// Iterate entries. The first call establishes the first entry. Returns false
// if no entry found, otherwise returns true and sets mpa->entry/entry_filename.
bool mp_archive_next_entry(struct mp_archive *mpa)
{
mpa->entry = NULL;
talloc_free(mpa->entry_filename);
mpa->entry_filename = NULL;
while (!mp_cancel_test(mpa->primary_src->cancel)) {
struct archive_entry *entry;
int r = archive_read_next_header(mpa->arch, &entry);
if (r == ARCHIVE_EOF)
break;
if (r < ARCHIVE_OK)
MP_ERR(mpa, "%s\n", archive_error_string(mpa->arch));
if (r < ARCHIVE_WARN) {
MP_FATAL(mpa, "could not read archive entry\n");
break;
}
if (archive_entry_filetype(entry) != AE_IFREG)
continue;
// Some archives may have no filenames, or libarchive won't return some.
const char *fn = archive_entry_pathname(entry);
char buf[64];
if (!fn || bstr_validate_utf8(bstr0(fn)) < 0) {
snprintf(buf, sizeof(buf), "mpv_unknown#%d", mpa->entry_num);
fn = buf;
}
mpa->entry = entry;
mpa->entry_filename = talloc_strdup(mpa, fn);
mpa->entry_num += 1;
return true;
}
return false;
}
static const char *enca_guess(struct mp_log *log, bstr buf, const char *language)
{
// Do our own UTF-8 detection, because ENCA seems to get it wrong sometimes
// (suggested by divVerent). Explicitly allow cut-off UTF-8.
if (bstr_validate_utf8(buf) > -8)
return "UTF-8";
if (!language || !language[0])
language = "__"; // neutral language
const char *detected_cp = NULL;
EncaAnalyser analyser = enca_analyser_alloc(language);
if (analyser) {
enca_set_termination_strictness(analyser, 0);
EncaEncoding enc = enca_analyse_const(analyser, buf.start, buf.len);
const char *tmp = enca_charset_name(enc.charset, ENCA_NAME_STYLE_ICONV);
if (tmp && enc.charset != ENCA_CS_UNKNOWN)
detected_cp = tmp;
enca_analyser_free(analyser);
} else {
mp_err(log, "ENCA doesn't know language '%s'\n", language);
size_t langcnt;
const char **languages = enca_get_languages(&langcnt);
mp_err(log, "ENCA supported languages:");
for (int i = 0; i < langcnt; i++)
mp_err(log, " %s", languages[i]);
mp_err(log, "\n");
free(languages);
}
return detected_cp;
}
// Runs charset auto-detection on the input buffer, and returns the result.
// If auto-detection fails, NULL is returned.
// If user_cp doesn't refer to any known auto-detection (for example because
// it's a real iconv codepage), user_cp is returned without even looking at
// the buf data.
const char *mp_charset_guess(struct mp_log *log, bstr buf, const char *user_cp,
int flags)
{
if (!mp_charset_requires_guess(user_cp))
return user_cp;
// Do our own UTF-8 detection, because at least ENCA seems to get it
// wrong sometimes (suggested by divVerent).
int r = bstr_validate_utf8(buf);
if (r >= 0 || (r > -8 && (flags & MP_ICONV_ALLOW_CUTOFF)))
return "UTF-8";
bstr params[3] = {{0}};
split_colon(user_cp, 3, params);
bstr type = params[0];
char lang[100];
snprintf(lang, sizeof(lang), "%.*s", BSTR_P(params[1]));
const char *fallback = params[2].start; // last item, already 0-terminated
const char *res = NULL;
#if HAVE_ENCA
if (bstrcasecmp0(type, "enca") == 0)
res = enca_guess(log, buf, lang);
#endif
#if HAVE_LIBGUESS
if (bstrcasecmp0(type, "guess") == 0)
res = libguess_guess(log, buf, lang);
#endif
if (bstrcasecmp0(type, "utf8") == 0 || bstrcasecmp0(type, "utf-8") == 0) {
if (!fallback)
fallback = params[1].start; // must be already 0-terminated
}
if (res) {
mp_dbg(log, "%.*s detected charset: '%s'\n",
BSTR_P(type), res);
} else {
res = fallback;
mp_dbg(log, "Detection with %.*s failed: fallback to %s\n",
BSTR_P(type), res && res[0] ? res : "broken UTF-8/Latin1");
}
if (!res && !(flags & MP_STRICT_UTF8))
res = "UTF-8-BROKEN";
return res;
}
// Runs charset auto-detection on the input buffer, and returns the result.
// If auto-detection fails, NULL is returned.
// If user_cp doesn't refer to any known auto-detection (for example because
// it's a real iconv codepage), user_cp is returned without even looking at
// the buf data.
// The return value may (but doesn't have to) be allocated under talloc_ctx.
const char *mp_charset_guess(void *talloc_ctx, struct mp_log *log, bstr buf,
const char *user_cp, int flags)
{
if (!mp_charset_requires_guess(user_cp))
return user_cp;
bool use_auto = strcasecmp(user_cp, "auto") == 0;
if (use_auto) {
#if HAVE_UCHARDET
user_cp = "uchardet";
#elif HAVE_ENCA
user_cp = "enca";
#else
user_cp = "UTF-8:UTF-8-BROKEN";
#endif
}
bstr params[3] = {{0}};
split_colon(user_cp, 3, params);
bstr type = params[0];
char lang[100];
snprintf(lang, sizeof(lang), "%.*s", BSTR_P(params[1]));
const char *fallback = params[2].start; // last item, already 0-terminated
const char *res = NULL;
if (use_auto) {
res = ms_bom_guess(buf);
if (res)
type = bstr0("auto");
}
#if HAVE_ENCA
if (bstrcasecmp0(type, "enca") == 0)
res = enca_guess(log, buf, lang);
#endif
#if HAVE_LIBGUESS
if (bstrcasecmp0(type, "guess") == 0)
res = libguess_guess(log, buf, lang);
#endif
#if HAVE_UCHARDET
if (bstrcasecmp0(type, "uchardet") == 0)
res = mp_uchardet(talloc_ctx, log, buf);
#endif
if (bstrcasecmp0(type, "utf8") == 0 || bstrcasecmp0(type, "utf-8") == 0) {
if (!fallback)
fallback = params[1].start; // must be already 0-terminated
int r = bstr_validate_utf8(buf);
if (r >= 0 || (r > -8 && (flags & MP_ICONV_ALLOW_CUTOFF)))
res = "utf-8";
}
if (res) {
mp_dbg(log, "%.*s detected charset: '%s'\n", BSTR_P(type), res);
} else {
res = fallback;
mp_dbg(log, "Detection with %.*s failed: fallback to %s\n",
BSTR_P(type), res && res[0] ? res : "broken UTF-8/Latin1");
}
if (!res && !(flags & MP_STRICT_UTF8))
res = "UTF-8-BROKEN";
mp_verbose(log, "Using charset '%s'.\n", res);
return res;
}
