Extracting Information from Strings with NumPy's char.partition()

Functionality

• Returns a new array with three elements for each input element:
  • The part before the separator (leftmost portion)
  • The separator itself
  • The part after the separator (rightmost portion)
• Splits elements in a NumPy array of strings (or a single string) at the first occurrence of a specified separator.

Syntax

import numpy as np

output = np.char.partition(input_array, separator)

Parameters

• separator: The substring (delimiter) used to split the strings. Can be a string or a regular expression.
• input_array: A NumPy array containing strings or a single string.

Return Value

• A new NumPy array with the same shape as the input array, but each element is a tuple (or list) containing the three parts:
  • The part before the separator (leftmost portion)
  • The separator itself
  • The part after the separator (rightmost portion)

Behavior if Separator Not Found

• If the separator is not found in a string element, the function returns a tuple containing:
  • The original string itself
  • Two empty strings for the separator and the part after the separator

Example

import numpy as np

data = np.array(['apple-banana-cherry', 'grapefruit', 'orange'])
separator = '-'

result = np.char.partition(data, separator)

print(result)

Output

[['apple' 'banana' 'cherry']
 ['grapefruit' '' '']
 ['orange' '' '']]

Key Points

• If you need to split at multiple occurrences of the separator, consider using str.split() or regular expressions with np.char.split().
• It's useful for splitting strings based on a delimiter and extracting specific parts.
• char.partition() operates element-wise on the input array.

• char.partition() might be deprecated in future NumPy versions, so be aware of potential updates and alternatives.
• While char.partition() is convenient for basic splitting at the first occurrence, for more complex splitting scenarios, explore str.split() or regular expressions with np.char.split().

---

Extracting File Extensions

import numpy as np

filenames = np.array(['image.jpg', 'data.csv', 'report.pdf', 'noname'])
separator = '.'

extensions = np.char.partition(filenames, separator)[:, -1]  # Extract only extensions

print(extensions)

This code splits filenames at the dot (.) to extract the file extensions and stores them in a separate array.

Handling Missing Separators

import numpy as np

data = np.array(['apple', 'banana-cherry', 'orange'])
separator = '-'

result = np.char.partition(data, separator)

# Check for missing separators (empty second element)
missing_separator = result[:, 1] == ''
print(data[missing_separator])  # Print elements without separators

This code identifies elements in the data array that lack the separator (-) using conditional indexing and prints them.

Using Regular Expressions (more advanced)

import numpy as np

text = np.array(['This is a sentence. Here is another.', 'No separators here'])
separator = r'\.'  # Raw string for literal dot (period)

result = np.char.split(text, separator)

print(result)

This code uses a regular expression (r'\.') to split strings at any occurrence of a period (.) and returns an array containing all the split substrings for each element.

---

str.split()

• Can be used with NumPy arrays by applying it element-wise using vectorized functions like np.vectorize().
• Operates on individual strings, not NumPy arrays directly.
• More widely used and flexible for splitting strings.

Example

import numpy as np

data = np.array(['apple-banana-cherry', 'grapefruit', 'orange'])
separator = '-'

def split_func(string, sep):
  return string.split(sep)

result = np.vectorize(split_func)(data, separator)

print(result)

np.char.split()

• Can handle more complex splitting scenarios using regular expressions.
• Similar to str.split() but works on NumPy arrays directly.

Example (basic split)

import numpy as np

data = np.array(['apple-banana-cherry', 'grapefruit', 'orange'])
separator = '-'

result = np.char.split(data, separator)

print(result)

Example (using regular expressions)

import numpy as np

text = np.array(['This is a sentence. Here is another.', 'No separators here'])
separator = r'\s+'  # Split on one or more whitespace characters

result = np.char.split(text, separator)

print(result)

• For more flexibility and control over splitting, especially at multiple occurrences or with regular expressions, use str.split() with vectorization or np.char.split().
• If you need basic splitting at the first occurrence and work with NumPy arrays, char.partition() might be sufficient (but consider potential deprecation).