Function to return subword of a camelcase string

Question

Given a camelcase string and an index, return the subword of the string that includes that index, e.g.:

find_word('CamelCaseString', 6) -> 'Case'
find_word('ACamelCaseString', 0) -> 'A'

My code:

def find_word(s, index):
    for i in range(index, 0, -1):
        if s[i].isupper():
            left = i
            break
    else:
        left = 0

    for i in range(index, len(s)-1):
        if s[i].islower() and s[i+1].isupper() or s[i:i+2].isupper():
            right = i
            break
    else:
        right = len(s) - 1

    return s[left:right+1]

Can this be made more concise/efficient?

Answer 1

Review

• Add docstrings and tests... or both in the form of doctests!

  def find_word(s, index):
      """
      Finds the CamalCased word surrounding the givin index in the string
  
      >>> find_word('CamelCaseString', 6)
      'Case'
      >>> find_word('ACamelCaseString', 0)
      'A'
      """
  
      ...
  

• Loop like a native.

  Instead of going over the indexes we can loop over the item directly

  range(index, 0, -1)
  

  We can loop over the item and index at the same time using enumerate

  for i, s in enumerate(string[index:0:-1])
  

  However this would be slower since it will create a new string object with every slice.

• If we can be sure that the givin string is a CamalCase string

  Then we can drop some of your second if statement

  if s[i].islower() and s[i+1].isupper() or s[i:i+2].isupper():
  

  Would be

   if s[i+1].isupper():
  

• Actually your code (from a performance aspect) is quite good

  We could however use a while loop to increment both side at once, for a little performance gain.

(slower, yet more readable) Alternative

A different approach to finding CamalCase words can be done with regex,

We can find all CamalCase words with the following regex: r"([A-Z][a-z]*)"

And we can use re.finditer to create a generator for our matches and loop over them, and return when our index is in between the end and the start.

import re

def find_word_2(string, index):
    for match in re.finditer(r"([A-Z][a-z]*)", string):
        if match.start() <= index < match.end():
            return match.group()

NOTE This yields more readable code, but it should be alot slower for large inputs.

Answer 2

An alternative approach may involve trading space for time and pre-calculate mappings between letter indexes and the individual words. That would make the actual lookup function perform at \$O(1)\$ with \$O(n)\$ sacrifice for space. This may especially be useful if this function would be executed many times and needs a constant time response for the same word.

And, as this is tagged with interview-questions, I personally think it would be beneficial for a candidate to mention this idea of pre-calculating indexes for future constant-time lookups.

We could use a list to store the mappings between indexes and words:

import re


class Solver:
    def __init__(self, word):
        self.indexes = []
        for match in re.finditer(r"([A-Z][a-z]*)", word):
            matched_word = match.group()
            for index in range(match.start(), match.end()):
                self.indexes.append(matched_word)

    def find_word(self, index):
        return self.indexes[index]


solver = Solver('CamelCaseString')
print(solver.find_word(2))  # prints "Camel"
print(solver.find_word(5))  # prints "Case"

Answer 3

Actually you found an example where looping over indices is ok. What you messed up is the search for the right end. When doing slicing the second value is not included 'abc[0:2]gives 'ab'. So your right shall be past the last included character, that is the next uppercase one. We rewrite the second loop to follow the style of the first one

for i in range(index+1, len(s)):
    if s[i].isupper():
        right = i
        break
else:
    right = len(s)

and return the slice

return s[left:right]

That is IMHO also the most readable solution following the KISS principle (and some python Zen)