Looking for help, including consulting, with font selection
There are a number of historical bugzillas dealing with the font-selection-strategy issue. Given the age of some of them, I sent the following to j...@apache.org. I was also advised that it would be reasonable to post here ... Basis Technology Corp is looking to engage someone on a consulting basis to make some improvements to Apache FOP. The missing killer feature is the ability to render the string "A=E3=81=82" with the "A" from font "Candara" and the "=E3=81=82" from font "MS Gothic". We believe that capability we need is http://www.w3.org/TR/2001/REC-xsl-20011015/slice7.html#font-selection-strategy and that's not yet implemented in FOP. We are looking for someone who (a) has experience with the FOP code sufficient to implement this, and (b) had engagement with the community sufficient to refine to the point where it is acceptable to the community for inclusion. Please just send email to me (bimargul...@gmail.com). - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org
Re: Looking for help, including consulting, with font selection
We don't want to specify fonts char-by-char, we want them picked. On Mon, Jan 4, 2010 at 7:35 AM, Abel Braaksma wrote: > I'm under the impression that Benson means selection strategy based on > availability. Suppose the "A" is available in Candara, but the other > letters/symbols are not, and the needed font selection strategy is > "character by character" then the result must be similar to what you wrote, > but automatically so. According to > http://xmlgraphics.apache.org/fop/trunk/fonts.html#selection this is not > implemented yet. > > Note that "auto" (which is implemented) means "the selection criterion is > implementation defined". For FOP that means (afaik) that word boundaries and > element boundaries work and that the "largest portion of a text or word > that's available in a particular font" will be chosen. In the example above, > the Gothic font will be used, because that has the largest part of the > word/text. > > Changing to "character-by-character" means that each character is considered > individually. > > Naturally, if Candara would contain all capital letters only, choosing > character-by-character in the following: > > A=E3=81=82 > > then the rendering with character-by-character would be equivalent with the > following: > > A > = > E > 3=81=82 > > Obviously, either this or the original example from Benson cannot be > achieved with current means (unless you add some rather prolific XSLT 2.0 > preprocessing, i.e., give it a list of fonts and do the selection strategy > in the earlier processing step, using some home-brewed extension functions, > which could be a possible alternative while waiting for the implementation). > > Cheers, > Abel > > > Jason Harrop wrote: >> >> I don't have a font called "MS Gothic" on my XP / Word 2007 PC, so >> i've used "Century Gothic" instead, but subject to that, and assuming >> the relevant fonts are available, isn't this just: >> >> >> A >> =E3=81=82 >> >> >> with a config such as: >> >> >> true >> >> >> >> >> > weight="normal"/> >> >> >> > weight="normal"/> >> >> >> >> >> >> >> cheers .. Jason >> >> >> > > - > To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org > For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org > > - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org
Re: Looking for help, including consulting, with font selection
I see how I misunderstood your email. I will get you a test case. On Mon, Jan 4, 2010 at 7:46 AM, Abel Braaksma wrote: > Can you clarify? Because that's precisely what I wrote. Perhaps we have some > misunderstanding on what the font-selection-strategy means and/or what > "picked" means (who picks what by what criteria?). > > Can you provide a simple input (XML) output (XSL-FO) example that currently > renders incorrectly, but which is a little bit more elaborate then the > "A=E3=81=82" string example? I.e., how would you deal with the selection of > the font and what do you expect FOP to do for you? > > Benson Margulies wrote: >> >> We don't want to specify fonts char-by-char, we want them picked. >> >> On Mon, Jan 4, 2010 at 7:35 AM, Abel Braaksma >> wrote: >> >>> >>> I'm under the impression that Benson means selection strategy based on >>> availability. Suppose the "A" is available in Candara, but the other >>> letters/symbols are not, and the needed font selection strategy is >>> "character by character" then the result must be similar to what you >>> wrote, >>> but automatically so. According to >>> http://xmlgraphics.apache.org/fop/trunk/fonts.html#selection this is not >>> implemented yet. >>> >>> Note that "auto" (which is implemented) means "the selection criterion is >>> implementation defined". For FOP that means (afaik) that word boundaries >>> and >>> element boundaries work and that the "largest portion of a text or word >>> that's available in a particular font" will be chosen. In the example >>> above, >>> the Gothic font will be used, because that has the largest part of the >>> word/text. >>> >>> Changing to "character-by-character" means that each character is >>> considered >>> individually. >>> >>> Naturally, if Candara would contain all capital letters only, choosing >>> character-by-character in the following: >>> >>> A=E3=81=82 >>> >>> then the rendering with character-by-character would be equivalent with >>> the >>> following: >>> >>> A >>> = >>> E >>> 3=81=82 >>> >>> Obviously, either this or the original example from Benson cannot be >>> achieved with current means (unless you add some rather prolific XSLT 2.0 >>> preprocessing, i.e., give it a list of fonts and do the selection >>> strategy >>> in the earlier processing step, using some home-brewed extension >>> functions, >>> which could be a possible alternative while waiting for the >>> implementation). >>> >>> Cheers, >>> Abel >>> >>> >>> Jason Harrop wrote: >>> >>>> >>>> I don't have a font called "MS Gothic" on my XP / Word 2007 PC, so >>>> i've used "Century Gothic" instead, but subject to that, and assuming >>>> the relevant fonts are available, isn't this just: >>>> >>>> >>>> A >>>> =E3=81=82 >>>> >>>> >>>> with a config such as: >>>> >>>> >>>> true >>>> >>>> >>>> >>>> >>>> >>> weight="normal"/> >>>> >>>> >>>> >>> weight="normal"/> >>>> >>>> >>>> >>>> >>>> >>>> >>>> cheers .. Jason >>>> >>>> >>>> >>>> >>> >>> - >>> To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org >>> For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org >>> >>> >>> >> >> - >> To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org >> For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org >> >> >> >> > > - > To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org > For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org > > - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org
Re: Looking for help, including consulting, with font selection
Please see bz https://issues.apache.org/bugzilla/show_bug.cgi?id=48481. On Mon, Jan 4, 2010 at 7:46 AM, Abel Braaksma wrote: > Can you clarify? Because that's precisely what I wrote. Perhaps we have some > misunderstanding on what the font-selection-strategy means and/or what > "picked" means (who picks what by what criteria?). > > Can you provide a simple input (XML) output (XSL-FO) example that currently > renders incorrectly, but which is a little bit more elaborate then the > "A=E3=81=82" string example? I.e., how would you deal with the selection of > the font and what do you expect FOP to do for you? > > Benson Margulies wrote: >> >> We don't want to specify fonts char-by-char, we want them picked. >> >> On Mon, Jan 4, 2010 at 7:35 AM, Abel Braaksma >> wrote: >> >>> >>> I'm under the impression that Benson means selection strategy based on >>> availability. Suppose the "A" is available in Candara, but the other >>> letters/symbols are not, and the needed font selection strategy is >>> "character by character" then the result must be similar to what you >>> wrote, >>> but automatically so. According to >>> http://xmlgraphics.apache.org/fop/trunk/fonts.html#selection this is not >>> implemented yet. >>> >>> Note that "auto" (which is implemented) means "the selection criterion is >>> implementation defined". For FOP that means (afaik) that word boundaries >>> and >>> element boundaries work and that the "largest portion of a text or word >>> that's available in a particular font" will be chosen. In the example >>> above, >>> the Gothic font will be used, because that has the largest part of the >>> word/text. >>> >>> Changing to "character-by-character" means that each character is >>> considered >>> individually. >>> >>> Naturally, if Candara would contain all capital letters only, choosing >>> character-by-character in the following: >>> >>> A=E3=81=82 >>> >>> then the rendering with character-by-character would be equivalent with >>> the >>> following: >>> >>> A >>> = >>> E >>> 3=81=82 >>> >>> Obviously, either this or the original example from Benson cannot be >>> achieved with current means (unless you add some rather prolific XSLT 2.0 >>> preprocessing, i.e., give it a list of fonts and do the selection >>> strategy >>> in the earlier processing step, using some home-brewed extension >>> functions, >>> which could be a possible alternative while waiting for the >>> implementation). >>> >>> Cheers, >>> Abel >>> >>> >>> Jason Harrop wrote: >>> >>>> >>>> I don't have a font called "MS Gothic" on my XP / Word 2007 PC, so >>>> i've used "Century Gothic" instead, but subject to that, and assuming >>>> the relevant fonts are available, isn't this just: >>>> >>>> >>>> A >>>> =E3=81=82 >>>> >>>> >>>> with a config such as: >>>> >>>> >>>> true >>>> >>>> >>>> >>>> >>>> >>> weight="normal"/> >>>> >>>> >>>> >>> weight="normal"/> >>>> >>>> >>>> >>>> >>>> >>>> >>>> cheers .. Jason >>>> >>>> >>>> >>>> >>> >>> - >>> To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org >>> For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org >>> >>> >>> >> >> - >> To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org >> For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org >> >> >> >> > > - > To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org > For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org > > - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org
Re: Looking for help, including consulting, with font selection
Does this add up to a reason to reopen the bz? On Mon, Jan 4, 2010 at 11:19 AM, Pascal Sancho wrote: > Abel Braaksma a écrit : >> >> To recap: apparently the problem was that for individual letters not >> available in a font, the wrong selection was made, resulting in the >> question mark (for missing letter) appearing. >> > > No, FOP replaces missing characters "#" character. I suspect a wrong > character selection >> >> While this was apparently fixed, note that character-by-character is >> still not supported. Meaning, in "AXB", where X is a Japanese (or other) >> character not available in font Candara (or other), but available in MS >> PMincho, it will still display as a question mark. Change it to "A X B" >> and it should work. >> >> If your only issue is with Japanese text, this can easily be solved in >> the XSLT step (using codepoints ranges), in case you have this type of >> exceptional cases in your text. >> >> Good to know that the issue is solved, Pascal. >> >> -- Abel -- > > Pascal > > - > To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org > For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org > > - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org
Re: Looking for help, including consulting, with font selection
Simon, The claim is that it's fixed on trunk. I did a trunk build, and my colleague who started all this will test it. If his test fails, I'll reopen it. --benson On Mon, Jan 4, 2010 at 3:12 PM, Simon Pepping wrote: > Pascal and Abel, > > Why was this bug closed? I believe it is Benson's intention to report > a known issue. He is looking to get it resolved and is willing to pay > for the effort. Unless Benson is raising a non-existent problem, I am > happy with his initiative. And I do not have the impression that it is > a non-existent problem. > > Benson, > > Can you submit the resulting XSL-FO file, and possibly the resulting > HTML file for comparison of the intended result? I have a problem > running your transformations, and I do not wish to debug my XSLT2 > system now. > > Regards, Simon > > On Mon, Jan 04, 2010 at 06:41:00PM +0100, Abel Braaksma wrote: >> It concludes what was said. And Pascal's remark is about the issue >> that was solved. But you try it with FOP Trunk and decide whether it >> is sufficiently solved. If not, provide a reproducible example >> (reproducible with the latest build) . >> >> My talk about character-by-character is a known issue and is >> unrelated to your bug report and is sufficiently covered on other >> places. >> >> Benson Margulies wrote: >> >Does this add up to a reason to reopen the bz? >> > > > -- > Simon Pepping > home page: http://www.leverkruid.eu > > - > To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org > For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org > > - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org
Stack Overflow Question
Apache FOP - strange image in PDF export http://stackoverflow.com/q/11037960/131433?sem=2 - To unsubscribe, e-mail: fop-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: fop-users-h...@xmlgraphics.apache.org