aniani/elinks
1
0
Fork 0
mirror of https://github.com/rkd77/elinks.git synced 2024-12-04 14:46:47 -05:00
Code

Merge with git+ssh://pasky.or.cz/srv/git/elinks.git

Browse Source
This commit is contained in:
Adam Golebiowski 2005-11-27 21:25:17 +01:00 committed by
commit 6edbc346e0
7 changed files with 298 additions and 314 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

187
src/document/html/parser.c
View File

@ -262,193 +262,6 @@ html_skip(struct html_context *html_context, unsigned char *a)
html_top.type = ELEMENT_DONT_KILL;
}
#ifdef CONFIG_ECMASCRIPT
int
do_html_script(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
/* TODO: <noscript> processing. Well, same considerations apply as to
* CSS property display: none processing. */
/* TODO: Charsets for external scripts. */
unsigned char *type, *language, *src;
int in_comment = 0;
html_skip(html_context, a);
/* We try to process nested <script> if we didn't process the parent
* one. That's why's all the fuzz. */
/* Ref:
* http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
*/
type = get_attr_val(a, "type", html_context->options);
if (type) {
unsigned char *pos = type;
if (!strncasecmp(type, "text/", 5)) {
pos += 5;
} else if (!strncasecmp(type, "application/", 12)) {
pos += 12;
} else {
mem_free(type);
not_processed:
/* Permit nested scripts and retreat. */
html_top.invisible++;
return 1;
}
if (!strncasecmp(pos, "javascript", 10)) {
int len = strlen(pos);
if (len > 10 && !isdigit(pos[10])) {
mem_free(type);
goto not_processed;
}
} else if (strcasecmp(pos, "ecmascript")
&& strcasecmp(pos, "jscript")
&& strcasecmp(pos, "livescript")
&& strcasecmp(pos, "x-javascript")
&& strcasecmp(pos, "x-ecmascript")) {
mem_free(type);
goto not_processed;
}
mem_free(type);
}
/* Check that the script content is ecmascript. The value of the
* language attribute can be JavaScript with optional version digits
* postfixed (like: ``JavaScript1.1'').
* That attribute is deprecated in favor of type by HTML 4.01 */
language = get_attr_val(a, "language", html_context->options);
if (language) {
int languagelen = strlen(language);
if (languagelen < 10
|| (languagelen > 10 && !isdigit(language[10]))
|| strncasecmp(language, "javascript", 10)) {
mem_free(language);
goto not_processed;
}
mem_free(language);
}
if (html_context->part->document
&& (src = get_attr_val(a, "src", html_context->options))) {
/* External reference. */
unsigned char *import_url;
struct uri *uri;
if (!get_opt_bool("ecmascript.enable")) {
mem_free(src);
goto not_processed;
}
/* HTML <head> urls should already be fine but we can.t detect them. */
import_url = join_urls(html_context->base_href, src);
mem_free(src);
if (!import_url) goto imported;
uri = get_uri(import_url, URI_BASE);
if (!uri) goto imported;
/* Request the imported script as part of the document ... */
html_context->special_f(html_context, SP_SCRIPT, uri);
done_uri(uri);
/* Create URL reference onload snippet. */
insert_in_string(&import_url, 0, "^", 1);
add_to_string_list(&html_context->part->document->onload_snippets,
import_url, -1);
imported:
/* Retreat. Do not permit nested scripts, tho'. */
if (import_url) mem_free(import_url);
return 1;
}
/* Positive, grab the rest and interpret it. */
/* First position to the real script start. */
while (html < eof && *html <= ' ') html++;
if (eof - html > 4 && !strncmp(html, "<!--", 4)) {
in_comment = 1;
/* We either skip to the end of line or to -->. */
for (; *html != '\n' && *html != '\r' && eof - html >= 3; html++) {
if (!strncmp(html, "-->", 3)) {
/* This means the document is probably broken.
* We will now try to process the rest of
* <script> contents, which is however likely
* to be empty. Should we try to process the
* comment too? Currently it seems safer but
* less tolerant to broken pages, if there are
* any like this. */
html += 3;
in_comment = 0;
break;
}
}
}
*end = html;
/* Now look ahead for the script end. The <script> contents is raw
* CDATA, so we just look for the ending tag and need not care for
* any quote marks counting etc - YET, we are more tolerant and permit
* </script> stuff inside of the script if the whole <script> element
* contents is wrapped in a comment. See i.e. Mozilla bug 26857 for fun
* reading regarding this. */
for (; *end < eof; (*end)++) {
unsigned char *name;
int namelen;
if (in_comment) {
/* TODO: If we ever get some standards-quirk mode
* distinction, this should be disabled in the
* standards mode (and we should just look for CDATA
* end, which is "</"). --pasky */
if (eof - *end >= 3 && !strncmp(*end, "-->", 3)) {
/* Next iteration will jump passed the ending '>' */
(*end) += 2;
in_comment = 0;
}
continue;
/* XXX: Scan for another comment? That's admittelly
* already stretching things a little bit to an
* extreme ;-). */
}
if (**end != '<')
continue;
/* We want to land before the closing element, that's why we
* don't pass @end also as the appropriate parse_element()
* argument. */
if (parse_element(*end, eof, &name, &namelen, NULL, NULL))
continue;
if (strlcasecmp(name, namelen, "/script", 7))
continue;
/* We have won! */
break;
}
if (*end >= eof) {
/* Either the document is not completely loaded yet or it's
* broken. At any rate, run away screaming. */
*end = eof; /* Just for sanity. */
return 1;
}
if (html_context->part->document && *html != '^') {
add_to_string_list(&html_context->part->document->onload_snippets,
html, *end - html);
}
return 0;
}
#endif
void
process_head(struct html_context *html_context, unsigned char *head)
{

4
src/document/html/parser.h
View File

@ -207,8 +207,4 @@ void ln_break(struct html_context *html_context, int n);
int get_color(struct html_context *html_context, unsigned char *a, unsigned char *c, color_T *rgb);
#ifdef CONFIG_ECMASCRIPT
int do_html_script(struct html_context *html_context, unsigned char *attr, unsigned char *html, unsigned char *eof, unsigned char **end);
#endif
#endif

212
src/document/html/parser/forms.c
View File

@ -298,105 +298,9 @@ hid:
html_context->special_f(html_context, SP_CONTROL, fc);
}
void
html_select(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
unsigned char *al;
if (!do_html_select(a, html, eof, end, html_context))
return;
al = get_attr_val(a, "name", html_context->options);
if (!al) return;
html_focusable(html_context, a);
html_top.type = ELEMENT_DONT_KILL;
mem_free_set(&format.select, al);
format.select_disabled = has_attr(a, "disabled", html_context->options)
? FORM_MODE_DISABLED
: FORM_MODE_NORMAL;
}
void
html_option(struct html_context *html_context, unsigned char *a,
unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
struct form_control *fc;
unsigned char *val;
if (!format.select) return;
val = get_attr_val(a, "value", html_context->options);
if (!val) {
struct string str;
unsigned char *p, *r;
unsigned char *name;
int namelen;
for (p = a - 1; *p != '<'; p--);
if (!init_string(&str)) goto end_parse;
if (parse_element(p, html_context->eoff, NULL, NULL, NULL, &p)) {
INTERNAL("parse element failed");
val = str.source;
goto end_parse;
}
se:
while (p < html_context->eoff && isspace(*p)) p++;
while (p < html_context->eoff && !isspace(*p) && *p != '<') {
sp:
add_char_to_string(&str, *p ? *p : ' '), p++;
}
r = p;
val = str.source; /* Has to be before the possible 'goto end_parse' */
while (r < html_context->eoff && isspace(*r)) r++;
if (r >= html_context->eoff) goto end_parse;
if (r - 2 <= html_context->eoff && (r[1] == '!' || r[1] == '?')) {
p = skip_comment(r, html_context->eoff);
goto se;
}
if (parse_element(r, html_context->eoff, &name, &namelen, NULL, &p)) goto sp;
if (strlcasecmp(name, namelen, "OPTION", 6)
&& strlcasecmp(name, namelen, "/OPTION", 7)
&& strlcasecmp(name, namelen, "SELECT", 6)
&& strlcasecmp(name, namelen, "/SELECT", 7)
&& strlcasecmp(name, namelen, "OPTGROUP", 8)
&& strlcasecmp(name, namelen, "/OPTGROUP", 9))
goto se;
}
end_parse:
fc = init_form_control(FC_CHECKBOX, a, html_context);
if (!fc) {
mem_free_if(val);
return;
}
fc->name = null_or_stracpy(format.select);
fc->default_value = val;
fc->default_state = has_attr(a, "selected", html_context->options);
fc->mode = has_attr(a, "disabled", html_context->options)
? FORM_MODE_DISABLED
: format.select_disabled;
put_chrs(html_context, " ", 1);
html_stack_dup(html_context, ELEMENT_KILLABLE);
format.form = fc;
format.style.attr |= AT_BOLD;
put_chrs(html_context, "[ ]", 3);
kill_html_stack_item(html_context, &html_top);
put_chrs(html_context, " ", 1);
html_context->special_f(html_context, SP_CONTROL, fc);
}
static struct list_menu lnk_menu;
int
static void
do_html_select(unsigned char *attr, unsigned char *html,
unsigned char *eof, unsigned char **end,
struct html_context *html_context)
@ -414,7 +318,6 @@ do_html_select(unsigned char *attr, unsigned char *html,
int group = 0;
int i, max_width;
if (has_attr(attr, "multiple", html_context->options)) return 1;
html_focusable(html_context, attr);
init_menu(&lnk_menu);
@ -440,7 +343,7 @@ abort:
}
destroy_menu(&lnk_menu);
*end = en;
return 0;
return;
}
if (lbl.source) {
@ -563,20 +466,115 @@ end_parse:
kill_html_stack_item(html_context, &html_top);
put_chrs(html_context, "]", 1);
html_context->special_f(html_context, SP_CONTROL, fc);
return 0;
}
void
html_textarea(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof, unsigned char **end)
static void
do_html_select_multiple(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof,
unsigned char **end)
{
do_html_textarea(a, html, eof, end, html_context);
unsigned char *al = get_attr_val(a, "name", html_context->options);
if (!al) return;
html_focusable(html_context, a);
html_top.type = ELEMENT_DONT_KILL;
mem_free_set(&format.select, al);
format.select_disabled = has_attr(a, "disabled", html_context->options)
? FORM_MODE_DISABLED
: FORM_MODE_NORMAL;
}
void
do_html_textarea(unsigned char *attr, unsigned char *html, unsigned char *eof,
unsigned char **end, struct html_context *html_context)
html_select(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
if (has_attr(a, "multiple", html_context->options))
do_html_select_multiple(html_context, a, html, eof, end);
else
do_html_select(a, html, eof, end, html_context);
}
void
html_option(struct html_context *html_context, unsigned char *a,
unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
struct form_control *fc;
unsigned char *val;
if (!format.select) return;
val = get_attr_val(a, "value", html_context->options);
if (!val) {
struct string str;
unsigned char *p, *r;
unsigned char *name;
int namelen;
for (p = a - 1; *p != '<'; p--);
if (!init_string(&str)) goto end_parse;
if (parse_element(p, html_context->eoff, NULL, NULL, NULL, &p)) {
INTERNAL("parse element failed");
val = str.source;
goto end_parse;
}
se:
while (p < html_context->eoff && isspace(*p)) p++;
while (p < html_context->eoff && !isspace(*p) && *p != '<') {
sp:
add_char_to_string(&str, *p ? *p : ' '), p++;
}
r = p;
val = str.source; /* Has to be before the possible 'goto end_parse' */
while (r < html_context->eoff && isspace(*r)) r++;
if (r >= html_context->eoff) goto end_parse;
if (r - 2 <= html_context->eoff && (r[1] == '!' || r[1] == '?')) {
p = skip_comment(r, html_context->eoff);
goto se;
}
if (parse_element(r, html_context->eoff, &name, &namelen, NULL, &p)) goto sp;
if (strlcasecmp(name, namelen, "OPTION", 6)
&& strlcasecmp(name, namelen, "/OPTION", 7)
&& strlcasecmp(name, namelen, "SELECT", 6)
&& strlcasecmp(name, namelen, "/SELECT", 7)
&& strlcasecmp(name, namelen, "OPTGROUP", 8)
&& strlcasecmp(name, namelen, "/OPTGROUP", 9))
goto se;
}
end_parse:
fc = init_form_control(FC_CHECKBOX, a, html_context);
if (!fc) {
mem_free_if(val);
return;
}
fc->name = null_or_stracpy(format.select);
fc->default_value = val;
fc->default_state = has_attr(a, "selected", html_context->options);
fc->mode = has_attr(a, "disabled", html_context->options)
? FORM_MODE_DISABLED
: format.select_disabled;
put_chrs(html_context, " ", 1);
html_stack_dup(html_context, ELEMENT_KILLABLE);
format.form = fc;
format.style.attr |= AT_BOLD;
put_chrs(html_context, "[ ]", 3);
kill_html_stack_item(html_context, &html_top);
put_chrs(html_context, " ", 1);
html_context->special_f(html_context, SP_CONTROL, fc);
}
void
html_textarea(struct html_context *html_context, unsigned char *attr,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
struct form_control *fc;
unsigned char *p, *t_name, *wrap_attr;

3
src/document/html/parser/forms.h
View File

@ -13,7 +13,4 @@ element_handler_T html_select;
element_handler_T html_option;
element_handler_T html_textarea;
int do_html_select(unsigned char *attr, unsigned char *html, unsigned char *eof, unsigned char **end, struct html_context *html_context);
void do_html_textarea(unsigned char *attr, unsigned char *html, unsigned char *eof, unsigned char **end, struct html_context *html_context);
#endif

195
src/document/html/parser/general.c
View File

@ -23,6 +23,8 @@
#include "document/html/parser/stack.h"
#include "document/html/parser/parse.h"
#include "document/html/parser.h"
#include "document/html/renderer.h"
#include "document/html/tables.h"
#include "document/options.h"
#include "intl/charsets.h"
#include "protocol/uri.h"
@ -164,9 +166,186 @@ html_script(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
#ifdef CONFIG_ECMASCRIPT
do_html_script(html_context, a, html, eof, end);
#else
/* TODO: <noscript> processing. Well, same considerations apply as to
* CSS property display: none processing. */
/* TODO: Charsets for external scripts. */
unsigned char *type, *language, *src;
int in_comment = 0;
#endif
html_skip(html_context, a);
#ifdef CONFIG_ECMASCRIPT
/* We try to process nested <script> if we didn't process the parent
* one. That's why's all the fuzz. */
/* Ref:
* http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
*/
type = get_attr_val(a, "type", html_context->options);
if (type) {
unsigned char *pos = type;
if (!strncasecmp(type, "text/", 5)) {
pos += 5;
} else if (!strncasecmp(type, "application/", 12)) {
pos += 12;
} else {
mem_free(type);
not_processed:
/* Permit nested scripts and retreat. */
html_top.invisible++;
return;
}
if (!strncasecmp(pos, "javascript", 10)) {
int len = strlen(pos);
if (len > 10 && !isdigit(pos[10])) {
mem_free(type);
goto not_processed;
}
} else if (strcasecmp(pos, "ecmascript")
&& strcasecmp(pos, "jscript")
&& strcasecmp(pos, "livescript")
&& strcasecmp(pos, "x-javascript")
&& strcasecmp(pos, "x-ecmascript")) {
mem_free(type);
goto not_processed;
}
mem_free(type);
}
/* Check that the script content is ecmascript. The value of the
* language attribute can be JavaScript with optional version digits
* postfixed (like: ``JavaScript1.1'').
* That attribute is deprecated in favor of type by HTML 4.01 */
language = get_attr_val(a, "language", html_context->options);
if (language) {
int languagelen = strlen(language);
if (languagelen < 10
|| (languagelen > 10 && !isdigit(language[10]))
|| strncasecmp(language, "javascript", 10)) {
mem_free(language);
goto not_processed;
}
mem_free(language);
}
if (html_context->part->document
&& (src = get_attr_val(a, "src", html_context->options))) {
/* External reference. */
unsigned char *import_url;
struct uri *uri;
if (!get_opt_bool("ecmascript.enable")) {
mem_free(src);
goto not_processed;
}
/* HTML <head> urls should already be fine but we can.t detect them. */
import_url = join_urls(html_context->base_href, src);
mem_free(src);
if (!import_url) goto imported;
uri = get_uri(import_url, URI_BASE);
if (!uri) goto imported;
/* Request the imported script as part of the document ... */
html_context->special_f(html_context, SP_SCRIPT, uri);
done_uri(uri);
/* Create URL reference onload snippet. */
insert_in_string(&import_url, 0, "^", 1);
add_to_string_list(&html_context->part->document->onload_snippets,
import_url, -1);
imported:
/* Retreat. Do not permit nested scripts, tho'. */
if (import_url) mem_free(import_url);
return;
}
/* Positive, grab the rest and interpret it. */
/* First position to the real script start. */
while (html < eof && *html <= ' ') html++;
if (eof - html > 4 && !strncmp(html, "<!--", 4)) {
in_comment = 1;
/* We either skip to the end of line or to -->. */
for (; *html != '\n' && *html != '\r' && eof - html >= 3; html++) {
if (!strncmp(html, "-->", 3)) {
/* This means the document is probably broken.
* We will now try to process the rest of
* <script> contents, which is however likely
* to be empty. Should we try to process the
* comment too? Currently it seems safer but
* less tolerant to broken pages, if there are
* any like this. */
html += 3;
in_comment = 0;
break;
}
}
}
*end = html;
/* Now look ahead for the script end. The <script> contents is raw
* CDATA, so we just look for the ending tag and need not care for
* any quote marks counting etc - YET, we are more tolerant and permit
* </script> stuff inside of the script if the whole <script> element
* contents is wrapped in a comment. See i.e. Mozilla bug 26857 for fun
* reading regarding this. */
for (; *end < eof; (*end)++) {
unsigned char *name;
int namelen;
if (in_comment) {
/* TODO: If we ever get some standards-quirk mode
* distinction, this should be disabled in the
* standards mode (and we should just look for CDATA
* end, which is "</"). --pasky */
if (eof - *end >= 3 && !strncmp(*end, "-->", 3)) {
/* Next iteration will jump passed the ending '>' */
(*end) += 2;
in_comment = 0;
}
continue;
/* XXX: Scan for another comment? That's admittelly
* already stretching things a little bit to an
* extreme ;-). */
}
if (**end != '<')
continue;
/* We want to land before the closing element, that's why we
* don't pass @end also as the appropriate parse_element()
* argument. */
if (parse_element(*end, eof, &name, &namelen, NULL, NULL))
continue;
if (strlcasecmp(name, namelen, "/script", 7))
continue;
/* We have won! */
break;
}
if (*end >= eof) {
/* Either the document is not completely loaded yet or it's
* broken. At any rate, run away screaming. */
*end = eof; /* Just for sanity. */
return;
}
if (html_context->part->document && *html != '^') {
add_to_string_list(&html_context->part->document->onload_snippets,
html, *end - html);
}
#endif
}
@ -411,12 +590,20 @@ html_hr(struct html_context *html_context, unsigned char *a,
}
void
html_table(struct html_context *html_context, unsigned char *a,
html_table(struct html_context *html_context, unsigned char *attr,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
if (html_context->options->tables
&& html_context->table_level < HTML_MAX_TABLE_LEVEL) {
format_table(attr, html, eof, end, html_context);
ln_break(html_context, 2);
return;
}
par_format.leftmargin = par_format.rightmargin = html_context->margin;
par_format.align = ALIGN_LEFT;
html_linebrk(html_context, a, html, eof, end);
html_linebrk(html_context, attr, html, eof, end);
format.style.attr = 0;
}

9
src/document/html/parser/parse.c
View File

@ -20,7 +20,6 @@
#include "document/html/parser/parse.h"
#include "document/html/parser/stack.h"
#include "document/html/parser.h"
#include "document/html/tables.h"
#include "document/options.h"
#include "intl/charsets.h"
#include "util/conv.h"
@ -782,14 +781,6 @@ start_element(struct element_info *ei,
restore_format = html_is_preformatted();
old_format = par_format;
if (ei->func == html_table && html_context->options->tables
&& html_context->table_level < HTML_MAX_TABLE_LEVEL) {
ELEMENT_RENDER_PROLOGUE
format_table(attr, html, eof, &html, html_context);
ln_break(html_context, 2);
return html;
}
/* Support for <meta refresh="..."> inside <body>. (bug 700) */
if (ei->func == html_meta && html_context->was_body) {
html_handle_body_meta(html_context, name - 1, eof);

2
src/osdep/stub.h
View File

@ -155,8 +155,10 @@ extern char *__builtin_stpcpy(char *dest, const char *src);
extern void *__builtin_mempcpy(void *dest, const void *src, size_t n);
#endif
#if 0
#ifndef __builtin_va_copy
#define __builtin_va_copy(dest, src) do { dest = src; } while (0)
#endif
#endif
#endif