/* Copyright 1997 Acorn Computers Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/***************************************************/
/* File : Fetch.c */
/* */
/* Purpose: Mid-level fetch functions, concerned */
/* mostly with HStreams but not low level */
/* HTMLLib interfacing. Compare with */
/* FetchPage.c, which provides a much */
/* higher level interface. */
/* */
/* Author : A.D.Hodgkinson */
/* */
/* History: 25-Nov-96: Created. */
/* 17-Aug-97: Split up to form the */
/* URLveneer.c and FetchHTML.c */
/* sources. */
/***************************************************/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include "swis.h"
#include "flex.h"
#include "HTMLLib.h" /* HTML library API, Which will include html2_ext.h, tags.h and struct.h */
#include "wimp.h"
#include "wimplib.h"
#include "event.h"
#include "svcprint.h"
#include "Global.h"
#include "MiscDefs.h"
#include "Utils.h"
#include "Authorise.h"
#include "Browser.h"
#include "Filetypes.h"
#include "FontManage.h"
#include "Forms.h"
#include "FetchHTML.h"
#include "Frames.h"
#include "History.h"
#include "Images.h"
#include "Memory.h"
#include "Meta.h"
#include "Redraw.h"
#include "Reformat.h"
#include "SaveFile.h"
#include "SaveObject.h"
#include "Toolbars.h"
#include "URLutils.h"
#include "URLveneer.h"
#include "Windows.h"
#include "Fetch.h" /* (Which itself includes URLstat.h) */
/* Local definitons */
#define AuthorisationStr "Authorization: Basic "
/* Local compilation options */
#define FRAMES_SUPPORT
/* Static function prototypes */
static HStream * fetch_find_anchor_token_r (browser_data * b, HStream * streambase, char * anchor);
/*************************************************/
/* fetch_start() */
/* */
/* Initiate a fetch for some URL. */
/* */
/* Parameters: Pointer to a browser_data struct */
/* for the browser window that the */
/* fetch relates to; the URL is */
/* pointed to in that structure, as */
/* the last item in the history. */
/* */
/* Returns: A pointer to a _kernel_oserror */
/* structure if an error occured, or */
/* NULL if there was no error. */
/*************************************************/
_kernel_oserror * fetch_start(browser_data * b)
{
int handle, method;
_kernel_oserror * e;
#ifdef TRACE
if (tl & (1u<<6)) Printf("\nfetch_start: Called\n");
#endif
/* (Order of evaluation ensures the check for the contents of */
/* the memory pointed to by browser_fetch_url only occurs if */
/* the pointer isn't null) */
if (!browser_fetch_url(b) || !*browser_fetch_url(b))
{
b->fetch_status = BS_IDLE;
toolbars_cancel_status(b, Toolbars_Status_Fetching);
return NULL;
}
/* URL method is set to POST if there is forms data, or GET */
/* if not (see Fetch.c for the definitions) */
method = b->extradata ? URL_Method_http_POST : URL_Method_http_GET;
/* Reset the encoding priority to default, awaiting any */
/* information in the HTTP header. Don't actually change */
/* the encoding yet as we don't want the menu to change */
/* at this point. */
if (
b->encoding_priority > priority_link &&
b->encoding_priority < priority_user
)
b->encoding_priority = priority_default;
/* Find out if this is an internal URL, and if so, */
/* set the 'displayed' field in the browser_data */
/* struct appropriately. */
/* */
/* If we're saving out data in a link, then don't */
/* do this, as the page contents aren't actually */
/* changing (so leave the flag alone). */
if (!b->save_link) urlutils_set_displayed(b, b->urlfdata);
/* Reset the data size counter */
b->data_size = 0;
/* Get, and start parsing the document */
e = html_get(b->urlfdata, /* Required document */
b->extradata, /* Extra bits to append for POST etc */
&handle, /* The library's handle for request */
method, /* See above - POST or GET at this point */
NULL, /* User name for Mailserv */
1, /* Allow HTML parsing, 1 = yes, 0 = no */
!b->reloading); /* If 0, don't go through a proxy - e.g. for a reload */
#ifdef TRACE
if (b->extradata)
{
flexcount -= flex_size((flex_ptr) &b->extradata);
if (tl & (1u<<13)) Printf("** flexcount: %d\n",flexcount);
}
#endif
if (b->extradata) flex_free((flex_ptr) &b->extradata);
if (e)
{
b->fetch_status = BS_IDLE;
toolbars_cancel_status(b, Toolbars_Status_Fetching);
return e;
}
/* No error, so signal that the fetch has started. */
b->fetch_handle = handle;
b->fetch_status = BS_STARTED;
toolbars_update_status(b, Toolbars_Status_Connecting);
/* At this point e will always be NULL but that might change, */
/* so the full trace code is being left in for now */
#ifdef TRACE
if (tl & (1u<<6))
{
if (e) Printf("fetch_start: Exiting with error\n");
else Printf("fetch_start: Successful\n");
}
#endif
return e;
}
/*************************************************/
/* fetch_fetching() */
/* */
/* Returns 1 if there is a fetch in progress */
/* according to the contents of the data that */
/* was pointed to (see Parameters), else 0. */
/* */
/* Parameters: Pointer to a browser_data struct */
/* relevant to the inquiry. */
/* */
/* Returns: 1 if a fetch is in progress, or */
/* 0 if a fetch is not in progress. */
/*************************************************/
int fetch_fetching(browser_data * b)
{
/* This is currently very simple - a fetch is considered to be in */
/* progress so long as the fetch_status doesn't indicate BS_IDLE. */
return (b->fetch_status != BS_IDLE);
}
/*************************************************/
/* fetch_find_name_tag() */
/* */
/* Finds the # separating an anchor name in a */
/* URL. */
/* */
/* Parameters: A pointer to the URL string. */
/* */
/* Returns: A pointer to the anchor string, */
/* including the leading # */
/*************************************************/
char * fetch_find_name_tag(char * url)
{
char * p;
p = strchr(url,'/'); /* Get past the first /, as in http:/ */
if (p) p = strchr(p + 1, '/'); /* Get past second /, as in http:// */
if (p) p = strchr(p + 1, '/'); /* Get past site specifier, as in http://www.this.that/ */
if (p) p = strchr(p + 1, '#'); /* Find # in the document path */
return p;
}
/*************************************************/
/* fetch_find_anchor_token() */
/* */
/* Returns the address of the first token in the */
/* token list which has the given anchor name */
/* associated with it, or NULL if none can be */
/* found. */
/* */
/* Parameters: Pointer to a browser_data struct */
/* relevant to the token list; */
/* */
/* Pointer to the anchor name. */
/* */
/* Returns: Pointer to the token associated */
/* with the given anchor name, or */
/* NULL if none is found. */
/*************************************************/
HStream * fetch_find_anchor_token(browser_data * b, char * anchor)
{
return fetch_find_anchor_token_r(b, b->stream, anchor);
}
/*************************************************/
/* fetch_find_anchor_token_r() */
/* */
/* Recursive back-end to fetch_find_anchor_token */
/* - takes an extra parameter giving the top of */
/* the HStream list to scan. */
/* */
/* Parameters: Pointer to a browser_data struct */
/* relevant to the token list; */
/* */
/* Pointer to first item in HStream */
/* list to scan; */
/* */
/* Pointer to the anchor name. */
/* */
/* Returns: As fetch_find_anchor_token. */
/*************************************************/
static HStream * fetch_find_anchor_token_r(browser_data * b, HStream * streambase, char * anchor)
{
HStream * tp;
tp = streambase;
/* Go down the token list, checking if a token represents an */
/* anchor, has a name, and that name matches the given one. If */
/* so, return the token address, else go onto the next token. */
while (tp && (tp->flags & HFlags_DealtWithToken))
{
/* A table token? */
if (
tp->tag == TABLE &&
ISBODY(tp)
)
{
table_stream * table = (table_stream *) tp;
table_row * row = NULL;
table_headdata * head = NULL;
HStream * tf;
int cellcount = 0;
int cellmax = table->ColSpan * table->RowSpan;
/* Scan the table for the token, using a recursive */
/* call to this function for each cell. */
if (table->cells)
{
row = table->List;
while (row && cellcount < cellmax)
{
head = row->List;
while (head && cellcount < cellmax)
{
switch (head->Tag)
{
case TagTableData:
case TagTableHead:
{
tf = fetch_find_anchor_token_r(b, (HStream *) head->List, anchor);
if (tf) return tf;
}
break;
}
cellcount ++;
head = head->Next;
/* Closure of 'while (head && ...)' */
}
row = row->Next;
/* Closure of 'while (row && ...)' */
}
/* Closure of 'if (table->cells)' */
}
/* Closure of check to see if token represents a table */
}
else if (
(tp->style & A) &&
tp->name &&
!strcmp(tp->name, anchor)
)
return tp;
tp = tp->next;
}
/* No match found - return NULL. */
return NULL;
}
/*************************************************/
/* fetch_preprocess_token() */
/* */
/* Takes a token for a given browser_data struct */
/* and preprocesses it - e.g. tells the image */
/* library about image tokens so fetches can */
/* start for those images. */
/* */
/* Parameters: A pointer to a browser_data */
/* structure relevant to the token; */
/* */
/* Pointer to the token. */
/*************************************************/
void fetch_preprocess_token(browser_data * b, HStream * tptr)
{
int reprocess_table = 0;
/* final_token keeps track of the last token dealt with by this */
/* routine, in the main token stream. */
if (!tptr->parent) b->final_token = tptr;
/* Are we reprocessing the contents of a table tag which has been */
/* dealt with before? */
if (tptr->tag == TABLE && ISBODY(tptr)) reprocess_table = 1;
/* Deal with smart quotes etc. */
// This could have side effects placed here...! Sort it out! - the function just returns at present.
reformat_change_text(b, tptr);
/* Deal with document body tags (not within HEAD, FRAMESET etc. containers) */
if (ISBODY(tptr))
{
/* Don't reprocess the token generally - this may accidentally */
/* clear form_flag, say, if the high level table structures */
/* we are passing through on the way to the lower level ones */
/* don't have the FORM bit set. Other such consequences are */
/* avoided by only doing the table handling if this is a table */
/* tag with the HFlags_DealtWithToken bit set in its flags. */
if (!reprocess_table)
{
/* If the 'style' entry has the image (IMG) bit set, ask */
/* the image library to handle a new image. tptr->src */
/* will be a char * to the URL of the image. */
if (tptr->style & IMG)
{
if (fetch_chkerror(b,
image_new_image(b,
tptr->src,
tptr,
0))) return;
}
/* Handle some form tags */
if (tptr->style & FORM)
{
/* If there are no forms in this fetch so far, create a new one */
if (!b->form_flag)
{
if (fetch_chkerror(b, form_new_form(b, tptr))) return;
}
/* Deal with creating a new field as appropriate */
if (tptr->tagno == TAG_INPUT)
{
switch(HtmlINPUTtype(tptr))
{
case inputtype_TEXT: if (fetch_chkerror(b, form_new_field(b, tptr, form_text, tptr->text ))) return; break;
case inputtype_PASSWORD:if (fetch_chkerror(b, form_new_field(b, tptr, form_password, tptr->text ))) return; break;
case inputtype_CHECKBOX:if (fetch_chkerror(b, form_new_field(b, tptr, form_checkbox, (char *) HtmlINPUTchecked(tptr)))) return; break;
case inputtype_RADIO: if (fetch_chkerror(b, form_new_field(b, tptr, form_radio, (char *) HtmlINPUTchecked(tptr)))) return; break;
case inputtype_IMAGE: if (fetch_chkerror(b, form_new_field(b, tptr, form_image, NULL ))) return; break;
case inputtype_HIDDEN: if (fetch_chkerror(b, form_new_field(b, tptr, form_hidden, NULL ))) return; break;
case inputtype_SUBMIT: if (fetch_chkerror(b, form_new_field(b, tptr, form_submit, NULL ))) return; break;
case inputtype_RESET: if (fetch_chkerror(b, form_new_field(b, tptr, form_reset, NULL ))) return; break;
}
}
/* Handle text areas */
if (tptr->tagno == TAG_TEXTAREA)
{
if (fetch_chkerror(b, form_new_field(b, tptr, form_textarea, tptr->text))) return;
}
/* Handle selection buttons */
if (tptr->tagno == TAG_SELECT)
{
if (fetch_chkerror(b, form_new_field(b, tptr, form_select, (char *) HtmlSELECToptions(tptr)))) return;
}
b->form_flag = 1;
}
else b->form_flag = 0;
}
/* Tables - need to preprocess any HStreams attached as part of a table */
/* tag. Because any one token is only run through this preprocessor */
/* once, and because when this table tag is run through it all of the */
/* HStreams within may not have arrived yet (the page is only partially */
/* fetched), it is still necessary to rescan the attached HStreams at a */
/* later date (e.g. as part of the reformatting process) to ensure they */
/* are all preprocessed correctly. */
if (tptr->tag == TABLE)
{
table_stream * table = (table_stream *) tptr;
table_row * R;
table_headdata * D;
HStream * attached;
R = table->List;
/* Scan the rows and cells */
while (R)
{
D = R->List;
while (D)
{
if (D->Tag)
{
switch (D->Tag)
{
case TagTableData:
case TagTableHead:
{
attached = (HStream *) D->List;
/* Preprocess any attached HStream list - must */
/* check table tags even if they've been done */
/* before to look for new HStreams, otherwise */
/* avoid preprocessing the same thing twice. */
while (attached)
{
if (
(
ISBODY(attached) &&
attached->tag == TABLE
)
||
(
!(attached->flags & HFlags_DealtWithToken)
)
)
fetch_preprocess_token(b, attached);
attached = attached->next;
}
}
break;
}
}
D = D->Next;
}
R = R->Next;
}
}
/* Closure of long 'if' to see if the HStream structure represented */
/* a body tag or header information - the code is run if it's a */
/* body tag. */
}
#ifdef FRAMES_SUPPORT
else if ISFRAMESET(tptr)
{
browser_data * parent;
browser_data * child = b;
parent = b->parent;
if (!parent) parent = b;
if (tptr->size)
{
int level = tptr->size;
/* If filling_frame is equal to the number of children, then */
/* they've all been filled - the frameset must be broken. */
if (
!child->nchildren ||
(
child->nchildren &&
child->filling_frame < child->nchildren
)
)
{
while (level > 1)
{
/* If in a nested frameset, find out what browser_data struct */
/* to put the frames in. This should go in the next frame that */
/* is to be filled in according to the parent. */
child = (browser_data *) child->children[child->filling_frame];
level--;
}
/* If stepping down a level, i.e. after a /frameset tag, */
/* will want to increment the filling_frame field for */
/* this browser to say that the child we just stepped */
/* down for has been filled with a frameset. There's the */
/* complication of a /frameset being followed by another */
/* frameset and the level therefore staying the same; */
/* this is dealt with in the frameset section below. */
if (tptr->size < parent->nesting_level)
{
child->filling_frame++;
}
/* The aforementioned frameset section... */
if (!(tptr->style & FRAME))
{
/* Define a new frameset. */
if (tptr->size == parent->nesting_level && child->parent)
{
/* If at the same level as before on receiving a frameset */
/* tag, must be doing nested frames and just had a */
/* /frameset before this tag came along. So need to */
/* increment the filling_frame counter of the *parent* */
/* (remember, we're at the level of the frame to fill in, */
/* not in the level below as with the code above that */
/* checked the level had stepped down). Therefore, need */
/* to find out again what browser_data struct is to be */
/* given the frameset based on the new filled_frame value. */
child->parent->filling_frame++;
child = (browser_data *) child->parent->children[child->parent->filling_frame];
}
/* Must force scrollbars off in this current view, */
/* as a frameset is about to appear over it. */
windows_check_tools(child, NULL);
/* Finally, define the frameset at the required depth. */
frames_define_frameset(child, tptr);
}
else
{
/* Fill in details of a frame. */
frames_define_frame(child, tptr);
}
parent->nesting_level = tptr->size;
}
#ifdef STRICT_PARSER
else
{
erb.errnum = Utils_Error_Custom_Message;
StrNCpy0(erb.errmess,
lookup_token("FramNest:Frames definition is badly nested; could not complete the frames layout.",
0,0));
show_error_ret(&erb);
}
#endif
}
}
#endif
else if ISHEAD(tptr)
{
/* Deal with header (HEAD) tags */
if (tptr->tagno == TAG_TITLE && tptr->text)
{
/* The tag is TITLE, and there is title text. */
char title[Limits_Title];
char * p = title;
char * end;
/* Can't overflow maximum length so just crop the string to fit */
StrNCpy0(title, tptr->text);
/* Strip any spaces at the start */
while (*p == ' ') p++;
/* Strip any spaces at the end */
end = (char *) ((int) p + strlen(p) - 1);
if (end > p) while (*end == ' ') *end-- = 0;
/* If there's anything left now... */
if (*p != 0)
{
/* Set the title */
if (!b->ancestor && fetch_chkerror(b, window_set_title(0, b->self_id, p))) return;
/* Try adding this title to the history, ignoring any errors */
history_add_title(p, browser_fetch_url(b));
}
}
if (tptr->tag == BODY)
{
/* The BODY tag. All sorts of exciting stuff in here... */
if (HtmlBODYbackground(tptr))
{
/* If there's a URL for the image, ask the image library for it */
/* and remember the image number in the browser_data structure */
image_new_image(b, HtmlBODYbackground(tptr), tptr, 2);
}
/* Get the 24-bit background colour, if any. */
if (HtmlBODYbgcolour(tptr) != NULL_COLOUR)
{
b->background_colour = HtmlBODYbgcolour(tptr);
#ifdef TRACE
if (tl & (1u<<6)) Printf("fetch_preprocess_token: Background colour set to %d\n", b->background_colour);
#endif
/* If there's no actual background image, set the anti-alias */
/* colour to be the same as the background colour. */
if (b->background_image < 0) b->antialias_colour = b->background_colour;
browser_update_bottom(b, 0);
}
/* Get the rest of the colour info out. */
if (HtmlBODYtext (tptr) != NULL_COLOUR) b->text_colour = HtmlBODYtext (tptr);
if (HtmlBODYlink (tptr) != NULL_COLOUR) b->link_colour = HtmlBODYlink (tptr);
if (HtmlBODYvlink(tptr) != NULL_COLOUR) b->used_colour = HtmlBODYvlink(tptr);
if (HtmlBODYalink(tptr) != NULL_COLOUR) b->followed_colour = HtmlBODYalink(tptr);
/* Also pull out the onload and onunload scripts. */
if (HtmlBODYonload (tptr)) b->onload = HtmlBODYonload (tptr);
if (HtmlBODYonunload(tptr)) b->onunload = HtmlBODYonunload(tptr);
}
/* Deal with META... tags */
if (tptr->tag == META)
{
meta_process_tag(b, tptr);
}
/* Closure of long else to see if the HStream structure represented */
/* a body tag or header information - the code is run if it's a */
/* head tag. */
}
/* If we've reached here, the token has been dealt with */
/* successfully - so mark this in its flags word. */
tptr->flags |= HFlags_DealtWithToken;
return;
}
/*************************************************/
/* fetch_fetcher() */
/* */
/* The main part of the fetch routine. Handles */
/* the processing of data from the URL module, */
/* after fetch_start has asked it to start */
/* getting data from a server. */
/* */
/* Parameters: A pointer to a browser_data */
/* structure, to which the fetch */
/* relates. */
/*************************************************/
void fetch_fetcher(browser_data * b)
{
HStream * tptr;
int start = -1;
int i, remain, sofar, waiting;
/* It really helps to understand this function if you realise that */
/* it's coded more or less backwards (Merlyn was weird...). For */
/* example, the BS_DATAFETCH code below isn't executed unless the */
/* fetch state reaches BS_DATAFETCH, but it can't do that unless */
/* fetcher code much further down is run. So you really need to */
/* read this all in one go before trying to piece it together, or */
/* maybe even read it from the bottom up... */
/* */
/* BS stands for Browser Status, by the way, and nothing else ;-) */
for (i = 0; i < 10; i ++) /* Get several tokens on each null event */
{
/* For BS_DATAFETCH, save the file */
if (b->fetch_status == BS_DATAFETCH)
{
#ifdef TRACE
if (tl & (1u<<6)) Printf("fetch_fetcher: fetch_status = BS_DATAFETCH / BS_DATAWHERE.\n");
#endif
/* This code gets called by the stuff further down advancing */
/* the status to BS_DATAFETCH. */
/* */
/* If the save_file field is NULL, we're still waiting for */
/* the user to pull their finger out and say where to save */
/* the object. */
if (b->save_file) /* Proceed if there's a file to save to */
{
char buffer[2048];
int success, bytes = -1, done = 0;
_kernel_oserror * e;
/* Get a chunk of data */
e = fetch_get_raw_data(NULL,
b->fetch_handle,
buffer,
sizeof(buffer),
&done,
&bytes);
/* If there's an error, show it but continue */
if (e) show_error_ret(e);
success = !e;
/* If there was not an error, write a chunk of file */
if (success && bytes) success = fwrite(buffer, 1, bytes, b->save_file);
/* If the expected number of bytes was not written, */
/* show whatever error fwrite generated */
if (success != bytes && bytes)
{
success = 0;
erb = *_kernel_last_oserror();
show_error_ret(&erb);
}
/* If apparently successful and finished, read the pathname */
/* of the file so the filetype can be set. */
if (success && done)
{
_swix(OS_Args,
_INR(0,2) | _IN(5),
7, /* Read pathname of open file */
b->save_file->__file,
buffer,
sizeof(buffer));
}
/* If finished or there was some error above, stop the fetch */
if (!success || done) fetch_stop(b, 0); /* This closes the output file, too */
/* If successful and finished, set the filetype */
if (success && done)
{
_swix(OS_File,
_INR(0,2),
18,
buffer,
b->save_type);
}
/* Finally, ensure toolbars are up to date. */
toolbars_update_progress(b);
}
return;
}
#ifdef TRACE
if (tl & (1u<<6)) Printf("fetch_fetcher: Get next token\n");
#endif
/* Get the next token, with fetch_chkerror allowing us to exit */
/* relatively cleanly should an error occur. */
if (
fetch_chkerror(
b, html_get_next_token(
b,
b->fetch_handle,
&remain,
&sofar,
&tptr,
&waiting,
(flex_ptr) &b->source,
browser_fetch_url(b),
0
)
)
)
return;
/* Show the fetch's progress */
toolbars_update_progress(b);
/* If waiting = 3 the data being fetched isn't parseable, or has */
/* been marked as not for parsing so that it may be saved. */
if (waiting == 3)
{
/* Full screen browsers can't save objects out */
if (choices.full_screen && b->full_screen)
{
erb.errnum = Utils_Error_Custom_Message;
StrNCpy0(erb.errmess,
lookup_token("NotInternal:Can't save objects when running in full screen mode.",
0,
0));
fetch_chkerror(b, &erb);
return;
}
else
{
#ifdef TRACE
if (tl & (1u<<6)) Printf("fetch_fetcher: fetch_status moved to BS_DATAFETCH\n");
#endif
b->fetch_status = BS_DATAFETCH;
b->save_type = remain;
b->save_link = 1;
if (b->save_type == FileType_DATA || b->save_type == 0x000)
{
/* If we've been given data or unknown, see if we can have a better guess! */
b->save_type = urlutils_filetype_from_url(browser_fetch_url(b));
}
/* Open a save dialogue for the object */
if (fetch_chkerror(b, saveobject_open_for(b))) return;
}
return;
}
/* If waiting = 2, a redirect has occurred */
else if (waiting == 2)
{
char * url;
int internal = 0;
#ifdef TRACE
if (tl & (1u<<6)) Printf("fetch_fetcher: Redirect to %s\n",(char *) remain);
#endif
/* Get the new URL pointed to by 'url' */
url = (char *) remain;
if (b->displayed != Display_Fetched_Page) internal = 1;
/* Record the pre-redirection URL in the global history */
if (!internal)
{
history_record_global(browser_fetch_url(b));
/* Allocate space for new URL and copy it into that space */
#ifdef TRACE
if (tl & (1u<<12)) Printf("fetch_fetcher: Chunk CK_FURL set to %d\n",strlen(url) + 1);
#endif
if (fetch_chkerror(b, memory_set_chunk_size(b, NULL, CK_FURL, strlen(url) + 1))) return;
strcpy(b->urlfdata, url);
}
else
{
char furl[Limits_URL];
/* Allocate space for new URL plus old URL and separator, */
/* and copy them into that space */
#ifdef TRACE
if (tl & (1u<<12)) Printf("fetch_fetcher: Chunk CK_FURL set to %d\n",strlen(url) + strlen(furl) + 2);
#endif
StrNCpy0(furl, browser_fetch_url(b));
if (fetch_chkerror(b, memory_set_chunk_size(b, NULL, CK_FURL, strlen(url) + strlen(furl) + 2))) return;
strcpy(b->urlfdata, url);
strcat(b->urlfdata, ":");
strcat(b->urlfdata, furl);
}
/* Reflect the new URL in the status and URL bars */
toolbars_update_status(b, Toolbars_Status_Redirected);
toolbars_update_url(b);
}
/* If we're waiting for new tokens, don't sit here in a loop */
/* single tasking - break out, allowing more external polling. */
else if (waiting)
{
break;
}
/* We're not waiting for data. */
else
{
/* If it isn't already non-zero, set 'start' to the number of */
/* the last line in the line list (i.e. nlines - 1). */
if (start < 0) start = b->cell->nlines - 1;
/* We're not waiting, have we got a token? */
if (b->fetch_status == BS_STARTED)
{
/* Yes - this is the first token on this page. Get the window */
/* ready for the new page - this includes ditching old data. */
int l;
/* Make the current display URL = current fetch URL... */
l = strlen(browser_fetch_url(b)); /* Get the fetching URL string length */
/* Allocate memory for it, and copy the string across */
#ifdef TRACE
if (tl & (1u<<12)) Printf("fetch_fetcher: Chunk CK_DURL set to %d\n",l + 1);
#endif
if (fetch_chkerror(b, memory_set_chunk_size(b, NULL, CK_DURL, l + 1))) return;
strcpy(b->urlddata, browser_fetch_url(b));
toolbars_hide_internal(b->urlddata);
/* Write to the global history */
history_record_global(b->urlddata);
/* Update the title bar */
if (!b->ancestor) /* Child windows don't have title bars... */
{
char title[Limits_Title];
StrNCpy0(title, b->urlddata);
if (fetch_chkerror(b, window_set_title(0, b->self_id, title))) return;
}
/* Set status to FETCHING instead of STARTED */
b->fetch_status = BS_FETCHING;
/* If there was previous display data present, get rid of it */
if ((b->display_handle) && (b->display_handle != b->fetch_handle))
{
html_close(b->display_handle);
b->display_handle = 0;
}
b->save_oldstore = 0;
/* Signal that the display data is coming from the fetch data, */
/* so that the fetch data doesn't get accidentally ditched */
/* until it's finished with */
b->display_handle = b->fetch_handle;
/* Initialise various things inside the browser_data structure */
/* to do with colours and so-forth */
#ifdef TRACE
if (tl & (1u<<18)) Printf("New fetch for %p, stream %p\n",b,tptr);
#endif
b->stream = tptr; /* Pointer to list of HStream structures */
b->final_token = NULL; /* Last HStream structure dealt with */
b->last_char = ' '; /* Last character dealt with */
b->background_colour = -1; /* Background colour, or -1 for default */
b->background_image = -1; /* Image no. of background image, 0=none */
b->text_colour = choices.text_colour; /* Body text default colour */
b->link_colour = choices.link_colour; /* Link text default colour */
b->used_colour = choices.used_colour; /* Followed link default colour */
b->antialias_colour = redraw_backcol(b); /* Colour to anti-alias to, or -1=none */
b->followed_colour = choices.followed_colour; /* Following link default colour */
b->selected_colour = choices.selected_colour; /* Selected (highlighted) link colour */
b->onload = NULL; /* <BODY onload> attribute */
b->onunload = NULL; /* <BODY onunload> attribute */
/* Ensure the nesting level and filling frame counters are reset */
b->nesting_level = 0;
b->filling_frame = 0;
/* Cancel any pending automatic fetches */
if (b->meta_refresh_at) deregister_null_claimant(Wimp_ENull, (WimpEventHandler *) meta_check_refresh, b);
b->meta_refresh_at = 0;
b->meta_refresh_url = NULL;
/* Cancel pending reformats */
reformat_stop_pending(b);
/* Hideously long comment alert... */
/* */
/* Although if a frame loads a document containing another frameset */
/* this is in one sense a nested frame defintion, in another the */
/* second document is independent of the first; certainly as far as */
/* incrementing the filling_frame field of the parent goes, the */
/* <frame> tag that loaded this document into the frame in the */
/* first place will already have done that. */
/* */
/* Consequently, whilst all child frames have an ancestor - the */
/* original, base browser that defined the first of possibly many */
/* framesets - only genuinely nested frameset arrays have parents. */
/* That is, a parent can only have children; it may not also be a */
/* child (i.e. have a parent), it may only have an ancestor. */
/* */
/* Genuinely nested frames consist of one document with more than */
/* one set of <frameset> tags. Here, filling_frame considerations */
/* demand the use of a nested_level count and a parent as well as */
/* an ancestor. For those single documents, we won't be running */
/* this code when second or further framesets come in, so the */
/* parent field will get estabilshed and remain as long as needed */
/* by the frames routines. */
b->parent = NULL;
/* Yup - that whole comment for one tiny line of code. Woo... */
/* */
/* 'Course, that said, it's useful for every child to know who its */
/* parent is. That's what the real_parent field is for. */
// {
// WimpGetWindowStateBlock state;
//
// state.window_handle = b->window_handle;
//
// if (!wimp_get_window_state(&state))
// {
// b->display_width = b->display_extent = state.visible_area.xmax - state.visible_area.xmin;
// }
// }
/* Don't want to set the pointer_over field, as then it may not seem to */
/* have changed from one fetch to another; the pointer can get 'stuck' */
/* in the 'link' shape. */
b->highlight = NULL; /* No tokens are highlighted */
b->selected = NULL; /* No tokens are selected */
b->selected_owner = NULL;
#ifdef TRACE
if (tl & (1u<<6)) Printf("\nfetch_fetcher: Document colours etc. set to default values\n");
#endif
/* Clear the status bar contents block for an ancestor */
/* window beginning a new fetch. */
if (!b->ancestor && b->nstatus)
{
#ifdef TRACE
if (
(tl & (1u<<1)) ||
(tl & (1u<<6))
)
Printf("fetch_fetcher: Freeing status_contents array\n");
#endif
b->nstatus = 0;
memory_set_chunk_size(b, NULL, CK_STAT, 0);
}
/* Clear allocated memory for the forms, and tell */
/* the font library that the fonts aren't needed */
// /* anymore. Images are cleared after the fetch, */
// /* so that any images common between the two can */
// /* be preserved. */
form_discard(b);
fm_lose_fonts(b);
/* Flag that images need to be garbage collected later */
// Um... ToDo list time...
image_discard(b);
// b->clear_images = 1;
/* IMPORTANT, must call the reformatter here to ensure that all various */
/* line list data is invalidated, discarded, and any new stuff is valid. */
/* Otherwise, could have bits of the application subsequently using old */
/* line data and things will go very wrong very quickly. */
/* */
/* DON'T put anything that might try and read line data before this call! */
b->display_extent = b->display_width; /* Ensure a new fetch starts with the horizontal extent matching the visible area */
start = -1;
reformat_format_from(b, -1, 1, -1);
reformat_check_extent(b);
/* If there's a save dialogue open for this frame, get rid of it */
savefile_close(b->self_id, 0);
/* Collapse any frames within this browser */
frames_collapse_set(b);
/* Ensure window tools are up to date */
if (b->ancestor || b->full_screen) windows_set_tools(b, NULL, !b->ancestor, 0, 0, 0);
/* If there's a # inside the URL (i.e. we're supposed to jump to an */
/* anchor) then set the token to display first to be DISPLAY_NAMED, */
/* a large number which acts as a flag to say 'jump to anchor'. The */
/* fetch polling routine (see FetchPage.c) should notice this and */
/* start looking for a token with the appropriate name, and if it */
/* finds it, display that token. */
if (fetch_find_name_tag(browser_current_url(b))) b->display_request = DISPLAY_NAMED;
/* Ensure the pointer shape is correct */
browser_pointer_check(0, NULL, NULL, b);
/* Reflect the new browser status */
toolbars_update_status(b, Toolbars_Status_Fetching);
/* Since the new fetch is now official, update the current and previous */
/* page variables */
if (!b->ancestor)
{
/* Not speed critical, so avoid lots of nasty C-isms with malloc */
/* and so-on, by running through OS_CLI. */
_swix(OS_CLI,
_IN(0),
"Set Browse$PreviousPage <Browse$CurrentPage>");
_swix(OS_SetVarVal,
_INR(0,4),
"Browse$CurrentPage",
b->urlfdata,
strlen(b->urlfdata),
0,
4);
}
/* (Initialisation to an empty state is now complete, so we're */
/* ready to fetch a new page). */
}
/* We're not waiting, but if there's also no data left to fetch, */
/* then we're just chugging through the list of tokens that the */
/* library has generated, telling various bits of the code about */
/* their contents (e.g. a new image, a new form). In this case, */
/* change the fetch status so the status bar can reflect the new */
/* situation. */
if (!remain)
{
b->fetch_status = BS_PROCESS;
toolbars_update_status(b, Toolbars_Status_Processing);
toolbars_update_progress(b);
}
/* If tptr is null, there are no HStream structures (see the */
/* html_get_next_token call). But we're not waiting either, */
/* so must be at the end of the file - stop the fetch. */
if (!tptr)
{
if (b->last_token->tag == TABLE && ISBODY(b->last_token))
{
/* If the last thing the reformatter dealt with was a table, */
/* then extra table structures could have been added by */
/* HTMLLib. It is important to ensure that any tokens that */
/* were added to the token stream are preprocessed before */
/* starting a reformat. */
fetch_preprocess_token(b, b->last_token);
/* Make sure the page is fully reformatted */
start = -1;
reformat_format_from(b, b->cell->nlines - 2, 1, -1);
}
#ifdef TRACE
if (tl & (1u<<6)) Printf("\nfetch_fetcher: Finished, so stopping and exiting.\n");
#endif
fetch_stop(b, 1);
break;
}
else fetch_preprocess_token(b, tptr);
/* Closure of series of ifs that checked the state of 'waiting' */
/* amongst other things, to handle redirections etc. The bulk */
/* of the code deals with a conventional fetch. */
}
/* Closure of for loop that deals with several fetches per null */
}
/* If start is >= 0, there is data that can be used for */
/* displaying the page; so start a reformat based on that */
/* data. Start from 'one line up' as the last line may have */
/* been only partially finished when it was last redrawn. */
/* The reformat session can be deferred if the reformatter */
/* is not running, but if the reformatter is still going, */
/* push this request through immediately. Otherwise */
/* problems with long-delayed reformats way after the page */
/* has been more or less completely formatted can occur. */
if (start >= 0)// && !reformat_formatting(b))
{
reformat_format_from(b,
start - 1, /* '-1' as this parameter is the *last valid* line number */
/* we want to keep, and 'start' holds the first line to */
/* start the reformat at. */
reformat_formatting(b),
-1);
}
}
/*************************************************/
/* fetch_chkerror() */
/* */
/* Called by low level fetch routines instead of */
/* the ChkError macro, as it stops the current */
/* fetch correctly before reporting the error. */
/* */
/* Parameters: Pointer to a browser_data */
/* structure relevant to the fetch; */
/* */
/* Pointer to a _kernel_oserror */
/* structure, which contains the */
/* error to report (or NULL). */
/* */
/* Returns: 0 if there was no error, else 1. */
/*************************************************/
int fetch_chkerror(browser_data * b, _kernel_oserror * e)
{
if (e)
{
/* There is an error - cancel the fetch */
fetch_cancel(b);
/* Report the error */
show_error_ret(e);
/* Flag the error in the returned value */
return 1;
}
return 0;
}
/*************************************************/
/* fetch_cancel() */
/* */
/* Aborts a fetch, closing any relevant streams, */
/* freeing up any claimed memory that was only */
/* relevant to the fetch, but leaves the page */
/* fetched so far visible. */
/* */
/* Parameters: A pointer to the browser_data */
/* structure relevant to the fetch */
/* to be cancelled. */
/*************************************************/
_kernel_oserror * fetch_cancel(browser_data * b)
{
/* If there is a fetch, and the HTML data isn't being used by the */
/* display routines, close the fetch handle and free up any memory */
/* associated with it. */
if ((b->fetch_handle) && ((b->fetch_handle) != (b->display_handle))) html_close(b->fetch_handle);
b->fetch_handle = 0;
/* If a META tag is about to do a reload, cancel this */
if (b->meta_refresh_at) deregister_null_claimant(Wimp_ENull, (WimpEventHandler *) meta_check_refresh, b);
b->meta_refresh_at = 0;
/* If not fetching, exit here */
if (!fetch_fetching(b)) return NULL;
/* Stop everything else */
fetch_stop(b, 1);
/* Ensure the page is correctly formatted */
if (b->cell->nlines) reformat_format_from(b, b->cell->nlines - 1, 1, -1);
return(NULL);
}
/*************************************************/
/* fetch_stop() */
/* */
/* Stops a fetch, optionally discarding the */
/* HTML source, making sure the browser window */
/* state (buttons, status bar animation etc.) is */
/* correct, any open files are closed, and so */
/* forth. In the UI sense this is higher level */
/* than fetch_cancel, though fetch_cancel calls */
/* this as part of doing other cancel actions, */
/* and is therefore the higher level function. */
/* */
/* Parameters: A pointer to the browser_data */
/* structure relevant to the fetch */
/* to be stopped; */
/* */
/* 1 to keep the HTML source, 0 to */
/* destroy it. */
/*************************************************/
void fetch_stop(browser_data * b, int keep_source)
{
/* Destroy the source, provided the browser was fetching any */
if (fetch_fetching(b) && !keep_source) browser_destroy_source(b);
/* Set the fetch status to idle */
b->fetch_status = BS_IDLE;
/* The save_link flag tells the browser to save the next fetch as data, */
/* even if it is parsable. Want to make sure that flag is clear now to */
/* avoid complications later on. */
b->save_link = 0;
/* If data was being saved to a file, close that file */
if (b->save_file)
{
char buffer[Limits_OS_Pathname];
/* Leave the file intact, but set it to a Data filetype */
if (
!_swix(OS_Args,
_INR(0,2) | _IN(5),
7, /* Read pathname of open file */
b->save_file->__file,
buffer,
sizeof(buffer))
)
{
_swix(OS_File,
_INR(0,2),
18,
buffer,
FileType_DATA);
}
fclose(b->save_file);
b->save_file = NULL;
}
if (b->save_dbox) saveobject_close(b);
/* If there is a fetch, and the associated HTML document isn't being */
/* used by the display routines, close that fetch handle and free */
/* any memory associated with it. */
if ((b->fetch_handle) && (b->fetch_handle != b->display_handle)) html_close(b->fetch_handle);
b->fetch_handle = 0;
/* Discard the URL being fetched */
#ifdef TRACE
if (tl & (1u<<12)) Printf("fetch_stop: Chunk CK_FURL set to 0\n");
#endif
memory_set_chunk_size(b, NULL, CK_FURL, 0);
b->reloading = 0;
/* Update the status bar */
toolbars_cancel_status(b, Toolbars_Status_Fetching);
/* Check that the window extent is large enough to fit the whole page in */
reformat_check_extent(b);
/* Set up the window buttons */
toolbars_set_button_states(b);
}
/*************************************************/
/* fetch_authorisation_proceed() */
/* */
/* Given a browser_data structure with a URL */
/* containing a host and a pointer to a realm */
/* string, proceed with an authorisation */
/* request based on the data in the global */
/* 'authorise' flex block (handled by the */
/* functions in Authorise.c). */
/* */
/* Parameters: Pointer to a browser_data struct */
/* relevant to the authorisation */
/* request; */
/* */
/* Pointer to a urlstat structure */
/* giving the fetch context, or NULL */
/* to get it from the browser_data */
/* structure's fetch_handle field */
/* (no good for images, obviously); */
/* */
/* Pointer to a string containing */
/* the realm for the request; */
/* */
/* Pointer to the request URL. */
/*************************************************/
void fetch_authorisation_proceed(browser_data * b, urlstat * context, char * realm, char * url)
{
int ok, l, s, offset;
char host [Limits_HostName];
char base64 [(Limits_AuthUserWrit + Limits_AuthPassWrit + 2) * 4 / 3];
char authcode[(Limits_AuthUserWrit + Limits_AuthPassWrit + 2)];
urlstat * up;
_kernel_oserror * e;
/* Clear the 'authorising' flag */
authorising = 0;
/* If required, find out the session handle */
if (!context)
{
up = urlstat_find_entry(b->fetch_handle);
if (!up)
{
fetch_cancel(b);
erb.errnum = Utils_Error_Custom_Normal; /* Nasty error but can recover from it here */
StrNCpy0(erb.errmess,
lookup_token("StrNotFd:Internal error: Can't find structure in %0.",
0,
"fetch_authorisation_proceed"));
show_error_ret(&erb);
return;
}
}
else up = context;
/* Mark this fetch as authorised once already - if the server */
/* resends an authorisation request the fetcher will know */
/* that the authorisation failed (see html_get_next_token). */
up->authorised = 2;
if (up->extradata) s = flex_size((flex_ptr) &up->extradata);
else s = 0;
/* Work out the host name */
urlutils_host_name_from_url(url, host, sizeof(host));
/* Store the details in the authcode block */
offset = authorise_find_user_name(host, realm);
if (offset < 0)
{
fetch_authorisation_fail(b);
return;
}
strcpy(authcode, authorise + offset);
strcat(authcode, ":");
offset = authorise_find_password(host, realm);
if (offset < 0)
{
fetch_authorisation_fail(b);
return;
}
strcat(authcode, authorise + offset);
/* Encode the block */
l = encode_base64(authcode, strlen(authcode), base64);
base64[l] = 0;
/* Allocate memory for the encoded data as a */
/* header entry. */
/* */
/* +2 accounts for CR + LF termination. */
l += strlen(AuthorisationStr) + 2;
if (s) ok = flex_extend((flex_ptr) &up->extradata, s + l);
else ok = flex_alloc((flex_ptr) &up->extradata, l + 1);
if (!ok)
{
fetch_cancel(b);
show_error_ret(make_no_fetch_memory_error(12));
return;
}
if (s) memmove(up->extradata + l, up->extradata, s);
/* Copy the data in */
strcpy(up->extradata, AuthorisationStr);
strncpy(up->extradata + strlen(AuthorisationStr), base64, l - 23);
up->extradata[l - 2] = '\r';
up->extradata[l - 1] = '\n';
if (!s) up->extradata[l] = 0;
/* Restart the fetch with authentication */
e = url_get_url(0,
up->session,
up->method,
url,
up->extradata,
NULL,
2);
if (e)
{
fetch_cancel(b);
show_error_ret(e);
}
return;
}
/*************************************************/
/* fetch_authorisation_fail() */
/* */
/* Called when authorisation for a URL fails in */
/* some way. Reports an appropriate error and */
/* stops the fetch. */
/* */
/* Parameters: Pointer to a browser_data struct */
/* relevant to the fetch. */
/*************************************************/
void fetch_authorisation_fail(browser_data * b)
{
/* Cancel the fetch */
authorising = 0;
fetch_cancel(b);
/* Give the error */
erb.errnum = Utils_Error_Custom_Message;
StrNCpy0(erb.errmess,
lookup_token("BadAuthor:Authorisation failed; you must use a valid user name and password.",
0,
0));
show_error_ret(&erb);
}
/*************************************************/
/* html_get_raw_data() */
/* */
/* Gets a chunk of data from a stream, assuming */
/* that it is *not* HTML. */
/* */
/* Parameters: Pointer to a browser_data struct */
/* relevant to the fetch; */
/* */
/* The fetch handle; */
/* */
/* Pointer to buffer into which the */
/* fetched data will be placed (as */
/* a char *); */
/* */
/* Size of the buffer; */
/* */
/* Pointer to an int into which 1 is */
/* placed if the fetch is complete, */
/* else 0 is returned (this pointer */
/* may be NULL); */
/* */
/* Pointer to an int into which the */
/* number of bytes fetched is placed */
/* (which may also be NULL). */
/* */
/* Assumes: That if the browser_data struct */
/* pointer is NULL, the fetch is not */
/* for an internal URL; */
/* */
/* The int pointer to take the */
/* number of bytes fetched may not */
/* be NULL. */
/*************************************************/
_kernel_oserror * fetch_get_raw_data(browser_data * b, unsigned int handle, char * buffer,
int size, int * done, int * bytes)
{
_kernel_oserror * e;
urlstat * up;
int s, t;
//
// This function does not know about internal URLs yet (so parameter 'b' is currently unused)...
// BEWARE when using this, as of course this URL fetch may not be for page data under an
// internal URL.
//
/* Find the urlstat structure for the fetch handle */
up = urlstat_find_entry(handle);
if (!up)
{
erb.errnum = Utils_Error_Custom_Fatal;
StrNCpy0(erb.errmess,
lookup_token("StrNotFd:Internal error: Can't find structure in %0.",
0,
"fetch_get_raw_data"));
return &erb;
}
/* Read some data */
e = url_read_data(0,
handle,
buffer,
size,
NULL,
bytes,
&t);
if (e) return e;
/* Get the fetch status */
e = url_status(0,
handle,
&s,
NULL,
NULL);
if (e) return e;
/* Fill in 'done' as appropriate to the fetch status and exit */
if (done) *done = (s & URL_Status_Done) ? 1 : 0;
return NULL;
}