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.
27 # Instead of a variable declaration, the generated header will define an
28 # object-like macro that can be used as the initializer for a char array.
30 # A string is defined by beginning a line with one or two & characters, which
31 # must be immediately followed by a C identifier. Two & characters indicates
32 # a string that should not be translated, as described below. A nonempty
33 # sequence of whitespace (with at most one newline) separates the identifier
34 # from the beginning of the string itself. This whitespace is never included
37 # The string is then interpreted as follows:
39 # - Leading blanks on each line are ignored.
40 # - The sequences \\, \a, \b, \t, \n, \v, \f and \r can be entered and
41 # mean the same as they do in C string literals. The "\\" sequence
42 # prevents any special interpretation of the second backslash.
43 # - Newlines in the input are included in the output, except for the
44 # where the entire string (including its identifier) are on one line.
45 # - If this is not desired, a newline which is immediately preceded by an
46 # unescaped backslash will deleted, along with the backslash.
47 # - All other backslashes are deleted. This can be used to prevent special
48 # handling of whitespace, # or & characters at the beginning of a line.
50 # Unless the @macro option is specified, the output defines a variable,
51 # strtab, which contains all of the strings, and each identifier in the input
52 # is declared as an emumeration constant whose value is the offset of the
53 # associated string within strtab. Otherwise, if the @macro option is
54 # specified, no variables are defined and STRTAB_INITIALIZER object-like macro
55 # may be used to initialize a char array with static storage duration.
57 # Normally, the generated source code wraps strings using the identity macro
58 # N_(x), which has no effect on the resulting data structures but enables tools
59 # such as xgettext to extract translatable strings from the source code. An
60 # identifier preceded by two ampersands (&&) suppresses this output to allow
61 # a single string table to also contain both translateable strings as well as
62 # ones that should not be translated.
64 # The object-like macro STRTAB_MAX_OFFSET is defined and expands to the
65 # greatest string offset, suitable for use in #if preprocessing directives.
67 # License WTFPL2: Do What The Fuck You Want To Public License, version 2.
68 # This is free software: you are free to do what the fuck you want to.
69 # There is NO WARRANTY, to the extent permitted by law.
74 print " * Automatically generated by gen-strtab.awk from " FILENAME
76 print " * Automatically generated by gen-strtab.awk"
78 print " * Do not edit."
83 # Check if "\\\\" in substitutions gives just one backslash.
84 bs = "x"; sub(/x/, "\\\\", bs);
85 bs = (length(bs) == 1 ? "\\\\" : "\\");
89 collected = ident = ""
90 startline = endline = 0
95 NF == 0 || $0 ~ /^[#]/ { next }
102 val = !sub(/^no_?/, "", $1);
106 print "error: unrecognized option: @" orig | "cat 1>&2"
115 finish_string_input(strings, ident, collected);
116 vars[num_vars++] = ident;
119 current_l10n = !sub(/^[&]/, "", $1);
129 sep = collected != "" ? "\n" : "";
130 collected = collected sep $0;
136 finish_string_input(strings, ident, collected)
137 vars[num_vars++] = ident
144 count = bucketsort(sorted_strings, strings)
147 print "\n#define STR_L10N_(x)"
149 print "# define N_(x) x"
153 print "\n#define STRTAB_INITIALIZER" cont;
155 print "\nstatic const char strtab[] =";
158 for (i = 0; i < count; i++) {
159 s = sorted_strings[i]
160 gsub(/\\\\/, "\2", s)
161 if ((n = index(strtab "\1", s "\1")) > 0) {
162 offsets[sorted_strings[i]] = real_length(substr(strtab, 1, n-1));
163 if (!(sorted_strings[i] in nol10n))
164 print "\tSTR_L10N_(N_(\"" sorted_strings[i] "\"))" cont;
166 strtab = strtab "\1" s
167 offsets[sorted_strings[i]] = strtab_len + 1
168 strtab_len += real_length(s) + 1
171 offsets[sorted_strings[i]] = 0
172 strtab_len += real_length(s)
176 gsub("\2", bs bs, strtab);
177 n = split(strtab, split_strtab, "\1");
178 for (i = 1; i <= n; i++) {
179 printf("\t%4s ", i > !!opts["zero"] ? "\"\\0\"" : "");
181 if (split_strtab[i] in nol10n) {
182 print "\"" split_strtab[i] "\"" cont;
184 print "N_(\"" split_strtab[i] "\")" cont;
187 print "\t\"\"" substr(";", 1, !opts["macro"]);
190 for (i = 0; i < num_vars; i++) {
191 sep = (i+1) != num_vars ? "," : ""
193 o = offsets[strings[s]] + (!opts["zero"])
194 print "\t" s " = " o sep
200 print "\n#define STRTAB_MAX_OFFSET " max
203 # finish_string_input(strings, ident, val)
205 # Deal with backslash-escapes and special characters in val, then set
206 # strings[ident] = val.
207 function finish_string_input(strings, ident, val, n, tmpval)
209 gsub(/\\\\/, "\1", val);
210 if (endline > startline)
212 gsub(/\\\n/, "", val);
215 while ((n = match(val, /\\[^abtnvfr]/)) > 0) {
216 tmpval = tmpval substr(val, 1, n-1)
217 val = substr(val, n+1)
221 # Escape special characters
222 gsub(/"/, bs"\"", tmpval)
223 gsub(/\t/, bs"t", tmpval)
224 gsub(/\n/, bs"n", tmpval)
225 gsub("\1", bs bs, tmpval)
227 strings[ident] = tmpval
233 function real_length(s, t)
236 return t - gsub(/\\./, "&", s)
239 # bucketsort(dst, src)
241 # Sort the elements of src by descending string length,
242 # placing them into dst[0] ... dst[n].
244 # Returns the number of elements.
245 function bucketsort(dst, src, max, count, i, t)
247 # Note: ULTRIX 4.5 nawk does not support local array parameters
248 split("", bucketsort_buckets);
252 if (i > max) { max = i }
253 bucketsort_buckets[i]++
256 for (i = max; i > 0; i--) {
257 if (i in bucketsort_buckets) {
258 t = bucketsort_buckets[i]
259 bucketsort_buckets[i] = count
265 i = length(t = src[t])
266 dst[bucketsort_buckets[i]++] = t