code for computational search

gistfile1.txt

query = 'ALS agencies in King County'

def a_vs_b(a, b):
    return

def agencies_in_location(type, location):
    return 

query_formats_to_functions = {}
non_values_to_query_formats = {} # given tuple of the non values return query format

def extract_non_values_from_format(format):
    # find indexes of {% and %} and store as pairs
    start = 0
    start_end_indexes_for_variable_names = []
    for i in range(format.count('{%')):
        start_end_indexes_for_variable_names.append((format.find('{%', start), format.find('%}', start)))
        start = start_end_indexes_for_variable_names[-1][1]
        
    # find starting and ending indexes for the non variable elements
    start_end_indexes_for_non_variable_elements = []
    for i in range(len(start_end_indexes_for_variable_names) - 1):
        start = start_end_indexes_for_variable_names[i][1] + 3
        end = start_end_indexes_for_variable_names[i + 1][0] - 1
        start_end_indexes_for_non_variable_elements.append((start, end))
        
    non_variable_elements = [format[start:end] for start, end in start_end_indexes_for_non_variable_elements]
    #return start_end_indexes_for_variable_names
    #print non_variable_elements
    return non_variable_elements

def add_format(format, function):
    query_formats_to_functions.update({format: function})
    non_values_to_query_formats.update({tuple(extract_non_values_from_format(format)): format})

add_format('{% a %} vs. {% b %}', a_vs_b)
add_format('{% agency_type %} agencies in {% location %}', agencies_in_location)

def find_matching_format(query):
    for non_values_tuple in list(non_values_to_query_formats):
        if not False in [non_value in query for non_value in non_values_tuple]:
            return non_values_to_query_formats[non_values_tuple]
    raise LookupError, 'No matching format found'

def extract_values_from_query(format, query):
    non_values = extract_non_values_from_format(format)
    
    # find start and end indexes of non values
    start_end_indexes_of_non_values = []
    for i, non_value in enumerate(non_values):
        start = query.find(non_value)
        end = start + len(non_value)
        start_end_indexes_of_non_values.append((start, end))
        
    # find start and end indexes of values
    start_end_indexes_of_values = []
    if start_end_indexes_of_non_values[0][0] > 0 and start_end_indexes_of_non_values[-1][1] < len(query) - 1:
        if len(non_values) == 1:
            start_end_indexes_of_values = [(0, start_end_indexes_of_non_values[0][0] - 1), (start_end_indexes_of_non_values[0][1] + 1, None)]
            
    
    return [query[start:end] for start, end in start_end_indexes_of_values]
        
print find_matching_format('Boston vs. Seattle')
print find_matching_format('BLS agencies in New York')
print find_matching_format('should raise an error')