program GetWordBeforeLastDemo;
{$mode objfpc}{$H+}{$J-} // Use Object Pascal mode, long strings, and typed constants
uses
SysUtils, Classes;
{
Splits a string into words based on ASCII and common Unicode separators.
Pascal's ExtractStrings is great for ASCII, but for Unicode delimiters
like '、', a manual scan is more reliable in a cross-platform way.
}
function GetWordBeforeLast(const Text: string): string;
var
Words: TStringList;
CurrentWord: string;
I: Integer;
U: string;
begin
Result := '';
Words := TStringList.Create;
try
CurrentWord := '';
I := 1;
while I <= Length(Text) do
begin
// Check for Japanese Comma (UTF-8: E3 80 81)
if (I <= Length(Text) - 2) and
(Byte(Text[I]) = $E3) and (Byte(Text[I+1]) = $80) and (Byte(Text[I+2]) = $81) then
begin
if CurrentWord <> '' then Words.Add(CurrentWord);
CurrentWord := '';
Inc(I, 3);
end
// Check for ASCII separators (Space and Punctuation)
else if Text[I] in [' ', #9, #10, #13, ',', '.', '!', '?', ';', ':', '"', '''', '(', ')', '[', ']', '{', '}', '-', '/', '\'] then
begin
if CurrentWord <> '' then Words.Add(CurrentWord);
CurrentWord := '';
Inc(I);
end
else
begin
CurrentWord := CurrentWord + Text[I];
Inc(I);
end;
end;
// Add the final word if exists
if CurrentWord <> '' then Words.Add(CurrentWord);
// Return the second to last word if available
if Words.Count >= 2 then
Result := Words[Words.Count - 2];
finally
Words.Free;
end;
end;
var
Tests: array[0..10] of string = (
'python pascal',
' many spaces here now ',
'OneWord',
'',
' ',
'Hello, world!',
'Tabs'#9'and'#10'newlines work too',
'Unicode 世界、こんにちは',
'Ends with punctuation.',
'Multiple words, with punctuation, here!',
'state-of-the-art program example'
);
T: string;
Output: string;
begin
WriteLn('=== Testing: Get Word Before Last ===');
WriteLn;
for T in Tests do
begin
Output := GetWordBeforeLast(T);
if Output = '' then Output := 'null';
WriteLn('Input: "', T, '"');
WriteLn('Output: ', Output);
WriteLn('----------------------------------------');
end;
end.
{
OUTPUT:
=== Testing: Get Word Before Last ===
Input: "python pascal"
Output: python
----------------------------------------
Input: " many spaces here now "
Output: here
----------------------------------------
Input: "OneWord"
Output: null
----------------------------------------
Input: ""
Output: null
----------------------------------------
Input: " "
Output: null
----------------------------------------
Input: "Hello, world!"
Output: Hello
----------------------------------------
Input: "Tabs and
newlines work too"
Output: work
----------------------------------------
Input: "Unicode 世界、こんにちは"
Output: 世界
----------------------------------------
Input: "Ends with punctuation."
Output: with
----------------------------------------
Input: "Multiple words, with punctuation, here!"
Output: punctuation
----------------------------------------
Input: "state-of-the-art program example"
Output: program
----------------------------------------
}
