A File Name the OS Accepts

A File Name the OS Accepts

Your extension attaches PDF exports to customer records, and the file is named after whatever the user typed: the document number, the customer name, sometimes a whole subject line. Every few days one of those names carries a / or a : — and the export dies, because Windows refuses nine characters in a file name: \ / : * ? " < > |. On top of that, a name with a trailing dot or space is silently changed by the file system, and a 300-character subject line does not fit at all.

Write the one procedure that turns any proposed name into a file name the operating system accepts — every time, deterministically.

Requirements

Create a codeunit named "Safe File Name" with a public procedure:

procedure ToSafeFileName(Proposed: Text; Extension: Text; MaxLength: Integer): Text

The result is the cleaned stem (the name without extension), a single dot, and Extension. Clean the stem in this order:

1. Illegal characters. Replace each of the nine characters \ / : * ? " < > | with one underscore _ — one underscore per character, so a/b becomes a_b and a name made of nothing but the nine illegal characters becomes nine underscores (not document).
2. Whitespace. A tab (character 9) counts as a space. Every run of two or more consecutive spaces or tabs collapses to a single space.
3. Edges. Remove every leading and every trailing space and dot from the stem — any number of them, in any mix: ' ..v1.2 notes . ' becomes v1.2 notes. Dots and spaces inside the stem stay where they are.
4. Empty stem. If nothing is left after steps 1-3 (the input was empty, or only spaces, tabs and dots), the stem is document.
5. Length. Append . + Extension. If the whole result is longer than MaxLength characters, cut the stem from the end so that the complete file name — stem, dot and extension — is exactly MaxLength characters long. The extension always survives intact; the length is measured after cleaning, so leading spaces you removed never count against it.

Everything else passes through untouched: letters of any alphabet (Müller, Straße, 日本語), digits, upper- and lowercase, and characters such as &, -, (, ), ' and , that the operating system allows.

You can rely on these guarantees about the input: Extension is never empty, never contains a dot and contains only letters and digits; MaxLength is always at least the length of the extension plus two; Proposed contains no control characters other than tabs; and the tests never cut a stem at a position where a space or a dot would end up trailing, so whether you trim again after cutting is your call (not graded).

The built-in Text methods do nearly all of this. ConvertStr maps characters one-to-one — its two character lists must be the same length or it raises a runtime error. DelChr removes characters at the start (<), at the end (>) or everywhere (=): the Where argument names positions, and the Which argument is a set of characters (not a literal string), so one call strips spaces and dots from both ends. Watch out for a plausible dead end: Text.Trim() removes whitespace only, never dots.

Pick object IDs in the range 50100–50199, and reference other objects by name, never by ID.

Examples

What the tests check

The grading tests call ToSafeFileName and compare the returned text character for character: the Invoice 1001/2026: Müller & Co example, a name containing all nine illegal characters (each must become one underscore), a name made of nothing but illegal characters (nine underscores), non-ASCII letters that must survive unchanged, tabs and runs of spaces inside the name, leading and trailing spaces and dots in a mix, an empty name and a name of only spaces, tabs and dots (both document), a 300-character stem cut so that the whole name is exactly MaxLength long with the extension intact, a name already exactly at MaxLength (untouched), a name one character over it (one character cut), and a padded name whose cleaned stem fits exactly (so cutting before cleaning fails). Several tests build their stems from generated letters, so returning a constant or matching the examples passes nothing.

Learn More

• Text.ConvertStr method — one-to-one character mapping, and why the two character lists must be the same length.
• Text.DelChr method — the Where positions (<, >, =) and the Which character set.
• Text.CopyStr method — cutting the stem down to a length without an overflow error.
• Text.Replace method — replacing a substring such as two spaces with one.