# BST-TIMEOUT: 600
#
# ~20 xmllint runs, each of which parses a whole DocBook DTD (five modules plus
# up to twenty ISO entity sets, ~450-540 kB per version) before it validates
# anything, and ~25 xmlcatalog runs that are near-instant beside them. The
# sibling 62-docbook-xsl suite, which loads a comparable amount of XML per run,
# was measured at ~42 s for 28 cases on an idle host and declares 600; this file
# does more runs of cheaper documents, so the same bound with the same ~3x load
# margin is the right one rather than a nominal figure.
#
# WHAT THIS SUITE IS FOR. docbook-xml-dtd4 is pure data -- allarch, empty
# do_compile, nothing linked -- so "the port works" cannot be shown by running
# it. It is shown by a VALIDATOR reaching the DTDs through the XML catalog and
# accepting a conforming document while rejecting a non-conforming one. Both
# halves are needed: a validator that accepts everything passes a suite built
# only from positive cases and is worthless, and a catalog that resolves nothing
# fails identically to one that resolves wrongly unless the negative control
# says which. Sections 6-8 are that structure.
#
# THIS IMAGE HAS A ROOT CATALOG AND STOCK YOCTO DOES NOT. oe-core's
# xmlcatalog.bbclass wires the NATIVE sysroot only and says so in its own header
# (Yocto #13271), so on a stock image /etc/xml/docbook-xml.xml lands with nothing
# chaining it. Here brintos-xml-catalog ships /etc/xml/catalog, which chains it.
# Section 4 asserts the sub-catalog on its own terms (XML_CATALOG_FILES pointed
# straight at it, which is what works anywhere), and section 5 asserts the root
# chain separately -- so the two facts stay distinguishable and a regression in
# either one names itself.
. "$BST_LIB"

# THE CANARY, which asserts nothing about docbook-xml-dtd4.
#
# A suite's FIRST BST result line is lost between the guest's printf and the
# host's read -- BRINTOS-1149, which established the loss as per-suite,
# deterministic and platform-side, and taught brintos_local_verify.py to report
# the deficit rather than removing it. The harness computes its pass count as
# total-minus-failures, so THE DROPPED LINE IS COUNTED AS A PASS: a suite whose
# first case is a genuine `not ok` loses exactly the line that mattered. Here
# that slot would hold `tree-4.1.2`, i.e. "the payload is on the image at all".
#
# So a nothing-case stands in that position. Nothing was moved out of the slot;
# a case was added in front of it. Same idiom and reason as 99-mailcap,
# 54-ninja, 56-snappy, 82-double-conversion and 83-flatbuffers.
#
# It sits ahead of the `require` guards deliberately: a `require` that skips
# emits its skip as the suite's first line, and that line would be eaten too.
#
# Remove this case when the loss itself is fixed -- BRINTOS-1149 closed on the
# attribution, not on the platform defect underneath it.
bst_ok docbook-dtd4/suite-canary

require xmllint    docbook-dtd4/xmllint-present
require xmlcatalog docbook-dtd4/xmlcatalog-present

DTD=/usr/share/xml/docbook/schema/dtd
SUBCAT=/etc/xml/docbook-xml.xml
ROOTCAT=/etc/xml/catalog
VERSIONS="4.1.2 4.2 4.3 4.4 4.5"

t=${TMPDIR:-/tmp}/bst-docbook-dtd4.$$
if ! mkdir -p "$t" 2>/dev/null; then
  bst_notok docbook-dtd4/tmpdir "cannot create $t"
  exit 0
fi

# Nothing below chdirs and nothing redirects stdin from the rootfs: both
# intermittently answer the kernel's restart sentinel as "Unknown error 513" and
# end the whole suite where they happen (brintos/linux#73). Documents are written
# into $t and named by absolute path.

# --- 1. the payload the recipe installs --------------------------------------
#
# Five versions, each a self-contained DTD tree. The ticket asks for 4.1.2
# through 4.5 and the recipe's do_install loop names exactly those, so a version
# that silently stopped unpacking is a case of its own rather than a count.
for v in $VERSIONS; do
  check docbook-dtd4/tree-$v       test -d "$DTD/$v"
  check docbook-dtd4/driver-$v     test -s "$DTD/$v/docbookx.dtd"
  check docbook-dtd4/subcatalog-$v test -s "$DTD/$v/catalog.xml"
done

# --- 2. every file the DTD's own entity declarations reference ----------------
#
# A half-unpacked payload is the characteristic failure of a five-zip recipe, and
# it is invisible to a presence check on the directory: docbookx.dtd is a driver
# that pulls in five .mod modules, and dbcentx.mod pulls in up to twenty ISO
# entity sets whose file names differ BETWEEN versions (4.1.2 ships iso-amsa.ent,
# 4.5 ships isoamsa.ent). So the reference list is derived from each version's
# own text rather than written out here -- a hardcoded list would be wrong for at
# least one version and would go stale on the next.
#
# The count is asserted as well as the misses. An extraction that matched nothing
# reports zero missing and would pass vacuously, which is the failure mode a
# presence guard most easily hides: 26 is the smallest count any of the five
# versions yields (4.1.2 and 4.2; 4.3-4.5 yield 27, htmltblx.mod being the extra).
for v in $VERSIONS; do
  refs=$(cat "$DTD/$v"/*.dtd "$DTD/$v"/*.mod 2>/dev/null \
         | awk -f "$BST_DATA/docbook-dtd4-strip-comments.awk" 2>/dev/null \
         | grep -oE '"[A-Za-z0-9._-]+(/[A-Za-z0-9._-]+)*\.(mod|dtd|ent)"' \
         | tr -d '"' | sort -u)
  n=$(printf '%s\n' "$refs" | grep -c '.')
  if [ "$n" -ge 26 ]; then bst_ok docbook-dtd4/refs-found-$v
  else bst_notok docbook-dtd4/refs-found-$v "extracted only $n references from $DTD/$v -- expected at least 26"; fi

  miss=
  for r in $refs; do
    [ -f "$DTD/$v/$r" ] || miss="$miss $r"
  done
  if [ -z "$miss" ]; then bst_ok docbook-dtd4/refs-present-$v
  else bst_notok docbook-dtd4/refs-present-$v "missing under $DTD/$v:$(_bst_1line "$miss")"; fi
done

# --- 3. the sub-catalog the recipe builds -------------------------------------
#
# do_install creates /etc/xml/docbook-xml.xml with xmlcatalog from libxml2-native
# and adds one nextCatalog per version. Each link is asserted separately: a
# version dropped from the chain is perfectly invisible until something asks for
# that version's DTD, and then it looks like a missing payload rather than a
# missing catalog entry.
check docbook-dtd4/catalog-file test -s "$SUBCAT"
for v in $VERSIONS; do
  check_out docbook-dtd4/catalog-chains-$v "$DTD/$v/catalog.xml" cat "$SUBCAT"
done

# --- 4. resolution through the sub-catalog ------------------------------------
#
# The ticket's pass criterion for the catalog: a public identifier must answer a
# LOCAL path. Both spellings are covered, because a document may name either and
# they take different routes through the catalog -- and the three older versions
# ship a catalog.xml with no prefer="public", so their PUBLIC and SYSTEM entries
# are consulted in a different order than 4.4/4.5's.
for v in $VERSIONS; do
  check_out docbook-dtd4/resolve-public-$v "$DTD/$v/docbookx.dtd" \
      xmlcatalog "$SUBCAT" "-//OASIS//DTD DocBook XML V$v//EN"
  check_out docbook-dtd4/resolve-system-$v "$DTD/$v/docbookx.dtd" \
      xmlcatalog "$SUBCAT" "http://www.oasis-open.org/docbook/xml/$v/docbookx.dtd"
  # An answer naming a file that is not there is worse than no answer: it is the
  # shape a stale or mis-versioned catalog takes, and it fails later and further
  # away. So the answer is followed to the file.
  res=$(xmlcatalog "$SUBCAT" "-//OASIS//DTD DocBook XML V$v//EN" 2>/dev/null)
  check docbook-dtd4/resolve-target-$v test -f "${res#file://}"
done

# A public identifier this package does not ship must resolve to nothing. Without
# it, a catalog that answered everything with one path would pass every case
# above.
uout=$(xmlcatalog "$SUBCAT" "-//OASIS//DTD DocBook XML V9.9//EN" 2>&1)
case "$uout" in
  *"No entry"*) bst_ok docbook-dtd4/resolve-unknown-none ;;
  *) bst_notok docbook-dtd4/resolve-unknown-none "resolved to '$(_bst_1line "$uout")'" ;;
esac

# --- 5. the root catalog: this image's, not oe-core's -------------------------
#
# Stock Yocto ships no /etc/xml/catalog (Yocto #13271, quoted in
# xmlcatalog.bbclass's own header), so a caller there must name the sub-catalog
# explicitly. brintos-xml-catalog fills that gap on this image, and section 6's
# default-search-path case is the consequence. Asserted here rather than assumed,
# because the whole difference between this image and a stock one is one file.
check     docbook-dtd4/root-catalog        test -s "$ROOTCAT"
check_out docbook-dtd4/root-catalog-chains "$SUBCAT" cat "$ROOTCAT"
check_out docbook-dtd4/root-resolves-45 "$DTD/4.5/docbookx.dtd" \
    xmlcatalog "$ROOTCAT" "-//OASIS//DTD DocBook XML V4.5//EN"
check_out docbook-dtd4/root-resolves-412 "$DTD/4.1.2/docbookx.dtd" \
    xmlcatalog "$ROOTCAT" "-//OASIS//DTD DocBook XML V4.1.2//EN"

# --- 6. validation: a conforming document, against each version ---------------
#
# The port's reason to exist. --nonet is what makes a pass mean anything: with
# it, an identifier the catalog does not rewrite is refused rather than fetched,
# so no case below can be passing because something reached the network. (This
# guest has none either way -- belt and braces, cheaply. Section 8 makes the
# offline claim explicitly.)
mkdoc() {  # mkdoc <version> <body> > file
  printf '%s\n' '<?xml version="1.0" encoding="UTF-8"?>'
  printf '%s\n' "<!DOCTYPE article PUBLIC \"-//OASIS//DTD DocBook XML V$1//EN\""
  printf '%s\n' "  \"http://www.oasis-open.org/docbook/xml/$1/docbookx.dtd\">"
  printf '%s\n' "<article><title>BSTDBTITLE</title>$2</article>"
}

GOOD='<sect1><title>SEC</title><para>BODYSENTINEL text.</para></sect1>'
# An element outside its allowed content model: sect1 is a perfectly valid
# element that may not appear inside para. The document stays WELL-FORMED, so a
# rejection can only come from the DTD -- section 7 pins that distinction.
BADCHILD='<sect1><title>SEC</title><para><sect1><title>N</title><para>x</para></sect1></para></sect1>'
# The other class of content-model violation: right children, wrong order.
BADORDER='<sect1><para>BODYSENTINEL text.</para><title>SEC</title></sect1>'

for v in $VERSIONS; do
  mkdoc "$v" "$GOOD" > "$t/good-$v.xml"
  check docbook-dtd4/validate-$v \
      env XML_CATALOG_FILES="$SUBCAT" xmllint --nonet --noout --valid "$t/good-$v.xml"
done

# The same document with no XML_CATALOG_FILES at all, i.e. through libxml2's
# default search path. This is the case that is true HERE and false on a stock
# Yocto image, and it is the one a downstream consumer (xmlto, xsltproc, any
# Makefile) actually takes -- none of them set XML_CATALOG_FILES.
check docbook-dtd4/validate-default-path \
    xmllint --nonet --noout --valid "$t/good-4.5.xml"

# --- 7. the negative cases: the validator must REJECT -------------------------
#
# Without these every case in section 6 passes against a validator that accepts
# anything, and the port would be certified by a suite that cannot fail.
for v in $VERSIONS; do
  mkdoc "$v" "$BADCHILD" > "$t/badchild-$v.xml"
  bout=$(XML_CATALOG_FILES="$SUBCAT" xmllint --nonet --noout --valid "$t/badchild-$v.xml" 2>&1); brc=$?
  if [ "$brc" -ne 0 ]; then bst_ok docbook-dtd4/reject-badchild-$v
  else bst_notok docbook-dtd4/reject-badchild-$v "rc=0 -- sect1 inside para was accepted against V$v"; fi
  case "$bout" in
    *"Element sect1 is not declared in para list of possible children"*)
      bst_ok docbook-dtd4/reject-badchild-diag-$v ;;
    *) bst_notok docbook-dtd4/reject-badchild-diag-$v "$(_bst_1line "$bout")" ;;
  esac
done

mkdoc 4.5 "$BADORDER" > "$t/badorder.xml"
oout=$(XML_CATALOG_FILES="$SUBCAT" xmllint --nonet --noout --valid "$t/badorder.xml" 2>&1); orc=$?
if [ "$orc" -ne 0 ]; then bst_ok docbook-dtd4/reject-badorder
else bst_notok docbook-dtd4/reject-badorder "rc=0 -- para before title in sect1 was accepted"; fi
case "$oout" in
  *"Element sect1 content does not follow the DTD"*) bst_ok docbook-dtd4/reject-badorder-diag ;;
  *) bst_notok docbook-dtd4/reject-badorder-diag "$(_bst_1line "$oout")" ;;
esac

# The control that gives section 7 its meaning: the SAME invalid document parses
# clean without --valid. So the rejections above are the DTD's verdict on the
# content model, not a well-formedness error that would fail with no DTD at all.
check docbook-dtd4/badchild-wellformed \
    env XML_CATALOG_FILES="$SUBCAT" xmllint --nonet --noout "$t/badchild-4.5.xml"

# --- 8. offline: the whole reason a local catalog exists ----------------------
#
# The positive half is every --nonet pass above. The negative half is here: with
# the catalog emptied, the identical document must FAIL, and must fail by
# reaching for the network. Without this case section 6 would pass on an image
# whose catalog does nothing, because the DTD could have been found some other
# way -- and the failure would only appear on a machine with no network, which is
# every machine this image runs on.
printf '%s\n%s\n' '<?xml version="1.0"?>' \
    '<catalog xmlns="urn:oasis:names:tc:entity:xmlns:xml:catalog"/>' > "$t/empty-catalog.xml"
eout=$(XML_CATALOG_FILES="$t/empty-catalog.xml" \
       xmllint --nonet --noout --valid "$t/good-4.5.xml" 2>&1); erc=$?
if [ "$erc" -ne 0 ]; then bst_ok docbook-dtd4/nocatalog-fails
else bst_notok docbook-dtd4/nocatalog-fails "rc=0 with the catalog emptied -- the DTD resolved by some path this suite does not control"; fi
case "$eout" in
  *"Attempt to load network entity"*) bst_ok docbook-dtd4/nocatalog-diagnosed ;;
  *) bst_notok docbook-dtd4/nocatalog-diagnosed "$(_bst_1line "$eout")" ;;
esac

# The contrast, and the reason this suite does not validate by local path: the
# DTD files themselves are fine when the catalog is not, so a path-driven
# validation passes either way and would hide section 8 entirely.
printf '%s\n' '<?xml version="1.0" encoding="UTF-8"?>' > "$t/bypath.xml"
printf '%s\n' "<!DOCTYPE article SYSTEM \"$DTD/4.5/docbookx.dtd\">" >> "$t/bypath.xml"
printf '%s\n' "<article><title>BSTDBTITLE</title>$GOOD</article>" >> "$t/bypath.xml"
check docbook-dtd4/nocatalog-path-works \
    env XML_CATALOG_FILES="$t/empty-catalog.xml" xmllint --nonet --noout --valid "$t/bypath.xml"

# --- 9. the entity sets, which are half the payload by file count -------------
#
# ent/ is ~20 files per version and nothing above touches it: docbookx.dtd loads
# the modules, but the ISO entity sets are only read when a document actually
# references one. A payload that unpacked the modules and lost ent/ therefore
# passes every case above. These do reference them, and assert the SUBSTITUTED
# text -- so a set that resolved to the wrong file fails too.
for v in $VERSIONS; do
  mkdoc "$v" '<sect1><title>SEC</title><para>ENT&mdash;SENT&copy;2026</para></sect1>' \
      > "$t/ent-$v.xml"
  check_out docbook-dtd4/entities-$v 'ENT—SENT©2026' \
      env XML_CATALOG_FILES="$SUBCAT" xmllint --nonet --valid --noent "$t/ent-$v.xml"
done

# And its control: with no DTD loaded the same entities are undefined, so the
# expansions above came from ent/ and not from libxml2's built-in five.
nout=$(XML_CATALOG_FILES="$SUBCAT" xmllint --nonet --noout --noent "$t/ent-4.5.xml" 2>&1)
case "$nout" in
  *"Entity 'mdash' not defined"*) bst_ok docbook-dtd4/entities-need-dtd ;;
  *) bst_notok docbook-dtd4/entities-need-dtd "expected an undefined-entity error without --valid, got: $(_bst_1line "$nout")" ;;
esac

rm -rf "$t"