#!/bin/bash

# This script checks that the LaTeX sources stick to the rules.

failed=0

# Ignore files where rules may be violated within macro definitions.
texfiles=$(ls *.tex | grep -v macros.tex | grep -v layout.tex | grep -v tables.tex)
texlibdesc="support.tex concepts.tex diagnostics.tex memory.tex meta.tex utilities.tex containers.tex iterators.tex ranges.tex algorithms.tex strings.tex text.tex numerics.tex time.tex iostreams.tex threads.tex exec.tex"
texlib="lib-intro.tex $texlibdesc"

# Filter that reformats the error message as a "workflow command",
# for native handling by github actions.
# Prefixes each line of input with "$*: ".
function fail() {
    # echo "::error file=app.js,line=10,col=15::Something went wrong"

    # For some reason, the file/line/col information is not shown in the GUI.
    # Make sure to leave it in the message proper.
    ! sed 's@^\(.\+\.tex\):\([0-9]\+\):@::error file=source/\1,line=\2::\1:\2:'"$*"': @; s@^\([^:][^:]\)@::error ::'"$*"': \1@' |
    grep .
}


# We require GNU tools.
sed --version | grep -Fqe "GNU sed" || { echo "sed is not GNU sed"; exit 1; }

# Find non-ASCII (Unicode) characters in the source
LC_ALL=C grep -ne '[^ -~	]' *.tex |
    fail 'non-ASCII character' || failed=1

# Trailing whitespace in a line.
grep -ne '\s$' *.tex |
    fail 'trailing whitespace' || failed=1

# Trailing empty lines
for f in *.tex; do
    [ $(tail -c 2 $f | wc -l) -eq 1 ] ||
	echo "$f" | fail 'trailing empty lines' || failed=1
done

# indented \begin{codeblock} / \end{codeblock} (causes unwanted empty space)
grep -ne '^.\+\\\(begin\|end\){codeblock}' $texfiles |
    fail 'indented codeblock env' || failed=1

# \pnum not alone on a line.
grep -ne '^[^%]\+\\pnum' $texfiles |
    fail "pnum not alone on line" || failed=1
grep -ne '\\pnum.\+$' $texfiles |
    fail "pnum not alone on line" || failed=1
# Fixup: sed '/\\pnum.\+$/s/\\pnum\s*/\\pnum\n/'

# Two consecutive \pnum
for f in $texfiles; do
    awk 'prev == $0 && /^\\pnum/ { print FILENAME ":" FNR ":" } { prev = $0 }' $f
done |
    fail 'two consecutive \\pnum' || failed=1

# punctuation after the footnote marker
grep -n '\\end{footnote' $texfiles | grep -v '}[@)%]\?$' |
    fail "punctuation after footnote marker" || failed=1

# \opt used incorrectly.
grep -n '\\opt[^{]' $texfiles |
    fail '\\opt used incorrectly' || failed=1
grep -n 'opt{}' *.tex |
    fail '\\opt used incorrectly' || failed=1

# Use \expos insted of "exposition only"
grep -n "// exposition only" $texfiles |
    fail 'use \\expos instead' || failed=1

# Use \notdef instead of "not defined".
grep -n "// not defined" $texfiles |
    fail "use \\notdef instead" || failed=1

# Use \Cpp{} instead of C++
grep -n '^[^%]*[^{"]C++[^"}]' $texfiles |
    fail 'use \\Cpp{} instead' || failed=1

# Use \caret instead of \^
grep -F -n '\^' $texfiles |
    fail 'use \\caret instead' || failed=1

# Use \unicode instead of U+nnnn
grep -n 'U+' $texfiles |
    fail 'use \\unicode or \\ucode or \\uname instead' || failed=1

# Discourage double-wrapping \tcode{\exposid{data_}}
grep -n '\\tcode{\\exposid{[a-zA-Z0-9_]*}}' $texfiles |
    fail 'double-wrapped \\exposid in \\tcode' || failed=1

# Hex digits inside \ucode and \unicode must be lowercase so that \textsc works
grep -n 'ucode{[^}]*[^0-9a-f}][^}]*}' $texfiles |
    fail 'use lowercase hex digits inside \\ucode' || failed=1
grep -n 'unicode{[^}]*[^0-9a-f}][^}]*}' $texfiles |
    fail 'use lowercase hex digits inside \\unicode' || failed=1

# Use \iref instead of "(\ref", except for subclause ranges
grep -n '.(\\ref{' $texfiles  | grep -v -- "--" |
    fail 'use \\iref instead of (\\ref' || failed=1

# \iref cannot be at the start of a line
grep -n '^\\iref' $texfiles |
    fail '\\iref must be flush against the preceding word, not at the start of a line' || failed=1

# \iref cannot be preceded by space
grep -n ' \\iref' $texfiles |
    fail '\\iref must not be preceded by space' || failed=1

# Use \xrefc instead of "ISO C x.y.z"
grep -n "^ISO C [0-9]*\." $texfiles |
    fail 'use \\xrefc instead' || failed=1

# Library element introducer followed by stuff.
grep -ne '^\\\(constraints\|mandates\|expects\|effects\|sync\|ensures\|returns\|throws\|complexity\|remarks\|errors\).\+$' $texlibdesc |
    fail 'stuff after library element' || failed=1
# Fixup: sed 's/^\\\(constraints\|mandates\|expects\|effects\|sync\|ensures\|returns\|throws\|complexity\|remarks\|errors\)\s*\(.\)/\\\1\n\2/'
# Fixup: sed 's/^\\ //'

# Order of library elements.
../tools/element-order.awk *.tex |
    fail 'incorrect ordering of library elements' || failed=1

# Change marker in [diff] followed by stuff.
grep -Hne '^\\\(change\|rationale\|effect\|difficulty\|howwide\)\s.\+$' compatibility.tex |
    fail "change marker in [diff] followed by stuff" || failed=1
# Fixup: sed 's/^\\\(change\|rationale\|effect\|difficulty\|howwide\)\s\(.\)/\\\1\n\2/'q

# "template <" (with space) in library clause.
grep -ne 'template\s\+<' $texlib |
    fail 'space between "template" and "<"' || failed=1

# In library declarations, constexpr should not follow explicit
grep -ne '\bexplicit\b.*\bconstexpr\b' $texlib |
    fail 'wrong order: explicit constexpr' || failed=1

# In library declarations, static should not follow constexpr
grep -ne '\bconstexpr\b.*\sstatic\s' $texlib |
    fail 'wrong order: constexpr static' || failed=1

# In library declarations, type aliases should not use typename
grep -ne "using.*= typename" $texlib |
    fail 'type alias with typename' || failed=1

# "Class" heading without namespace
for f in $texlib; do
    sed -n '/rSec[0-9].*{Class/,/\\end{codeblock}/{/\\begin{example}/,/\\end{example}/b;/\\begin{codeblock}/,/\(^namespace\)\|\(\\end{codeblock}\)/{s/template<[^>]*>//;/\(class\|struct\)[A-Za-z0-9_: ]*{/{=;p;};};}' $f |
    # prefix output with filename and line
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/"
done |
    fail 'No namespace around class definition' || failed=1

# ref-qualifier on member functions with no space, e.g. "const&"
grep -F -ne ') const&' $texlib |
    fail 'no space between cv-qualifier and ref-qualifier' || failed=1

grep -n '\\\(def\)\?\(lib\|expos\)concept{[a-z0-9_-]*[^a-z0-9_}-][a-z0-9_-]*}' $texlib |
    fail 'bad concept name' || failed=1

# \begin{example/note} with non-whitespace in front on the same line.
grep -ne '^.*[^ ]\s*\\\(begin\|end\){\(example\|note\)}' $texfiles |
    fail "non-whitespace before note/example begins" || failed=1
# Fixup: sed 's/^\(.*[^ ]\)\s*\(\\\(begin\|end\){\(example\|note\)}\)/\1\n\2/'

# \begin/end{example/note} with stuff (except %) following on the same line.
grep -ne '\\\(begin\|end\){\(example\|note\)}[^%]\+$' $texfiles |
    fail "content following note/example env" || failed=1
# Fixup: sed 's/\(\\\(begin\|end\){\(example\|note\)}\)\s*\([^ ].*\)$/\1\n\4/'

# \end{note} or \end{example} at the end of a table cell
grep -n -A1 '\\end{\(example\|note\)}' $texfiles | grep -- '- *\(\\\\\|&\)' |
    fail "needs tailnote or tailexample" || failed=1

# Blank line between "begin example" and "begin codeblock"
for f in $texfiles; do
    sed -n '/\\begin{example}/{N;N;/\n\n\\begin{codeblock}$/{=;p;};}' $f |
    # prefix output with filename and line
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/"
done |
    fail 'blank line between "begin example" and "begin codeblock"' || failed=1
# Fixup: sed '/\\begin{example}/{N;s/\n$//;}'

# Comment not aligned to multiple of four. (Ignore lines with "@".)
for f in $texfiles; do
    sed -n '/begin{codeblock\(tu\)\?}/,/end{codeblock\(tu\)\?}/{/^[^@]*[^ @][^@]*\/\//{=;p;};}' $f |
    # prefix output with filename and line
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/" |
    awk '{ match($0,"^[-a-z0-9]*[.]tex:[0-9]*:"); n=match(substr($0,RLENGTH+1),"[ ;]//"); if (n % 4 != 0) print "comment starts in column " n ": " $0; }'
done |
    fail "comment not aligned to multiple of 4" || failed=1

# Deleted special member function with a parameter name.
grep -n "&[ 0-9a-z_]\+) = delete" $texfiles |
    fail 'named parameter in deleted special member' || failed=1
# to fix: sed '/= delete/s/&[ 0-9a-z_]\+)/\&)/'

# Bad characters in label.
grep -n '^\\rSec.\[[^]]*[^a-z.0-9][^]]*\]{' $texfiles |
    fail 'bad character in label' || failed=1

# Use of parenthesized \ref
grep -n '[ ~](\\ref{[.a-z0-9]*})' $texfiles |
    fail 'replace \\ref with \\iref' || failed=1

# "shall", "may", or "should" inside a note
for f in $texfiles; do
    sed -n '/begin{\(note\|footnote\)}/,/end{\(note\|footnote\)}/{/\(shall\|may\|should\)[^a-zA-Z]/{=;p;};}' $f |
    # prefix output with filename and line
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/"
done |
    fail '"shall", "should", or "may" inside a note' || failed=1

# Comma after e.g. and i.e.
grep -n "e\.g\.[^,]" $texfiles |
    fail '"e.g." must be followed by a comma'
grep -n "i\.e\.[^,]" $texfiles |
    fail '"i.e." must be followed by a comma'


# \logop should use lowercase arguments
grep -n '\\logop{[^}]*[^andor}][^}]*}' $texfiles |
    fail 'bad argument for \\logop' || failed=1

# Bad indexing for ::iterator and ::sentinel exposition-only classes
grep -Hn '\\indexlibrary\(ctor\|member\).*::\(iterator\|sentinel\)}.*' ranges.tex |
    fail 'use \\exposid for iterator/sentinel member indexing' || failed=1
# Fixup: sed -i '/indexlibrary\(member\|ctor\)/s/::\(iterator\|sentinel\)/::\\exposid{\1}/' ranges.tex

# Do not index the exposition-only ::iterator and ::sentinel class names
grep -Hn '\\indexlibraryglobal.*::\(iterator\|sentinel\)}.*' ranges.tex |
    fail 'do not index exposition-only ::iterator and ::sentinel class names' ranges.tex || failed=1
# Fixup: sed -i '/indexlibraryglobal.*::\(iterator\|sentinel\)}.*/d' ranges.tex

# Hanging paragraphs
for f in $texfiles; do
    sed -n '/^\\rSec/{=;p;};/^\\pnum/{s/^.*$/x/;=;p;}' $f |
    # prefix output with filename and line
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/" |
    awk -F: 'BEGIN { prevlevel = 0 } $3 ~ /^\\rSec./ { match($3, "[0-9]"); level=substr($3, RSTART, 1); if (text && level > prevlevel) { print prevsec " <-- Hanging paragraph follows" } prevlevel = level; prevsec = $3; text = 0 } $3 == "x" { text = 1 }'
done |
    fail 'hanging paragraph' || failed=1

# Subclauses without siblings
for f in $texfiles; do
    sed -n '/^\\rSec/{=;p;}' $f |
    # prefix output with filename and line
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/" |
    awk -F: 'BEGIN { prevlevel = 0 }
      {
	match($3, "[0-9]");
	level = substr($3, RSTART, 1);
	if (level < prevlevel && secs[prevlevel] == 1) { print title[prevlevel] }
	++secs[level];
	title[level] = $0;
	secs[level + 1] = 0;
	prevlevel = level;
      }'
done | fail 'subclause without siblings' || failed=1


# Library descriptive macros not immediately preceded by \pnum.
for f in $texlibdesc; do
    sed -n '/begin{itemdescr}/,/end{itemdescr}/{=;p;}' < $f |
    sed '/^[0-9]\+$/{N;s/\n/:/;}' | sed "s/.*/$f:&/" |
    awk -F: '$3 ~ /^\\pnum/ { seenpnum=1; next } $3 ~ /^\\index/ { next } $3 ~ /^\\(constraints|mandates|constantwhen|expects|hardexpects|effects|sync|ensures|returns|throws|complexity|remarks|errors|recommended)/ { if(seenpnum == 0) { print $0 } } { seenpnum=0 }'
done |
    fail '\\pnum missing' || failed=1

# Cross-references pointing to their own section.
for f in $texfiles; do
    sed -n '/^\\rSec/{s/^.rSec.\[/S /;s/\].*$//;=;p;};/\\iref{/{s/^.*\\.\?ref{\([-a-z.0-9]\+\)}.*/R \1/g;=;p;}' $f |
    sed '/^[0-9]\+$/{N;s/\n/: /;}' | sed "s/.*/$f:&/" |
    awk '$2 == "S" { seclabel = $3 } $2 == "R" && $3 == seclabel { print $1 " section self-reference to [" $3 "]" }'
done |
    fail "cross-reference to its own section" || failed=1

# \placeholder before (
#egrep 'placeholder{[-A-Za-z]*}@?\(' *.tex
# to fix: sed -i 's/placeholder\({[-A-Za-z]*}@\?(\)/placeholdernc\1/g' *.tex

# \placeholdernc before <
#egrep 'placeholdernc{[-A-Za-z]*}@?<' *.tex
# to fix: sed -i 's/placeholdernc\({[-A-Za-z]*}@\?<\)/placeholder\1/g' *.tex

# \placeholder before . or ,
# egrep 'placeholder{[-A-Za-z]*}@?[,.]' *.tex
# to fix: sed -i 's/placeholder\({[-A-Za-z]*}@\?[.,]\)/placeholdernc\1/g' *.tex

# We can't reliably check if the PDF is up to date, because we don't have a
# deterministic rebuild process, and different versions of dot produce
# different files anyway. So just check the timestamp.
for f in *.dot; do
  if [ "$f" -nt "${f%.dot}.pdf" ]; then
    echo -e "need to rebuild ${f%.dot}.pdf:\nmake clean-figures && make figures"
  fi
done |
    fail 'outdated figure'  || failed=1

exit $failed