contiki/apps/webbrowser/www.c

1013 lines
28 KiB
C
Raw Normal View History

/*
* Copyright (c) 2002, Adam Dunkels.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials provided
* with the distribution.
* 3. The name of the author may not be used to endorse or promote
* products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS
* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
* DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
* GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
* NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* This file is part of the Contiki desktop environment
*
*
*/
#include <string.h>
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#include <stddef.h>
#include "ctk/ctk.h"
#include "ctk/ctk-textentry-cmdline.h"
#include "contiki-net.h"
#include "lib/petsciiconv.h"
#include "sys/arg.h"
#if WWW_CONF_WITH_WGET
#include "program-handler.h"
#endif /* WWW_CONF_WITH_WGET */
#include "webclient.h"
#include "htmlparser.h"
#include "http-strings.h"
#include "www.h"
#if 1
#define PRINTF(x)
#else
#include <stdio.h>
#define PRINTF(x) printf x
#endif
/* The array that holds the current URL. */
static char url[WWW_CONF_MAX_URLLEN + 1];
/* The array that holds the web page text. */
static char webpage[WWW_CONF_WEBPAGE_WIDTH *
WWW_CONF_WEBPAGE_HEIGHT + 1];
/* The CTK widgets for the main window. */
static struct ctk_window mainwindow;
#if WWW_CONF_HISTORY_SIZE > 0
static struct ctk_button backbutton =
{CTK_BUTTON(0, 0, 4, "Back")};
static struct ctk_button downbutton =
{CTK_BUTTON(10, 0, 4, "Down")};
#else /* WWW_CONF_HISTORY_SIZE > 0 */
static struct ctk_button downbutton =
{CTK_BUTTON(0, 0, 4, "Down")};
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
static struct ctk_button stopbutton =
{CTK_BUTTON(WWW_CONF_WEBPAGE_WIDTH - 16, 0, 4, "Stop")};
static struct ctk_button gobutton =
{CTK_BUTTON(WWW_CONF_WEBPAGE_WIDTH - 4, 0, 2, "Go")};
static struct ctk_separator sep1 =
{CTK_SEPARATOR(0, 2, WWW_CONF_WEBPAGE_WIDTH)};
static char editurl[WWW_CONF_MAX_URLLEN + 1];
static struct ctk_textentry urlentry =
{CTK_TEXTENTRY(0, 1, WWW_CONF_WEBPAGE_WIDTH - 2,
1, editurl, WWW_CONF_MAX_URLLEN)};
static struct ctk_label webpagelabel =
{CTK_LABEL(0, 3, WWW_CONF_WEBPAGE_WIDTH,
WWW_CONF_WEBPAGE_HEIGHT, webpage)};
static char statustexturl[WWW_CONF_WEBPAGE_WIDTH];
static struct ctk_label statustext =
{CTK_LABEL(0, WWW_CONF_WEBPAGE_HEIGHT + 4,
WWW_CONF_WEBPAGE_WIDTH, 1, "")};
static struct ctk_separator sep2 =
{CTK_SEPARATOR(0, WWW_CONF_WEBPAGE_HEIGHT + 3,
WWW_CONF_WEBPAGE_WIDTH)};
#if WWW_CONF_WITH_WGET || defined(WWW_CONF_WGET_EXEC)
#if CTK_CONF_WINDOWS
static struct ctk_window wgetdialog;
static struct ctk_label wgetlabel1 =
{CTK_LABEL(1, 1, 34, 1, "This web page cannot be displayed.")};
static struct ctk_label wgetlabel2 =
{CTK_LABEL(1, 3, 35, 1, "Would you like to download instead?")};
static struct ctk_button wgetnobutton =
{CTK_BUTTON(1, 5, 6, "Cancel")};
static struct ctk_button wgetyesbutton =
{CTK_BUTTON(11, 5, 24, "Close browser & download")};
#else /* CTK_CONF_WINDOWS */
static struct ctk_button wgetnobutton =
{CTK_BUTTON((WWW_CONF_WEBPAGE_WIDTH - 38) / 2 + 1,
11, 6, "Cancel")};
static struct ctk_button wgetyesbutton =
{CTK_BUTTON((WWW_CONF_WEBPAGE_WIDTH - 38) / 2 + 11,
11, 24, "Close browser & download")};
#endif /* CTK_CONF_WINDOWS */
#endif /* WWW_CONF_WITH_WGET || WWW_CONF_WGET_EXEC */
#if WWW_CONF_HISTORY_SIZE > 0
/* The char arrays that hold the history of visited URLs. */
static char history[WWW_CONF_HISTORY_SIZE][WWW_CONF_MAX_URLLEN];
static unsigned char history_last;
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
struct linkattrib {
struct ctk_hyperlink hyperlink;
char url[1];
};
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
struct inputattrib;
struct formattrib {
struct inputattrib *nextptr;
char action[1];
};
struct inputattrib {
struct inputattrib *nextptr;
struct formattrib *formptr;
struct ctk_widget widget;
};
struct textattrib {
struct inputattrib *nextptr;
struct formattrib *formptr;
struct ctk_textentry textentry;
char name[1];
};
struct submitattrib {
struct inputattrib *nextptr;
struct formattrib *formptr;
struct ctk_button button;
char name[1];
};
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static char pageattribs[WWW_CONF_PAGEATTRIB_SIZE];
static char *pageattribptr;
#if WWW_CONF_FORMS
static struct formattrib *formptr;
static struct inputattrib *currptr;
#endif /* WWW_CONF_FORMS */
#define ISO_nl 0x0a
#define ISO_space 0x20
#define ISO_hash 0x23
#define ISO_ampersand 0x26
#define ISO_plus 0x2b
#define ISO_slash 0x2f
#define ISO_eq 0x3d
#define ISO_questionmark 0x3f
/* The state of the rendering code. */
static char *webpageptr;
static unsigned char x, y;
static unsigned char loading;
static unsigned short firsty, pagey;
static unsigned char newlines;
static unsigned char count;
static char receivingmsgs[4][23] = {
"Receiving web page ...",
"Receiving web page. ..",
"Receiving web page.. .",
"Receiving web page... "
};
PROCESS(www_process, "Web browser");
AUTOSTART_PROCESSES(&www_process);
static void CC_FASTCALL formsubmit(struct inputattrib *trigger);
/*-----------------------------------------------------------------------------------*/
/* make_window()
*
* Creates the web browser's window.
*/
static void
make_window(void)
{
#if WWW_CONF_HISTORY_SIZE > 0
CTK_WIDGET_ADD(&mainwindow, &backbutton);
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
CTK_WIDGET_ADD(&mainwindow, &downbutton);
CTK_WIDGET_ADD(&mainwindow, &stopbutton);
CTK_WIDGET_ADD(&mainwindow, &gobutton);
CTK_WIDGET_ADD(&mainwindow, &urlentry);
CTK_WIDGET_ADD(&mainwindow, &sep1);
CTK_WIDGET_ADD(&mainwindow, &webpagelabel);
CTK_WIDGET_SET_FLAG(&webpagelabel, CTK_WIDGET_FLAG_MONOSPACE);
CTK_WIDGET_ADD(&mainwindow, &sep2);
CTK_WIDGET_ADD(&mainwindow, &statustext);
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
pageattribptr = pageattribs;
}
/*-----------------------------------------------------------------------------------*/
/* redraw_window():
*
* Convenience function that calls upon CTK to redraw the browser
* window. */
static void
redraw_window(void)
{
ctk_window_redraw(&mainwindow);
}
/*-----------------------------------------------------------------------------------*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static char * CC_FASTCALL
add_pageattrib(unsigned size)
{
char *ptr;
if(pageattribptr + size > pageattribs + sizeof(pageattribs)) {
return NULL;
}
ptr = pageattribptr;
pageattribptr += size;
return ptr;
}
/*-----------------------------------------------------------------------------------*/
#if WWW_CONF_FORMS
static void CC_FASTCALL
add_forminput(struct inputattrib *inputptr)
{
inputptr->nextptr = NULL;
currptr->nextptr = inputptr;
currptr = inputptr;
}
#endif /* WWW_CONF_FORMS */
/*-----------------------------------------------------------------------------------*/
static void
clear_page(void)
{
ctk_window_clear(&mainwindow);
make_window();
memset(webpage, 0, WWW_CONF_WEBPAGE_WIDTH * WWW_CONF_WEBPAGE_HEIGHT);
redraw_window();
}
/*-----------------------------------------------------------------------------------*/
static void
show_url(void)
{
memcpy(editurl, url, WWW_CONF_MAX_URLLEN);
strncpy(editurl, "http://", 7);
petsciiconv_topetscii(editurl + 7, WWW_CONF_MAX_URLLEN - 7);
CTK_WIDGET_REDRAW(&urlentry);
}
/*-----------------------------------------------------------------------------------*/
static void
start_loading(void)
{
loading = 1;
x = y = 0;
pagey = 0;
newlines = 0;
webpageptr = webpage;
clear_page();
}
/*-----------------------------------------------------------------------------------*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static void CC_FASTCALL
show_statustext(char *text)
{
ctk_label_set_text(&statustext, text);
CTK_WIDGET_REDRAW(&statustext);
}
/*-----------------------------------------------------------------------------------*/
/* open_url():
*
* Called when the URL present in the global "url" variable should be
* opened. It will call the hostname resolver as well as the HTTP
* client requester.
*/
static void
open_url(void)
{
unsigned char i;
static char host[32];
char *file;
register char *urlptr;
static uip_ipaddr_t addr;
/* Trim off any spaces in the end of the url. */
urlptr = url + strlen(url);
while(urlptr > url) {
if(*(urlptr - 1) == ' ') {
*--urlptr = 0;
} else {
break;
}
}
/* Don't even try to go further if the URL is empty. */
if(urlptr == url) {
return;
}
/* See if the URL starts with http://, otherwise prepend it. */
if(strncmp(url, http_http, 7) != 0) {
while(urlptr >= url) {
*(urlptr + 7) = *urlptr;
--urlptr;
}
strncpy(url, http_http, 7);
}
/* Find host part of the URL. */
urlptr = &url[7];
for(i = 0; i < sizeof(host); ++i) {
if(*urlptr == 0 ||
*urlptr == '/' ||
*urlptr == ' ' ||
*urlptr == ':') {
host[i] = 0;
break;
}
host[i] = *urlptr;
++urlptr;
}
/* XXX: Here we should find the port part of the URL, but this isn't
currently done because of laziness from the programmer's side
:-) */
/* Find file part of the URL. */
while(*urlptr != '/' && *urlptr != 0) {
++urlptr;
}
if(*urlptr == '/') {
file = urlptr;
} else {
file = "/";
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#if UIP_UDP
/* Try to lookup the hostname. If it fails, we initiate a hostname
lookup and print out an informative message on the statusbar. */
if(uiplib_ipaddrconv(host, &addr) == 0) {
core/net/resolv: IPv6 and mDNS ("Bonjour") support. Major refactor. This patch updates the DNS resolver to support IPv6 and introduces an improved API for looking up DNS entries. This patch also adds optional support for mDNS lookups and responses to the DNS resolver. Here is a quick summary of the changes: * Added support for IPv6 lookups. * DNS queries now honor record expiration. * Added support for mDNS, compatible with "Bonjour". * Implemented a new lookup api, `resolv_lookup2()`, which provides more information about the state of the record(error, expired, looking-up, etc.). About mDNS/Bonjour Support -------------------------- This patch adds basic support for mDNS/Bonjour, which allows you to refer to the name of a device instead of its IP address. This is incredibly convenient for IPv6 addresses because they tend to be very long and difficult to remember. It is especially important for link-local IPv6 addresses, since not all programs support the '%' notation for indicating a network interface (required on systems with more than one network interface to disambiguate). In other words, instead of typing in this: * `http://[fe80::58dc:d7ed:a644:628f%en1]/` You can type this instead: * `http://contiki.local/` Huge improvement, no? The convenience extends beyond that: this mechanism can be used for nodes to talk to each other based on their human-readable names instead of their IPv6 addresses. So instead of a switch on `aaaa::58dc:d7ed:a644:628f` triggering an actuator on `aaaa::ed26:19c1:4bd2:f95b`, `light-switch.local` can trigger the actuator on `living-room-lights.local`. What you need to do to be able to look up `.local` names on your workstation depends on a few factors: * Your machine needs to be able to send and receive multicast packets to and from the LoWPAN. You can do this easily with the Jackdaw firmware on an RZUSBStick. If you have a border router, you will need it to bridge the mDNS multicast packets across the border. * If you are using a Mac, you win. All Apple devices support mDNS lookups. * If you are using Windows, you can install Apple's Bonjour for Windows package. (This may be already installed on your machine if you have installed iTunes) After you install this you can easily do `.local` lookups. * If you are using a Unix machine, you can install Avahi. The default hostname is set to `contiki.local.`. You can change the hostname programmatically by calling `resolv_set_hostname()`. You can change the default hostname by changing `CONTIKI_CONF_DEFAULT_HOSTNAME`. You may disable mDNS support by setting `RESOLV_CONF_SUPPORTS_MDNS` to `0`. --------------------------------- core/net/resolv: `resolv_lookup2()` -> `resolv_lookup()` Note that this patch should fix several `resolv_lookup()` bugs that already existed. There were many cases where `resolv_lookup()` was being called and the IP address ignored, but later code assumed that the IP address had been fetched... ANYWAY, those should be fixed now. --------------------------------- examples/udp-ipv6: Updated client to use MDNS to lookup the server. Also updated the Cooja regression test simulation.
2012-05-17 15:24:08 +00:00
uip_ipaddr_t *addrptr;
if(resolv_lookup(host, &addrptr) != RESOLV_STATUS_CACHED) {
resolv_query(host);
show_statustext("Resolving host...");
return;
}
core/net/resolv: IPv6 and mDNS ("Bonjour") support. Major refactor. This patch updates the DNS resolver to support IPv6 and introduces an improved API for looking up DNS entries. This patch also adds optional support for mDNS lookups and responses to the DNS resolver. Here is a quick summary of the changes: * Added support for IPv6 lookups. * DNS queries now honor record expiration. * Added support for mDNS, compatible with "Bonjour". * Implemented a new lookup api, `resolv_lookup2()`, which provides more information about the state of the record(error, expired, looking-up, etc.). About mDNS/Bonjour Support -------------------------- This patch adds basic support for mDNS/Bonjour, which allows you to refer to the name of a device instead of its IP address. This is incredibly convenient for IPv6 addresses because they tend to be very long and difficult to remember. It is especially important for link-local IPv6 addresses, since not all programs support the '%' notation for indicating a network interface (required on systems with more than one network interface to disambiguate). In other words, instead of typing in this: * `http://[fe80::58dc:d7ed:a644:628f%en1]/` You can type this instead: * `http://contiki.local/` Huge improvement, no? The convenience extends beyond that: this mechanism can be used for nodes to talk to each other based on their human-readable names instead of their IPv6 addresses. So instead of a switch on `aaaa::58dc:d7ed:a644:628f` triggering an actuator on `aaaa::ed26:19c1:4bd2:f95b`, `light-switch.local` can trigger the actuator on `living-room-lights.local`. What you need to do to be able to look up `.local` names on your workstation depends on a few factors: * Your machine needs to be able to send and receive multicast packets to and from the LoWPAN. You can do this easily with the Jackdaw firmware on an RZUSBStick. If you have a border router, you will need it to bridge the mDNS multicast packets across the border. * If you are using a Mac, you win. All Apple devices support mDNS lookups. * If you are using Windows, you can install Apple's Bonjour for Windows package. (This may be already installed on your machine if you have installed iTunes) After you install this you can easily do `.local` lookups. * If you are using a Unix machine, you can install Avahi. The default hostname is set to `contiki.local.`. You can change the hostname programmatically by calling `resolv_set_hostname()`. You can change the default hostname by changing `CONTIKI_CONF_DEFAULT_HOSTNAME`. You may disable mDNS support by setting `RESOLV_CONF_SUPPORTS_MDNS` to `0`. --------------------------------- core/net/resolv: `resolv_lookup2()` -> `resolv_lookup()` Note that this patch should fix several `resolv_lookup()` bugs that already existed. There were many cases where `resolv_lookup()` was being called and the IP address ignored, but later code assumed that the IP address had been fetched... ANYWAY, those should be fixed now. --------------------------------- examples/udp-ipv6: Updated client to use MDNS to lookup the server. Also updated the Cooja regression test simulation.
2012-05-17 15:24:08 +00:00
uip_ipaddr_copy(&addr, addrptr);
}
#else /* UIP_UDP */
uiplib_ipaddrconv(host, &addr);
#endif /* UIP_UDP */
/* The hostname we present in the hostname table, so we send out the
initial GET request. */
if(webclient_get(host, 80, file) == 0) {
show_statustext("Out of memory error");
} else {
show_statustext("Connecting...");
}
}
/*-----------------------------------------------------------------------------------*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
/* set_link(link):
*
* Will format a link from the current web pages so that it suits the
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
* open_url() function.
*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static void CC_FASTCALL
set_link(char *link)
{
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
register char *urlptr;
if(strncmp(link, http_http, 7) == 0) {
/* The link starts with http://. We just copy the contents of the
link into the url string and jump away. */
strncpy(url, link, WWW_CONF_MAX_URLLEN);
} else if(*link == ISO_slash &&
*(link + 1) == ISO_slash) {
/* The link starts with //, so we'll copy it into the url
variable, starting after the http (which already is present in
the url variable since we were able to open the web page on
which this link was found in the first place). */
strncpy(&url[5], link, WWW_CONF_MAX_URLLEN);
} else if(*link == ISO_slash) {
/* The link starts with a slash, so it is a non-relative link
within the same web site. We find the start of the filename of
the current URL and paste the contents of this link there, and
head off to the new URL. */
for(urlptr = &url[7];
*urlptr != 0 && *urlptr != ISO_slash;
++urlptr);
strncpy(urlptr, link, WWW_CONF_MAX_URLLEN - (urlptr - url));
} else {
/* A fully relative link is found. We find the last slash in the
current URL and paste the link there. */
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
/* XXX: we should really parse any ../ in the link as well. */
for(urlptr = url + strlen(url);
urlptr != url && *urlptr != ISO_slash;
--urlptr);
++urlptr;
strncpy(urlptr, link, WWW_CONF_MAX_URLLEN - (urlptr - url));
}
}
/*-----------------------------------------------------------------------------------*/
#if WWW_CONF_HISTORY_SIZE > 0
/* log_back():
*
* Copies the current URL from the url variable and into the log for
* the back button.
*/
static void
log_back(void)
{
if(strncmp(url, history[(int)history_last], WWW_CONF_MAX_URLLEN) != 0) {
memcpy(history[(int)history_last], url, WWW_CONF_MAX_URLLEN);
++history_last;
if(history_last >= WWW_CONF_HISTORY_SIZE) {
history_last = 0;
}
}
}
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
/*-----------------------------------------------------------------------------------*/
static void
quit(void)
{
ctk_window_close(&mainwindow);
process_exit(&www_process);
LOADER_UNLOAD();
}
/*-----------------------------------------------------------------------------------*/
/* www_process():
*
* The program's signal dispatcher function. Is called whenever a signal arrives.
*/
PROCESS_THREAD(www_process, ev, data)
{
static struct ctk_widget *w;
#if WWW_CONF_WITH_WGET
static char *argptr;
#endif /* WWW_CONF_WITH_WGET */
w = (struct ctk_widget *)data;
PROCESS_BEGIN();
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
/* Create the main window. */
memset(webpage, 0, sizeof(webpage));
ctk_window_new(&mainwindow, WWW_CONF_WEBPAGE_WIDTH,
WWW_CONF_WEBPAGE_HEIGHT+5, "Web browser");
make_window();
#ifdef WWW_CONF_HOMEPAGE
strncpy(editurl, WWW_CONF_HOMEPAGE, sizeof(editurl));
#endif /* WWW_CONF_HOMEPAGE */
CTK_WIDGET_FOCUS(&mainwindow, &urlentry);
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#if WWW_CONF_WITH_WGET || defined(WWW_CONF_WGET_EXEC)
#if CTK_CONF_WINDOWS
/* Create download dialog.*/
ctk_dialog_new(&wgetdialog, 38, 7);
CTK_WIDGET_ADD(&wgetdialog, &wgetlabel1);
CTK_WIDGET_ADD(&wgetdialog, &wgetlabel2);
CTK_WIDGET_ADD(&wgetdialog, &wgetnobutton);
CTK_WIDGET_ADD(&wgetdialog, &wgetyesbutton);
#endif /* CTK_CONF_WINDOWS */
#endif /* WWW_CONF_WITH_WGET || WWW_CONF_WGET_EXEC */
ctk_window_open(&mainwindow);
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
while(1) {
PROCESS_WAIT_EVENT();
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(ev == tcpip_event) {
webclient_appcall(data);
} else if(ev == ctk_signal_widget_activate) {
if(w == (struct ctk_widget *)&gobutton ||
w == (struct ctk_widget *)&urlentry) {
start_loading();
firsty = 0;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#if WWW_CONF_HISTORY_SIZE > 0
log_back();
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
memcpy(url, editurl, WWW_CONF_MAX_URLLEN);
petsciiconv_toascii(url, WWW_CONF_MAX_URLLEN);
open_url();
CTK_WIDGET_FOCUS(&mainwindow, &gobutton);
#if WWW_CONF_HISTORY_SIZE > 0
} else if(w == (struct ctk_widget *)&backbutton) {
firsty = 0;
start_loading();
--history_last;
/* Note: history_last is unsigned ! */
if(history_last > WWW_CONF_HISTORY_SIZE) {
history_last = WWW_CONF_HISTORY_SIZE - 1;
}
memcpy(url, history[(int)history_last], WWW_CONF_MAX_URLLEN);
*history[(int)history_last] = 0;
open_url();
CTK_WIDGET_FOCUS(&mainwindow, &backbutton);
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
} else if(w == (struct ctk_widget *)&downbutton) {
firsty = pagey + WWW_CONF_WEBPAGE_HEIGHT - 2;
start_loading();
open_url();
CTK_WIDGET_FOCUS(&mainwindow, &downbutton);
} else if(w == (struct ctk_widget *)&stopbutton) {
loading = 0;
webclient_close();
#if WWW_CONF_WITH_WGET || defined(WWW_CONF_WGET_EXEC)
} else if(w == (struct ctk_widget *)&wgetnobutton) {
#if CTK_CONF_WINDOWS
ctk_dialog_close();
#else /* CTK_CONF_WINDOWS */
clear_page();
#endif /* CTK_CONF_WINDOWS */
} else if(w == (struct ctk_widget *)&wgetyesbutton) {
#if CTK_CONF_WINDOWS
ctk_dialog_close();
#else /* CTK_CONF_WINDOWS */
clear_page();
#endif /* CTK_CONF_WINDOWS */
#if WWW_CONF_WITH_WGET
quit();
2007-11-18 01:24:39 +00:00
argptr = arg_alloc((char)WWW_CONF_MAX_URLLEN);
if(argptr != NULL) {
strncpy(argptr, url, WWW_CONF_MAX_URLLEN);
}
program_handler_load("wget.prg", argptr);
#else /* WWW_CONF_WITH_WGET */
petsciiconv_topetscii(url, sizeof(url));
/* Clear screen */
ctk_restore();
WWW_CONF_WGET_EXEC(url);
redraw_window();
show_statustext("Cannot exec wget");
#endif /* WWW_CONF_WITH_WGET */
#endif /* WWW_CONF_WITH_WGET || WWW_CONF_WGET_EXEC */
#if WWW_CONF_FORMS
} else {
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
/* Assume form widget. */
formsubmit((struct inputattrib *)
(((char *)w) - offsetof(struct inputattrib, widget)));
#endif /* WWW_CONF_FORMS */
}
} else if(ev == ctk_signal_hyperlink_activate) {
firsty = 0;
#if WWW_CONF_HISTORY_SIZE > 0
log_back();
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
set_link(w->widget.hyperlink.url);
show_url();
open_url();
start_loading();
CTK_WIDGET_FOCUS(&mainwindow, &stopbutton);
} else if(ev == ctk_signal_hyperlink_hover) {
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(CTK_WIDGET_TYPE((struct ctk_widget *)data) == CTK_WIDGET_HYPERLINK) {
strncpy(statustexturl, w->widget.hyperlink.url,
sizeof(statustexturl));
petsciiconv_topetscii(statustexturl, sizeof(statustexturl));
show_statustext(statustexturl);
}
#if UIP_UDP
} else if(ev == resolv_event_found) {
/* Either found a hostname, or not. */
core/net/resolv: IPv6 and mDNS ("Bonjour") support. Major refactor. This patch updates the DNS resolver to support IPv6 and introduces an improved API for looking up DNS entries. This patch also adds optional support for mDNS lookups and responses to the DNS resolver. Here is a quick summary of the changes: * Added support for IPv6 lookups. * DNS queries now honor record expiration. * Added support for mDNS, compatible with "Bonjour". * Implemented a new lookup api, `resolv_lookup2()`, which provides more information about the state of the record(error, expired, looking-up, etc.). About mDNS/Bonjour Support -------------------------- This patch adds basic support for mDNS/Bonjour, which allows you to refer to the name of a device instead of its IP address. This is incredibly convenient for IPv6 addresses because they tend to be very long and difficult to remember. It is especially important for link-local IPv6 addresses, since not all programs support the '%' notation for indicating a network interface (required on systems with more than one network interface to disambiguate). In other words, instead of typing in this: * `http://[fe80::58dc:d7ed:a644:628f%en1]/` You can type this instead: * `http://contiki.local/` Huge improvement, no? The convenience extends beyond that: this mechanism can be used for nodes to talk to each other based on their human-readable names instead of their IPv6 addresses. So instead of a switch on `aaaa::58dc:d7ed:a644:628f` triggering an actuator on `aaaa::ed26:19c1:4bd2:f95b`, `light-switch.local` can trigger the actuator on `living-room-lights.local`. What you need to do to be able to look up `.local` names on your workstation depends on a few factors: * Your machine needs to be able to send and receive multicast packets to and from the LoWPAN. You can do this easily with the Jackdaw firmware on an RZUSBStick. If you have a border router, you will need it to bridge the mDNS multicast packets across the border. * If you are using a Mac, you win. All Apple devices support mDNS lookups. * If you are using Windows, you can install Apple's Bonjour for Windows package. (This may be already installed on your machine if you have installed iTunes) After you install this you can easily do `.local` lookups. * If you are using a Unix machine, you can install Avahi. The default hostname is set to `contiki.local.`. You can change the hostname programmatically by calling `resolv_set_hostname()`. You can change the default hostname by changing `CONTIKI_CONF_DEFAULT_HOSTNAME`. You may disable mDNS support by setting `RESOLV_CONF_SUPPORTS_MDNS` to `0`. --------------------------------- core/net/resolv: `resolv_lookup2()` -> `resolv_lookup()` Note that this patch should fix several `resolv_lookup()` bugs that already existed. There were many cases where `resolv_lookup()` was being called and the IP address ignored, but later code assumed that the IP address had been fetched... ANYWAY, those should be fixed now. --------------------------------- examples/udp-ipv6: Updated client to use MDNS to lookup the server. Also updated the Cooja regression test simulation.
2012-05-17 15:24:08 +00:00
if((char *)data != NULL &&
resolv_lookup((char *)data, NULL) == RESOLV_STATUS_CACHED) {
open_url();
} else {
show_statustext("Host not found");
}
#endif /* UIP_UDP */
} else if(ev == ctk_signal_window_close ||
ev == PROCESS_EVENT_EXIT) {
quit();
}
}
PROCESS_END();
}
/*-----------------------------------------------------------------------------------*/
/* set_url():
*
* Constructs an URL from the arguments and puts it into the global
* "url" variable and the visible "editurl" (which is shown in the URL
* text entry widget in the browser window).
*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static void CC_FASTCALL
set_url(char *host, uint16_t port, char *file)
{
char *urlptr;
memset(url, 0, WWW_CONF_MAX_URLLEN);
if(strncmp(file, http_http, 7) == 0) {
strncpy(url, file, sizeof(url));
} else {
strncpy(url, http_http, 7);
urlptr = url + 7;
strcpy(urlptr, host);
urlptr += strlen(host);
strcpy(urlptr, file);
}
show_url();
}
/*-----------------------------------------------------------------------------------*/
/* webclient_aborted():
*
* Callback function. Called from the webclient when the HTTP
* connection was abruptly aborted.
*/
void
webclient_aborted(void)
{
show_statustext("Connection reset by peer");
}
/*-----------------------------------------------------------------------------------*/
/* webclient_timedout():
*
* Callback function. Called from the webclient when the HTTP
* connection timed out.
*/
void
webclient_timedout(void)
{
show_statustext("Connection timed out");
}
/*-----------------------------------------------------------------------------------*/
/* webclient_closed():
*
* Callback function. Called from the webclient when the HTTP
* connection was closed after a request from the "webclient_close()"
* function.
*/
void
webclient_closed(void)
{
show_statustext("Stopped");
petsciiconv_topetscii(webpageptr - x, x);
CTK_WIDGET_FOCUS(&mainwindow, &downbutton);
redraw_window();
}
/*-----------------------------------------------------------------------------------*/
/* webclient_connected():
*
* Callback function. Called from the webclient when the HTTP
* connection is connected.
*/
void
webclient_connected(void)
{
start_loading();
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
show_statustext("Request sent...");
set_url(webclient_hostname(), webclient_port(), webclient_filename());
htmlparser_init();
}
/*-----------------------------------------------------------------------------------*/
/* webclient_datahandler():
*
* Callback function. Called from the webclient module when HTTP data
* has arrived.
*/
void
webclient_datahandler(char *data, uint16_t len)
{
if(len > 0) {
if(strstr(webclient_mimetype(), http_html + 1) != 0) {
count = (count + 1) & 3;
show_statustext(receivingmsgs[count]);
htmlparser_parse(data, len);
redraw_window();
} else {
uip_abort();
#if WWW_CONF_WITH_WGET || defined(WWW_CONF_WGET_EXEC)
#if CTK_CONF_WINDOWS
ctk_dialog_open(&wgetdialog);
#else /* CTK_CONF_WINDOWS */
strcpy(webpage + WWW_CONF_WEBPAGE_WIDTH * 5,
(80 - WWW_CONF_WEBPAGE_WIDTH) / 2 +
" This web page cannot be displayed.");
strcpy(webpage + WWW_CONF_WEBPAGE_WIDTH * 6,
(80 - WWW_CONF_WEBPAGE_WIDTH) / 2 +
" Would you like to download instead?");
CTK_WIDGET_ADD(&mainwindow, &wgetnobutton);
CTK_WIDGET_ADD(&mainwindow, &wgetyesbutton);
redraw_window();
#endif /* CTK_CONF_WINDOWS */
#endif /* WWW_CONF_WITH_WGET || WWW_CONF_WGET_EXEC */
}
} else {
/* Clear remaining parts of page. */
loading = 0;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(data == NULL) {
loading = 0;
show_statustext("Done");
petsciiconv_topetscii(webpageptr - x, x);
CTK_WIDGET_FOCUS(&mainwindow, &urlentry);
redraw_window();
}
}
/*-----------------------------------------------------------------------------------*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static void CC_FASTCALL
add_pagewidget(char *text, unsigned char size, char *attrib, unsigned char type,
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
unsigned char border)
{
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
char *wptr;
static unsigned char maxwidth;
newlines = 0;
if(!loading) {
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
return;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
maxwidth = size ? WWW_CONF_WEBPAGE_WIDTH - (1 + 2 * border)
2013-07-30 20:48:04 +00:00
: WWW_CONF_WEBPAGE_WIDTH;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
/* If the text of the link is too long so that it does not fit into
the width of the current window, counting from the current x
coordinate, we first try to jump to the next line. */
if(size + x > maxwidth) {
htmlparser_newline();
if(!loading) {
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
return;
}
}
/* If the text of the link still is too long, we just chop it off!
XXX: this is not really the right thing to do, we should probably
either make a link into a multiline link, or add multiple
buttons. But this will do for now. */
if(size > maxwidth) {
text[maxwidth] = 0;
size = maxwidth;
}
if(firsty == pagey) {
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
unsigned char attriblen = strlen(attrib);
wptr = webpageptr;
/* To save memory, we'll copy the widget text to the web page
drawing area and reference it from there. */
wptr[0] = 0;
wptr += border;
memcpy(wptr, text, size);
wptr[size] = 0;
wptr[size + border] = ' ';
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
switch(type) {
2015-06-04 20:39:16 +00:00
case CTK_WIDGET_HYPERLINK: {
struct linkattrib *linkptr =
(struct linkattrib *)add_pageattrib(sizeof(struct linkattrib) /* incl 1 attrib char */ + attriblen);
if(linkptr != NULL) {
CTK_HYPERLINK_NEW(&linkptr->hyperlink, x, y + 3, size, wptr, linkptr->url);
strcpy(linkptr->url, attrib);
CTK_WIDGET_SET_FLAG(&linkptr->hyperlink, CTK_WIDGET_FLAG_MONOSPACE);
CTK_WIDGET_ADD(&mainwindow, &linkptr->hyperlink);
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
}
2015-06-04 20:39:16 +00:00
break;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#if WWW_CONF_FORMS
2015-06-04 20:39:16 +00:00
case CTK_WIDGET_BUTTON: {
struct submitattrib *submitptr =
(struct submitattrib *)add_pageattrib(sizeof(struct submitattrib) /* incl 1 attrib char */ + attriblen);
if(submitptr != NULL) {
CTK_BUTTON_NEW((struct ctk_button *)&submitptr->button, x, y + 3, size, wptr);
add_forminput((struct inputattrib *)submitptr);
submitptr->formptr = formptr;
strcpy(submitptr->name, attrib);
CTK_WIDGET_SET_FLAG(&submitptr->button, CTK_WIDGET_FLAG_MONOSPACE);
CTK_WIDGET_ADD(&mainwindow, &submitptr->button);
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
}
2015-06-04 20:39:16 +00:00
break;
}
case CTK_WIDGET_TEXTENTRY: {
struct textattrib *textptr =
(struct textattrib *)add_pageattrib(sizeof(struct textattrib) /* incl 1 attrib char */ + attriblen
+ (size ? WWW_CONF_MAX_INPUTVALUELEN : strlen(text)) + 1);
if(textptr != NULL) {
CTK_TEXTENTRY_NEW((struct ctk_textentry *)&textptr->textentry, x, y + 3, size, 1,
textptr->name + attriblen + 1, WWW_CONF_MAX_INPUTVALUELEN);
add_forminput((struct inputattrib *)textptr);
textptr->formptr = formptr;
strcpy(textptr->textentry.text, text);
strcpy(textptr->name, attrib);
if(size) {
CTK_WIDGET_SET_FLAG(&textptr->textentry, CTK_WIDGET_FLAG_MONOSPACE);
CTK_WIDGET_ADD(&mainwindow, &textptr->textentry);
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
}
}
2015-06-04 20:39:16 +00:00
break;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#endif /* WWW_CONF_FORMS */
}
}
/* Increase the x coordinate with the length of the link text plus
the extra space behind it and the CTK button markers. */
if(size) {
size += 1 + 2 * border;
}
x += size;
if(firsty == pagey) {
webpageptr += size;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(x == WWW_CONF_WEBPAGE_WIDTH) {
htmlparser_newline();
}
}
/*-----------------------------------------------------------------------------------*/
void
htmlparser_newline(void)
{
char *wptr;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(++newlines > 2) {
return;
}
if(pagey < firsty) {
++pagey;
x = 0;
return;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(!loading) {
return;
}
webpageptr += (WWW_CONF_WEBPAGE_WIDTH - x);
++y;
x = 0;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
wptr = webpageptr - WWW_CONF_WEBPAGE_WIDTH;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
petsciiconv_topetscii(wptr, WWW_CONF_WEBPAGE_WIDTH);
if(y == WWW_CONF_WEBPAGE_HEIGHT) {
loading = 0;
webclient_close();
}
}
/*-----------------------------------------------------------------------------------*/
void
htmlparser_word(char *word, unsigned char wordlen)
{
newlines = 0;
if(loading) {
if(wordlen + 1 > WWW_CONF_WEBPAGE_WIDTH - x) {
htmlparser_newline();
}
if(loading) {
if(pagey == firsty) {
memcpy(webpageptr, word, wordlen);
webpageptr += wordlen;
*webpageptr = ' ';
++webpageptr;
}
x += wordlen + 1;
if(x == WWW_CONF_WEBPAGE_WIDTH) {
htmlparser_newline();
}
}
}
}
/*-----------------------------------------------------------------------------------*/
void
htmlparser_link(char *text, unsigned char textlen, char *url)
{
/* No link for https or fragment-only as we would't be able to handle it anyway. */
if(url[0] == ISO_hash || strncmp(url, http_https, sizeof(http_https) - 1) == 0) {
htmlparser_word(text, textlen);
} else {
add_pagewidget(text, textlen, url, CTK_WIDGET_HYPERLINK, 0);
}
}
/*-----------------------------------------------------------------------------------*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#if WWW_CONF_FORMS
void
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
htmlparser_form(char *action)
{
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
formptr = (struct formattrib *)add_pageattrib(sizeof(struct formattrib) + strlen(action));
if(formptr != NULL) {
formptr->nextptr = NULL;
currptr = (struct inputattrib *)formptr;
strcpy(formptr->action, action);
}
}
/*-----------------------------------------------------------------------------------*/
void
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
htmlparser_submitbutton(char *text, char *name)
{
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
add_pagewidget(text, (unsigned char)strlen(text), name, CTK_WIDGET_BUTTON, 1);
}
/*-----------------------------------------------------------------------------------*/
void
htmlparser_inputfield(unsigned char type, unsigned char size, char *text, char *name)
{
if(type == HTMLPARSER_INPUTTYPE_HIDDEN) {
add_pagewidget(text, 0, name, CTK_WIDGET_TEXTENTRY, 0);
} else {
add_pagewidget(text, size, name, CTK_WIDGET_TEXTENTRY, 1);
}
}
/*-----------------------------------------------------------------------------------*/
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static void CC_FASTCALL
add_query(char delimiter, char *string)
{
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
static char *query;
unsigned char length;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
if(delimiter == ISO_questionmark) {
query = url + strlen(url);
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
length = strlen(string);
if(query - url + WWW_CONF_MAX_URLLEN - 1 /* delimiter */ < length) {
return;
}
*query++ = delimiter;
strcpy(query, string);
if(delimiter == ISO_eq) {
char *space = query;
petsciiconv_toascii(query, length);
while((space = strchr(space, ISO_space)) != NULL) {
*space = ISO_plus;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
}
query += length;
}
/*-----------------------------------------------------------------------------------*/
static void CC_FASTCALL
formsubmit(struct inputattrib *trigger)
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
{
struct inputattrib *input;
struct formattrib *form = trigger->formptr;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
char delimiter = ISO_questionmark;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
set_link(form->action);
/* No button pressed so prepare to look for default button. */
if(trigger->widget.type == CTK_WIDGET_TEXTENTRY) {
trigger = NULL;
}
for(input = form->nextptr; input != NULL; input = input->nextptr) {
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
char *name;
char *value;
if(input->widget.type == CTK_WIDGET_TEXTENTRY) {
name = ((struct textattrib *)input)->name;
value = ((struct textattrib *)input)->textentry.text;
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
} else {
/* Consider first button as default button. */
if(trigger == NULL) {
trigger = input;
}
if(input != trigger) {
continue;
}
name = ((struct submitattrib *)input)->name;
value = ((struct submitattrib *)input)->button.text;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
add_query(delimiter, name);
add_query(ISO_eq, value);
delimiter = ISO_ampersand;
}
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
#if WWW_CONF_HISTORY_SIZE > 0
log_back();
#endif /* WWW_CONF_HISTORY_SIZE > 0 */
Reorganized web page attribute data handling. - Up to now the web browser used several fixed size arrays to hold the various types attribute data of the web page. This turned out to be way to inflexible for any non-trivial web page. Therefore now all attribute data is stored in a single buffer one after the other as they arrive from the parser only occupying the memory actually needed. This allows for pages with many links with rather short URLs as well as pages with few link with long URLs as well as pages with several simple forms as well as pages with one form with many form inputs. - Using the actual web page buffer to hold the text buffers of text entry fields was in general a cool idea but in reality it is often necessary to enter text longer than the size of the text entry field. Therefore the text buffer is now stored in the new unified attribute data buffer. - Splitting up the process of canonicalizing a link URL and actually navigating to the resulting URL allowed to get rid of the 'tmpurl' buffer used during form submit. Now the form action is canonicalized like a usual link, then the form input name/value pairs are written right into the 'url' buffer and afterwards the navigation is triggered. - Support for the 'render states' was completely removed. The only actually supported render state was centered output. The new unified attribute buffer would have complicated enumerating all widgets added to the page in order to adjust their position. Therefore I decided to drop the whole feature as the <center> tag is barely used anymore and newer center attributes are to hard to parse.
2013-03-06 15:29:36 +00:00
show_url();
open_url();
start_loading();
}
#endif /* WWW_CONF_FORMS */
/*-----------------------------------------------------------------------------------*/