HTMLv 23.70

git-svn-id: svn://kolibrios.org@2228 a494cfbc-eb01-0410-851d-a64ba20cac60
This commit is contained in:
Kirill Lipatov (Leency) 2011-09-22 11:18:01 +00:00
parent f20c948791
commit db9a61710d
6 changed files with 227 additions and 28 deletions

View File

@ -1,3 +1,9 @@
22.09.11 -- v0.70 -- 11.3 Кб
- улучшен и оптимизирован код обработки юникодовских тегов (&), исправлен
и дополнен их список;
- устранён вылет программы при нажатии клавиши CapsLock (спасибо Mario);
- заглушка на вылет программы при загрузке страниц из и-нэта (спасибо Mario).
13.09.11 -- v0.69 -- 11.2 Кб
- нормальная прокрутка изображений;
- работа с файлами любого размера;

View File

@ -18,7 +18,7 @@ dword j,
char download_path[]="/rd/1/.download";
//char search_path[]="http://nova.rambler.ru/search?words=";
char search_path[]="http://nigma.ru/index.php?s=";
char version[]=" Text-based Browser 23.69";
char version[]=" Text-based Browser 23.70";
struct TWebBrowser {
@ -37,6 +37,7 @@ TWebBrowser WB1;
#include "include\history.h--"
#include "include\colors.h--"
#include "include\unicode_tags.h--"
void TWebBrowser::Scan(dword id) {
IF (id > 399) {
@ -57,7 +58,6 @@ void TWebBrowser::Scan(dword id) {
} //åñëè ìàëî ñòðîê èãíîðèðóåì íåêîòîðûå êíîïêè
switch (id) {
case 011:
case 004:
ReadHtml();
break;
case BACK:
@ -185,6 +185,7 @@ void ReadHtml(byte DO_LOAD)
IF (URL[strlen(#URL)-1]=='/') URL[strlen(#URL)-1]='';
downloader_id = RunProgram("/sys/network/downloader", #URL);
IF (downloader_id<0) RunProgram("@notify", "Error running Downloader. Internet unavilable.");
RunProgram("/sys/network/downloader", #URL);
RETURN;
}
file_size stdcall (#download_path);
@ -192,6 +193,7 @@ void ReadHtml(byte DO_LOAD)
ELSE file_size stdcall (#URL);
filesize = EBX;
IF (!filesize) return; //Lee 22.09
mem_Free(buf);
buf = mem_Alloc(filesize);
ReadFile(0, filesize, buf, #URL);
@ -221,6 +223,9 @@ void TWebBrowser::Load(dword adress) {
edit_box_draw stdcall(#edit1); //ðèñóåì ñòðîêó àäðåñà
ReadHtml(LETS_LOAD);
IF (!filesize) return; //Lee 22.09
wintodos(buf);
ParseHTML(buf, filesize);
IF(!strlen(buf)) {
@ -266,6 +271,7 @@ void TWebBrowser::ParseHTML(dword bword, fsize){
line = '';
copystr("|", #page_links);
IF(!strcmp(#URL + strlen(#URL) - 4, ".txt")) pre_text = 1; //çà÷¸òíîå îòîáðàæåíèå òåêñòà
//IF(!strcmp(#URL + strlen(#URL) - 4, ".rtf")) pre_text = 1;
IF(!strcmp(#URL + strlen(#URL) - 4, ".mht")) ignor_text = 1;
IF(za_kadrom == 0) || (WindowRePaint) DrawBar(left, top, width - 15, 15, 0xFFFFFF); //çàêðàøèâàåì ïåðâóþ ñòðîêó
for (; buf + fsize > bword; bword++;) {
@ -339,43 +345,79 @@ void TWebBrowser::ParseHTML(dword bword, fsize){
bukva=Hex2Symb(#temp);
IF (bukva) goto DEFAULT_MARK;
break;
case '&':
IF(ignor_text) break;
/*case '\\': //ïîääåğæêà rtf
IF(strcmp(#URL + strlen(#URL) - 4, ".rtf")<>0) goto DEFAULT_MARK;
bword++;
j = 0;
WHILE(ESBYTE[bword] <>';') && (j < 7) {
IF (ESBYTE[bword] =='\'')
{
bword++;
bukva=ESBYTE[bword];
copystr(#bukva, #temp);
bword++;
bukva=ESBYTE[bword];
copystr(#bukva, #temp + strlen(#temp));
bukva=Hex2Symb(#temp);
IF (bukva) goto DEFAULT_MARK;
}
else
{
FOR (j=0; (ESBYTE[bword] <>'\\') && (buf + fsize < bword); j++; bword++;)
{
bukva = ESBYTE[bword];
copystr(#bukva, #tag + strlen(#tag));
bword++;
j++;
}
bukva = ''; //âñş ıòó õğåíü íàäî çàëàáàòü ÷åğåç ìàñèâ
IF(!strcmp("nbsp", #tag)) copystr(" ", #line + strlen(#line));
IF(!strcmp("copy", #tag)) || (!strcmp("#169", #tag)) copystr("(c)", #line + strlen(#line));
IF(!strcmp("reg", #tag)) || (!strcmp("#174", #tag)) copystr("(r)", #line + strlen(#line));
IF ((!strcmp(#tag, "par\0x0a")) || (!strcmp(#tag, "par\0x0b"))) stroka++;
tag='';
bword--;
}
break;*/
case '&': //îáğàáîòêà òåãîâ òèïà &nbsp;
IF(ignor_text) break;
bword++;
tag='';
FOR (j=0; (ESBYTE[bword] <>';') && (j < 7); j++; bword++;)
{
bukva = ESBYTE[bword];
copystr(#bukva, #tag + strlen(#tag));
}
FOR (j=0; unicode_tags[j]!=0; j+=2;)
{
IF(!strcmp(#tag, unicode_tags[j]))
{
copystr(unicode_tags[j+1], #line + strlen(#line));
break 1;
}
}
rez = StrToInt(#tag + 1) - 1040;
IF(tag[1] == '1') && (strlen(#tag) == 5) bukva = utf100tmmv_mas[rez];
IF(!strcmp("bdquo", #tag)) copystr(",,", #line + strlen(#line));
IF(!strcmp("amp", #tag)) || (!strcmp("#38", #tag)) bukva = '&';
IF(!strcmp("lt", #tag)) || (!strcmp("#60", #tag)) bukva = '<';
IF(!strcmp("gt", #tag)) || (!strcmp("#62", #tag)) bukva = '>';
IF(!strcmp("minus", #tag)) || (!strcmp("#8722", #tag)) bukva = '-';
IF(!strcmp("#151", #tag)) || (!strcmp("#149", #tag)) || (!strcmp("ndash", #tag)) bukva = '-';
IF(!strcmp("#39", #tag)) || (!strcmp("#96", #tag)) bukva = '\'';
IF(!strcmp("quot", #tag)) || (!strcmp("#34", #tag)) || (!strcmp("ldquo", #tag)) || (!strcmp("rdquo", #tag)) bukva = '"';
tag = '';
//break;
IF(tag[1] == '1') && (rez>1040) && (rez<1118) && (strlen(#tag) == 5)
{
bukva = utf100tmmv_mas[rez];
copystr(#bukva, #line + strlen(#line));
break;
}
WriteDebug("Unknown tag");
WriteDebug(#tag);
copystr(#tag, #line + strlen(#line)); //âûâîäèì íà ıêğàí íåîáğàáîòàííûé òåã, òàê áğàóçåğû çà÷åì-òî äåëàşò
break;
default:
DEFAULT_MARK:
IF(ignor_text) break;
IF(pre_text == 0) && (bukva == ' ') && (strcmp(#line + strlen(#line) - 1, " ") == 0) continue;
//
if (stolbec + strlen(#line) > max_kolvo_stolbcov) && (w_title == 0) {
if (stolbec + strlen(#line) > max_kolvo_stolbcov) && (w_title == 0)
{
copystr(#line + find_symbol(#line, ' '), #temp); //ïåðåíîñ ïî ñëîâàì
line[find_symbol(#line, ' ')] = 0x00;
NEXT_MARK: IF(stroka - 1 > max_kolvo_strok) && (za_kadrom <>0) break 1; //óõîäèì...
WhatTextStyle(left + 5, stroka * 10 + top + 5, width - 20); //âûâîä ñòðîêè
TextGoDown(left + 5, stroka * 10 + top + 5, width - 20); //çàêğàøèâàåì ñòğîêó âêîíöå
TextGoDown(left + 5, stroka * 10 + top + 5, width - 20); //çàêğàøèâàåì ñëåäóùóş ñòğîêó
copystr(#temp, #line);
}
IF(pre_text == 0) && (bukva == ' ') && (stolbec == 0) && (strlen(#line) == 0) CONTINUE;

View File

@ -0,0 +1,80 @@
<!--doctype html public "-//w3c//dtd html 4.0 transitional//ru"-->
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=windows-1251">
<title>Çàãîëîâîê ýòîãî äîêóìåíòà</title>
</head>
<body bgcolor="#000000" text="#FFFFFF">
<b>Ýòî òåñòîâàÿ ñòðàíèöà äëÿ ïðîâåðêè HTML Viewer</b><br />
<br />
Ýòà ïðîãðàììà ïîääåðæèâàåò êîäèðîâêó cp1251(Win), koi-8 (Linux, *BSD, àäåïòû Ñâÿòîãî Ïàòðèêà),
cp866 (DOS) è åù¸ äâå Þíèêîäîâñêèõ.<br><br>
<b>
<font color="#FF0000">C</font>
<font color="#FFC005">r</font>
<font color="#F6F574">e</font>
<font color="#9EFC36">a</font>
<font color="#29FDFD">t</font>
<font color="#36A8FC">e</font>
<font color="#C911FF">d </font>
<font color="#C911FF">b</font>
<font color= #C911FF >y </font>
<font color="#C911FF">L</font>
<font color="#36A8FC">e</font>
<font color="#29FDFD">e</font>
<font color="#9EFC36">n</font>
<font color="#FFC005">c</font>
<font color="#FF0000">y</font><br>
</b>
Æèòîìèð 2008
<p>
Íåáîëüøîé ñïèñîê:<ol>
<li><q>Ýòîò òåêñò â êàâû÷êàõ</q></li>
<li><u>Ýòî ïîä÷åðêíóòûé òåêñò</u></li>
<li><s>Ýòî ïåðå÷åðêíóòûé òåêñò</s></li>
<li><i>Ýòî íàêëîííûé òåêñò</i></li>
<li><strong>Ýòî òåêñò ñ óñèëåííûì âûäåëåíèåì</strong></li>
<li><b><u><s><font color="#6495ED">Êó÷à òåãîâ</font></s></u></b></ol>
</p>
<br>
<pre>
Àòü!
Äâà!
Òðè!
Àâòîð ñåãî òâîðåíèÿ íå èçâåñòåí...
</pre>
<!--ýòîãî òåêñòà çäåñü <íåò>-->
<hr color="#758999">
 ýòîì òåêñòå åñòü ïåðåõîä íà ñëåäóþùóþ ñòðîêó, íî áðàóçåð
åãî äîëæåí ïðîèãíîðèðîâàòü. Åùå ìíîãî ïðîáåëîâ. À òóò åñòü
òàáû. Êîå-êàêèå ñèìâîëû: &quot; &amp; &lt; &gt; &#149; &nbsp; &copy; &reg; &bdquo; &ldquo; &#151; ¯æà÷îê ³íøîãî áîêó ð. ªâôðàò ç'¿â ´ðóíò òà ºâðåéñüêèé ´åðáàëàéô.
<p><a href='index.htm'>Íà ãëàâíóþ ñòðàíèöó</a></p>
<p><a href="index.htm" id="#end">×òåíèå íå ïîñëåäíåãî òåãà</a></p>
<p><a href="img_mv.files/logo.jpg" id="#end">Ïîñìîòðèòå íÿøíîå èçîáðàæåíèå</a></p>
1 &#39;
2 &quot;
3 &uarr;
4 &uarr
5 &darr;
6 &laquo;
7 &raquo;
8 &gt;
9 &minus;
10 &lt;
11 &gt;
12 &ndash;
13 &rsquo;
14 &ldquo;
</body>
</html>

View File

@ -17,8 +17,6 @@
#define DONT_LOAD 0
#define LETS_LOAD 1
byte utf100tmmv_mas[77] = "€亗儎厗噲墛媽崕彁憭摂晼棙櫄洔潪煚、¥ウЖ┆<D096><E29486><EFBFBD>徕沅彐玷殛腱眍镳駂<E995B3>i\244j";
dword get_URL_part(byte len) {
char temp1[1000];

View File

@ -0,0 +1,73 @@
byte utf100tmmv_mas[77] = "€<>ƒ„…†‡ˆ‰ŠŒ<E280B9>Ž<EFBFBD><C5BD>“”•˜™šœ<E280BA>žŸ ¡¢£¤¥¦§¨©ª«¬­®¯àáâãäåæçèéêëìíîïðñh£\243i\105\244\0";
/* IF (BL==178) ESBYTE[ESI] = 73; //I
IF (BL==179) ESBYTE[ESI] = 105; //i
IF (BL==175) ESBYTE[ESI] = 244; //J
IF (BL==191) ESBYTE[ESI] = 245; //j
IF (BL==170) ESBYTE[ESI] = 242; //E
IF (BL==186) ESBYTE[ESI] = 243; //e
IF (BL==168) ESBYTE[ESI] = 240; //à
IF (BL==184) ESBYTE[ESI] = 'e'; //e
IF (BL==180) ESBYTE[ESI] = 254; //£
IF ((BL==147) || (BL==148) || (BL==171) || (BL==187)) ESBYTE[ESI] = 34;
IF ((BL==150) || (BL==151)) ESBYTE[ESI] = 45;
*/
char *unicode_tags[]={
"nbsp", " ",
"#38", " ",
"copy", "(c)",
"#169", "(c)",
"reg", "(r)",
"#174", "(r)",
"bdquo", ",,",
"amp", "&",
"#38", "&",
"lt", "<",
"#60", "<",
"gt", ">",
"#62", ">",
"minus", "-",
"ndash", "-",
"#8722", "-",
"#8211", "-",
"#151", "-",
"#149", "-",
"rsquo", "'",
"#39", "'",
"#96", "'",
"#8217", "'",
"quot", "\"",
"#34", "\"",
"ldquo", "\"",
"rdquo", "\"",
"#8222","\"",
"#8221", "\"",
"laquo", "<<",
"raquo", ">>",
"uarr", "\24",
"darr", "\25",
"rarr", "\26",
"larr", "\27",
"#1028","\242",
"#1030","I",
"#1031","\244",
"#8470","N",
"percnt","%",
0};