.HWP File Format – Sections, notes and other controls

HWPTAG_CTRL_HEADER

An ID is four characters packed as MAKE_4CHID(a, b, c, d) = (a << 24) | (b << 16) | (c << 8) | d and stored little-endian, so the bytes in the file read backwards: the column definition cold is 0x636F6C64, stored 64 6C 6F 63.

The table, drawing object, equation, picture, OLE and group controls are on the tables, pictures and drawings page; the field IDs are listed on the paragraphs page.

HWPTAG_LIST_HEADER

Starts a list of paragraphs inside a control: a table cell, a header, a note, a text box. The paragraphs follow as PARA_HEADER records at the same level as the list header. Table 65:

Six bytes; a control may add its own fields after them, such as a table cell’s cell properties.

HWPTAG_CTRL_DATA

Free data of a control, such as a field’s name or a hyperlink, stored as a parameter set (table 66). A bookmark keeps its name here.

Section definition (secd)

Stored in the first paragraph of each section. Table 129:

Child records, in order: HWPTAG_PAGE_DEF (40 bytes), HWPTAG_FOOTNOTE_SHAPE for footnotes and again for endnotes (26 bytes each), HWPTAG_PAGE_BORDER_FILL, and, when the section has master pages for both, odd or even pages, the master page information with its paragraph list. The specification counts 140 bytes for all of this. The end of each section also holds the extended master pages (last page, any page), and the end of the last section the memos; their layout is not given.

HWPTAG_PAGE_DEF

An A4 sheet, 210 × 297 mm, is about 59,528 × 84,189 HWPUNIT (210 ÷ 25.4 × 7200 and 297 ÷ 25.4 × 7200).

HWPTAG_FOOTNOTE_SHAPE

HWPTAG_PAGE_BORDER_FILL

The fields take 14 bytes, as the specification’s table 5 says; its table 135 gives 12.

Master page information

Followed by the master page’s paragraph list.

Column definition (cold)

Stored in the first paragraph of each run of columns. Table 138:

12 + 2 × cnt bytes. The meaning of bits 16–31 is not given.

Header and footer (head, foot)

Followed by the header’s or footer’s paragraph list.

Footnote and endnote (fn  , en  )

A note has no properties of its own, only its paragraph list, but 8 bytes are written after the ID so that no stray values are stored and updates are avoided. Their meaning is not given. How notes look and are numbered comes from the section’s footnote shape.

Automatic number (atno)

New number (nwno)

Restarts a count from here (table 144): a UINT32 whose bits 0–3 say what is counted, with the same values as the automatic number, then a UINT16 new number. The specification gives the length as 8; the fields take 6.

Hide (pghd)

Hides parts of the current page (table 145). One value, typed UINT but given a length of 2, whose bits say what to hide:

Odd/even adjustment (pgct)

A UINT32 (table 146) whose bits 0–1 choose the page: 0 both, 1 even page, 2 odd page. The next page is moved so that it falls on that kind of page.

Page number position (pgnp)

Index mark (idxm)

Bookmark (bokm)

A bookmark has only its name, which is kept in a control data record. A bookmark over a block of text is a field instead (%bmk).

Overlapping characters (tcps)

10 + 2 × len + 4 × cnt bytes. The border type values are not listed.

Ruby text (tdut)

Small text set above or below main text (Korean dutmal). Table 151:

The specification prints the two texts as single WCHARs with a length before each, and gives a total of 18 bytes; the texts are strings of the stated length.

Hidden comment (tcmt)

Only a paragraph list. Depending on the document’s security level, hidden comment data may be voided.

Field start

Fields start with control character 3 and end with control character 4; the field control’s ID says the kind (see field IDs). Table 152:

15 + 2 × len bytes. The command syntax of each field kind is not given.

Sources

• Hangul document file format 5.0, revision 1.3 – Hancom’s specification (PDF, in Korean), sections 4.3.6–4.3.8 and 4.3.10