1term(5) File Formats Manual term(5)
2
6 term - format of compiled term file.
7
9 term
10
12 STORAGE LOCATION
13 Compiled terminfo descriptions are placed under the directory
14 /usr/share/terminfo. Two configurations are supported (when building
15 the ncurses libraries):
16 directory tree
18 A two-level scheme is used to avoid a linear search of a huge UNIX
19 system directory: /usr/share/terminfo/c/name where name is the
20 name of the terminal, and c is the first character of name. Thus,
21 act4 can be found in the file /usr/share/terminfo/a/act4. Syn‐
22 onyms for the same terminal are implemented by multiple links to
23 the same compiled file.
24 hashed database
26 Using Berkeley database, two types of records are stored: the ter‐
27 minfo data in the same format as stored in a directory tree with
28 the terminfo's primary name as a key, and records containing only
29 aliases pointing to the primary name.
30 If built to write hashed databases, ncurses can still read ter‐
32 minfo databases organized as a directory tree, but cannot write
33 entries into the directory tree. It can write (or rewrite) en‐
34 tries in the hashed database.
35 ncurses distinguishes the two cases in the TERMINFO and TER‐
37 MINFO_DIRS environment variable by assuming a directory tree for
38 entries that correspond to an existing directory, and hashed data‐
39 base otherwise.
40 LEGACY STORAGE FORMAT
42 The format has been chosen so that it will be the same on all hardware.
43 An 8 or more bit byte is assumed, but no assumptions about byte order‐
44 ing or sign extension are made.
45 The compiled file is created with the tic program, and read by the rou‐
47 tine setupterm(3X). The file is divided into six parts:
48 a) header,
50 b) terminal names,
52 c) boolean flags,
54 d) numbers,
56 e) strings, and
58 f) string table.
60 The header section begins the file. This section contains six short
62 integers in the format described below. These integers are
63 (1) the magic number (octal 0432);
65 (2) the size, in bytes, of the terminal names section;
67 (3) the number of bytes in the boolean flags section;
69 (4) the number of short integers in the numbers section;
71 (5) the number of offsets (short integers) in the strings section;
73 (6) the size, in bytes, of the string table.
75 The capabilities in the boolean flags, numbers, and strings sections
77 are in the same order as the file <term.h>.
78 Short integers are signed, in the range -32768 to 32767. They are
80 stored as two 8-bit bytes. The first byte contains the least signifi‐
81 cant 8 bits of the value, and the second byte contains the most signif‐
82 icant 8 bits. (Thus, the value represented is 256*second+first.) This
83 format corresponds to the hardware of the VAX and PDP-11 (that is, lit‐
84 tle-endian machines). Machines where this does not correspond to the
85 hardware must read the integers as two bytes and compute the little-en‐
86 dian value.
87 Numbers in a terminal description, whether they are entries in the num‐
89 bers or strings table, are positive integers. Boolean flags are
90 treated as positive one-byte integers. In each case, those positive
91 integers represent a terminal capability. The terminal compiler tic
92 uses negative integers to handle the cases where a capability is not
93 available:
94 • If a capability is absent from this terminal, tic stores a -1 in
96 the corresponding table.
97 The integer value -1 is represented by two bytes 0377, 0377.
99 Absent boolean values are represented by the byte 0 (false).
100 • If a capability has been canceled from this terminal, tic stores a
102 -2 in the corresponding table.
103 The integer value -2 is represented by two bytes 0377, 0376.
105 The boolean value -2 is represented by the byte 0376.
106 • Other negative values are illegal.
108 The terminal names section comes after the header. It contains the
110 first line of the terminfo description, listing the various names for
111 the terminal, separated by the “|” character. The terminal names sec‐
112 tion is terminated with an ASCII NUL character.
113 The boolean flags section has one byte for each flag. Boolean capabil‐
115 ities are either 1 or 0 (true or false) according to whether the termi‐
116 nal supports the given capability or not.
117 Between the boolean flags section and the number section, a null byte
119 will be inserted, if necessary, to ensure that the number section be‐
120 gins on an even byte This is a relic of the PDP-11's word-addressed ar‐
121 chitecture, originally designed to avoid traps induced by addressing a
122 word on an odd byte boundary. All short integers are aligned on a
123 short word boundary.
124 The numbers section is similar to the boolean flags section. Each ca‐
126 pability takes up two bytes, and is stored as a little-endian short in‐
127 teger.
128 The strings section is also similar. Each capability is stored as a
130 short integer. The capability value is an index into the string table.
131 The string table is the last section. It contains all of the values of
133 string capabilities referenced in the strings section. Each string is
134 null-terminated. Special characters in ^X or \c notation are stored in
135 their interpreted form, not the printing representation. Padding in‐
136 formation $<nn> and parameter information %x are stored intact in unin‐
137 terpreted form.
138 EXTENDED STORAGE FORMAT
140 The previous section describes the conventional terminfo binary format.
141 With some minor variations of the offsets (see PORTABILITY), the same
142 binary format is used in all modern UNIX systems. Each system uses a
143 predefined set of boolean, number or string capabilities.
144 The ncurses libraries and applications support extended terminfo binary
146 format, allowing users to define capabilities which are loaded at run‐
147 time. This extension is made possible by using the fact that the other
148 implementations stop reading the terminfo data when they have reached
149 the end of the size given in the header. ncurses checks the size, and
150 if it exceeds that due to the predefined data, continues to parse ac‐
151 cording to its own scheme.
152 First, it reads the extended header (5 short integers):
154 (1) count of extended boolean capabilities
156 (2) count of extended numeric capabilities
158 (3) count of extended string capabilities
160 (4) count of the items in extended string table
162 (5) size of the extended string table in bytes
164 The count- and size-values for the extended string table include the
166 extended capability names as well as extended capability values.
167 Using the counts and sizes, ncurses allocates arrays and reads data for
169 the extended capabilities in the same order as the header information.
170 The extended string table contains values for string capabilities. Af‐
172 ter the end of these values, it contains the names for each of the ex‐
173 tended capabilities in order, e.g., booleans, then numbers and finally
174 strings.
175 Applications which manipulate terminal data can use the definitions de‐
177 scribed in term_variables(3X) which associate the long capability names
178 with members of a TERMTYPE structure.
179 EXTENDED NUMBER FORMAT
181 On occasion, 16-bit signed integers are not large enough. With ncurses
182 6.1, a new format was introduced by making a few changes to the legacy
183 format:
184 • a different magic number (octal 01036)
186 • changing the type for the number array from signed 16-bit integers
188 to signed 32-bit integers.
189 To maintain compatibility, the library presents the same data struc‐
191 tures to direct users of the TERMTYPE structure as in previous formats.
192 However, that cannot provide callers with the extended numbers. The
193 library uses a similar but hidden data structure TERMTYPE2 to provide
194 data for the terminfo functions.
195
197 setupterm
198 Note that it is possible for setupterm to expect a different set of ca‐
199 pabilities than are actually present in the file. Either the database
200 may have been updated since setupterm has been recompiled (resulting in
201 extra unrecognized entries in the file) or the program may have been
202 recompiled more recently than the database was updated (resulting in
203 missing entries). The routine setupterm must be prepared for both pos‐
204 sibilities - this is why the numbers and sizes are included. Also, new
205 capabilities must always be added at the end of the lists of boolean,
206 number, and string capabilities.
207 Binary format
209 X/Open Curses does not specify a format for the terminfo database.
210 UNIX System V curses used a directory-tree of binary files, one per
211 terminal description.
212 Despite the consistent use of little-endian for numbers and the other‐
214 wise self-describing format, it is not wise to count on portability of
215 binary terminfo entries between commercial UNIX versions. The problem
216 is that there are at least three versions of terminfo (under HP-UX,
217 AIX, and OSF/1) which diverged from System V terminfo after SVr1, and
218 have added extension capabilities to the string table that (in the bi‐
219 nary format) collide with System V and XSI Curses extensions. See ter‐
220 minfo(5) for detailed discussion of terminfo source compatibility is‐
221 sues.
222 This implementation is by default compatible with the binary terminfo
224 format used by Solaris curses, except in a few less-used details where
225 it was found that the latter did not match X/Open Curses. The format
226 used by the other Unix versions can be matched by building ncurses with
227 different configuration options.
228 Magic codes
230 The magic number in a binary terminfo file is the first 16-bits (two
231 bytes). Besides making it more reliable for the library to check that
232 a file is terminfo, utilities such as file also use that to tell what
233 the file-format is. System V defined more than one magic number, with
234 0433, 0435 as screen-dumps (see scr_dump(5)). This implementation uses
235 01036 as a continuation of that sequence, but with a different high-or‐
236 der byte to avoid confusion.
237 The TERMTYPE structure
239 Direct access to the TERMTYPE structure is provided for legacy applica‐
240 tions. Portable applications should use the tigetflag and related
241 functions described in curs_terminfo(3X) for reading terminal capabili‐
242 ties.
243 Mixed-case terminal names
245 A small number of terminal descriptions use uppercase characters in
246 their names. If the underlying filesystem ignores the difference be‐
247 tween uppercase and lowercase, ncurses represents the “first character”
248 of the terminal name used as the intermediate level of a directory tree
249 in (two-character) hexadecimal form.
250
252 As an example, here is a description for the Lear-Siegler ADM-3, a pop‐
253 ular though rather stupid early terminal:
254 adm3a|lsi adm3a,
256 am,
257 cols#80, lines#24,
258 bel=^G, clear= 32$<1>, cr=^M, cub1=^H, cud1=^J,
259 cuf1=^L, cup=\E=%p1%{32}%+%c%p2%{32}%+%c, cuu1=^K,
260 home=^^, ind=^J,
261 and a hexadecimal dump of the compiled terminal description:
264 0000 1a 01 10 00 02 00 03 00 82 00 31 00 61 64 6d 33 ........ ..1.adm3
266 0010 61 7c 6c 73 69 20 61 64 6d 33 61 00 00 01 50 00 a|lsi ad m3a...P.
267 0020 ff ff 18 00 ff ff 00 00 02 00 ff ff ff ff 04 00 ........ ........
268 0030 ff ff ff ff ff ff ff ff 0a 00 25 00 27 00 ff ff ........ ..%.'...
269 0040 29 00 ff ff ff ff 2b 00 ff ff 2d 00 ff ff ff ff ).....+. ..-.....
270 0050 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
271 0060 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
272 0070 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
273 0080 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
274 0090 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
275 00a0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
276 00b0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
277 00c0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
278 00d0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
279 00e0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
280 00f0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
281 0100 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
282 0110 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........ ........
283 0120 ff ff ff ff ff ff 2f 00 07 00 0d 00 1a 24 3c 31 ....../. .....$<1
284 0130 3e 00 1b 3d 25 70 31 25 7b 33 32 7d 25 2b 25 63 >..=%p1% {32}%+%c
285 0140 25 70 32 25 7b 33 32 7d 25 2b 25 63 00 0a 00 1e %p2%{32} %+%c....
286 0150 00 08 00 0c 00 0b 00 0a 00 ........ .
287
290 Some limitations:
291 • total compiled entries cannot exceed 4096 bytes in the legacy for‐
293 mat.
294 • total compiled entries cannot exceed 32768 bytes in the extended
296 format.
297 • the name field cannot exceed 128 bytes.
299 Compiled entries are limited to 32768 bytes because offsets into the
301 strings table use two-byte integers. The legacy format could have sup‐
302 ported 32768-byte entries, but was limited a virtual memory page's 4096
303 bytes.
304
306 /usr/share/terminfo/*/* compiled terminal capability data base
307
309 curses(3X), terminfo(5).
310
312 Thomas E. Dickey
313 extended terminfo format for ncurses 5.0
314 hashed database support for ncurses 5.6
315 extended number support for ncurses 6.1
316 Eric S. Raymond
318 documented legacy terminfo format, e.g., from pcurses.
319 term(5)