# BST-TIMEOUT: 300   # ~60 cases; 8 upstream programs + ~30 ikslint/iksperf runs
#
# iksemel 1.5+git -- "fast and portable XML parser and Jabber protocol library".
#
# THE EVIDENCE HERE IS OF TWO KINDS, AND THEY COVER DIFFERENT THINGS.
#
# 1. UPSTREAM'S OWN RELEASE GATE, RUN ON TARGET. test/ ships eight conformance
#    programs that `make all` already builds and `make install` then discards;
#    the bbappend installs them instead (the qcbortest precedent -- an image with
#    no compiler has no other way to run a library's own tests). They are the
#    first eight cases below, one case each rather than one lump, so a failure
#    names the subsystem: the arena allocator, the node/attribute API, the
#    incremental SAX parser, DOM build/serialize, SHA1, MD5, the stanza filter,
#    and JID parsing. Each is silent + rc=0 on success and prints expected-vs-got
#    then exit(1) on failure, so its own diagnostic is what lands in the detail.
#
# 2. CROSS-ARM AGREEMENT ON A FIXED CORPUS. data/iksemel-corpus.xml is a 48 kB
#    deterministic document (src/iksemel_mkfixture.py) and data/iksemel-native.txt
#    is what an x86_64 build of the SAME SRCREV, configured the same way, reports
#    about it (src/iksemel_mknative.sh). The target is asked to reproduce those
#    numbers. This is what upstream's corpus does NOT cover: it uses only
#    in-memory buffers, so nothing in it reads a file, and the read path is
#    exactly where this arm differs.
#
# THE ONE ASSERTION THAT CANNOT BE FAKED BY A HOST BINARY is wasm32-stack-smaller:
# ikstack occupancy for the same document is dominated by pointer-sized fields, so
# a 4-byte-pointer build MUST occupy strictly less than the 8-byte control. Every
# other number here is arch-independent by design, and would pass just as happily
# against an x86_64 binary that had somehow been packaged into the image.
#
# WHAT IS NOT COVERED, DELIBERATELY: the plaintext socket transport
# (src/io-posix.c, USE_DEFAULT_IO). It compiles and links, but driving it wants a
# Jabber server to connect to, which the gate does not have. It is stated as a
# gap in the test-plan entry rather than papered over with a connect-to-nothing
# case that would only assert that the connection failed.

. "$BST_LIB"
require ikslint iksemel/present

NAT="$BST_DATA/iksemel-native.txt"
COR="$BST_DATA/iksemel-corpus.xml"
[ -r "$NAT" ] || { bst_skip iksemel/payload "missing control $NAT"; exit 0; }
[ -r "$COR" ] || { bst_skip iksemel/payload "missing corpus $COR"; exit 0; }

# Read one key out of the x86_64 control. Values may contain spaces.
nat() { sed -n "s/^$1 //p" "$NAT" | head -1; }

W=${TMPDIR:-/tmp}/bst-iksemel.$$
mkdir -p "$W" 2>/dev/null || { bst_notok iksemel/workdir "cannot mkdir $W"; exit 0; }
trap 'rm -rf "$W"' EXIT

# ---------------------------------------------------------------------------
# 1. Upstream's eight conformance programs.
#
# libexecdir is /usr/libexec on this distro, but the suite locates the directory
# rather than hardcoding it: a layer that moves libexecdir should make these
# cases skip with a clear reason, not fail as though the library were broken.
IKST=
for d in /usr/libexec/iksemel-tests /usr/lib/iksemel-tests /usr/libexec/iksemel; do
  [ -d "$d" ] && { IKST=$d; break; }
done

UPSTREAM_TESTS="tst-ikstack tst-iks tst-sax tst-dom tst-sha tst-md5 tst-filter tst-jid"
if [ -z "$IKST" ]; then
  # Not a skip: the bbappend installs these, so their absence is a packaging
  # regression, and a skip here would hide exactly the deliverable this port
  # exists to add.
  for t in $UPSTREAM_TESTS; do
    bst_notok "iksemel/upstream-$t" "iksemel-tests dir not installed (looked in /usr/libexec, /usr/lib)"
  done
else
  for t in $UPSTREAM_TESTS; do
    if [ ! -x "$IKST/$t" ]; then
      bst_notok "iksemel/upstream-$t" "not installed in $IKST"
    else
      out=$("$IKST/$t" 2>&1) && rc=0 || rc=$?
      if [ "$rc" -eq 0 ]; then
        bst_ok "iksemel/upstream-$t"
      else
        bst_notok "iksemel/upstream-$t" "rc=$rc $(_bst_1line "$out")"
      fi
    fi
  done
fi

# ---------------------------------------------------------------------------
# 2. The corpus arrived intact. Everything below compares against numbers taken
#    from THIS document, so a truncated payload would produce a page of
#    mismatches whose real cause is the payload, not the parser.
sz=$(wc -c < "$COR" | tr -d ' ')
check_eq iksemel/corpus-bytes "$(nat IKS-CORPUS-BYTES)" "$sz"

# ---------------------------------------------------------------------------
# 3. ikslint: well-formedness and the document's shape.
check    iksemel/lint-wellformed ikslint "$COR"
check_eq iksemel/lint-version "$(nat IKS-VER-lint)" "$(ikslint -V 2>&1 | head -1)"

ls_out=$(ikslint -s "$COR" 2>&1)
check_eq iksemel/lint-pairs  "$(nat IKS-LINT-PAIRS)" \
  "$(printf '%s' "$ls_out" | sed -n 's/^Tags: \([0-9]*\) pairs.*/\1/p')"
check_eq iksemel/lint-single "$(nat IKS-LINT-SINGLE)" \
  "$(printf '%s' "$ls_out" | sed -n 's/^Tags: [0-9]* pairs, \([0-9]*\) single.*/\1/p')"
check_eq iksemel/lint-depth  "$(nat IKS-LINT-DEPTH)" \
  "$(printf '%s' "$ls_out" | sed -n 's/.*single, \([0-9]*\) max depth.*/\1/p')"
check_eq iksemel/lint-cdata  "$(nat IKS-LINT-CDATA)" \
  "$(printf '%s' "$ls_out" | sed -n 's/^Total size of character data: \([0-9]*\) bytes.*/\1/p')"

# The histogram walks the whole document and counts by name -- it is the check
# that the parser did not merely survive the file but saw every element in it.
hist=$(ikslint -t "$COR" 2>&1)
check_eq iksemel/hist-unique "$(nat IKS-HIST-UNIQUE)" \
  "$(printf '%s' "$hist" | sed -n 's/^Histogram of \([0-9]*\) unique tags.*/\1/p')"
for tag in record name value note flag level catalog bottom; do
  got=$(printf '%s\n' "$hist" | sed -n "s/^<$tag> \([0-9]*\) times\..*/\1/p")
  check_eq "iksemel/hist-$tag" "$(nat "IKS-HIST-$tag")" "${got:-0}"
done

# ---------------------------------------------------------------------------
# 4. The incremental SAX parser, fed at six read-block sizes from 1 byte up.
#
# THE TAG HOOK COUNT MUST NOT MOVE. Feeding the parser one byte at a time forces
# every tag, attribute and entity in a 48 kB document to be reassembled from
# saved state across a resume boundary; if the saved state is wrong -- which on
# an arm with a 4-byte size_t is the failure to expect -- the count changes or
# the parse dies. The CDATA hook count DOES move (character data is delivered in
# chunks, so smaller blocks mean more calls), and it is pinned per block size
# from the control rather than being waved through as "expected to vary".
for b in 1 7 64 512 4096 65536; do
  o=$(iksperf -s -b "$b" "$COR" 2>&1)
  check_eq "iksemel/taghook-b$b" "$(nat "IKS-PERF-TAGHOOK-b$b")" \
    "$(printf '%s' "$o" | sed -n 's/^SAX: tag hook called \([0-9]*\),.*/\1/p')"
  check_eq "iksemel/cdatahook-b$b" "$(nat "IKS-PERF-CDATAHOOK-b$b")" \
    "$(printf '%s' "$o" | sed -n 's/.*cdata hook called \([0-9]*\) times.*/\1/p')"
done

# An independent cross-check between two different tools' counters: at -b 1 the
# cdata hook fires once per character-data byte, so it must equal the total
# ikslint reported. Two code paths agreeing is worth more than either alone.
cd1=$(iksperf -s -b 1 "$COR" 2>&1 | sed -n 's/.*cdata hook called \([0-9]*\) times.*/\1/p')
check_eq iksemel/cdatahook-b1-equals-lint-cdata "$(nat IKS-LINT-CDATA)" "$cd1"

# ---------------------------------------------------------------------------
# 5. SHA1 over the corpus, computed by iksemel's own src/sha.c.
sha=$(iksperf -1 "$COR" 2>&1 | sed -n 's/^SHA: hash \[\([0-9a-f]*\)\].*/\1/p')
check_eq iksemel/sha1 "$(nat IKS-PERF-SHA1)" "$sha"

# And against a SECOND implementation on the target, when the image has one --
# this is what turns "matches the control" into "is actually SHA1".
if command -v sha1sum >/dev/null 2>&1; then
  check_eq iksemel/sha1-vs-sha1sum "$(sha1sum "$COR" | cut -d' ' -f1)" "$sha"
else
  bst_skip iksemel/sha1-vs-sha1sum "no sha1sum on this image"
fi

# ---------------------------------------------------------------------------
# 6. THE WASM32 DISCRIMINATOR. ikstack occupancy for one document is dominated by
#    pointer-sized fields in the iks node and attribute structs, so a build with
#    4-byte pointers MUST use strictly less arena than the 8-byte control for the
#    identical input. This is the one case here that an x86_64 binary packaged
#    into the image by mistake could not pass.
dom=$(iksperf -d "$COR" 2>&1)
alloc=$(printf '%s' "$dom" | sed -n 's/.*ikstack: \([0-9]*\) bytes allocated.*/\1/p')
used=$(printf '%s' "$dom" | sed -n 's/.*allocated, \([0-9]*\) bytes used.*/\1/p')
nat_used=$(nat IKS-STACK-USED)

if [ -z "$used" ] || [ -z "$alloc" ]; then
  bst_notok iksemel/stack-reported "iksperf -d printed no ikstack line: $(_bst_1line "$dom")"
elif [ "$used" -le 0 ] 2>/dev/null; then
  bst_notok iksemel/stack-reported "ikstack used=$used"
else
  bst_ok iksemel/stack-reported
fi

if [ -n "$used" ] && [ -n "$nat_used" ] && [ "$used" -lt "$nat_used" ] 2>/dev/null; then
  bst_ok iksemel/wasm32-stack-smaller
else
  bst_notok iksemel/wasm32-stack-smaller \
    "used=$used not < x86_64 $nat_used -- 32-bit pointers must shrink the arena"
fi

# The arena must also still be a superset of what it handed out; a used > alloc
# would mean iks_stack_stat is reading past its own bookkeeping.
if [ -n "$used" ] && [ -n "$alloc" ] && [ "$used" -le "$alloc" ] 2>/dev/null; then
  bst_ok iksemel/stack-used-within-alloc
else
  bst_notok iksemel/stack-used-within-alloc "used=$used alloc=$alloc"
fi

# ---------------------------------------------------------------------------
# 7. Well-formedness verdicts on one violation per file.
#
# These pin OBSERVED behaviour on both arms, including where iksemel is LAX:
# it does not enforce entity declaration, a single root element, or a construct
# left open at EOF, so three of the five are rc=0 with no message. That is not a
# brintOS defect and not something this port fixes -- it is upstream's parser,
# and pinning it is what makes a real change in the error paths show up as a red.
for n in bad-entity double-root mismatch stray-lt unclosed-attr; do
  f="$BST_DATA/iksemel-bad-$n.xml"
  if [ ! -r "$f" ]; then
    bst_notok "iksemel/bad-$n-rc" "missing payload $f"
    continue
  fi
  out=$(ikslint "$f" 2>&1) && rc=0 || rc=$?
  check_eq "iksemel/bad-$n-rc" "$(nat "IKS-BAD-$n-RC")" "$rc"
  # The fixture path is stripped on both sides: the payload sits at a different
  # absolute path here than it did on the build host.
  msg=$(printf '%s' "$out" | head -1 | sed "s| in file '.*'||")
  check_eq "iksemel/bad-$n-msg" "$(nat "IKS-BAD-$n-MSG")" "$msg"
done

# ---------------------------------------------------------------------------
# 8. The TLS boundary the bbappend chose (PACKAGECONFIG = "").
#
# gnutls is not ported, so iks_has_tls() is 0 and the library has no STARTTLS.
# Asserted from the target through the only path that reports it without a
# compiler, so enabling TLS later is a CHANGED ASSERTION, not a silent gain.
if command -v iksroster >/dev/null 2>&1; then
  check_out iksemel/tls-absent-note "$(nat IKS-TLS-HELPNOTE)" iksroster --help
  rout=$(iksroster -s -b test@example.org 2>&1) && rrc=0 || rrc=$?
  check_eq iksemel/tls-secure-rc  "$(nat IKS-TLS-SECURE-RC)" "$rrc"
  check_eq iksemel/tls-secure-msg "$(nat IKS-TLS-SECURE-MSG)" \
    "$(printf '%s' "$rout" | sed -n '2p')"
  check_eq iksemel/roster-version "$(nat IKS-VER-roster)" "$(iksroster -V 2>&1 | head -1)"
else
  for c in tls-absent-note tls-secure-rc tls-secure-msg roster-version; do
    bst_notok "iksemel/$c" "iksroster not installed (iksemel-bin missing from the image?)"
  done
fi
check_eq iksemel/perf-version "$(nat IKS-VER-perf)" "$(iksperf -V 2>&1 | head -1)"

# ---------------------------------------------------------------------------
# 9. THE READ PATH ACROSS TWO FILESYSTEMS.
#
# ikslint sizes its read buffer from st_blksize (tools/ikslint.c:161), so it is
# one of the few tools on the image whose behaviour is a direct function of what
# the filesystem reports. The payload lives on the virtiofs root; a tmpfs copy
# is the control. libmaxminddb's port set this precedent after psmisc found the
# virtiofs root refusing an operation tmpfs allowed -- the class is real here,
# not hypothetical, and a parser that reads short on one of the two would show up
# as a changed character-data total rather than as an error.
cp "$COR" "$W/corpus.xml" 2>/dev/null
if [ -r "$W/corpus.xml" ]; then
  a=$(ikslint -s "$COR"        2>&1 | sed -n 's/^Total size of character data: \([0-9]*\).*/\1/p')
  b=$(ikslint -s "$W/corpus.xml" 2>&1 | sed -n 's/^Total size of character data: \([0-9]*\).*/\1/p')
  check_eq iksemel/read-virtiofs-vs-tmpfs "$a" "$b"
  check_eq iksemel/read-tmpfs-cdata "$(nat IKS-LINT-CDATA)" "$b"
else
  bst_notok iksemel/read-virtiofs-vs-tmpfs "could not copy corpus to $W"
fi

# ---------------------------------------------------------------------------
# 10. Packaging: what a consumer of this port actually links against, which
#     nothing above exercises.
#
# THIS ARM IS STATIC ONLY. The distro builds no shared libraries for the guest,
# so libtool produces libiksemel.a and no libiksemel.so -- which is why the
# upstream conformance programs above are plain wasm modules rather than libtool
# wrappers. Asserting a .so here would be asserting a distro property that is
# false; asserting the .a is asserting the deliverable.
if [ -f /usr/lib/libiksemel.a ]; then
  bst_ok iksemel/libiksemel-a-installed
else
  bst_notok iksemel/libiksemel-a-installed "no /usr/lib/libiksemel.a"
fi

# The absence of a .so is pinned deliberately, not left unstated: if this distro
# ever gains shared libraries, this case turns red and someone re-reads the port
# rather than discovering the change through a link failure downstream.
if ls /usr/lib/libiksemel.so* >/dev/null 2>&1; then
  bst_notok iksemel/no-shared-lib "a libiksemel.so appeared -- this arm was static-only when the port landed"
else
  bst_ok iksemel/no-shared-lib
fi

check iksemel/pkgconfig-installed test -f /usr/lib/pkgconfig/iksemel.pc
check iksemel/header-installed    test -f /usr/include/iksemel.h