ResumeUniversities.py

# Define multiple universities names patterns
sub_patterns = [
    
                '[A-Z][a-z]* [A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* at [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]*',
    
                '[A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* at [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* University of [A-Z][a-z]*',
    
                '[A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]*',
                '[A-Z][a-z]* University of [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* University of [A-Z][a-z]* at [A-Z][a-z]*',
                '[A-Z][a-z]* University of [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* University of [A-Z][a-z]*',
    
                'University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                'University of [A-Z][a-z]* [A-Z][a-z]* at [A-Z][a-z]*',
                'University of [A-Z][a-z]* at [A-Z][a-z]* [A-Z][a-z]*',
                'University of [A-Z][a-z]* at [A-Z][a-z]*',
                'University of [A-Z][a-z]* [A-Z][a-z]*',
                'University of [A-Z][a-z]*',
    
                '[A-Z][a-z]* [A-Z][a-z]* University',
                '[A-Z][a-z]* University',
                '[A-Z]* University',
    
                '[A-Z][a-z]* Institute of [A-Z][a-z]* & [A-Z][a-z]*',
                '[A-Z][a-z]* Institute of [A-Z][a-z]* [A-Z][a-z]*',
                '[A-Z][a-z]* Institute of [A-Z][a-z]*',
    
                '[A-Z][a-z]* [A-Z][a-z]* Community College',
                '[A-Z][a-z]* Community College',
    
                'College of [A-Z][a-z]* & [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* College of [A-Z][a-z]*',
                '[A-Z]* College of [A-Z][a-z]*',
                '[A-Z][a-z]* College of [A-Z][a-z]*',
                '[A-Z][a-z]* [A-Z][a-z]* College',
                '[A-Z][a-z]* College',
    
                '[A-Z][a-z]* [A-Z][a-z]* High School',
                '[A-Z][a-z]* High School',
    
                '[A-Z][a-z]* [A-Z][a-z]* Military Academy',
                '[A-Z][a-z]* Military Academy',
    
                'Universidad de los [A-Z][a-z]*',
                'Universidad de las [A-Z][a-z]*',
                'Universidad de [A-Z][a-z]*',

                'Universidad [A-Z][a-z]* de [A-Z][a-z]* [A-Z][a-z]*',
                'Universidad [A-Z][a-z]* de los [A-Z][a-z]*',
                'Universidad [A-Z][a-z]* de las [A-Z][a-z]*',
                'Universidad [A-Z][a-z]* de [A-Z][a-z]*',
                ]

# Join university patterns
university_patterns = '({})'.format('|'.join(sub_patterns))

def get_univerisities(self):
    
    # Find all strings in text that follow university names patterns
    university_matches = re.findall(university_patterns, self.raw_text)
    
    # Declare list of university names to remove if found
    universities_to_remove = ["University","University "," University"," University ","College"]
  
    # Assign list of universities names
    self.universities = list(set(university_matches))

    # Loop through the list of universities to remove
    for university in universities_to_remove:
        try:
            self.universities.remove(university)
        except:
            pass