NLP: Split text into sentences

Build a rule-based sentence splitter that copes with abbreviations, decimals and quotes.

What to build

A program that splits a line of text into numbered sentences.

Requirements

  • Line 1 lists abbreviations (such as Dr. and e.g.), separated by spaces. Line 2 is the text; split it into words on whitespace.
  • A word ends a sentence when, after removing any closing ", ' or ) from its end, it ends with ., ! or ? and is not one of the abbreviations, and either it is the last word or the next word, after removing any opening ", ' or (, starts with a capital letter.
  • Words left over at the end form a final sentence even without a closing mark.
  • Print each sentence as N: sentence, numbered from 1, with its words joined by single spaces.

Input

Line 1: abbreviations (at least one). Line 2: the text.

Output

One line per sentence.

Constraints

The text has at most 2,000 characters.

Example

Input:

Dr. Mr.
Dr. Smith arrived at 3.30 today. He was late! Was Mr. Jones there?

Output:

1: Dr. Smith arrived at 3.30 today.
2: He was late!
3: Was Mr. Jones there?

Why it works

Real sentence splitters combine exactly these clues: a closing mark, a following capital letter, and a list of known abbreviations that do not end sentences.

Examples

Example 1 · Input
Dr. Mr.
Dr. Smith arrived at 3.30 today. He was late! Was Mr. Jones there?
Expected output
1: Dr. Smith arrived at 3.30 today.
2: He was late!
3: Was Mr. Jones there?
Example 2 · Input
e.g.
Wait... what? I said "Stop." Then I left
Expected output
1: Wait... what?
2: I said "Stop."
3: Then I left

Prompts inside input() are ignored when checking, so input("n: ") is fine.

Hints

Hint 1

Work word by word: add each word to the current sentence, then decide whether that word ends it.

Hint 2

`word.rstrip("\"')")` removes closing quotes and brackets; `words[i + 1].lstrip("\"'(")` removes opening ones from the next word.

Solution

Show one correct solution (try it yourself first)
abbreviations = input().split()
words = input().split()
sentences, current = [], []
for i, word in enumerate(words):
    current.append(word)
    core = word.rstrip("\"')")
    if not core or core[-1] not in ".!?" or core in abbreviations:
        continue
    if i + 1 < len(words):
        following = words[i + 1].lstrip("\"'(")
        if not following[:1].isupper():
            continue
    sentences.append(" ".join(current))
    current = []
if current:
    sentences.append(" ".join(current))
for n, sentence in enumerate(sentences, 1):
    print(f"{n}: {sentence}")

Lesson: What natural language processing is →