/**
* Intuit the media type they are searching based on the first XML tag
* we find in the meta data string, using simplistic lexical parsing which
* will encompass 99% of the cases.
*/
static uint32
g2_node_intuit_media_type(const char *md)
{
const char *p = md;
const char *start;
int c;
uint32 flags;
while ('<' != (c = *p++) && c != 0)
/* empty */;
if (0 == c)
return 0; /* Did not find any tag opening */
start = p = skip_ascii_spaces(p);
while (0 != (c = *p)) {
if (is_ascii_space(c) || '/' == c || '>' == c) {
char *name;
/* Found end of word, we got the tag name */
name = h_strndup(start, p - start);
flags = TOKENIZE(name, g2_q2_md);
if (0 == flags) {
g_warning("%s(): unknown tag \"%s\", XML string was \"%s\"",
G_STRFUNC, name, md);
}
hfree(name);
return flags;
}
p++;
}
return 0;
}
/**
* Try to connect to the list of nodes given by in following form:
*
* list = <node> | <node>, 1*<node>
* port = 1..65535
* hostname = 1*[a-zA-Z0-9.-]
* node = hostname [":" <port>]
* | <IPv4 address>[":" <port>]
* | <IPv6 address>
* | "[" <IPv6 address> "]:" <port>
* peer = ["tls:"]["g2:"]<node>
*
* If the port is omitted, the default port (GTA_PORT: 6346) is used.
* The case-insensitive prefix "tls:" requests a TLS (encrypted) connection.
*/
void
nodes_gui_common_connect_by_name(const gchar *line)
{
const gchar *q;
g_assert(line);
q = line;
while ('\0' != *q) {
const gchar *endptr, *hostname;
size_t hostname_len;
host_addr_t addr;
guint32 flags;
guint16 port;
bool g2;
q = skip_ascii_spaces(q);
if (',' == *q) {
q++;
continue;
}
addr = zero_host_addr;
port = GTA_PORT;
flags = SOCK_F_FORCE;
endptr = NULL;
hostname = NULL;
hostname_len = 0;
endptr = is_strcaseprefix(q, "tls:");
if (endptr) {
flags |= SOCK_F_TLS;
q = endptr;
}
endptr = is_strcaseprefix(q, "g2:");
if (endptr) {
g2 = TRUE;
q = endptr;
} else {
g2 = FALSE;
}
if (!string_to_host_or_addr(q, &endptr, &addr)) {
g_message("expected hostname or IP address");
break;
}
if (!is_host_addr(addr)) {
hostname = q;
hostname_len = endptr - q;
}
q = endptr;
if (':' == *q) {
gint error;
port = parse_uint16(&q[1], &endptr, 10, &error);
if (error || 0 == port) {
g_message("cannot parse port");
break;
}
q = skip_ascii_spaces(endptr);
} else {
q = skip_ascii_spaces(endptr);
if ('\0' != *q && ',' != *q) {
g_message("expected \",\" or \":\"");
break;
}
}
if (!hostname) {
if (g2) {
guc_node_g2_add(addr, port, flags);
} else {
guc_node_add(addr, port, flags);
}
} else {
struct add_node_context *ctx;
gchar *p;
if ('\0' == hostname[hostname_len]) {
p = NULL;
} else {
size_t n = 1 + hostname_len;
g_assert(n > hostname_len);
p = halloc(n);
g_strlcpy(p, hostname, n);
hostname = p;
}
WALLOC(ctx);
ctx->port = port;
ctx->flags = flags;
ctx->g2 = g2;
guc_adns_resolve(hostname, add_node_helper, ctx);
HFREE_NULL(p);
}
}
}
/**
* Append a new line of text at the end of the header.
* A private copy of the text is made.
*
* @return an error code, or HEAD_OK if appending was successful.
*/
int
header_append(header_t *o, const char *text, int len)
{
char buf[MAX_LINE_SIZE];
const char *p = text;
uchar c;
header_field_t *hf;
header_check(o);
g_assert(len >= 0);
if (o->flags & HEAD_F_EOH)
return HEAD_EOH_REACHED;
/*
* If empty line, we reached EOH.
*/
if (len == 0) {
o->flags |= HEAD_F_EOH; /* Mark we reached EOH */
return HEAD_EOH;
}
/*
* Sanity checks.
*/
if (o->size >= HEAD_MAX_SIZE)
return HEAD_TOO_LARGE;
if (++(o->num_lines) >= HEAD_MAX_LINES)
return HEAD_MANY_LINES;
/*
* Detect whether line is a new header or a continuation.
*/
c = *p;
if (is_ascii_space(c)) {
/*
* It's a continuation.
*
* Make sure we already have recorded something, or we have
* an unexpected continuation line.
*/
if (NULL == o->fields)
return HEAD_CONTINUATION; /* Unexpected continuation */
/*
* When a previous header line was malformed, we cannot accept
* further continuation lines.
*/
if (o->flags & HEAD_F_SKIP)
return HEAD_SKIPPED;
/*
* We strip leading spaces of all continuations before storing
* them. If we have to dump the header, we will have to put
* some spaces, but we don't guarantee we'll put the same amount.
*/
p++; /* First char is known space */
while ((c = *p)) {
if (!is_ascii_space(c))
break;
p++;
}
/*
* If we've reached the end of the line, then the continuation
* line was made of spaces only. Weird, but we can ignore it.
* Note that it's not an EOH mark.
*/
if (*p == '\0')
return HEAD_OK;
/*
* Save the continuation line by appending into the last header
* field we handled.
*/
hf = slist_tail(o->fields);
hfield_append(hf, p);
add_continuation(o, hf->name, p);
o->size += len - (p - text); /* Count only effective text */
/*
* Also append the data in the hash table.
*/
} else {
char *b;
bool seen_space = FALSE;
/*
* It's a new header line.
*/
o->flags &= ~HEAD_F_SKIP; /* Assume this line will be OK */
/*
* Parse header field. Must be composed of ascii chars only.
* (no control characters, no space, no ISO Latin or other extension).
* The field name ends with ':', after possible white spaces.
*/
for (b = buf, c = *p; c; c = *(++p)) {
if (c == ':') {
*b++ = '\0'; /* Reached end of field */
break; /* Done, buf[] holds field name */
}
if (is_ascii_space(c)) {
seen_space = TRUE; /* Only trailing spaces allowed */
continue;
}
if (
seen_space || (c != '-' && c != '.' &&
(!isascii(c) || is_ascii_cntrl(c) || is_ascii_punct(c)))
) {
o->flags |= HEAD_F_SKIP;
return HEAD_BAD_CHARS;
}
*b++ = c;
}
/*
* If buf[] does not end with a NUL, we did not fully recognize
* the header: we reached the end of the line without encountering
* the ':' marker.
*
* If the buffer starts with a NUL char, it's also clearly malformed.
*/
g_assert(b > buf || (b == buf && *text == '\0'));
if (b == buf || *(b-1) != '\0') {
o->flags |= HEAD_F_SKIP;
return HEAD_MALFORMED;
}
/*
* We have a valid header field in buf[].
*/
hf = hfield_make(buf);
/*
* Strip leading spaces in the value.
*/
g_assert(*p == ':');
p++; /* First char is field separator */
p = skip_ascii_spaces(p);
/*
* Record field value.
*/
hfield_append(hf, p);
add_header(o, buf, p);
if (!o->fields) {
o->fields = slist_new();
}
slist_append(o->fields, hf);
o->size += len - (p - text); /* Count only effective text */
}
return HEAD_OK;
}
/**
* Creates a valid and sanitized filename from the supplied string. For most
* Unix-like platforms anything goes but for security reasons, shell meta
* characters are replaced by harmless characters.
*
* @param filename the suggested filename.
* @param no_spaces if TRUE, spaces are replaced with underscores.
* @param no_evil if TRUE, "evil" characters are replaced with underscores.
*
* @returns a newly allocated string using halloc() or ``filename''
* if it was a valid filename already.
*/
char *
filename_sanitize(const char *filename, bool no_spaces, bool no_evil)
{
const char *p;
const char *s;
char *q;
g_assert(filename);
/* Almost all evil characters are forbidden on Windows, anyway */
no_evil |= is_running_on_mingw();
/* Leading spaces are just confusing */
p = skip_ascii_spaces(filename);
/* Make sure the filename isn't too long */
if (strlen(p) >= FILENAME_MAXBYTES) {
q = halloc(FILENAME_MAXBYTES);
filename_shrink(p, q, FILENAME_MAXBYTES);
s = q;
} else {
s = p;
q = NULL;
}
/*
* Replace shell meta characters and likely problematic characters.
*
* Although parentheses are not evil per se, they make it a pain to
* copy-n-paste filenames without going through the shell's auto-
* completion (which normally does the necessary escaping).
*
* To keep things "readable", we replace parentheses with brackets.
* Although brackets are meaningful for the shells, they are only
* interpreted in the presence of "*" or "?", two characters that we
* strip already.
* --RAM, 2013-11-03
*/
{
size_t i;
uchar c;
for (i = 0; '\0' != (c = s[i]); i++) {
if (
c < 32
|| is_ascii_cntrl(c)
|| G_DIR_SEPARATOR == c
|| '/' == c
|| (0 == i && ('.' == c || '-' == c))
|| (no_spaces && is_ascii_space(c))
|| (no_evil && filename_is_evil_char(c))
) {
if (!q)
q = h_strdup(s);
q[i] = '_'; /* replace undesired char with underscore */
} else if ('(' == c) {
if (!q)
q = h_strdup(s);
q[i] = '[';
} else if (')' == c) {
if (!q)
q = h_strdup(s);
q[i] = ']';
}
}
/**
* Windows does not like filenames ending with a space or period.
*/
while (i-- > 0 && (is_ascii_space(s[i]) || '.' == s[i])) {
if (!q)
q = h_strdup(s);
q[i] = '\0'; /* truncate string */
}
}
if (filename_is_reserved(q ? q : s)) {
HFREE_NULL(q);
q = h_strdup("noname");
}
if (NULL == q && s != filename)
q = h_strdup(s); /* Trimmed leading white space, must copy */
return q ? q : deconstify_gchar(s);
}
G_GNUC_COLD void
upload_stats_load_history(void)
{
FILE *upload_stats_file;
file_path_t fp;
char line[FILENAME_MAX + 64];
guint lineno = 0;
gcu_upload_stats_gui_freeze();
file_path_set(&fp, settings_config_dir(), ul_stats_file);
/* open file for reading */
upload_stats_file = file_config_open_read(ul_stats_what, &fp, 1);
if (upload_stats_file == NULL)
goto done;
/* parse, insert names into ul_stats_clist */
while (fgets(line, sizeof(line), upload_stats_file)) {
static const struct ul_stats zero_item;
struct ul_stats item;
struct sha1 sha1_buf;
const char *p;
size_t i;
lineno++;
if (line[0] == '#' || line[0] == '\n')
continue;
p = strchr(line, '\t');
if (NULL == p)
goto corrupted;
line[p - line] = '\0'; /* line is now the URL-escaped file name */
p++;
/* URL-unescape in-place */
if (!url_unescape(line, TRUE))
goto corrupted;
item = zero_item;
item.pathname = line;
for (i = 0; i < 8; i++) {
guint64 v;
int error;
const char *endptr;
p = skip_ascii_spaces(p);
/* SVN versions up to 15322 had only 6 fields in the history */
if (5 == i && '\0' == *p)
break;
switch (i) {
case 7:
/* We have a SHA1 or '*' if none known */
if ('*' != *p) {
size_t len = clamp_strlen(p, SHA1_BASE32_SIZE);
error = !parse_base32_sha1(p, len, &sha1_buf);
item.sha1 = error ? NULL : &sha1_buf;
} else {
error = FALSE;
}
p = skip_ascii_non_spaces(p);
v = 0;
break;
default:
v = parse_uint64(p, &endptr, 10, &error);
p = deconstify_gchar(endptr);
}
if (error || !is_ascii_space(*endptr))
goto corrupted;
switch (i) {
case 0: item.size = v; break;
case 1: item.attempts = v; break;
case 2: item.complete = v; break;
case 3: item.bytes_sent |= ((guint64) (guint32) v) << 32; break;
case 4: item.bytes_sent |= (guint32) v; break;
case 5: item.rtime = MIN(v + (time_t) 0, TIME_T_MAX + (guint64) 0);
case 6: item.dtime = MIN(v + (time_t) 0, TIME_T_MAX + (guint64) 0);
case 7: break; /* Already stored above */
default:
g_assert_not_reached();
goto corrupted;
}
}
/*
* We store the filenames UTF-8 encoded but the file might have been
* edited or corrupted.
*/
if (is_absolute_path(item.pathname)) {
item.filename = lazy_filename_to_utf8_normalized(
filepath_basename(item.pathname), UNI_NORM_NFC);
} else {
item.filename = lazy_unknown_to_utf8_normalized(
filepath_basename(item.pathname), UNI_NORM_NFC, NULL);
}
if (upload_stats_find(NULL, item.pathname, item.size)) {
g_warning("upload_stats_load_history():"
" Ignoring line %u due to duplicate file.", lineno);
} else if (upload_stats_find(item.sha1, item.pathname, item.size)) {
g_warning("upload_stats_load_history():"
" Ignoring line %u due to duplicate file.", lineno);
} else {
upload_stats_add(item.pathname, item.size, item.filename,
item.attempts, item.complete, item.bytes_sent,
item.rtime, item.dtime, item.sha1);
}
continue;
corrupted:
g_warning("upload statistics file corrupted at line %u.", lineno);
}
/* close file */
fclose(upload_stats_file);
done:
gcu_upload_stats_gui_thaw();
return;
}
