xref: /plan9/sys/src/cmd/aux/antiword/wordconst.h (revision 25b329d522281a8cdd35da0dcc08c3fc621059a9)
1 /*
2  * wordconst.h
3  * Copyright (C) 1998-2004 A.J. van Os; Released under GNU GPL
4  *
5  * Description:
6  * Constants and macros for the interpretation of MS Word files
7  */
8 
9 #if !defined(__wordconst_h)
10 #define __wordconst_h 1
11 
12 /*
13  * A bit odd definition of the type Boolean, but RISC OS insists
14  * on this and Linux/Unix doesn't mind.
15  */
16 #if !defined(BOOL)
17 #define BOOL int
18 #define TRUE 1
19 #define FALSE 0
20 #endif /* !BOOL */
21 
22 /* Block sizes */
23 #define HEADER_SIZE			768
24 #define BIG_BLOCK_SIZE			512
25 #define PROPERTY_SET_STORAGE_SIZE	128
26 #define SMALL_BLOCK_SIZE		 64
27 /* Switch size of Depot use */
28 #define MIN_SIZE_FOR_BBD_USE		0x1000
29 /* Table sizes */
30 #define TABLE_COLUMN_MAX		 31
31 /* Maximum number of tabs positions in a paragraph */
32 #define NUMBER_OF_TABS_MAX		 64
33 /* Font sizes (in half-points) */
34 #define MIN_FONT_SIZE			  8
35 #define DEFAULT_FONT_SIZE		 20
36 #define MAX_FONT_SIZE			240
37 #define MIN_TABLEFONT_SIZE		 16
38 #define MAX_TABLEFONT_SIZE		 20
39 /* Font styles */
40 #define FONT_REGULAR			0x0000
41 #define FONT_BOLD			0x0001
42 #define FONT_ITALIC			0x0002
43 #define FONT_UNDERLINE			0x0004
44 #define FONT_CAPITALS			0x0008
45 #define FONT_SMALL_CAPITALS		0x0010
46 #define FONT_STRIKE			0x0020
47 #define FONT_HIDDEN			0x0040
48 #define FONT_MARKDEL			0x0080
49 #define FONT_SUPERSCRIPT		0x0100
50 #define FONT_SUBSCRIPT			0x0200
51 /* Font colors */
52 #define FONT_COLOR_DEFAULT		 0
53 #define FONT_COLOR_BLACK		 1
54 #define FONT_COLOR_BLUE			 2
55 #define FONT_COLOR_CYAN			 3
56 #define FONT_COLOR_GREEN		 4
57 #define FONT_COLOR_MAGENTA		 5
58 #define FONT_COLOR_RED			 6
59 #define FONT_COLOR_YELLOW		 7
60 #define FONT_COLOR_WHITE		 8
61 /* Special block numbers */
62 #define END_OF_CHAIN			0xfffffffeUL
63 #define UNUSED_BLOCK			0xffffffffUL
64 /* Blocksize (512 bytes) and maximum filesize (4 GB) gives 0..7fffff */
65 #define MAX_BLOCKNUMBER			0x007fffffUL
66 /* Invalid character position */
67 #define CP_INVALID			0xffffffffUL
68 /* Invalid file offset */
69 #define FC_INVALID			0xffffffffUL
70 /* Special istd values */
71 #define ISTD_INVALID			USHRT_MAX
72 #define ISTD_NORMAL			0
73 /* Properties modifier without value */
74 #define IGNORE_PROPMOD			0
75 /* Types of lists */
76 #define LIST_ARABIC_NUM			0x00
77 #define LIST_UPPER_ROMAN		0x01
78 #define LIST_LOWER_ROMAN		0x02
79 #define LIST_UPPER_ALPHA		0x03
80 #define LIST_LOWER_ALPHA		0x04
81 #define LIST_ORDINAL_NUM		0x05
82 #define LIST_NUMBER_TXT			0x06
83 #define LIST_ORDINAL_TXT		0x07
84 #define LIST_OUTLINE_NUM		0x16
85 #define LIST_SPECIAL			0x17
86 #define LIST_SPECIAL2			0x19
87 #define LIST_BULLETS			0xff
88 /* Types of paragraph alignment */
89 #define ALIGNMENT_LEFT			0x00
90 #define ALIGNMENT_CENTER		0x01
91 #define ALIGNMENT_RIGHT			0x02
92 #define ALIGNMENT_JUSTIFY		0x03
93 /* Minimum vertical space before and after a heading line */
94 #define HEADING_GAP			120	/* twips */
95 /* Style identifier */
96 #define STI_USER			0xffe
97 #define STI_NIL				0xfff
98 /* Table border style codes */
99 #define TABLE_BORDER_TOP		0x01
100 #define TABLE_BORDER_LEFT		0x02
101 #define TABLE_BORDER_BOTTOM		0x04
102 #define TABLE_BORDER_RIGHT		0x08
103 
104 /* Macros */
105 	/* Get macros */
106 #define ucGetByte(i,a)		((unsigned char)(a[i]))
107 #define usGetWord(i,a)		((unsigned short)\
108 					((unsigned int)(a[(i)+1])<<8|\
109 					 (unsigned int)(a[i])))
110 #define ulGetLong(i,a)		((unsigned long)(a[i])|\
111 					(unsigned long)(a[(i)+1])<<8|\
112 					(unsigned long)(a[(i)+2])<<16|\
113 					(unsigned long)(a[(i)+3])<<24)
114 #define usGetWordBE(i,a)	((unsigned short)\
115 					((unsigned int)(a[i])<<8|\
116 					 (unsigned int)(a[(i)+1])))
117 #define ulGetLongBE(i,a)	((unsigned long)(a[(i)+3])|\
118 					(unsigned long)(a[(i)+2])<<8|\
119 					(unsigned long)(a[(i)+1])<<16|\
120 					(unsigned long)(a[i])<<24)
121 	/* Font style macros */
122 #define bIsBold(x)		(((x) & FONT_BOLD) == FONT_BOLD)
123 #define bIsItalic(x)		(((x) & FONT_ITALIC) == FONT_ITALIC)
124 #define bIsUnderline(x)		(((x) & FONT_UNDERLINE) == FONT_UNDERLINE)
125 #define bIsCapitals(x)		(((x) & FONT_CAPITALS) == FONT_CAPITALS)
126 #define bIsSmallCapitals(x)	(((x) & FONT_SMALL_CAPITALS) == FONT_SMALL_CAPITALS)
127 #define bIsStrike(x)		(((x) & FONT_STRIKE) == FONT_STRIKE)
128 #define bIsHidden(x)		(((x) & FONT_HIDDEN) == FONT_HIDDEN)
129 #define bIsMarkDel(x)		(((x) & FONT_MARKDEL) == FONT_MARKDEL)
130 #define bIsSuperscript(x)	(((x) & FONT_SUPERSCRIPT) == FONT_SUPERSCRIPT)
131 #define bIsSubscript(x)		(((x) & FONT_SUBSCRIPT) == FONT_SUBSCRIPT)
132 	/* Table border style code macros */
133 #define bIsTableBorderTop(x)	(((x) & TABLE_BORDER_TOP) == TABLE_BORDER_TOP)
134 #define bIsTableBorderLeft(x)	(((x) & TABLE_BORDER_LEFT) == TABLE_BORDER_LEFT)
135 #define bIsTableBorderBottom(x)	(((x) & TABLE_BORDER_BOTTOM) == TABLE_BORDER_BOTTOM)
136 #define bIsTableBorderRight(x)	(((x) & TABLE_BORDER_RIGHT) == TABLE_BORDER_RIGHT)
137 	/* Computation macros */
138 #if defined(__riscos)
139 /* From Words half-points to draw units (plus a percentage) */
140 #define lWord2DrawUnits00(x)	((long)(x) * 320)
141 #define lWord2DrawUnits20(x)	((long)(x) * 384)
142 #define lToBaseLine(x)		((long)(x) *  45)
143 #endif /* __riscos */
144 /* From twips (1/20 of a point) to millipoints */
145 #define lTwips2MilliPoints(x)	((long)(x) * 50)
146 /* From twips (1/20 of a point) to points */
147 #define dTwips2Points(x)	((double)(x) / 20.0)
148 /* From default characters (16 OS units wide) to millipoints */
149 #define lChar2MilliPoints(x)	((long)(x) * 6400)
150 #define iMilliPoints2Char(x)	(int)(((long)(x) + 3200) / 6400)
151 #define iDrawUnits2Char(x)	(int)(((long)(x) + 2048) / 4096)
152 /* From draw units (1/180*256 inch) to millipoints (1/72*1000 inch) */
153 #define lDrawUnits2MilliPoints(x)	(((long)(x) * 25 +  8) / 16)
154 #define lMilliPoints2DrawUnits(x)	(((long)(x) * 16 + 12) / 25)
155 #define lPoints2DrawUnits(x)		((long)(x) * 640)
156 #define dDrawUnits2Points(x)		((double)(x) / 640.0)
157 
158 /* Special characters */
159 #define IGNORE_CHARACTER	0x00	/* ^@ */
160 #define PICTURE			0x01	/* ^A */
161 #define FOOTNOTE_OR_ENDNOTE	0x02	/* ^B */
162 #define FOOTNOTE_SEPARATOR	0x03	/* ^C */
163 #define FOOTNOTE_CONTINUATION	0x04	/* ^D */
164 #define ANNOTATION		0x05	/* ^E */
165 #define TABLE_SEPARATOR		0x07	/* ^G */
166 #define FRAME			0x08	/* ^H */
167 #define TAB			0x09	/* ^I */
168 /* End of line characters */
169 #define LINE_FEED		0x0a	/* ^J */
170 #define HARD_RETURN		0x0b	/* ^K */
171 #define PAGE_BREAK		0x0c	/* ^L */
172 #define PAR_END			0x0d	/* ^M */
173 #define COLUMN_FEED		0x0e	/* ^N */
174 /* Embedded stuff */
175 #define START_EMBEDDED		0x13	/* ^S */
176 #define END_IGNORE		0x14	/* ^T */
177 #define END_EMBEDDED		0x15	/* ^U */
178 /* Special characters */
179 #if defined(DEBUG)
180 #define FILLER_CHAR		'~'
181 #else
182 #define FILLER_CHAR		' '
183 #endif /* DEBUG */
184 #define TABLE_SEPARATOR_CHAR	'|'
185 /* Pseudo characters. These must be outside the Unicode range */
186 #define FOOTNOTE_CHAR		((unsigned long)0xffff + 1)
187 #define ENDNOTE_CHAR		((unsigned long)0xffff + 2)
188 #define UNKNOWN_NOTE_CHAR	((unsigned long)0xffff + 3)
189 
190 /* Charactercodes as used by Word */
191 #define WORD_UNBREAKABLE_JOIN		0x1e
192 #define WORD_SOFT_HYPHEN		0x1f
193 
194 /* Unicode characters */
195 #define UNICODE_DOUBLE_LEFT_ANGLE_QMARK	0x00ab
196 #define UNICODE_MIDDLE_DOT		0x00b7
197 #define UNICODE_DOUBLE_RIGHT_ANGLE_QMARK	0x00bb
198 #define UNICODE_CAPITAL_D_WITH_STROKE	0x0110
199 #define UNICODE_SMALL_D_WITH_STROKE	0x0111
200 #define UNICODE_CAPITAL_LIGATURE_OE	0x0152
201 #define UNICODE_SMALL_LIGATURE_OE	0x0153
202 #define UNICODE_SMALL_F_HOOK		0x0192
203 #define UNICODE_GREEK_CAPITAL_CHI	0x03a7
204 #define UNICODE_GREEK_SMALL_UPSILON	0x03c5
205 #define UNICODE_MODIFIER_CIRCUMFLEX	0x02c6
206 #define UNICODE_SMALL_TILDE		0x02dc
207 #define UNICODE_SMALL_LETTER_OMEGA	0x03c9
208 #define UNICODE_EN_QUAD			0x2000
209 #define UNICODE_EM_QUAD			0x2001
210 #define UNICODE_EN_SPACE		0x2002
211 #define UNICODE_EM_SPACE		0x2003
212 #define UNICODE_THREE_PER_EM_SPACE	0x2004
213 #define UNICODE_FOUR_PER_EM_SPACE	0x2005
214 #define UNICODE_SIX_PER_EM_SPACE	0x2006
215 #define UNICODE_FIGURE_SPACE		0x2007
216 #define UNICODE_PUNCTUATION_SPACE	0x2008
217 #define UNICODE_THIN_SPACE		0x2009
218 #define UNICODE_HAIR_SPACE		0x200a
219 #define UNICODE_ZERO_WIDTH_SPACE	0x200b
220 #define UNICODE_ZERO_WIDTH_NON_JOINER	0x200c
221 #define UNICODE_ZERO_WIDTH_JOINER	0x200d
222 #define UNICODE_LEFT_TO_RIGHT_MARK	0x200e
223 #define UNICODE_RIGHT_TO_LEFT_MARK	0x200f
224 #define UNICODE_HYPHEN			0x2010
225 #define UNICODE_NON_BREAKING_HYPHEN	0x2011
226 #define UNICODE_FIGURE_DASH		0x2012
227 #define UNICODE_EN_DASH			0x2013
228 #define UNICODE_EM_DASH			0x2014
229 #define UNICODE_HORIZONTAL_BAR		0x2015
230 #define UNICODE_DOUBLE_VERTICAL_LINE	0x2016
231 #define UNICODE_DOUBLE_LOW_LINE		0x2017
232 #define UNICODE_LEFT_SINGLE_QMARK	0x2018
233 #define UNICODE_RIGHT_SINGLE_QMARK	0x2019
234 #define UNICODE_SINGLE_LOW_9_QMARK	0x201a
235 #define UNICODE_SINGLE_HIGH_REV_9_QMARK	0x201b
236 #define UNICODE_LEFT_DOUBLE_QMARK	0x201c
237 #define UNICODE_RIGHT_DOUBLE_QMARK	0x201d
238 #define UNICODE_DOUBLE_LOW_9_QMARK	0x201e
239 #define UNICODE_DOUBLE_HIGH_REV_9_QMARK	0x201f
240 #define UNICODE_DAGGER			0x2020
241 #define UNICODE_DOUBLE_DAGGER		0x2021
242 #define UNICODE_BULLET			0x2022
243 #define UNICODE_TRIANGULAR_BULLET	0x2023
244 #define UNICODE_ONE_DOT_LEADER		0x2024
245 #define UNICODE_TWO_DOT_LEADER		0x2025
246 #define UNICODE_ELLIPSIS		0x2026
247 #define UNICODE_HYPHENATION_POINT	0x2027
248 #define UNICODE_LEFT_TO_RIGHT_EMBEDDING	0x202a
249 #define UNICODE_RIGHT_TO_LEFT_EMBEDDING	0x202b
250 #define UNICODE_POP_DIRECTIONAL_FORMATTING	0x202c
251 #define UNICODE_LEFT_TO_RIGHT_OVERRIDE	0x202d
252 #define UNICODE_RIGHT_TO_LEFT_OVERRIDE	0x202e
253 #define UNICODE_NARROW_NO_BREAK_SPACE	0x202f
254 #define UNICODE_PER_MILLE_SIGN		0x2030
255 #define UNICODE_PRIME			0x2032
256 #define UNICODE_DOUBLE_PRIME		0x2033
257 #define UNICODE_SINGLE_LEFT_ANGLE_QMARK	0x2039
258 #define UNICODE_SINGLE_RIGHT_ANGLE_QMARK	0x203a
259 #define UNICODE_UNDERTIE		0x203f
260 #define UNICODE_FRACTION_SLASH		0x2044
261 #define UNICODE_EURO_SIGN		0x20ac
262 #define UNICODE_CIRCLE			0x20dd
263 #define UNICODE_SQUARE			0x20de
264 #define UNICODE_DIAMOND			0x20df
265 #define UNICODE_NUMERO_SIGN		0x2116
266 #define UNICODE_TRADEMARK_SIGN		0x2122
267 #define UNICODE_KELVIN_SIGN		0x212a
268 #define UNICODE_LEFTWARDS_ARROW		0x2190
269 #define UNICODE_UPWARDS_ARROW		0x2191
270 #define UNICODE_RIGHTWARDS_ARROW	0x2192
271 #define UNICODE_DOWNWARDS_ARROW		0x2193
272 #define UNICODE_N_ARY_SUMMATION		0x2211
273 #define UNICODE_MINUS_SIGN		0x2212
274 #define UNICODE_DIVISION_SLASH		0x2215
275 #define UNICODE_ASTERISK_OPERATOR	0x2217
276 #define UNICODE_BULLET_OPERATOR		0x2219
277 #define UNICODE_RATIO			0x2236
278 #define UNICODE_TILDE_OPERATOR		0x223c
279 #define UNICODE_BD_LIGHT_HORIZONTAL	0x2500
280 #define UNICODE_BD_LIGHT_VERTICAL	0x2502
281 #define UNICODE_BD_LIGHT_DOWN_RIGHT	0x250c
282 #define UNICODE_BD_LIGHT_DOWN_AND_LEFT	0x2510
283 #define UNICODE_BD_LIGHT_UP_AND_RIGHT	0x2514
284 #define UNICODE_BD_LIGHT_UP_AND_LEFT	0x2518
285 #define UNICODE_BD_LIGHT_VERTICAL_AND_RIGHT	0x251c
286 #define UNICODE_BD_LIGHT_VERTICAL_AND_LEFT	0x2524
287 #define UNICODE_BD_LIGHT_DOWN_AND_HORIZONTAL	0x252c
288 #define UNICODE_BD_LIGHT_UP_AND_HORIZONTAL	0x2534
289 #define UNICODE_BD_LIGHT_VERTICAL_AND_HORIZONTAL	0x253c
290 #define UNICODE_BD_DOUBLE_HORIZONTAL	0x2550
291 #define UNICODE_BD_DOUBLE_VERTICAL	0x2551
292 #define UNICODE_BD_DOUBLE_DOWN_AND_RIGHT	0x2554
293 #define UNICODE_BD_DOUBLE_DOWN_AND_LEFT	0x2557
294 #define UNICODE_BD_DOUBLE_UP_AND_RIGHT	0x255a
295 #define UNICODE_BD_DOUBLE_UP_AND_LEFT	0x255d
296 #define UNICODE_BD_DOUBLE_VERTICAL_AND_RIGHT	0x2560
297 #define UNICODE_BD_DOUBLE_VERTICAL_AND_LEFT	0x2563
298 #define UNICODE_BD_DOUBLE_DOWN_AND_HORIZONTAL	0x2566
299 #define UNICODE_BD_DOUBLE_UP_AND_HORIZONTAL	0x2569
300 #define UNICODE_BD_DOUBLE_VERTICAL_AND_HORIZONTAL	0x256c
301 #define UNICODE_LIGHT_SHADE		0x2591
302 #define UNICODE_MEDIUM_SHADE		0x2592
303 #define UNICODE_DARK_SHADE		0x2593
304 #define UNICODE_BLACK_SQUARE		0x25a0
305 #define UNICODE_BLACK_CLUB_SUIT		0x2663
306 #define UNICODE_SMALL_LIGATURE_FI	0xfb01
307 #define UNICODE_SMALL_LIGATURE_FL	0xfb02
308 #define UNICODE_ZERO_WIDTH_NO_BREAK_SPACE	0xfeff
309 
310 #if defined(__riscos)
311 #define OUR_ELLIPSIS			0x8c
312 #define OUR_EM_DASH			0x98
313 #define OUR_UNBREAKABLE_JOIN		0x99
314 #else
315 #define OUR_ELLIPSIS			'.'
316 #define OUR_EM_DASH			'-'
317 #define OUR_UNBREAKABLE_JOIN		'-'
318 #endif /* __riscos */
319 #define OUR_DIAMOND			'-'
320 
321 #endif /* __wordconst_h */
322