mjgoldman16

## GATHERING INFORMATION BASED ON EDUATION LEVEL
bachelors = [re.findall("(?<![A-Z])B\.?S\.?c?(?![A-Z])|(?<![A-Z])B\.?A\.?(?![A-Z])|BACHELOR|UNDERGRAD.{0,40} DEGREE|ASSOCIATE'?S?.{20}DEGREE",i, re.IGNORECASE) for i in desc_col.values]
mba = [re.findall("([\s|-|/]MBA[\s|-|/]|[\s|-|/]MBUS[\s|-|/]|[\s|-|/]MBS[\s|-|/]|MASTERS? OF BUSINESS)",i,re.IGNORECASE) for i in desc_col.values]
masters = [re.findall("(MASTER'?S?.{0,40}DEGREE|GRADUATE.{0,40}DEGREE|(?<![A-Z])M\.?S\.?(?![A-Z]|\sDYNAMICS|,\sDSC)(?!-?~?\s?OFFICE|\sEXCEL|\sWORD|\sACCESS|-?\s?SQL)|ADVANCED?.{0,40}DEGREE)",i,re.IGNORECASE) for i in desc_col.values]
phd = [re.findall("(PH\.?D|ADVANCED?.{0,40}DEGREE|DOCTORA[TE|L]|POST-?\s?GRADUATE)",i,re.IGNORECASE) for i in desc_col.values]

## GATHERING INFORMATION BASED ON SKILLS
python = [re.findall("PYTHON",i,re.IGNORECASE) for i in desc_col.values]
R = [re.findall("[\s,\.\-(\[\\\]R[\s,\.\-)\]\\\]",i,re.IGNORECASE) for i in desc_col.values]
SQL = [re.findall("SQL",i,re.IGNORECASE) for i in desc_col.values]

mjgoldman16

start_urls = [("https://www.glassdoor.com/Job/" +
                   str(city[0]) +  # The first element of the city name
                   "-data-scientist-jobs-SRCH_IL.0," +  # The job search term
                   str(len(city[0])) +  # The last element of the city name
                   "_IC" +
                   str(city[1]) +  # The unique city ID
                   "_KO" +
                   str(len(city[0]) + 1) +  # The first element of the job search term
                   "," +
                   str(len(city[0]) + 15) +  # The last element of the job search term (data-scientist is 15 chr long)