.CDR File Format – Text and styles

How an object finds its text

1. The object’s loda has type 4 (artistic text) or 6 (paragraph text; 3 and 5 before version 4). Its 0xC8 argument names the style, its 0x1E argument holds the frame (below).
2. A spnd chunk in the object holds an unsigned number: the text id.
3. The text id is the frame id in a txsm record. libcdr keeps the last frame id of a record as the record’s text id.
4. Each paragraph of the record names a style from the style sheet and carries style overrides for runs of its characters.

The geometry argument of a text object is two coordinates for artistic text, which libcdr does not use (the text sits at the origin of the object’s transformation, the first baseline on the alignment point). For paragraph text it is a field as long as a coordinate (2 bytes before version 6, 4 later), then the frame’s width and height; the frame’s top left corner is the origin and it runs down. In CorelDRAW 4 an artistic text’s characters are in the object itself (below).

txsm: the text record

Version 5 and later. After the record’s header the layout depends on the version. All counts are as wide as an unsigned (16 bits up to version 5, 32 bits later).

Versions 7 to 15 (and 6, with fewer fields)

A paragraph

A style override

The rotation and shifts are what CorelDRAW stores for text fitted to a path and for characters moved with the Shape tool.

CorelDRAW 6

Frame flag (4), 24 unknown bytes, text on a path (4; 1 adds 32 bytes), number of frames (4), then per frame the frame id (4), 48 bytes and, for artistic text, 8 more. Number of paragraphs (4); each: style id (4), number of overrides (4) and for each override a flags byte, 3 unknown bytes, font id and character set (2 + 2, or 4 unknown bytes), 4 unknown, a size (a 32-bit coordinate, or 4 unknown), 44 unknown, the fill id (flags bit 0x10) and the outline id (0x20). Then the number of characters (4), 4 unknown bytes, and for each character 12 bytes: the character, 5 unknown bytes, a byte that is the override’s number halved, 5 unknown bytes.

CorelDRAW 5

2 unknown bytes, the number of frames (2), per frame an id (2) and 2 unknown bytes; the number of paragraphs (2); each: style id (2), number of overrides (2) and the overrides, 36 bytes each, then the number of characters (2) and 8 bytes a character.

A character of CorelDRAW 5: 4 unknown bytes, the character, 1 unknown byte and a 16-bit word whose bits 3 and up are the override number. One byte a character, so bit 0 of that word never means “two bytes” here. A CR LF pair counts as one line break.

CorelDRAW X6 and later

The first bytes of a real text record, from a file saved by CorelDRAW 2020 (the record is in content/data/page1.dat, 4002 bytes long, referred to by a txsm chunk of root.dat; the style string is UTF-8 JSON from version 17):

000: ff ff ff ff ff ff ff ff 00 00 00 00 ff ff ff ff
010: 08 00 00 00 0e 00 00 00 00 00 00 00 03 00 00 00
020: 00 00 00 00 08 07 00 00 00 01 00 00 00 80 b1 1b
030: 25 00 00 00 00 00 00 f0 3f 00 00 00 00 00 00 00
060: 41 00 00 00 00 00 00 00 00 00 00 00 00 01 00 00
070: 00 88 6d a0 25 00 00 a2 0e 00 00 7b 0a 20 20 22

Style strings (X6 and later)

A style string is JSON. libcdr reads these keys; the file has many more.

A colour string is a model, a palette, numbers and a GUID, separated by commas: RGB255,USER,255,255,255,100,00000000-0000-0000-0000-000000000000 is white. The models are CMYK and CMYK100 (CMYK, 0 to 100), CMYK255, RGB255, HSB, HLS, GRAY255, YIQ255, LAB, LAB255, REGCOLOR, SPOT and PANTONEHX; the palettes USER, FOCTONE, DUPONT, TOYO, DIC, PAN_HEX and HKS variants. The numbers are the components, then the opacity in percent. When fallback colours follow after ,~, libcdr uses the first of them. The same strings fill the style sheet of the package, styles/document.cdss.

stlt: the style sheet (CorelDRAW 7 and later)

From version 7 the styles are one record: a LIST stlt whose body is not chunks (in X6 and later packages a chunk stlt with a reference). It is a series of tables; each count is 32 bits. In order:

Spacing units: with the last field 0 the numbers are shares of the character height (1,000,000 is 100%); with 1 (version 14 and later) they are lengths in 1/254,000 inch. libcdr uses the line spacing, the spacing before and after a paragraph, and the character and word spacing (shares of a space’s width).

A style’s settings are found by following its parent ids upwards and applying the settings from the oldest ancestor down: a font, size, weight, shifts, rotation, alignment, indents, spacings, fill and outline each override only when set.

In X6 and later the whole stlt record may be inside the chunk, as data inline (stream number 0xFFFFFFFF).

styd: a style (CorelDRAW 3 to 6)

Up to version 6 each style is one chunk in the style list (LIST styl). It has the same shape as a loda:

CorelDRAW 4: text in the object

Version 4 has no txsm. An artistic text’s 0x1E argument holds 2 unknown bytes, the number of characters (2), 2 more bytes, then 3 bytes a character: flags, the character in the font’s code page, and a zero. Flag bit 0 adds 22 bytes of the character’s own style, which libcdr does not read. Lines end in CR LF. The text takes its style from the object and the fill and outline of the object itself. In a style (styd) of version 4 the font style is 1 normal, 2 bold, 4 italic.

Character sets

A font record, a style or an override names a Windows character set. 0 means none: libcdr then guesses from the text.

CorelDRAW 5 marks Central European text with the East Asian sets too; libcdr turns those to 0. A font whose name ends in a script name (for example “Arial Cyr”, “Times CE”, “Arial Greek”) has the set of that script, and a name beginning “GOST” is Cyrillic.

Laying the text out

CorelDRAW’s layout rules were measured by libcdr’s authors on PDF exports; the explorer applies only the simplest of them.

• A baseline is the line spacing times the previous line’s descent plus this line’s ascent below the previous one.
• Character spacing is a share of a space’s width added after each character; a space is the space width times the word spacing minus the character spacing.
• An empty line has the previous line’s height; a leading line break makes no line.
• Artistic text is anchored at the transformed origin of the object, its first baseline on the anchor; a stretched object keeps the font size and scales the text.
• Text on a path has the characters’ places in the record (the shifts and rotations), laid out left aligned.
• A text outline is drawn over the fill; a text with no fill has outlined characters only.

Worked example

The text “MPEG 3” in a real CorelDRAW 6 drawing, a txsm chunk of 240 bytes (the chunk’s own 8-byte header is not in the offsets):

The style, number 10, is a styd chunk of the style list; it gives the font, size and fill. The object’s loda names style 10 in its 0xC8 argument and the text id 9 is in its spnd chunk.

In the explorer

The tree shows each txsm, stlt and styd field by field, with the characters decoded in the code page of their style. The Contents tab lists the texts and the styles with their fonts, sizes and colours, and the Drawing tab draws the text approximately: it places each line from the style’s size and alignment with the browser’s own fonts, which are not CorelDRAW’s, so line breaks and widths differ.