Module step_encoding

ifc_lite_core

Module step_encoding 

Source
Expand description

STEP string escape decoding (ISO 10303-21 / IFC).

Decoding only. The one writer is ifc_lite_export::escape_step_string (a second, literal-unsafe encoder here was deleted: architecture review behind #4577, F6). The round trip decode_ifc_string(escape(s)) == s is pinned in rust/export/tests/step_escape_parity.rs.

IFC string attribute values encode non-ASCII characters with backslash escape sequences. This module decodes them to native UTF-8 so the Rust crates, CLI, and server surface the same text the browser parser does via decodeIfcString in @ifc-lite/encoding. The two decoders are pinned to a shared test-vector fixture (tests/fixtures/ifc_string_vectors.json).

Supported escapes:

  • \X2\HHHH..\X0\ UTF-16 code units, 4 hex digits each (surrogate pairs ok)
  • \X4\HHHHHHHH..\X0\ Unicode scalar values, 8 hex digits each
  • \X\HH single ISO-8859-1 byte (NOT code-page dependent: always the ISO 10646 row-0 value, per ISO 10303-21 6.4.3)
  • \S\C extended ASCII: code point of C plus 128, mapped through the currently selected \P?\ code page (default ISO 8859-1)
  • \PA\..\PI\ code-page directive (A=ISO 8859-1 .. I=ISO 8859-9): tracked for subsequent \S\ escapes, then dropped. Any other letter is dropped without changing the active page.

ISO 10303-21 also doubles the reverse solidus inside a string literal, so \\ decodes to one \. That arm sits AFTER the directive arms: a directive immediately followed by an escaped backslash ends in three backslashes (\X2\00FC\X0\ + \\), and collapsing pairs first would eat the directive’s own terminator.

Unknown or malformed escapes are passed through unchanged. The '' doubled-quote escape is NOT handled here — the tokenizer’s consumers strip the surrounding quotes and un-double before calling this.

Functions§

decode_ifc_string
Decode IFC STEP string escapes to UTF-8.
is_page_quote_operand
Whether the apostrophe at pos, inside a STEP string literal, is the operand of the PAGE directive \S\ (#7540). Tested before a doubled '': '5\S\'' is “5§” and closes at its last quote. No other legal string unit ends in \S\ right before an apostrophe, so this cannot misfire on a legal file.
step_string_literal_end
Index just past the apostrophe closing the STEP string literal whose opening apostrophe is at open, or None when it never closes (#7589).