Line Ending Detector (LF/CRLF/CR)

Line endings never settled on a single standard. Windows ended a line with two characters, carriage return and line feed — CRLF. Unix, Linux and modern macOS use just the line feed, LF. And Mac up to version 9 used the carriage return alone, a bare CR, which still turns up in old files and in exports from legacy systems.

Paste the text and the page counts each kind separately, naming the dominant one. The count treats CRLF as a unit before looking for stray CR and LF, so a normal Windows file shows a high CRLF count with the other two at zero. When all three show up mixed, the file passed through different editors without normalisation — a situation that breaks shell scripts, confuses diffs and makes Git flag the whole file as modified.

Where this bites: a script with CRLF on a Linux server returns "bad interpreter: /bin/bash^M", because the carriage return becomes part of the interpreter name; a configuration file read line by line carries an invisible character at the end of every value; and diffs show changes on every line even though nobody touched the content. The permanent fix is combining normalisation in .gitattributes with the editor setting.