- Dillo-dev - lists.mailman3.com

dillo CSS and charset

by furaisanjin＠gmail.com

Hi, 2009/1/22 Hofmann Johannes <Johannes.Hofmann(a)gmx.de>: > On Thu, Jan 22, 2009 at 09:18:00PM +0900, furaisanjin wrote: > However looking at japanese pages, it seems that word wrapping is > still an issue, see: > http://lists.auriga.wearlab.de/pipermail/dillo-dev/2008-November/005427.html > > Do you have an idea about a possible solution? No, I don't think there is good solution but I added these code in html.cc. --- a/src/html.cc Thu Jan 22 08:31:44 2009 -0300 +++ b/src/html.cc Thu Jan 22 21:33:56 2009 +0900 @@ -58,6 +58,8 @@ #define TAB_SIZE 8 +#define UTF8_NONE_1WORD(x) ((x)&0x80) + /*----------------------------------------------------------------------------- * Name spaces *---------------------------------------------------------------------------*/ @@ -111,6 +113,7 @@ DilloImage *image); static void Html_callback(int Op, CacheClient_t *Client); static void Html_tag_cleanup_at_close(DilloHtml *html, int TagIdx); +static int Html_utf8_size (char c); /*----------------------------------------------------------------------------- * Local Data @@ -1117,6 +1120,7 @@ { int i, j, start; char *Pword, ch; + int utf8size; DilloHtmlParseMode parse_mode = S_TOP(html)->parse_mode; if (parse_mode == DILLO_HTML_PARSE_MODE_STASH || @@ -1159,7 +1163,20 @@ } else { if (!memchr(word,'&', size)) { /* No entities */ - DW2TB(html->dw)->addText(word, html->styleEngine->wordStyle ()); + for (start = 0; start < (int) strlen (word); ) { + if (UTF8_NONE_1WORD(word[start])) { + utf8size = Html_utf8_size (word[start]); + DW2TB(html->dw)->addText(dStrndup(&word[start],utf8size), html->styleEngine->wordStyle()); + start += utf8size; + } else { + for (i=0; i < (int) strlen (word)-start; i++) { + if (UTF8_NONE_1WORD(word[start+i])) + break; + } + DW2TB(html->dw)->addText(dStrndup(&word[start],i), html->styleEngine->wordStyle()); + start += i; + } + } } else { /* Collapse white-space entities inside the word (except  ) */ Pword = a_Html_parse_entities(html, word, size); @@ -1167,7 +1184,20 @@ if (strchr("\t\f\n\r", Pword[i])) for (j = i; (Pword[j] = Pword[j+1]); ++j) ; - DW2TB(html->dw)->addText(Pword, html->styleEngine->wordStyle ()); + for (start = 0; start < (int) strlen (Pword); ) { + if (UTF8_NONE_1WORD(Pword[start])) { + utf8size = Html_utf8_size (Pword[start]); + DW2TB(html->dw)->addText(dStrndup(&Pword[start],utf8size), html->styleEngine->wordStyle()); + start += utf8size; + } else { + for (i=0; i < (int) strlen (Pword)-start; i++) { + if (UTF8_NONE_1WORD(Pword[start+i])) + break; + } + DW2TB(html->dw)->addText(dStrndup(&Pword[start],i), html->styleEngine->wordStyle()); + start += i; + } + } dFree(Pword); } } @@ -3700,4 +3730,20 @@ return token_start; } +static int Html_utf8_size (char c) +{ + if ((c&0xfc) == 0xfc) + return 6; + if ((c&0xf8) == 0xf8) + return 5; + if ((c&0xf0) == 0xf0) + return 4; + if ((c&0xe0) == 0xe0) + return 3; + if ((c&0xc0) == 0xc0) + return 2; + return 1; +} + Then dillo appearance is something I mentioned here. http://lists.auriga.wearlab.de/pipermail/dillo-dev/2008-November/005430.html By the way I notice bookmark (ctrl-b) doesn't work. Does anybody has this problem? Regards, furaisanjin

Jan. 22, 2009

dillo CSS and charset

by Johannes.Hofmann＠gmx.de

On Thu, Jan 22, 2009 at 09:18:00PM +0900, furaisanjin wrote: > Hi. > > And charset conversion works fine. BTW. now you can set vw_fontname in dillorc again. This font will then be used when CSS specifies "sans" as font, e.g: vw_fontname="VL Gothic" However looking at japanese pages, it seems that word wrapping is still an issue, see: http://lists.auriga.wearlab.de/pipermail/dillo-dev/2008-November/005427.html Do you have an idea about a possible solution? Cheers, Johannes

Jan. 22, 2009

dillo CSS and charset

by furaisanjin＠gmail.com

Hi. And charset conversion works fine. Regards, furaisanjin

Jan. 22, 2009

dillo CSS and charset

by Johannes.Hofmann＠gmx.de

On Thu, Jan 22, 2009 at 08:34:32AM -0300, Jorge Arellano Cid wrote: > On Thu, Jan 22, 2009 at 10:22:39AM +0100, Hofmann Johannes wrote: > > Hi, > > > > On Wed, Jan 21, 2009 at 07:06:06PM -0300, Jorge Arellano Cid wrote: > > > On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > > > > Hi. > > > > > > > > This is work around. > > > > > > This so called "work around" works very well. > > > > > > While working on this, I used the testcases in the following table: > > > > > > > > > load Reload Back > > > ---------------------------- > > > H__ OK OK OK H = has charset in HTTP header > > > HM_ OK OK OK M = has charset in META > > > H_C OK OK OK C = has remote CSS stylesheet > > > HMC OK OK OK _ = disabled > > > ___ OK OK OK > > > _M_ OK OK OK > > > __C OK OK OK > > > _MC OK OK OK > > > ---------------------------- > > > > > > Both patches pass all of them! > > > > > > I committed a similar approach that has a single code path for > > > calling a_Decode_charset_init(). > > > > > > Please test and comment. > > > > Starting with revision d29cdb5b842e I get repeatable crashes when > > clicking on the "Home" link in about:splash. > > Please try the current tip. Yes, it's fixed! Cheers, Johannes

Jan. 22, 2009

dillo CSS and charset

by jcid＠dillo.org

On Thu, Jan 22, 2009 at 10:22:39AM +0100, Hofmann Johannes wrote: > Hi, > > On Wed, Jan 21, 2009 at 07:06:06PM -0300, Jorge Arellano Cid wrote: > > On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > > > Hi. > > > > > > This is work around. > > > > This so called "work around" works very well. > > > > While working on this, I used the testcases in the following table: > > > > > > load Reload Back > > ---------------------------- > > H__ OK OK OK H = has charset in HTTP header > > HM_ OK OK OK M = has charset in META > > H_C OK OK OK C = has remote CSS stylesheet > > HMC OK OK OK _ = disabled > > ___ OK OK OK > > _M_ OK OK OK > > __C OK OK OK > > _MC OK OK OK > > ---------------------------- > > > > Both patches pass all of them! > > > > I committed a similar approach that has a single code path for > > calling a_Decode_charset_init(). > > > > Please test and comment. > > Starting with revision d29cdb5b842e I get repeatable crashes when > clicking on the "Home" link in about:splash. Please try the current tip. -- Cheers Jorge.-

Jan. 22, 2009

dillo CSS and charset

by Johannes.Hofmann＠gmx.de

Hi, On Wed, Jan 21, 2009 at 07:06:06PM -0300, Jorge Arellano Cid wrote: > On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > > Hi. > > > > This is work around. > > This so called "work around" works very well. > > While working on this, I used the testcases in the following table: > > > load Reload Back > ---------------------------- > H__ OK OK OK H = has charset in HTTP header > HM_ OK OK OK M = has charset in META > H_C OK OK OK C = has remote CSS stylesheet > HMC OK OK OK _ = disabled > ___ OK OK OK > _M_ OK OK OK > __C OK OK OK > _MC OK OK OK > ---------------------------- > > Both patches pass all of them! > > I committed a similar approach that has a single code path for > calling a_Decode_charset_init(). > > Please test and comment. Starting with revision d29cdb5b842e I get repeatable crashes when clicking on the "Home" link in about:splash. Cheers, Johannes Program received signal SIGBUS, Bus error. DilloHtml::getCurTagLineNumber (this=0x28778500) at html.cc:596 596 if (p[i] == '\n') (gdb) bt #0 DilloHtml::getCurTagLineNumber (this=0x28778500) at html.cc:596 #1 0x08063e38 in DilloHtml::bugMessage (this=0x28778500, format=0x80f8198 "HEAD section lacks the TITLE element\n") at html.cc:149 #2 0x08063f2a in Html_tag_close_head (html=0x28778500, TagIdx=32) at html.cc:1569 #3 0x080684de in Html_tag_cleanup_at_close (html=0x28778500, TagIdx=32) at html.cc:1306 #4 0x08068957 in DilloHtml::finishParsing (this=0x28778500, ClientKey=2) at html.cc:627 #5 0x0805a566 in Cache_process_queue (entry=0x287418c0) at cache.c:1153 #6 0x0805b378 in a_Cache_process_dbuf (Op=2, buf=0x0, buf_size=0, Url=0x28741c80) at cache.c:886 #7 0x0805c871 in a_Capi_ccc (Op=4, Branch=2, Dir=1, Info=0x2874fd60, Data1=0x0, Data2=0x0) at capi.c:618 #8 0x08058288 in a_Chain_fcb (Op=0, Info=0x287a3000, Data1=0x0, Data2=0x0) at chain.c:112 #9 0x08078cee in a_Dpi_ccc (Op=4, Branch=2, Dir=1, Info=0x2874f140, Data1=0x0, Data2=0x0) at dpi.c:671 #10 0x08058288 in a_Chain_fcb (Op=0, Info=0x287a3000, Data1=0x0, Data2=0x0) at chain.c:112 #11 0x0807969e in a_IO_ccc (Op=4, Branch=2, Dir=1, Info=0x2874f6c0, Data1=0x2874f8c0, Data2=0x0) ---Type <return> to continue, or q <return> to quit--- ---Type <return> to continue, or q <return> to quit--- at IO.c:418 #12 0x08079980 in IO_callback (io=0x2874f8c0) at IO.c:211 #13 0x080799ff in IO_fd_read_cb (fd=7, data=0x2) at IO.c:287 #14 0x080ca465 in fltk::wait () #15 0x080ca615 in fltk::run () #16 0x0804f03a in main (argc=1, argv=0xbfbff864) at dillo.cc:334 (gdb)

Jan. 22, 2009

[bug]: hang in Css_parse_delaration()

by onepoint＠starurchin.org

On Wed, Jan 21, 2009 at 09:41:45PM +0100, Hofmann Johannes wrote: > On Wed, Jan 21, 2009 at 04:49:27PM +0000, Jeremy Henty wrote: > > > > A simple example that hangs the CSS parser. > Please check current tip. It's fixed! > The page is still not looking pretty, That's OK. Dillo is not responsible for the ugliness of the world! ;-) Regards, Jeremy Henty

Jan. 22, 2009

dillo CSS and charset

by jcid＠dillo.org

On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > Hi. > > This is work around. This so called "work around" works very well. While working on this, I used the testcases in the following table: load Reload Back ---------------------------- H__ OK OK OK H = has charset in HTTP header HM_ OK OK OK M = has charset in META H_C OK OK OK C = has remote CSS stylesheet HMC OK OK OK _ = disabled ___ OK OK OK _M_ OK OK OK __C OK OK OK _MC OK OK OK ---------------------------- Both patches pass all of them! I committed a similar approach that has a single code path for calling a_Decode_charset_init(). Please test and comment. -- Cheers Jorge.-

Jan. 21, 2009

[bug]: hang in Css_parse_delaration()

by Johannes.Hofmann＠gmx.de

On Wed, Jan 21, 2009 at 04:49:27PM +0000, Jeremy Henty wrote: > > A simple example that hangs the CSS parser. The contents of the css > block are the critical thing: it doesn't seem to matter what the > selector is but changing either the parameter or the value inside the > block stops the hang. After aborting the stack trace varies, but > always seems to be inside Css_parse_declaration(). (I've attached > some example stacks.) > > The real-world page that triggered the bug: > > http://dialinf.wordpress.com/2009/01/21/ultraproducts-the-category-theoreti… Please check current tip. The page is still not looking pretty, but it no longer hangs. Cheers, Johannes

Jan. 21, 2009

[bug]: hang in Css_parse_delaration()

by onepoint＠starurchin.org

A simple example that hangs the CSS parser. The contents of the css block are the critical thing: it doesn't seem to matter what the selector is but changing either the parameter or the value inside the block stops the hang. After aborting the stack trace varies, but always seems to be inside Css_parse_declaration(). (I've attached some example stacks.) The real-world page that triggered the bug: http://dialinf.wordpress.com/2009/01/21/ultraproducts-the-category-theoreti… Regards, Jeremy Henty

Jan. 21, 2009