import numpy as np

# I have a sorted array which describes all possible values (distinct) for the field 'name'

names = np.array(["jacques", "john", "pete", "robert"], dtype="|S8")

# I have a recarray 'bank' which describes for all the people identified by their name the money they have.
# Not every names from 'names' are represented in 'bank'.
# Every name encountered in 'bank' should be found in 'names'.
# Some names from 'names' can be encountered more than once in 'bank'.
# 'bank' is sorted by name.

bank_descriptor = [('name', '|S8'), ('money', int)]
bank = np.array([("jacques", 100), ("jacques", 200), ("jacques", 550),
                  # (no line for john)
                  ("pete", 50), ("pete", 240),
                  ("robert", 1000)],
                  dtype=bank_descriptor)

# I want an array of integers called 'bank_encoded' such as
# bank['name'][i] == names[bank_encoded[i]]
# With such an array, I can forget the field 'name' and I will
# only work with arrays of integers to summarize data or merge some arrays together.

bank_encoded = np.zeros(bank['name'].shape[0], dtype=int)

i_bank, i_names = 0, 0

while i_bank < bank['name'].shape[0] and i_names < names.shape[0]:
    if bank['name'][i_bank] == names[i_names]:
        bank_encoded[i_bank] = i_names
        i_bank += 1
    elif bank['name'][i_bank] > names[i_names]:
        i_names += 1
    else:
        # bank['name'][i_bank] < names[i_names] means that the considered name is not in 'names'
        raise Exception, "Some names in bank are not in 'names'"

if i_bank != bank['name'].shape[0]:
    # there are still some values to examine in bank, but no more available values in 'names'
    raise Exception, "Some names in bank are not in 'names'"
else:
    # easy summarization of bank
    bank_summarized = np.zeros(names.shape[0], dtype=int)   
    for i in range(bank['money'].shape[0]):
        bank_summarized[bank_encoded[i]] += bank['money'][i]
    print bank_summarized