Book Production Guide

Open the manuscript file in LibreOffice, save as DOCX.

Replace double linebreaks with paragraph breaks.

Remove non-breaking space if it has no purpose.

The pandoc Docx reader can extract images, deals better with trailing spaced italics, footnotes and produces cleaner headers.

In the most simple case:

pandoc -f docx -t latex --wrap=none trav.docx > trav.docx.tex

Use --extract-media=./images if needed.

More involved:

./convert.sh

#!/bin/bash

for i in $(ls -1 *.docx); do
    echo "$i"
    pandoc -f docx -t latex+footnotes -o "./tex/$i.tex" "$i"
done

# remove \label{}
sed -i 's/\\label{[^}]\+}//' ./tex/*.tex

# fix {[} and {]}
sed -i -e 's/{\[}/[/g; s/{\]}/]/g;' ./tex/*.tex

# remove roman numerals from section titles
sed -i 's/\\section{[IV]\+\. */\\section{/' ./tex/*.tex

# remove double blanks
for i in $(ls -1 ./tex/*.tex); do
    cat -s "$i" > "$i.tmp"
    mv "$i.tmp" "$i"
done

After this, copy in the tex contents to the repository.

Just get it to compile. Fix only major errors at this time, or redo the conversion.

Open the tex and copy chapters into separate files in ./manuscript/tex/, clean up only the header and subheaders for now.

Comment out image references.

Fix up the text just enough to get it compile.

Just add \chapter, leave quote formatting, etc. for later.

Inkscape is easy to use when experimenting with fonts, sizes and layout.

Open a previous PDF with Inkscape, import a typical page and start changing elements. ’Save a Copy’ will save a PDF, print to see how it looks on paper.

Consider:

• chapter and section headings
• appendix chapters
• backmatter chapters
• running header and footer
• TOC
• footnotes
• endnotes
• conversations: \speaker{Ajahn Chah}
• Q & A: …
• Q: …
• A: …

Get the book to compile with the basic layout with the fonts you want to use.

Open (not Import) the the default page in Inkscape as vectors to get basic layout sizes and positions.

Try design ideas, export to image and send to the author for review. Don’t export PDF, it doesn’t display reliably.

Try Internal Import first, use Replace fonts. It is useful to have pieces as text elements instead of paths.

Use Poppler Import if there are tricky marks or shapes on the page that the navive import produces artefacts for. Maybe you’ll have to select them and copy over for best results.

Import:

• chapter page
• verso page
• recto page
• TOC page

Use separate .svg files, each page will need different guides for margin alignments.

Add a border rectangle with 0.4pt stroke, and Save a Copy as PDFs for printing.

Look for bad semantic formatting syntax, fix the easy fixes.

Fix typos and conversion artefacts.

Go through the chapter files one by one and fix the easy fixes and immediately apparent errors.

Wrap thai in \thai:

rg '\p{Thai}' ./*.tex

0-2 letter italics:

rg '\\emph{[.{0,2}]}'

\footnote before full stop, comma, quotemark.

\footnote before closing quotemark.

Pali words in italics.

Full stop, quote, etc. not included in the italics.

Convert Glossary items to glossarydescription.

\begin{glossarydescription}

\item[anicca] (Pali) Impermanence: one of the \emph{three characteristics of
    existence} along with not-self (\emph{anattā}) and unsatisfactoriness
  (\emph{dukkha}).

\end{glossarydescription}

Review basic typeface settings for the design in prophecy.cls.

Don’t reflow the text in emacs (Alt+Q), to make it easier to diff direct pandoc output.

Look for {[} from the coversion:

pt -e "{\[}|{\]}"

Use neovim when using macros.

Search for speaker names:

/^ *([[:alpha:]']\+ [[:alpha:]']\+

Moving speaker names to parens with a macro (pasted here in nvim):

"adi(0df {O

% Speaker: “ap

Replace \textsuperscript{th} with th.

Replace quote environment with lpchah:

perl -0777 -i -pe "s/(% Speaker: Luang Por\n+)\\\\begin\{quote\}(.*?)\\\\end\{quote\}/\1\\\\begin{lpchah}\2\\\\end{lpchah}/gs" ./*.tex

Review with meld, it will show the changes since the last commit.

Find when Luang Por is not wrapped in an environment:

pcregrep -M '^% Speaker: Luang Por\n\s[^\s\\]' ./*.tex

Find speakers followed by and environment, should be all Luang Por:

pcregrep -M '^% Speaker: [^\n]+\n\s\\begin' ./*.tex | grep -v 'Speaker: Luang Por' | grep -v 'lpchah' | uniq

Fix closing parens in \emph:

(\emph{Upasampadā)} to (\emph{Upasampadā})

perl -0777 -i -pe "s/(\(\\\\emph\{[^\\\\\)}]+)\)\}/\1})/gs" ./*.tex

Fix quote marks included in emph:

indicated by the old Thai monastic idiom referring to it: `\emph{prot
sat',} or `saving sentient beings'. Seeing the monks' restrained and

`\emph{Attā hi attano nātho.' `}

grep -E '[[:punct:]]\}' ./*.tex

Normalize \emph{} jumbles.

\emph{Mah}ā\emph{nik}ā\emph{ya} and \emph{Dhammayut nik}ā\emph{ya}
(\emph{pa}ṭ\emph{ipatti})

It will need two or more passes.

perl -0777 -i -pe "s/(\\\\emph\{[ [:alpha:]āīūṃṅñṭḍṇḷĀĪŪṂṄÑṬḌṆḶ]+)\}([[:alpha:]āīūṃṅñṭḍṇḷĀĪŪṂṄÑṬḌṆḶ]+)\\\\emph\{([ [:alpha:]āīūṃṅñṭḍṇḷĀĪŪṂṄÑṬḌṆḶ]+\})/\1\2\3/g" ./*.tex

Any more of this mess?

pcregrep '\}[-[:alpha:]āīūṃṅñṭḍṇḷĀĪŪṂṄÑṬḌṆḶ]+\\' ./*.tex

Find cases of 0-2 characters in \emph. Note that \emph{I} is legitimate.

grep -E '\\emph{.{0,2}}' ./*.tex

Remove emph from quotes, parens, commas:

\emph{.}
\emph{,}
\emph{'}
\emph{(`}
\emph{--}
\emph{. }
\emph{(}white-robed nuns)

perl -0777 -i -pe "s/\\\\emph\{([ ,:\.\'\`\(\)-]{0,2})\}/\1/g" ./*.tex

Find footnotes before punctuation.

Here we go again. Another \emph{kappatī}\footnote{A Pali
  word from the mealtime blessing. Luang Por's mnemonic for `a cup of
  tea'.}.

All this is called \emph{vitakka}\footnote{In this paragraph Luang Por
  explains the five mental states that define the first state of deep
  absorption concentration (\emph{jhāna}).}: raising something up

pcregrep -M '\\footnote\{[^\n]+\}[[:punct:]]' ./*.tex

Don’t italicize the parens wrapping the terms, unless already in an italic block. Follow Bringhurst, not Chicago. Use the form:

(\emph{nimitta})

https://tex.stackexchange.com/questions/13048/upright-parentheses-in-italic-text

https://tex.stackexchange.com/questions/14983/should-i-include-punctuation-marks-in-emph-or-should-i-place-those-marks-after

https://english.stackexchange.com/questions/9878/italics-and-punctuation

Find and fix terms with italicized parens:

grep -E '\\emph\{\(' ./*.tex

\emph{(MN 112)}
\emph{(nimitta)}
\emph{(parikamma-nimitta).}
\emph{(uggaha-nimitta).}
\emph{(paṭibhāga-nimitta).''}

perl -0777 -i -pe "s/\\\\emph\{\(([- \.0-9[:punct:][:alpha:]āīūṃṅñṭḍṇḷĀĪŪṂṄÑṬḌṆḶ]+)\)([^\}]*)\}/(\\\\emph{\1})\2/g" ./*.tex

Find Thai and wrap it in \thai{} to set the appropriate font.

pt -e '\p{Thai}' ./*.tex

Find weird whitespace.

grep -f ~/prods/weird-spaces ./*.tex

Find triple full stops instead of ellipses.

grep -E '\.\.\.' ./*.tex

Fix quotes ending in triple ''' with \thinspace. Otherwise you get ”’, when you need ’ ”.

grep -E "'''" ./*.tex

moment, looking at the babies, then laughed quietly to himself, `Oh
well, oh well.'\thinspace ''

Other bad quote patterns:

``'Yes, they all die.' I answered irritably.''
```Yes.' I replied sheepishly.''
good. `\emph{Attā hi attano nātho.' `}It's the self that is the refuge

Convert all smart-quote characters to LaTeX equivalents to avoid ambiguity in pattern matching.

sed -i -e "s/‘/\`/g; s/’/'/g; s/“/\`\`/g; s/”/''/g;" ./*.tex

Find un-smart single and double quotes

grep -E " '{1,2}\w" ./*.tex
sed -i "s/ '\(\w\)/ \`\1/g" ./*.tex

grep -E ' "{1,2}\w' ./*.tex
sed -i 's/ "\(\w\)/ \`\`\1/g' ./*.tex

grep -E '\w"{1,2}[^[:alpha:]-]' ./*.tex
sed -i "s/\(\w\)\"\([^[:alpha:]-]\)/\1''\2/g" ./*.tex

Or, if needing the unicode quotation marks instead, as in the case of ebooks:

sed -i -f ../../helpers/smart_quotes_unicode.sed ./*.adoc

Find - (minus) instead of – (en-dash)

grep -E ' - ' ./*.tex
grep -E '[^[:alpha:]0-9-]-[^[:alpha:]0-9-]' ./*.tex

Find places where nested quotes don’t alternate single and double quote marks.

pcregrep -M "[^\`]\`[^\`][^']*[^\`]\`[^\`]" ./*.tex

Review the spaces before and after \ldots{}

pcregrep -M "[^ ]\\\\ldots{} " ./*.tex

Design complex covers in Inkscape.

Simple covers can be OK to implement in LaTeX.

fixing spacing and italics in quoted passges

• check the tracking amount (\MakeUppercase, chapter, section title)

NOTE: Use regex on *.tex, in the manuscript/tex folder.

NOTE: Commit all changes before applying regex subs, so that you can revert by git checkout.

cd manuscript/tex

Multiple quotes:

rg -e "```|'''"

Keep an eye out for footnote positions and punct. marks.

• \ldots\ for ellipis instead of ...

sed -i 's/\.\.\./\\\\ldots{}/g' *.tex

• check weird spaces (tabs, zero-width space, etc)

grep -f ~/prods/books/manuscript-reference/weird-spaces.txt *.tex

Also check tabs:

rg -e '\\t' ./manuscript/**/*.tex

is your editor inserting tabs in quotes and multiline {} arguments like footnotes? some others might have been a conversion artifact from Word

rg -e '[{][^0-9\[\]]}[}]' ./manuscript/tex-pt/*.tex

• remove bold – probably a conversion artifact from Word

como o ``\emph{\textbf{kalyāṇamitta''}} ou ``bom amigo''\footnote{``\emph{Kalyāṇa}''

rg textbf

• remove thinspace shorthand \,

Veja o trecho ``\,'O mais lendário

Mas quanto tempo é ``razoável''\,'?
Mas quanto tempo é “razoável” ’?

• quote marks outside of braces ``\emph{ācariya}''

# ``\emph{mitta''}
sed -i "s/``\\\\emph[{]\\([^}]\\+\\)''[}]/``\\\\emph{\\1}''/g" *.tex

# \emph{``mitta''}
sed -i "s/\\\\emph[{]``\\([^}]\\+\\)''[}]/``\\\\emph{\\1}''/g" *.tex

# \emph{``mitta}''
sed -i "s/\\\\emph[{]``\\([^}]\\+\\)[}]''/``\\\\emph{\\1}''/g" *.tex

• thinspace

# Follow thinspace with a space
sed -i 's/\\\\thinspace */\\\\thinspace /g' *.tex

• fix straight quotes, double and single, at the beginning and ending of quotes

# double quote after thinspace
sed -i 's/\\\\thinspace "/\\\\thinspace \'\'/g' *.tex

# closing double quotes followed by punctuation, but not \ as in ="\emph{}"=, but yes as in ="peace",= .
sed -i 's/"\([.,:;!?]\)/\'\'\1/g' *.tex

# opening double quotes at the beginning of the line
sed -i 's/^"/``/' *.tex

# closing double quotes at the end of the line
sed -i 's/"$/\'\'/' *.tex

# all remaining opening double
sed -i 's/"\\b/``/' (find -iregex './[0-9].*\.tex')

# all remaining closing double
sed -i 's/\\b"/\'\'/' (find -iregex './[0-9].*\.tex')

Now single quotes, only need to fix opening mark to backtick:

# opening single quotes at the beginning of the line
sed -i 's/^\'/`/' *.tex

# all remaining opening single, but not within words as in "don't"
sed -i 's/\(\\W\)\'\(\\w\)/\1`\2/g' (find -iregex './[0-9].*\.tex')

• move quote mark out of braces `\emph{Tudong'}

sed -i 's/\([{][^`}]\+\)\'[}]/\1}\'/g' (find -iregex './[0-9].*\.tex')

./helpers/chapters_to_asciidoc.sh

It runs tex_to_asciidoc.sh on the numbered files in manuscript/tex/[0-9][0-9]*.tex.

Additional preface.tex and other files have to be added manually there.

tex_to_asciidoc.sh will check for LaTeX commands that are not recognized.

If pandoc can convert it, add the command to pandoc_known_latex.

Otherwise we have to convert it to some kind of canonical LaTeX which pandoc can then handle.

Add replacement rules in canonicalize_latex.sh and add it to the list in replace_known_latex.

Editing the LaTeX temporarily, or copy everything to a tex_for_adoc folder, for the conversion to pass, is also good. For example footnotes in headings are awkward in ebooks, it makes more sense to move the text of a \chapterFootnote to the end of the chapter, separated with a \sectionBreak. After the conversion, revert the .tex chapter.

The conversion doesn’t have to represent a perfect mapping of LaTeX to Asciidoc. It is just a step to establish the content of the ebooks.

The conversion only has to complete once. Modifications (typos, etc.) later on are best to add by hand.

Web version cover must include image, not PDF

run epubcheck

open with Sigil and F7

Don’t transform to sRGB, because the slightly blueish CMYK will be bluish in the sRGB. Also, Gray does not become sRGB.

Re-save them in GIMP instead, which will import the CMYK with applying the profile better, and the colors will be better grays.

The -quality 90 will produce larger files than the 90 quality setting in GIMP. That’s fine, GIMP can re-compress. Don’t use -compress LossLess -quality 100, it can make the format unreadable for some jpeg libraries.

Covert the photos:

for i in photos/300dpi/*.jpg; do echo "$i"; convert "$i" -resize 600x800\> -compress jpeg -quality 90 ./photos/ebook-600x800-sRGB/`basename $i`; done

• rename to -sGRB.jpg with wdired
• open it all in GIMP, re-export with settings: quality 90, 4x4x4 subsampling (best)
• crop the Aj Chah front photo (sitting)
• crop the Aj Chah back photo (smiling)
• replace the illustration photos with the screenshot captures from the PDF in ebook-92dpi-sRGB
• resize them to height 800px with GIMP
• copy quotebreak.png
• copy sectionbreak.png
• copy webcover.jpg

Check color spaces are all sRGB:

for i in ./*.jpg; do echo -n "$i ... "; identify -format '%[colorspace]\n' $i; done