Re: [docbook-apps] Odd Characters in UTF-8, Docbook 5.1

Hi,

 

Thank you everyone! With the info on that fop configuration for custom setup I was able to get fop to actually use the xconf file....so much trouble because I had thought fop had been using the standard system configuration file by default, but it does not. Edits to use the required font mentioned in earlier replies now works...it just goes to show...be careful what you edit, it might not be the file being used! Fonts now work as they should and include the previously missing glyphs.

 

The quick workaround is to add the "-c /usr/share/fop/conf/fop.xconf". That custom font script though adds a lot to making sure doc builds are more predictable.

 

I think I do need to ask the fop people why the dump of current configuration details fails. If this had been working I might have seen earlier that the expected xconf file was not used...I'm still curious as to where its configuration is really from. The base-14 fonts are standard so perhaps no config file was used at all and these are just built in...fop just doesn't want to tell me about its configuration.

 

Thanks!

 

----- Original Message -----
From: 
To: , 
Sent: Fri, 03 Feb 2017 01:38:19 -0000 (UTC)
Subject: Re: [docbook-apps] Odd Characters in UTF-8, Docbook 5.1

I use DejaVu fonts with FOP on both Windows and Linux.

 

I use my own config file via "-c configFile.xml" switch with FOPs command 
line to override the standard config. The integration with DejaVu fonts is 
pretty straight forward.

 

This may help you - attached is part of the script that I use on Linux to 
call FOP and also the configuration file.

 

You should be able to comment out the DejaVu parts of the config file and 
see your fonts return to default. If not, I would check the fop script that 
starts FOP.  (using 'which fop' to locate it)

 

Good luck.

 

Dean Nelson

 

 

In a message dated 2/2/2017 2:19:16 P.M. Pacific Standard Time, 
 writes:

This is turning out to be odd...I have to conclude that fop is not 
 reading any configuration files...especially none named "fop.xconf". The short 
 story is that I can intentionally introduce errors in fop.xconf (all 
 occurrences on the computer as verified by updatedb and locate .xconf), 
 introduce "<strict-configuration>true</strict-configuration>", and 
 no mention of anything different occurs. I can run this under strace, and I 
 see no calls at all referring to opening any xconf file (nor any attempt to do 
 so). Somehow I need to figure out how fop is configured since the installed 
 xconf files are ignored.

 

Here's a summary of how I go there:

 

The DejaVuSans fonts do contain the character glyphs, and this is what 
 I've been trying to switch to instead of Helvetica for the sans-serif.

 

It was determined that param.xsl picks between serif and sans-serif by 
 naming one of these:

<xsl:param name="body.font.family">serif</xsl:param>

<xsl:param 
 name="body.font.family">sans-serif</xsl:param>

 

When changing between serif and sans-serif vi param.xsl edits, it is 
 verified that that either Times or Helvetica are used upon switch. This works 
 as expected.

 

I added the fop option "-d" to debug...this did not indicate any error, 
 though it was more explicit about not finding glyphs. After that I tried the 
 strict-configuration modification. No change. I looked for any possible 
 occurrence of fop.xconf, these two exist (I don't know how they differ), I 
 intentionally tried to break both for warning information:

/usr/share/fop/conf/fop.xconf
/usr/share/publican/fop/fop.xconf
 

This had no effect, so this is when I tried strace. It seems fop does not 
 use external configuration file searches, and I don't know why...so I can't 
 actually add DejaVuSans. I'm currently trying to figure out where there is a 
 mailing list or other support for fop. Incidentally, the "fop -x" command to 
 dump configuration fails...it wants an input file or it won't tell me anything 
 about configuration.

 

Thanks everyone for the help!

 

----- Original Message -----
From: Dave Pawson 
 <>
To: Richard Hamilton 

<>
Cc: , Docbook 
 <>
Sent: Thu, 02 Feb 2017 06:37:36 
 -0000 (UTC)
Subject: Re: [docbook-apps] Odd Characters in UTF-8, Docbook 
 5.1

Perhaps start where the problem is. Look for a font which contains
the 
 glyphs for the Unicode code points you want to use. Check the
code points 
 produce the glyphs you want.
Find out how the app (fop and any others you 
 use in your tool chain)
selects fonts.
Run the app knowing that the font 
 has what you want and your code
points are correct.

HTH

On 1 February 2017 at 23:44, Richard Hamilton 
 <> wrote:
> I’m getting out of my depth on 
 fop, so maybe someone else on the list can help here.
>
> However, 
 diving in deeper than I should:-)I suspect the problem is number 
 1.
>
> I suggest running fop with the -d (debug) mode, and you 
 might also try -x, which will dump the configuration; that way you can see if 
 the configuration fop uses is what you think it is using.
>
> Good 
 luck,
> Dick
> -------
> XML Press
> XML for Technical 
 Communicators
> http://xmlpress.net
> 
 
>
>
>
>> On Feb 1, 2017, at 
 15:21,  wrote:
>>
>> 
 Hi,
>>
>> So what my experimenting shows so far for trying 
 to change font is that I can adjust param.xsl and switch between serif or 
 sans-serif. The result depends on what is available in fop.xconf which lists 
 Times-Roman for serif, and Helvetica for sans-serif. I've attempted to add an 
 additional font definition for DejaVuSans by the following inside the 
 <fonts> of fop.xconf:
>>
>> <!-- DejaVuSans 
 -->
>> <font 
 embed-url="">
>> 
 <font-triplet name="DejaVuSans" style="normal" 
 weight="normal"/>
>> </font>
>>
>> I am 
 unable to name "DejaVuSans" instead of "sans-serif" in param.xsl (it falls 
 back to Symbol). I see three possibilities...
>> 1. My DejaVuSans 
 declaration is wrong.
>> 2. I have to remove Helvetica in fop.xconf 
 from the sans-serif listings.
>> 3. The Docbook parameters you 
 mentioned are somewhere outside of fop.xconf or 
 param.xsl.
>>
>> It seems unlikely I need to remove 
 Helvetica, more likely there is another configuration elsewhere. However, I 
 don't see any occurrence of "Helvetica" in any of the other Docbook files 
 under the "stylesheet/docbook-xsl-ns-1.79.1/" directory tree which are related 
 to this case (for example, there is a subdirectory for "slides" which would 
 seem to be unrelated to the "book" being published to pdf). The full path to 
 the ttf font file is available and readable to everyone; this is verifiable 
 with a font editor (one reason to choose DejaVu is that the font editor 
 actually shows the desired box drawing font glyphs at index 0x2500 and 
 on).
>>
>> Am I missing another file to edit, or am I simply 
 using the wrong syntax for my addition of DejaVuSans?
>>
>> 
 Thanks!
>>
>> ----- Original Message -----
>> From: 
 Richard Hamilton <>
>> To: 
 
>> Cc: Docbook 
 <>
>> Sent: Wed, 01 Feb 2017 
 20:36:59 -0000 (UTC)
>> Subject: Re: [docbook-apps] Odd Characters in 
 UTF-8, Docbook 5.1
>> Hi,
>> There are two steps to the font 
 assignment process: 1) configuring fop’s fonts and 2) setting DocBook 
 parameters to call the fonts you have configured in fop.
>> In this 
 case, the mapping between serif and Times (and sans-serif and Helvetica) 
 happens in the fop configuration file, fop.xconf.
>> So, when you set 
 the DocBook parameter to sans-serif, fop defaults to Helvetica. If you want 
 sans-serif to default to something else, or if you want to add a font, you 
 need to change the configuration in fop.xconf. Once you have configured fop to 
 recognize a font, you can refer to it from the DocBook parameters using the 
 name you have associated with the font in fop.xconf. If you look in your 
 fop.xconf file, you will find that the definition for Times contains several 
 instances of the <font-triplet> element, each of which defines a name 
 you can use for that font, along with information about style and 
 weight.
>> This page has some information on configuring the 
 fop.xconf file for fonts:
>> 
 https://xmlgraphics.apache.org/fop/2.1/fonts.html
>> Best 
 regards,
>> Dick Hamilton
>> -------
>> XML 
 Press
>> XML for Technical Communicators
>> 
 http://xmlpress.net
>> 
>> > On Feb 
 1, 2017, at 11:53,  wrote:
>> >
>> 
 > Hi,
>> >
>> > I tested this by changing 
 body.font.family to sans-serif. The document did successfully switch, and used 
 Helvetica. Unfortunately, Helvetica also lacks the box drawing characters. Is 
 there a configuration file used for a given Docbook 5.1 installation which 
 maps "Times Roman" to serif, and "Helvetica" to sans-serif? I think this is 
 the key to getting those box drawing characters to work...perhaps if I can 
 remap serif and sans-serif to valid fonts other than Helvetica and Times Roman 
 it would work with those characters.
>> >
>> > 
 Thanks!
>> >
>> > ----- Original Message 
 -----
>> > From: Bob Stayton <>
>> 
 > To: , 
>> 
 > Sent: Wed, 01 Feb 2017 18:51:04 -0000 (UTC)
>> > Subject: Re: 
 [docbook-apps] Odd Characters in UTF-8, Docbook 5.1
>> > 
 Hi,
>> > If by system-wide you mean everyone sharing one DocBook 
 XSL
>> > installation, you could edit the parameters file for that 
 installation.
>> > Find the file named fo/param.xsl and change 
 these stylesheet parameters:
>> > <xsl:param 
 name="body.font.family">serif</xsl:param>
>> > 
 <xsl:param 
 name="title.font.family">sans-serif</xsl:param>
>> > All 
 of the stylesheet params are documented here:
>> > 
 http://docbook.sourceforge.net/release/xsl/current/doc/fo/index.html
>> 
 >
>> > Bob Stayton
>> > Sagehill 
 Enterprises
>> > 
>> > On 2/1/2017 
 10:22 AM,  wrote:
>> > > I just wanted to 
 add some information from reading build logs for the
>> > > 
 document. It turns out the error shows up under "fop" because the
>> 
 > > particular font does not support this character (I'm 
 surprised
>> > > Times-Roman does not have this by default 
 since just about every
>> > > application on the system has the 
 character):
>> > > WARNING: Glyph "├" (0x251c, SF080000) not 
 available in font "Times-Roman".
>> > >
>> > > 
 The trouble here is I now want to change this for the system as a 
 whole,
>> > > not just on the document (I want to avoid any 
 project using
>> > > Times-Roman). I am curious if there is 
 some simple way under Linux to
>> > > find out which fonts fop 
 has available, and how to change the default
>> > > font family 
 for all Docbook publishing via a system-wide setting? I've
>> > 
 > found many documents on modifying Docbook behavior within source 
 code,
>> > > but not much seems to exist on system-wide Docbook 
 configuration from a
>> > > system administrator's point of 
 view...if anyone has a URL for
>> > > information on specifics 
 of tuning Docbook defaults under Linux I'd be
>> > > very 
 happy!
>> > >
>> > > Thanks!
>> > 
 >
>> > > ----- Original Message -----
>> > > 
 From: 
>> > > To: 
 
>> > > Sent: Tue, 31 Jan 2017 
 23:20:24 -0000 (UTC)
>> > > Subject: [docbook-apps] Odd 
 Characters in UTF-8, Docbook 5.1
>> > > Hi,
>> > 
 >
>> > > I'm working on a Linux machine which has UTF-8 
 encoding. The docbook 5.1
>> > > I'm working with is 
 declared:
>> > > <?xml version="1.0" 
 encoding="utf-8"?>
>> > > <!DOCTYPE book [
>> 
 > > <!ENTITY version "0.1 unreleased">
>> > > 
 ]>
>> > >
>> > > Entities changed a while 
 back (relaxng changed things), but so far as I
>> > > know I 
 should still be able to work with numeric entities when using odd
>> 
 > > characters which are part of UTF-8. I'm trying to get some box 
 drawing
>> > > characters to work, specifically because I need 
 to illustrate a file
>> > > directory structure which has been 
 output like this from the "tree -d"
>> > > command. Here's a 
 short sample:
>> > >> tree -d -L 2 /etc | tail -n 20 | tail 
 -n 6
>> > > │ ├── pluginconf.d
>> > > │ ├── 
 protected.d
>> > > │ └── vars
>> > > └── 
 yum.repos.d
>> > >
>> > > As you can see the 
 vertical and horizontal box drawing characters are
>> > > used. 
 When I redirect this to a file or copy and paste into the document
>> 
 > > all I get is a substitute...each shows as "#". So I tried to use 
 numeric
>> > > entities instead. As an example "boxh" 
 (horizontal box line) should work
>> > > as "&#x2500;". 
 This does not work, and rendering still ends up as "#". I
>> > 
 > tried the numeric code for the plus/minus character, "|&#x00B1;|", 
 and
>> > > this does work. I don't know if it proves anything, 
 but since my UTF-8
>> > > terminal shows all of the above 
 characters just fine on a terminal, it
>> > > seems like the 
 ability to output is not in question...perhaps this is an
>> > 
 > invalid assumption.
>> > >
>> > > Btw, one 
 reference on the Unicode numeric value is from:
>> > > 
 http://www.oasis-open.org/docbook/documentation/reference/html/iso-box.html
>> 
 > >
>> > > Under Docbook 5.1, what do I need to do to use 
 this numeric entity other
>> > > than using UTF-8 and the 
 numeric entity syntax? Is there an additional
>> > > XML 
 declaration I need?
>> > >
>> > > 
 Thanks!
>
>
> 
 ---------------------------------------------------------------------
> 
 To unsubscribe, e-mail: 
> 
 For additional commands, e-mail: 
 
>

-- 
Dave Pawson
XSLT XSL-FO FAQ.
Docbook 
 FAQ.
http://www.dpawson.co.uk