooser_window and
load its data using xml_tools.load_annotated_xml function
"""
self._filechooser_window.hide()
filename = self._filechooser_window.get_filename()
self.current_open_file = filename
data,tags = xml_tools.load_annotated_xml(filename)
if not data:
self.show_message("PubMed Tagger was not able to parse %s" %filename)
return 0
self._update_interface(data, textbuffer=True)
self.update_annotation_liststore(tags)
self.tag_text(tags)
