comparison src/html.c @ 7094:2343c3aa1dec

[gaim-migrate @ 7659] grab_url() and parse_url() are gone, replaced with gaim_url_fetch() and gaim_url_parse(). They were also moved to util.[ch]. committer: Tailor Script <[e-mail]>

7093:3650612c7daa

			text2[j++] = text2[i];
		}
	}
	text2[j] = '\0';
	return text2;
}

struct g_url *parse_url(char *url)
{
	struct g_url *test = g_new0(struct g_url, 1);
	char scan_info[255];
	char port[5];
	int f;
	char* turl;
	/* hyphen at end includes it in control set */
	char addr_ctrl[] = "A-Za-z0-9.-";
	char port_ctrl[] = "0-9";
	char page_ctrl[] = "A-Za-z0-9.~_/:*!@&%%?=+^-";

	if((turl=strstr(url, "http://")) || (turl=strstr(url, "HTTP://")))
		url=turl+=7;

	snprintf(scan_info, sizeof(scan_info),
		 "%%[%s]:%%[%s]/%%[%s]",
		 addr_ctrl, port_ctrl, page_ctrl);

	f = sscanf(url, scan_info, test->address, port, test->page);
	if (f == 1) {
		snprintf(scan_info, sizeof(scan_info),
			 "%%[%s]/%%[%s]",
			 addr_ctrl, page_ctrl);
		f = sscanf(url, scan_info, test->address, test->page);
		snprintf(port, sizeof(port), "80");
	}
	if (f == 1)
		test->page[0] = '\0';

	sscanf(port, "%d", &test->port);
	return test;
}

struct grab_url_data {
	void (* callback)(gpointer, char *, unsigned long);
	gpointer data;
	struct g_url *website;
	char *url;
	gboolean full;
	char *user_agent;
	int http11;

	int inpa;

	gboolean sentreq;
	gboolean newline;
	gboolean startsaving;
	char *webdata;
	unsigned long len;
	unsigned long data_len;
};

static gboolean
parse_redirect(const char *data, size_t data_len, gint sock,
			   struct grab_url_data *gunk)
{
	gchar *s;

	if ((s = g_strstr_len(data, data_len, "Location: ")) != NULL) {
		gchar *new_url, *temp_url, *end;
		gboolean full;
		int len;

		s += strlen("Location: ");
		end = strchr(s, '\r');

		/* Just in case :) */
		if (end == NULL)
			end = strchr(s, '\n');

		len = end - s;

		new_url = g_malloc(len + 1);
		strncpy(new_url, s, len);
		new_url[len] = '\0';

		full = gunk->full;

		if (*new_url == '/' || g_strstr_len(new_url, len, "://") == NULL) {
			temp_url = new_url;

			new_url = g_strdup_printf("%s:%d%s", gunk->website->address,
									  gunk->website->port, temp_url);

			g_free(temp_url);

			full = FALSE;
		}

		/* Close the existing stuff. */
		gaim_input_remove(gunk->inpa);
		close(sock);

		gaim_debug(GAIM_DEBUG_INFO, "grab_url",
				   "Redirecting to %s\n", new_url);

		/* Try again, with this new location. */
		grab_url(new_url, full, gunk->callback, gunk->data, gunk->user_agent, gunk->http11);

		/* Free up. */
		g_free(new_url);
		g_free(gunk->webdata);
		g_free(gunk->website);
		g_free(gunk->url);
		g_free(gunk->user_agent);
		g_free(gunk);

		return TRUE;
	}

	return FALSE;
}

static size_t
parse_content_len(const char *data, size_t data_len)
{
	size_t content_len = 0;

	sscanf(data, "Content-Length: %d", &content_len);

	return content_len;
}

static void grab_url_callback(gpointer dat, gint sock, GaimInputCondition cond)
{
	struct grab_url_data *gunk = dat;
	char data;

	if (sock == -1) {
		gunk->callback(gunk->data, NULL, 0);
		g_free(gunk->website);
		g_free(gunk->url);
		g_free(gunk->user_agent);
		g_free(gunk);
		return;
	}

	if (!gunk->sentreq) {
		char buf[1024];

		if(gunk->user_agent) {
			if(gunk->http11)
				g_snprintf(buf, sizeof(buf), "GET %s%s HTTP/1.1\r\nUser-Agent: \"%s\"\r\nHost: %s\r\n\r\n", gunk->full ? "" : "/",
						gunk->full ? gunk->url : gunk->website->page, gunk->user_agent, gunk->website->address);
			else
				g_snprintf(buf, sizeof(buf), "GET %s%s HTTP/1.0\r\nUser-Agent: \"%s\"\r\n\r\n", gunk->full ? "" : "/",
						gunk->full ? gunk->url : gunk->website->page, gunk->user_agent);
		}
		else {
			if(gunk->http11)
				g_snprintf(buf, sizeof(buf), "GET %s%s HTTP/1.1\r\nHost: %s\r\n\r\n", gunk->full ? "" : "/",
						gunk->full ? gunk->url : gunk->website->page, gunk->website->address);
			else
				g_snprintf(buf, sizeof(buf), "GET %s%s HTTP/1.0\r\n\r\n", gunk->full ? "" : "/",
						gunk->full ? gunk->url : gunk->website->page);

		}
		gaim_debug(GAIM_DEBUG_MISC, "grab_url_callback",
				   "Request: %s\n", buf);

		write(sock, buf, strlen(buf));
		fcntl(sock, F_SETFL, O_NONBLOCK);
		gunk->sentreq = TRUE;
		gunk->inpa = gaim_input_add(sock, GAIM_INPUT_READ, grab_url_callback, dat);
		gunk->data_len = 4096;
		gunk->webdata = g_malloc(gunk->data_len);
		return;
	}

	if (read(sock, &data, 1) > 0 || errno == EWOULDBLOCK) {
		if (errno == EWOULDBLOCK) {
			errno = 0;
			return;
		}

		gunk->len++;

		if (gunk->len == gunk->data_len + 1) {
			gunk->data_len += (gunk->data_len) / 2;

			gunk->webdata = g_realloc(gunk->webdata, gunk->data_len);
		}

		gunk->webdata[gunk->len - 1] = data;

		if (!gunk->startsaving) {
			if (data == '\r')
				return;
			if (data == '\n') {
				if (gunk->newline) {
					size_t content_len;
					gunk->startsaving = TRUE;

					/* See if we can find a redirect. */
					if (parse_redirect(gunk->webdata, gunk->len, sock, gunk))
						return;

					/* No redirect. See if we can find a content length. */
					content_len = parse_content_len(gunk->webdata, gunk->len);

					if (content_len == 0) {
						/* We'll stick with an initial 8192 */
						content_len = 8192;
					}

					/* Out with the old... */
					gunk->len = 0;
					g_free(gunk->webdata);
					gunk->webdata = NULL;

					/* In with the new. */
					gunk->data_len = content_len;
					gunk->webdata = g_malloc(gunk->data_len);
				}
				else
					gunk->newline = TRUE;
				return;
			}
			gunk->newline = FALSE;
		}
	} else if (errno != ETIMEDOUT) {
		gunk->webdata = g_realloc(gunk->webdata, gunk->len + 1);
		gunk->webdata[gunk->len] = 0;

		gaim_debug(GAIM_DEBUG_MISC, "grab_url_callback",
				   "Received: '%s'\n", gunk->webdata);

		gaim_input_remove(gunk->inpa);
		close(sock);
		gunk->callback(gunk->data, gunk->webdata, gunk->len);
		if (gunk->webdata)
			g_free(gunk->webdata);
		g_free(gunk->website);
		g_free(gunk->url);
		g_free(gunk->user_agent);
		g_free(gunk);
	} else {
		gaim_input_remove(gunk->inpa);
		close(sock);
		gunk->callback(gunk->data, NULL, 0);
		if (gunk->webdata)
			g_free(gunk->webdata);
		g_free(gunk->website);
		g_free(gunk->url);
		g_free(gunk->user_agent);
		g_free(gunk);
	}
}

void grab_url(char *url, gboolean full, void callback(gpointer, char *, unsigned long),
		gpointer data, char *user_agent, int http11)
{
	int sock;
	struct grab_url_data *gunk = g_new0(struct grab_url_data, 1);

	gunk->callback = callback;
	gunk->data = data;
	gunk->url = g_strdup(url);
	gunk->user_agent = (user_agent) ? g_strdup(user_agent) : NULL;
	gunk->http11 = http11;
	gunk->website = parse_url(url);
	gunk->full = full;

	if ((sock = gaim_proxy_connect(NULL, gunk->website->address,
								   gunk->website->port, grab_url_callback,
								   gunk)) < 0) {
		g_free(gunk->website);
		g_free(gunk->url);
		g_free(gunk->user_agent);
		g_free(gunk);
		callback(data, g_strdup(_("g003: Error opening connection.\n")), 0);
	}
}

struct gaim_parse_tag {
	char *src_tag;
	char *dest_tag;

		*plain_out = g_strdup(plain->str);
	g_string_free(xhtml, TRUE);
	g_string_free(plain, TRUE);
}

int info_extract_field(char *original, char *add_to, char *start_tok,
		int skip, char *end_tok, char check_value, char *no_value_tok,
		char *display_name, int islink, char *link_prefix)
{
	char *p, *q;
	char buf[1024];

7094:2343c3aa1dec

			text2[j++] = text2[i];
		}
	}
	text2[j] = '\0';
	return text2;
}

struct gaim_parse_tag {
	char *src_tag;
	char *dest_tag;

		*plain_out = g_strdup(plain->str);
	g_string_free(xhtml, TRUE);
	g_string_free(plain, TRUE);
}

int info_extract_field(const char *original, char *add_to, char *start_tok,
		int skip, char *end_tok, char check_value, char *no_value_tok,
		char *display_name, int islink, char *link_prefix)
{
	char *p, *q;
	char buf[1024];