.XLS File Format – Drawing objects and comments

How the records are ordered

In a worksheet, macro sheet or dialog sheet substream the drawing records are (ABNF): OBJECTS = *(MSODRAWING *(TEXTOBJECT / OBJ)) [MsoDrawingSelection] and MSODRAWING = MsoDrawing *Continue. In the Globals substream a single MSODRAWINGGROUP = MsoDrawingGroup *Continue holds the file-wide drawing data. When a shape in an MsoDrawing has a client-data record, the very next record must be an Obj; when it has a client text box, the next record must be a TxO (the sizes of these client records do not include the Obj or TxO that follows).

The records inside rgChildRec use the OfficeArt record header (8 bytes, rh: recVer, recInstance, recType, recLen, from [MS-ODRAW] 2.2.1). The container types (OfficeArtDggContainer, OfficeArtDgContainer, shape and shape-property records, BLIP stores and pictures) are not defined in [MS-XLS], and [MS-ODRAW] was not read for this page; so this page lists only the structures [MS-XLS] itself defines for BIFF8, below. A drawing larger than a record’s limit is continued in Continue records.

OfficeArt records defined by [MS-XLS]

Each begins with an 8-byte OfficeArtRecordHeader whose recVer and recInstance are 0 and whose recType is 0xF010 (client anchor), 0xF011 (client data) or 0xF00D (client text box).

OfficeArtClientAnchorSheet (2.5.193)

The position of a shape on a sheet; recType 0xF010, recLen 18.

OfficeArtClientAnchorChart (2.5.191)

The position of a shape in a chart: rh (recType 0xF010, recLen 0x12), 2 bytes of flags (bit 0 fMove 0; bit 1 fSize: 0 the shape resizes with the chart area, 1 it does not; bits 2–4 reserved or undefined; 11 bits unused), then four signed 4-byte values lx1, ly1, lx2, ly2: the upper-left and lower-right corners relative to the upper-left corner of the chart area, in SPRC (1/4000 of the chart’s width or height).

OfficeArtClientAnchorHF (2.5.192)

The size of a header or footer picture: rh (recType 0xF010, recLen 8), width (4, signed, pixels, above 0) and height (4, signed, pixels, above 0).

OfficeArtClientData and OfficeArtClientTextbox

Each is just an 8-byte header with recLen 0: recType 0xF011 for the client data (2.5.194), after which the next record must be Obj; 0xF00D for the client text box (2.5.195), after which the next record must be TxO. Each must be the last structure in the rgChildRec of its MsoDrawing.

Obj (0x005D)

The properties of one object (2.4.181). It is a sequence of subrecords, each of which begins with a 2-byte ft and, for most, a 2-byte cb; the ot field of the first (FtCmo) decides which others are present.

FtCmo (2.5.143)

ft = 0x15, cb = 0x12, then:

Fixed subrecords

FtLbsData (2.5.147)

A list or drop-down list. ft = 0x0013, then cbFContinued (2: 0 means nothing else follows; otherwise at least the size of the rest of the structure when it fits in one record, and when it is continued in Continue records, the size of this record’s share minus 1), then:

LbsDropData (2.5.171): 2 bytes of flags (bits 0–1 wStyle: 0 combo dropdown, 1 combo edit dropdown, 2 simple dropdown, just the button; bit 2 unused; bit 3 fFiltered; 12 bits unused), cLine (2, lines shown, at most 0x7FFF), dxMin (2, the smallest dropdown width in pixels, at most 0x7FFF), str (XLUnicodeString, the current value) and one unused byte if the size of str in bytes is odd.

Formula subrecords

TxO (0x01B6)

The text in a text box or form control (2.4.329). If cchText is not 0, Continue records follow: first those holding the text (each an XLUnicodeStringNoCch with part of the string, the total being cchText characters), then those holding the formatting runs (a TxORuns, split across Continue records if longer than 8,224 bytes).

ControlInfo (2.5.61), for form controls on a dialog sheet: 2 bytes of flags (bit 0 fDefault dismisses the dialog and does the default action; 1 fHelp loads context help; 2 fCancel dismisses and does nothing; 3 fDismiss dismisses; each 0 unless the control is a button; 12 reserved bits), accel1 (2, signed, the accelerator character; 0 for none) and reserved (2, zero).

TxORuns (2.5.272) is an array of Run structures, of cbRuns / 8 - 1 elements, then lastRun, which is present only in the last Continue record. A Run (2.5.226) is a FormatRun (ich 2, the zero-based index of the first character of the run, strictly increasing and at most cchText; ifnt 2, a FontIndex), followed by 4 unused bytes. TxOLastRun (2.5.271): cchText (2, equal to the TxO’s), unused (2) and unused (4).

Note (0x001C)

A cell comment (2.4.179). In the Workbook stream, within a worksheet, dialog sheet or macro sheet substream, the body is a NoteSh; in a revision log it is a NoteRR (a revision of a comment; its layout is not given here). The comment text is in the TxO belonging to the Obj named by idObj.

BkHim (0x00E9)

The background image of a sheet (2.4.19): cf (2, signed: 0x0009 a bitmap, as in [MSDN-BMP]; 0x000E the native format of another application, which cannot be processed directly), reserved (2, 0x0001), lcb (4, signed, at least 1: the size of the image) and imageBlob (lcb bytes of image data).

Source

[MS-XLS] Excel Binary File Format (.xls) Structure, v20250819, sections 2.4.19, 2.4.93, 2.4.170–2.4.172, 2.4.179, 2.4.181, 2.4.187, 2.4.329, 2.5.61, 2.5.140–2.5.154, 2.5.171, 2.5.186–2.5.189, 2.5.191–2.5.195, 2.5.202–2.5.203, 2.5.226, 2.5.271–2.5.272 and the substream ABNF. Not covered: the OfficeArt records themselves ([MS-ODRAW]), the contents of ObjectParsedFormula, and NoteRR.