3 # Copyright © 2021, 2023 Nick Bowler
5 # Generate a C string table based on an input string specification file.
7 # A string table is a single large char single array containing all of
8 # the specified (0-terminated) strings, which is then offset to obtain
9 # the desired string. By storing these offsets instead of string pointers
10 # into read-only data structures, this can reduce the need for relocation
11 # processing at startup when programs are built in PIC mode.
13 # The string specification file is processed line by line. Comment
14 # lines may be included by beginning the line with a # character, which
15 # must be the very first character on the line. If a comment is encountered,
16 # processing immediately moves on to the next line and the result is as if
17 # the comment line were omitted from the input.
19 # Options may be used to alter the normal behaviour. An option is placed
20 # on a line by itself beginning with an @ character, and may appear anywhere
21 # in the input file. The following options are defined:
24 # All strings will have a non-zero offset in the strtab.
26 # A string is defined by beginning a line with one or two & characters, which
27 # must be immediately followed by a C identifier. Two & characters indicates
28 # a string that should not be translated, as described below. A nonempty
29 # sequence of whitespace (with at most one newline) separates the identifier
30 # from the beginning of the string itself. This whitespace is never included
33 # The string is then interpreted as follows:
35 # - Leading blanks on each line are ignored.
36 # - The sequences \\, \a, \b, \t, \n, \v, \f and \r can be entered and
37 # mean the same as they do in C string literals. The "\\" sequence
38 # prevents any special interpretation of the second backslash.
39 # - Newlines in the input are included in the output, except for the
40 # where the entire string (including its identifier) are on one line.
41 # - If this is not desired, a newline which is immediately preceded by an
42 # unescaped backslash will deleted, along with the backslash.
43 # - All other backslashes are deleted. This can be used to prevent special
44 # handling of whitespace, # or & characters at the beginning of a line.
46 # The output defines a variable, strtab, which contains all of the strings,
47 # and each identifier in the input is declared as an emumeration constant
48 # whose value is the offset of the associated string within strtab.
50 # Normally, the generated source code wraps strings using the identity macro
51 # N_(x), which has no effect on the resulting data structures but enables tools
52 # such as xgettext to extract translatable strings from the source code. An
53 # identifier preceded by two ampersands (&&) suppresses this output to allow
54 # a single string table to also contain both translateable strings as well as
55 # ones that should not be translated.
57 # The object-like macro STRTAB_MAX_OFFSET is defined and expands to the
58 # greatest string offset, suitable for use in #if preprocessing directives.
60 # License WTFPL2: Do What The Fuck You Want To Public License, version 2.
61 # This is free software: you are free to do what the fuck you want to.
62 # There is NO WARRANTY, to the extent permitted by law.
67 print " * Automatically generated by gen-strtab.awk from " FILENAME
69 print " * Automatically generated by gen-strtab.awk"
71 print " * Do not edit."
77 collected = ident = ""
78 startline = endline = 0
83 NF == 0 || $0 ~ /^[#]/ { next }
90 val = !sub(/^no_?/, "", $1);
94 print "error: unrecognized option: @" orig | "cat 1>&2"
103 finish_string_input(strings, ident, collected)
104 vars[num_vars++] = ident
107 current_l10n = !sub(/^[&]/, "", $1);
118 collected = collected "\n" $0
128 finish_string_input(strings, ident, collected)
129 vars[num_vars++] = ident
136 count = bucketsort(sorted_strings, strings)
139 print "\n#define STR_L10N_(x)"
141 print "# define N_(x) x"
143 print "\nstatic const char strtab[] ="
145 for (i = 0; i < count; i++) {
146 s = sorted_strings[i]
147 gsub(/\\\\/, "\2", s)
148 if ((n = index(strtab "\1", s "\1")) > 0) {
149 offsets[sorted_strings[i]] = real_length(substr(strtab, 1, n-1));
150 if (!(sorted_strings[i] in nol10n))
151 print "\tSTR_L10N_(N_(\"" sorted_strings[i] "\"))";
153 strtab = strtab "\1" s
154 offsets[sorted_strings[i]] = strtab_len + 1
155 strtab_len += real_length(s) + 1
158 offsets[sorted_strings[i]] = 0
159 strtab_len += real_length(s)
163 gsub(/\2/, "\\\\", strtab);
164 n = split(strtab, split_strtab, "\1");
165 for (i = 1; i <= n; i++) {
166 printf("\t%4s ", i > !!opts["zero"] ? "\"\\0\"" : "");
168 if (split_strtab[i] in nol10n) {
169 print "\"" split_strtab[i] "\"";
171 print "N_(\"" split_strtab[i] "\")";
177 for (i = 0; i < num_vars; i++) {
178 sep = (i+1) != num_vars ? "," : ""
180 o = offsets[strings[s]] + !opts["zero"]
181 print "\t" s " = " o sep
187 print "\n#define STRTAB_MAX_OFFSET " max
190 # finish_input_string(strings, ident, val)
192 # Deal with backslash-escapes and special characters in val, then set
193 # strings[ident] = val.
194 function finish_string_input(strings, ident, val, n, tmpval)
196 gsub(/\\\\/, "\1", val)
197 val = val (endline > startline ? "\n" : "")
198 gsub(/\\\n/, "", val)
201 while ((n = match(val, /\\[^abtnvfr]/)) > 0) {
202 tmpval = tmpval substr(val, 1, n-1)
203 val = substr(val, n+1)
207 # Escape special characters
208 gsub(/"/, "\\\"", tmpval)
209 gsub(/\t/, "\\t", tmpval)
210 gsub(/\n/, "\\n", tmpval)
211 gsub(/\1/, "\\\\", tmpval)
213 strings[ident] = tmpval
219 function real_length(s, t)
222 return t - gsub(/\\./, "&", s)
225 # bucketsort(dst, src)
227 # Sort the elements of src by descending string length,
228 # placing them into dst[0] ... dst[n].
230 # Returns the number of elements.
231 function bucketsort(dst, src, buckets, max, count, i, t)
235 if (i > max) { max = i }
239 for (i = max; i > 0; i--) {
248 i = length(t = src[t])
249 dst[buckets[i]++] = t