-- Standard C-style escape sequences inside `"..."` literals decode to
-- their control characters, not literal backslash-letter pairs. This is
-- what every PDF/log/ANSI-aware pipeline depends on:
-- \f form feed (0x0C, pdftotext page separator)
-- \b backspace (0x08, ANSI cursor edits)
-- \v vertical tab (0x0B)
-- \a bell (0x07)
-- \0 null (0x00, null-terminated formats)
-- \/ forward slash (JSON-style passthrough)
-- plus the long-standing \n \t \r \" \\ pair.
-- Original pdf-analyst case: count PDF pages by splitting on form-feed.
pages>n;len (spl "page1\fpage2\fpage3" "\f")
-- Every single-char escape is exactly one scalar:
formfeed-len>n;len "\f"
backspace-len>n;len "\b"
null-len>n;len "\0"
vtab-len>n;len "\v"
bell-len>n;len "\a"
slash-len>n;len "\/"
-- Mixed escapes in one literal all decode (six escapes, six chars):
mixed>n;len "\n\t\f\r\b\0"
-- Unknown escapes (e.g. \z) keep the pre-fix passthrough so existing
-- programs that put bare backslashes in strings are unaffected:
unknown-len>n;len "\z"
-- run: pages
-- out: 3
-- run: formfeed-len
-- out: 1
-- run: backspace-len
-- out: 1
-- run: null-len
-- out: 1
-- run: vtab-len
-- out: 1
-- run: bell-len
-- out: 1
-- run: slash-len
-- out: 1
-- run: mixed
-- out: 6
-- run: unknown-len
-- out: 2