int main(int argc, const char **argv)
{
FILE *fpout;
#ifdef _GENERATE_BUILTIN_DATA
psl_ctx_t *psl;
#endif
int ret = 0;
if (argc != 3) {
fprintf(stderr, "Usage: psl2c <infile> <outfile>\n");
fprintf(stderr, " <infile> is the 'effective_tld_names.dat' (aka Public Suffix List), lowercase UTF-8 encoded\n");
fprintf(stderr, " <outfile> is the the C filename to be generated from <infile>\n");
return 1;
}
#ifdef _GENERATE_BUILTIN_DATA
if (!(psl = psl_load_file(argv[1])))
return 2;
if ((fpout = fopen(argv[2], "w"))) {
FILE *pp;
struct stat st;
size_t cmdsize = 16 + strlen(argv[1]);
char *cmd = alloca(cmdsize), checksum[64] = "";
#if 0
/* include library code did not generate punycode, so let's do it for the builtin data */
_add_punycode_if_needed(psl->suffixes);
_add_punycode_if_needed(psl->suffix_exceptions);
#endif
_print_psl_entries(fpout, psl->suffixes, "suffixes");
_print_psl_entries(fpout, psl->suffix_exceptions, "suffix_exceptions");
snprintf(cmd, cmdsize, "sha1sum %s", argv[1]);
if ((pp = popen(cmd, "r"))) {
if (fscanf(pp, "%63[0-9a-zA-Z]", checksum) < 1)
*checksum = 0;
pclose(pp);
}
if (stat(argv[1], &st) != 0)
st.st_mtime = 0;
fprintf(fpout, "static time_t _psl_file_time = %lu;\n", st.st_mtime);
fprintf(fpout, "static time_t _psl_compile_time = %lu;\n", time(NULL));
fprintf(fpout, "static const char _psl_sha1_checksum[] = \"%s\";\n", checksum);
fprintf(fpout, "static const char _psl_filename[] = \"%s\";\n", argv[1]);
if (fclose(fpout) != 0)
ret = 4;
} else {
fprintf(stderr, "Failed to write open '%s'\n", argv[2]);
ret = 3;
}
psl_free(psl);
#else
if ((fpout = fopen(argv[2], "w"))) {
fprintf(fpout, "static _psl_entry_t suffixes[1];\n");
fprintf(fpout, "static _psl_entry_t suffix_exceptions[1];\n");
fprintf(fpout, "static time_t _psl_file_time;\n");
fprintf(fpout, "static time_t _psl_compile_time;\n");
fprintf(fpout, "static const char _psl_sha1_checksum[] = \"\";\n");
fprintf(fpout, "static const char _psl_filename[] = \"\";\n");
if (fclose(fpout) != 0)
ret = 4;
} else {
fprintf(stderr, "Failed to write open '%s'\n", argv[2]);
ret = 3;
}
#endif /* GENERATE_BUILTIN_DATA */
return ret;
}
int main(int argc, const char **argv)
{
FILE *fpout;
#ifdef _GENERATE_BUILTIN_DATA
psl_ctx_t *psl;
#endif
int ret = 0;
if (argc != 3) {
fprintf(stderr, "Usage: psl2c <infile> <outfile>\n");
fprintf(stderr, " <infile> is the 'public_suffix_list.dat', lowercase UTF-8 encoded\n");
fprintf(stderr, " <outfile> is the the C filename to be generated from <infile>\n");
return 1;
}
#ifdef _GENERATE_BUILTIN_DATA
if (!(psl = psl_load_file(argv[1])))
return 2;
/* look for ambigious or double entries */
/* if (_check_psl(psl)) {
psl_free(psl);
return 5;
}
*/
if ((fpout = fopen(argv[2], "w"))) {
FILE *pp;
struct stat st;
size_t cmdsize = 16 + strlen(argv[1]);
char *cmd = alloca(cmdsize), checksum[64] = "";
const char *source_date_epoch = NULL;
#if 0
/* include library code did not generate punycode, so let's do it for the builtin data */
_add_punycode_if_needed(psl->suffixes);
#endif
_print_psl_entries(fpout, psl->suffixes, "suffixes");
snprintf(cmd, cmdsize, "sha1sum %s", argv[1]);
if ((pp = popen(cmd, "r"))) {
if (fscanf(pp, "%63[0-9a-zA-Z]", checksum) < 1)
*checksum = 0;
pclose(pp);
}
if (stat(argv[1], &st) != 0)
st.st_mtime = 0;
fprintf(fpout, "static time_t _psl_file_time = %lu;\n", st.st_mtime);
if ((source_date_epoch = getenv("SOURCE_DATE_EPOCH")))
fprintf(fpout, "static time_t _psl_compile_time = %lu;\n", atol(source_date_epoch));
else
fprintf(fpout, "static time_t _psl_compile_time = %lu;\n", time(NULL));
fprintf(fpout, "static int _psl_nsuffixes = %d;\n", psl->nsuffixes);
fprintf(fpout, "static int _psl_nexceptions = %d;\n", psl->nexceptions);
fprintf(fpout, "static int _psl_nwildcards = %d;\n", psl->nwildcards);
fprintf(fpout, "static const char _psl_sha1_checksum[] = \"%s\";\n", checksum);
fprintf(fpout, "static const char _psl_filename[] = \"%s\";\n", argv[1]);
if (fclose(fpout) != 0)
ret = 4;
} else {
fprintf(stderr, "Failed to write open '%s'\n", argv[2]);
ret = 3;
}
psl_free(psl);
#else
if ((fpout = fopen(argv[2], "w"))) {
fprintf(fpout, "static _psl_entry_t suffixes[1];\n");
fprintf(fpout, "static time_t _psl_file_time;\n");
fprintf(fpout, "static time_t _psl_compile_time;\n");
fprintf(fpout, "static int _psl_nsuffixes = 0;\n");
fprintf(fpout, "static int _psl_nexceptions = 0;\n");
fprintf(fpout, "static int _psl_nwildcards = 0;\n");
fprintf(fpout, "static const char _psl_sha1_checksum[] = \"\";\n");
fprintf(fpout, "static const char _psl_filename[] = \"\";\n");
if (fclose(fpout) != 0)
ret = 4;
} else {
fprintf(stderr, "Failed to write open '%s'\n", argv[2]);
ret = 3;
}
#endif /* GENERATE_BUILTIN_DATA */
return ret;
}
/**
* psl_load_fp:
* @fp: FILE pointer
*
* This function loads the public suffixes from a FILE pointer.
* To free the allocated resources, call psl_free().
*
* The suffixes are expected to be lowercase UTF-8 encoded if they are international.
*
* Returns: Pointer to a PSL context or %NULL on failure.
*
* Since: 0.1
*/
psl_ctx_t *psl_load_fp(FILE *fp)
{
psl_ctx_t *psl;
_psl_entry_t suffix, *suffixp;
char buf[256], *linep, *p;
#ifdef WITH_LIBICU
UIDNA *idna;
UErrorCode status = 0;
#endif
if (!fp)
return NULL;
if (!(psl = calloc(1, sizeof(psl_ctx_t))))
return NULL;
#ifdef WITH_LIBICU
idna = uidna_openUTS46(UIDNA_USE_STD3_RULES, &status);
#endif
/*
* as of 02.11.2012, the list at http://publicsuffix.org/list/ contains ~6000 rules and 40 exceptions.
* as of 19.02.2014, the list at http://publicsuffix.org/list/ contains ~6500 rules and 19 exceptions.
*/
psl->suffixes = _vector_alloc(8*1024, _suffix_compare);
psl->suffix_exceptions = _vector_alloc(64, _suffix_compare);
while ((linep = fgets(buf, sizeof(buf), fp))) {
while (isspace(*linep)) linep++; /* ignore leading whitespace */
if (!*linep) continue; /* skip empty lines */
if (*linep == '/' && linep[1] == '/')
continue; /* skip comments */
/* parse suffix rule */
for (p = linep; *linep && !isspace(*linep);) linep++;
*linep = 0;
if (*p == '!') {
/* add to exceptions */
if (_suffix_init(&suffix, p + 1, linep - p - 1) == 0) {
suffixp = _vector_get(psl->suffix_exceptions, _vector_add(psl->suffix_exceptions, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
#ifdef WITH_LIBICU
_add_punycode_if_needed(idna, psl->suffix_exceptions, suffixp);
#elif defined(WITH_LIBIDN2) || defined(WITH_LIBIDN)
_add_punycode_if_needed(psl->suffix_exceptions, suffixp);
#endif
}
} else {
/* add to suffixes */
if (_suffix_init(&suffix, p, linep - p) == 0) {
suffixp = _vector_get(psl->suffixes, _vector_add(psl->suffixes, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
#ifdef WITH_LIBICU
_add_punycode_if_needed(idna, psl->suffixes, suffixp);
#elif defined(WITH_LIBIDN2) || defined(WITH_LIBIDN)
_add_punycode_if_needed(psl->suffixes, suffixp);
#endif
}
}
}
_vector_sort(psl->suffix_exceptions);
_vector_sort(psl->suffixes);
#ifdef WITH_LIBICU
if (idna)
uidna_close(idna);
#endif
return psl;
}
