# BST-TIMEOUT: 1800  # 17 documents, 915 KB, parsed three ways: sha256 over all
#                    # 24 installed files, one libxml2 well-formedness pass, and
#                    # a count(//*)+count(//@*) XPath pass per document (17 more
#                    # parses, icccm.xml alone 267 KB). At the 46-68x wall-clock
#                    # factor this platform carries (linux-in-the-browser#117)
#                    # that is minutes, not seconds. The bound is deliberately a
#                    # LOAD bound: it must fire when the guest is starved, not
#                    # when the package regressed.
. "$BST_LIB"

# xorg-docs is documentation and nothing else -- 17 DocBook 4.3 XML documents and
# MAINTAINERS under ${docdir}, six section-7 man pages under ${mandir}. There is
# no binary to run, so "the port works" is a statement about the INSTALLED SET,
# and the ticket says so: the declared documents are present under ${docdir} and
# the installed file list matches a native build of the same version.
#
# Both halves are asserted against evidence the recipe SHIPPED rather than
# against constants typed here:
#
#   CENSUS  gen-xorg-docs-census.py, at do_install: the expected set DERIVED from
#           the generated Makefiles (docbook.am gates four groups of generated
#           formats on what configure found, so the set is not a constant), per
#           file sha256 and size, and element/attribute counts from EXPAT.
#   NATIVE  xorg-docs-native-oracle.sh, at do_install: the same tarball
#           configured, built and installed on x86-64, with its file list carried
#           verbatim so THIS suite can compare the rootfs against it.
#
# The counts below are re-derived through LIBXML2, which shares no code with
# expat, so agreement is a statement about the bytes rather than about either
# parser.

DOC=/usr/share/doc/xorg-docs
MAN=/usr/share/man/man7
V=/usr/share/xorg-docs
CENSUS=$V/CENSUS
NATIVE=$V/NATIVE
# xorg-sgml-doctools' entity file. Every document but two declares it as an
# external parameter entity by RELATIVE system id, so whether it is on the image
# decides what can be said about validation -- see section 7.
DEFS=/usr/share/sgml/X11/defs.ent

# The first BST result line of a suite is dropped by the guest console
# (brintos/linux#96). This case exists to be the one that is lost.
bst_ok xorg-docs/suite-canary

if [ ! -d "$DOC" ]; then
  bst_skip xorg-docs/present "no $DOC on this image"
  exit 0
fi
bst_ok xorg-docs/present

t=${TMPDIR:-/tmp}/bst-xorg-docs.$$
if ! mkdir -p "$t" 2>/dev/null; then
  bst_notok xorg-docs/tmpdir "cannot create $t"
  exit 0
fi

# --- 1. the recipe's own evidence arrived, intact ---------------------------
#
# Every later section reads one of these two files. A suite that asserted
# agreement WITH A MISSING FILE would agree with nothing and say so as a pass, so
# the shape of each read below is "derive from the file, then compare", never
# "grep the file for the answer".
check xorg-docs/census-installed test -s "$CENSUS"
check xorg-docs/native-installed test -s "$NATIVE"
check xorg-docs/census-green grep -qx "CENSUS-END fails=0" "$CENSUS"
check xorg-docs/native-green grep -qx "NATIVE-END fails=0" "$NATIVE"
check_out xorg-docs/census-version "CENSUS-PV 1.7.2" cat "$CENSUS"

# --- 2. the declared set, in both directions --------------------------------
#
# ${docdir} belongs to this package alone, so the count is an EXACT assertion
# rather than a floor: a partially unpacked payload fails here and in
# declared-set, and nowhere else.
DECL=$t/declared.list
sed -n 's/^CENSUS-FILE \([^ ]*\) .*$/\1/p' "$CENSUS" | LC_ALL=C sort > "$DECL"
check_eq xorg-docs/declared-count 24 "$(wc -l < "$DECL" | tr -d ' ')"

missing=0
while IFS= read -r p; do [ -f "$p" ] || { missing=$((missing+1)); miss1=$p; }; done < "$DECL"
check_eq xorg-docs/declared-set 0 "$missing"
[ "$missing" -eq 0 ] || bst_notok xorg-docs/declared-set-first "$miss1 declared but not installed"

check_eq xorg-docs/docdir-count 18 "$(find "$DOC" -type f 2>/dev/null | wc -l | tr -d ' ')"

# Both directions: a file that appears under ${docdir} and is NOT declared is as
# much a fault as one that is declared and absent -- it is what a generated
# format leaking through would look like.
extra=0
for p in $(find "$DOC" -type f 2>/dev/null | LC_ALL=C sort); do
  grep -qxF "$p" "$DECL" || { extra=$((extra+1)); extra1=$p; }
done
check_eq xorg-docs/no-extra-files 0 "$extra"
[ "$extra" -eq 0 ] || bst_notok xorg-docs/no-extra-files-first "$extra1 installed but not declared"

# The eight subdirectories are the shape of the install; losing one takes its
# documents with it and is the failure a `cp` that lost its recursion makes.
sub_missing=0
for d in ctext fonts graphics icccm input platforms xlfd Xserver; do
  [ -d "$DOC/$d" ] || sub_missing=$((sub_missing+1))
done
check_eq xorg-docs/subdirs 0 "$sub_missing"

manmiss=0
for m in Consortium Standards X XOrgFoundation XProjectTeam Xsecurity; do
  [ -f "$MAN/$m.7" ] || manmiss=$((manmiss+1))
done
check_eq xorg-docs/man-pages-present 0 "$manmiss"

# --- 3. the bytes -----------------------------------------------------------
bad=0
while read -r _ p sz sum; do
  [ -f "$p" ] || { bad=$((bad+1)); continue; }
  [ "$(wc -c < "$p" | tr -d ' ')" = "$sz" ] || { bad=$((bad+1)); bad1="$p size"; continue; }
  [ "$(sha256sum "$p" 2>/dev/null | cut -d' ' -f1)" = "$sum" ] || { bad=$((bad+1)); bad1="$p sha256"; }
done <<EOF
$(grep '^CENSUS-FILE ' "$CENSUS")
EOF
check_eq xorg-docs/census-sha256 0 "$bad"
[ "$bad" -eq 0 ] || bst_notok xorg-docs/census-sha256-first "${bad1:-?} differs from the census"

check_eq xorg-docs/census-bytes-total "$(sed -n 's/^CENSUS-BYTES total=//p' "$CENSUS")" \
    "$(cat $(sed -n 's/^CENSUS-FILE \([^ ]*\) .*$/\1/p' "$CENSUS") 2>/dev/null | wc -c | tr -d ' ')"

# The sha256 check above passes on a file whose digest was never computed, so
# prove the comparison can go red: the same content plus one byte must not match
# the census entry it came from.
ref=$DOC/platforms/Darwin.xml
cp "$ref" "$t/mutated.xml" && printf 'x' >> "$t/mutated.xml"
refsum=$(sed -n "s|^CENSUS-FILE $ref [0-9]* ||p" "$CENSUS")
mutsum=$(sha256sum "$t/mutated.xml" 2>/dev/null | cut -d' ' -f1)
if [ -n "$refsum" ] && [ -n "$mutsum" ] && [ "$refsum" != "$mutsum" ]; then
  bst_ok xorg-docs/control-sha256-detects-change
else
  bst_notok xorg-docs/control-sha256-detects-change "census='$refsum' mutated='$mutsum'"
fi

# --- 4. the native comparison, asserted against THIS rootfs -----------------
#
# NATIVE-FILELIST records what do_install produced; this compares what do_rootfs
# DELIVERED. They are different claims: a package split, an image-level doc strip
# or a lost -doc entry changes the second without touching the first.
NAT=$t/native.list
sed -n 's/^NATIVE-LIST //p' "$NATIVE" | LC_ALL=C sort > "$NAT"
check_eq xorg-docs/native-list-count 24 "$(wc -l < "$NAT" | tr -d ' ')"

HAVE=$t/rootfs.list
{ find "$DOC" -type f 2>/dev/null
  grep "^$MAN/" "$NAT" | while IFS= read -r p; do [ -f "$p" ] && echo "$p"; done
} | LC_ALL=C sort > "$HAVE"
if diff "$NAT" "$HAVE" > "$t/list.diff" 2>&1; then
  bst_ok xorg-docs/rootfs-matches-native
else
  bst_notok xorg-docs/rootfs-matches-native "$(_bst_1line "$(head -6 "$t/list.diff")")"
fi

check_out xorg-docs/native-configure "NATIVE-CONFIGURE rc=0"        cat "$NATIVE"
check_out xorg-docs/native-filelist  "NATIVE-FILELIST identical=1"  cat "$NATIVE"
check_out xorg-docs/native-bytes     "NATIVE-BYTES same=24 differing=0" cat "$NATIVE"
check_out xorg-docs/native-count     "NATIVE-COUNT native=24 cross=24"  cat "$NATIVE"

# Both builds took the same branch of docbook.am, which is what makes the file
# lists comparable at all. Asserted rather than assumed: a native host that HAD
# xmlto would install up to five more files per document, and the comparison
# would then be measuring the two hosts' tooling rather than this arm.
gates=0
for g in XMLTO XSLTPROC FOP STYLESHEET_SRCDIR XORG_SGML_PATH; do
  grep -qx "NATIVE-TOOL $g=empty"  "$NATIVE" || gates=$((gates+1))
  grep -qx "CENSUS-TOOL $g=empty"  "$CENSUS" || gates=$((gates+1))
done
check_eq xorg-docs/doc-toolchain-gates-off 0 "$gates"
check_out xorg-docs/native-generated-none "NATIVE-GENERATED declared=0" cat "$NATIVE"
check_out xorg-docs/census-generated-none "CENSUS-GENERATED declared=0" cat "$CENSUS"

# --- 5. no generated format reached the rootfs ------------------------------
#
# The direct statement of the same thing, made against the tree rather than
# against a report: with xmlto, xsltproc and fop all absent at configure time
# there is no .html, .txt, .pdf, .ps or .db anywhere under ${docdir}.
check_eq xorg-docs/no-generated-formats 0 \
    "$(find "$DOC" -type f \( -name '*.html' -o -name '*.txt' -o -name '*.pdf' \
        -o -name '*.ps' -o -name '*.db' \) 2>/dev/null | wc -l | tr -d ' ')"

# --- 6. the documents, through libxml2 --------------------------------------
require xmllint xorg-docs/xmllint-present

XMLS=$t/xml.list
find "$DOC" -type f -name '*.xml' 2>/dev/null | LC_ALL=C sort > "$XMLS"
check_eq xorg-docs/xml-count 17 "$(wc -l < "$XMLS" | tr -d ' ')"

# Non-validating: the external DTD and defs.ent are declared but not loaded, so
# this is a well-formedness statement about the shipped bytes and nothing else.
# libxml2 reports the unresolved entities on stderr and still exits 0, which is
# why every case here is written on an exit status or an extracted number rather
# than on "no output".
nwf=0
while IFS= read -r f; do
  xmllint --nonet --noout "$f" >/dev/null 2>&1 || { nwf=$((nwf+1)); wf1=$f; }
done < "$XMLS"
check_eq xorg-docs/xml-wellformed 0 "$nwf"
[ "$nwf" -eq 0 ] || bst_notok xorg-docs/xml-wellformed-first "$wf1 is not well-formed"

# Element and attribute totals, re-derived here through libxml2's XPath and
# compared against the census's expat numbers. Agreement across two
# implementations is what makes these a measurement of the documents.
# _num defaults a non-number to 0. It is load-bearing, not defensive style: a
# document that fails to parse yields an EMPTY xpath result, `$((tel + ))` is a
# syntax error in dash, and the shell exits -- so one malformed file would end
# the run silently, taking sections 7-9 with it and reporting fewer cases rather
# than a red. Found by the mutation pass, which is the only thing that looks.
_num() { case "$1" in ''|*[!0-9]*) echo 0 ;; *) echo "$1" ;; esac; }

tel=0; tat=0
while IFS= read -r f; do
  pair=$(xmllint --nonet --xpath 'concat(count(//*)," ",count(//@*))' "$f" 2>/dev/null)
  e=${pair%% *}; a=${pair##* }
  tel=$((tel + $(_num "${e%%.*}"))); tat=$((tat + $(_num "${a%%.*}")))
done < "$XMLS"
check_eq xorg-docs/xml-elements "$(sed -n 's/^CENSUS-XML-TOTAL files=[0-9]* elements=\([0-9]*\).*$/\1/p' "$CENSUS")" "$tel"
check_eq xorg-docs/xml-attrs    "$(sed -n 's/^CENSUS-XML-TOTAL .*attrs=\([0-9]*\)$/\1/p' "$CENSUS")" "$tat"
check_eq xorg-docs/census-xml-files 17 "$(sed -n 's/^CENSUS-XML-TOTAL files=\([0-9]*\).*$/\1/p' "$CENSUS")"

# One DocBook version across the whole set. A document at another version is not
# a fault by itself, but it changes which DTD section 7 must resolve, so it is
# pinned rather than discovered later.
n43=0
while IFS= read -r f; do
  head -4 "$f" | grep -q 'DocBook XML V4.3//EN' || n43=$((n43+1))
done < "$XMLS"
check_eq xorg-docs/xml-doctype-43 0 "$n43"

# The control for the whole section: xmllint must be able to say no.
printf '<a><b></a>\n' > "$t/broken.xml"
# Non-zero, not a pinned code: libxml2 spells "unclassified", "DTD" and
# "validation" errors 1/2/3/4 and has moved which it uses. The control's claim is
# that it CAN say no, and a red here for a renumbered exit code would say nothing
# about this package.
if xmllint --nonet --noout "$t/broken.xml" >/dev/null 2>&1; then
  bst_notok xorg-docs/control-xmllint-detects-malformed "accepted a malformed document"
else
  bst_ok xorg-docs/control-xmllint-detects-malformed
fi

# --- 7. what the documents depend on, and what that costs -------------------
#
# 15 of the 17 declare `<!ENTITY % defs SYSTEM "defs.ent">` -- a RELATIVE system
# id, resolved by X.Org's build through --searchpath into xorg-sgml-doctools'
# tree. Installed as sources they carry that reference unresolved, so DTD
# validation of the real documents needs that package on the image. This is a
# property of what upstream ships, not a defect of the port, and it is pinned
# here so it cannot change unnoticed.
ndefs=0
while IFS= read -r f; do
  grep -q 'ENTITY % defs SYSTEM "defs.ent"' "$f" && ndefs=$((ndefs+1))
done < "$XMLS"
check_eq xorg-docs/docs-declare-defs-ent 15 "$ndefs"

# The DTD path itself IS exercised on this image, against a self-contained
# fixture, so the skip below is narrow: it is about defs.ent, not about whether
# validation works here at all.
check_out xorg-docs/dtd43-catalog-resolves "docbook" \
    xmlcatalog /etc/xml/catalog "-//OASIS//DTD DocBook XML V4.3//EN"
cat > "$t/fixture.xml" <<'XEOF'
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook XML V4.3//EN"
  "http://www.oasis-open.org/docbook/xml/4.3/docbookx.dtd">
<article id="bstxdocs">
  <articleinfo><title>brintOS xorg-docs fixture</title></articleinfo>
  <sect1 id="s"><title>SENTINEL</title><para>FIXTURESENTINEL</para></sect1>
</article>
XEOF
check xorg-docs/fixture-valid-43 xmllint --nonet --noout --valid "$t/fixture.xml"
# and it can go red: <sect1> is not permitted directly inside <articleinfo>.
sed 's|<articleinfo><title>|<articleinfo><sect1 id="bad"/><title>|' "$t/fixture.xml" > "$t/badfixture.xml"
if xmllint --nonet --noout --valid "$t/badfixture.xml" >/dev/null 2>&1; then
  bst_notok xorg-docs/control-fixture-invalid-43 "sect1 inside articleinfo validated"
else
  bst_ok xorg-docs/control-fixture-invalid-43
fi

if [ -f "$DEFS" ]; then
  # xorg-sgml-doctools is here, so the documents can be validated the way their
  # own build validates them: --path is xmllint's equivalent of --searchpath.
  nval=0
  while IFS= read -r f; do
    xmllint --nonet --noout --valid --path "$(dirname "$DEFS")" "$f" >/dev/null 2>&1 \
      || { nval=$((nval+1)); val1=$f; }
  done < "$XMLS"
  check_eq xorg-docs/docs-valid-43 0 "$nval"
  [ "$nval" -eq 0 ] || bst_notok xorg-docs/docs-valid-43-first "$val1 did not validate"
else
  bst_skip xorg-docs/docs-valid-43 \
    "xorg-sgml-doctools is not on this image, so defs.ent cannot be resolved and the documents cannot be DTD-validated as installed"
fi

# --- 8. the documents drive an XSLT engine on this target -------------------
#
# The strongest thing that can be said about a document is that something reads
# it. The stylesheet is written here rather than taken from docbook-xsl on
# purpose: it needs no catalog, so a green result is attributable to the SHIPPED
# XML and to libxslt on wasm32, and not to the image's DocBook stack.
if command -v xsltproc >/dev/null 2>&1; then
  cat > "$t/probe.xsl" <<'XEOF'
<?xml version="1.0"?>
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
  <xsl:output method="text"/>
  <xsl:template match="/">
    <xsl:text>TITLE=</xsl:text><xsl:value-of select="normalize-space((//title)[1])"/>
    <xsl:text>|ELEMENTS=</xsl:text><xsl:value-of select="count(//*)"/>
    <xsl:text>|ULINKS=</xsl:text><xsl:value-of select="count(//ulink)"/>
    <xsl:text>&#10;</xsl:text>
  </xsl:template>
</xsl:stylesheet>
XEOF
  tout=$(xsltproc --nonet "$t/probe.xsl" "$DOC/platforms/Darwin.xml" 2>/dev/null); trc=$?
  if [ "$trc" -eq 0 ]; then bst_ok xorg-docs/transform-rc
  else bst_notok xorg-docs/transform-rc "rc=$trc"; fi
  check_eq xorg-docs/transform-fields \
      "TITLE=X on Darwin and Mac OS X|ELEMENTS=17|ULINKS=7" "$tout"

  # The same transform over the largest document, scored on the count the census
  # recorded for it -- so libxslt's parse of a 267 KB DocBook book is compared
  # against expat's, not against a number typed here.
  bout=$(xsltproc --nonet "$t/probe.xsl" "$DOC/icccm/icccm.xml" 2>/dev/null)
  bel=${bout#*|ELEMENTS=}; bel=${bel%%|*}; bel=$(_num "$bel")
  check_eq xorg-docs/transform-large-elements \
      "$(sed -n "s|^CENSUS-XML $DOC/icccm/icccm.xml [a-z]* \([0-9]*\) .*$|\1|p" "$CENSUS")" "$bel"

  # The control: the same engine over something that is not XML must fail.
  printf 'this is not xml\n' > "$t/notxml"
    if xsltproc --nonet "$t/probe.xsl" "$t/notxml" >/dev/null 2>&1; then
    bst_notok xorg-docs/control-transform-rejects-nonxml "transformed a non-XML input"
  else
    bst_ok xorg-docs/control-transform-rejects-nonxml
  fi
else
  bst_skip xorg-docs/transform-rc "xsltproc not installed on this image"
fi

# --- 9. the man pages -------------------------------------------------------
#
# These six are the only GENERATED files in the package: sed output over .man
# sources, substituting xorg-macros' MAN_SUBSTS plus __datadir__, __libdir__ and
# __sysconfdir__. A substitution that resolved to the BUILD sysroot would leave
# the file list, every size and every count untouched, so it is checked directly.
nroff=0; ntok=0
for m in Consortium Standards X XOrgFoundation XProjectTeam Xsecurity; do
  grep -q '^\.TH ' "$MAN/$m.7" 2>/dev/null || nroff=$((nroff+1))
  grep -qE '__[a-zA-Z][a-zA-Z0-9_]*__' "$MAN/$m.7" 2>/dev/null && ntok=$((ntok+1))
done
check_eq xorg-docs/man-roff 0 "$nroff"
check_eq xorg-docs/man-no-unsubstituted-tokens 0 "$ntok"
# and the substitutions resolved to the TARGET prefix, which is the half that a
# token scan alone cannot show.
check_out xorg-docs/man-datadir-substituted "/usr/share/X11/rgb.txt" cat "$MAN/X.7"
check_out xorg-docs/man-libdir-substituted  "/usr/lib/X11/"          cat "$MAN/X.7"
# The control for the scan: a planted token must be found.
printf '.TH T 7\n.\\" __projectroot__\n' > "$t/planted.7"
check xorg-docs/control-token-scan-works grep -qE '__[a-zA-Z][a-zA-Z0-9_]*__' "$t/planted.7"

# man-db 2.12.0 and groff 1.23.0 are on this image, and they READ these pages:
# the lookup, the NAME-section parse and both groff preprocessors all succeed
# against X.7. Only the final formatting stage does not -- see below.
E="env MANPATH=/usr/share/man MANPAGER=cat MANWIDTH=80"
if command -v man >/dev/null 2>&1; then
  check_out xorg-docs/man-locate-x "$MAN/X.7" $E man -w 7 X
  locmiss=0
  for m in Consortium Standards X XOrgFoundation XProjectTeam Xsecurity; do
    [ "$($E man -w 7 "$m" 2>/dev/null)" = "$MAN/$m.7" ] || locmiss=$((locmiss+1))
  done
  check_eq xorg-docs/man-locate-all 0 "$locmiss"
else
  bst_skip xorg-docs/man-locate-x "man not installed on this image"
fi

# lexgrog parses the NAME section the way whatis/apropos do, so a green here says
# the page is well-formed man source and not merely well-formed roff.
if command -v lexgrog >/dev/null 2>&1; then
  check_out xorg-docs/man-lexgrog "network-transparent window system" lexgrog "$MAN/X.7"
else
  bst_skip xorg-docs/man-lexgrog "lexgrog not installed on this image"
fi

# soelim and preconv are the two groff preprocessors man pipes a page through
# before troff. Both run here, which is what makes the failure below specifically
# troff's and not the toolchain's.
if command -v soelim >/dev/null 2>&1 && command -v preconv >/dev/null 2>&1; then
  n=$(soelim "$MAN/X.7" 2>/dev/null | wc -c | tr -d ' '); n=$(_num "$n")
  sz=$(wc -c < "$MAN/X.7" | tr -d ' ')
  if [ "$n" -ge "$sz" ]; then bst_ok xorg-docs/man-soelim
  else bst_notok xorg-docs/man-soelim "soelim emitted $n bytes for a $sz-byte page"; fi
  check_out xorg-docs/man-preconv "network-transparent window system" \
      sh -c "preconv '$MAN/X.7' 2>/dev/null | head -200"
else
  bst_skip xorg-docs/man-soelim "soelim/preconv not installed on this image"
fi

# --- 10. the one platform defect this port meets ----------------------------
#
# THE CAUSE, ASSERTED FIRST AND ON ITS OWN: troff traps SIGILL (rc 132) in a
# static constructor of libgroff's uniuni.cpp, BEFORE main(), so even
# `troff --version` dies without printing. That is
# linux-in-the-browser#171, already tracked and already carried as a red by
# 89-man-db.test -- it is a property of groff on wasm32 and has nothing to do
# with this package. Measured here rather than inferred: `man 7 X` with LC_ALL=C
# fails identically, so the "can't set the locale" line man prints alongside it
# (this image ships no locales) is noise, not the cause.
#
# Both cases are written in the direction that is TRUE ONCE #171 LANDS, so they
# self-flip rather than needing an edit. Two reds, one cause: the second is the
# effect, and it is kept because it is the assertion a reader of this port cares
# about -- whether the six pages can be read as pages.
if command -v troff >/dev/null 2>&1; then
  check_out xorg-docs/man-troff-runs "GNU troff (groff) version 1.23.0" troff --version
else
  bst_skip xorg-docs/man-troff-runs "troff not installed on this image"
fi
if command -v man >/dev/null 2>&1; then
  check_out xorg-docs/man-renders-x "portable, network-transparent window system" \
      $E man 7 X
else
  bst_skip xorg-docs/man-renders-x "man not installed on this image"
fi

rm -rf "$t"