Extract numbers from text

Pulls every number out of a block of text, one per line, keeping decimals and negative signs. Both decimal conventions are recognised — a dot and a comma.

Recognising the comma as a decimal mark is necessary for European text and creates a genuine ambiguity that nothing can resolve automatically: `3,75` is three and three quarters in Amsterdam and three thousand seven hundred and fifty in London. Both are extracted as written and it is left to you, because guessing would be worse. Note that a number embedded in an identifier — an order number, a postcode — is extracted too, since there is no way to distinguish a quantity from a label without understanding the sentence. Sorting the result numerically afterwards is usually the quickest way to spot the ones that are not quantities.