.eng Documented

The ENG text format

All of a game's text lives in .eng files: one for strings, one ending _mm.eng for the pop-up messages, which carries dialog geometry alongside the words. Anyone translating one of these games, or reading its text out, works here.

Used by Caesar III Pharaoh Zeus Emperor

Two related files, both in three parts: a short header, a fixed-size index, and a data section of null-separated strings.

The wiki documents Caesar III as its worked example — its files identify themselves with a descriptor string like "C3 textfile." — and the later titles use the same structure with their own descriptor and their own quirks.

Text file header

28 bytes.

Offset	Size	Type	Field	Description
0	16	char[]	descriptor	Identifies the file and the game, e.g. "C3 textfile."
16	4	int32	num_groups	Highest group ID in use, plus one.
20	4	int32	num_strings	Total strings in the file.
24	4	int32	num_words	Word count. Not reliable — do not use it to size a buffer.

Text file index

1,000 entries of 8 bytes each, following the header.

Offset	Size	Type	Field	Description
0	4	int32	offset	Where this group's strings begin in the data section.
4	4	int32	in_use	In the earlier games a flag, 0 or 1. In the later ones, the number of strings in the group.

Text data and encoding

Strings run consecutively, separated by null bytes, and a string must contain at least one printable character — a byte of 32 or above — to count.

There is no encoding declared anywhere in the file; it follows from the language of the release. Western European editions use Windows-1252, Polish uses Windows-1250, Russian Windows-1251, and Korean Windows-950. Traditional Chinese uses an encoding of its own.

Message file header

24 bytes. Message files are the ones whose names end in _mm.eng.

Offset	Size	Type	Field	Description
0	16	char[]	descriptor	Identifies the file, e.g. "C3 MMfile".
16	4	int32	num_entries	Index entries present — typically 1,000.
20	4	int32	num_used	Highest index in use, plus one.

Message index entry

80 bytes each. This is where the format stops being a string table: each entry lays out the dialog the message appears in.

Offset	Size	Type	Field	Description
0	2	int16	type	Message type.
2	2	int16	message_type	Message subtype.
6	2	int16	x	Dialog X position.
8	2	int16	y	Dialog Y position.
10	2	int16	width_blocks	Dialog width, in 16-pixel blocks.
12	2	int16	height_blocks	Dialog height, in 16-pixel blocks.
14	2	int16	image1_id	Primary image ID.
16	2	int16	image1_x	Primary image X.
18	2	int16	image1_y	Primary image Y.
20	2	int16	image2_id	Secondary image ID.
22	2	int16	image2_x	Secondary image X.
24	2	int16	image2_y	Secondary image Y.
26	2	int16	title_x	Title X.
28	2	int16	title_y	Title Y.
30	2	int16	subtitle_x	Subtitle X.
32	2	int16	subtitle_y	Subtitle Y.
38	2	int16	video_x	Video X.
40	2	int16	video_y	Video Y.
56	4	int32	urgent	0 for a normal message, 1 for an urgent one.
60	4	int32	video_offset	Offset of the video filename string.
64	4	int32	unknown_offset	Offset of a string whose purpose is not established.
68	4	int32	title_offset	Offset of the title string.
72	4	int32	subtitle_offset	Offset of the subtitle string.
76	4	int32	content_offset	Offset of the body text, which may contain formatting.