# BST-TIMEOUT: 600   # 145 cases, measured 41s on a host under full build load
#                    # (==BST-SUITE libidn2 elapsed=41s cases=145). Cost is one
#                    # 2.6 MB wasm module start plus 145 shell-level compares, so
#                    # the figure is start-up dominated and does not grow with a
#                    # few more cases. The declared bound is well over the 60s
#                    # default because 41s already sits too close to it.
. "$BST_LIB"

# libidn2 2.3.7 -- GNU's IDNA2008 + Unicode TR46 implementation (poky
# recipes-extended/libidn). What ships is /usr/lib/libidn2.a, /usr/include/idn2.h,
# /usr/lib/pkgconfig/libidn2.pc (libidn2-dev + -staticdev) and /usr/bin/idn2
# (libidn2-bin).
#
# ---------------------------------------------------------------------------
# WHY THE CLI IS *NOT* THIS LIBRARY'S COVERAGE HERE, THOUGH IT IS FOR MOST PORTS
#
# 79-libsmi drives smilint/smidump because those tools ARE libsmi's callers and
# everything they print is a value the library computed. idn2(1) is the same shape
# of tool -- and on this image it cannot play that role, for a reason that is
# about the image and not about libidn2.
#
# Every entry point idn2(1) uses is a LOCALE-flavoured one -- idn2_to_ascii_lz,
# idn2_to_unicode_lzlz, idn2_register_ul -- and each begins by transcoding its
# argument from locale_charset() to UTF-8. This image ships no locale data at all:
# no /usr/lib/locale, no locale-archive, no gconv modules. So setlocale(LC_ALL,"")
# lands on "C", nl_langinfo(CODESET) answers ANSI_X3.4-1968, and every non-ASCII
# argument is refused with IDN2_ICONV_FAIL *before any IDNA code runs*. The CLI
# covers the tool and the error surface; it cannot reach one codepoint of the
# Unicode semantics libidn2 exists for.
#
# That boundary is ASSERTED below (idn2/locale-charset, idn2/nonascii-refused)
# rather than left to be discovered, and it is the image's boundary, not the
# library's: the day this image gains a UTF-8 locale those two cases change, and
# they are the prompt to widen the CLI coverage rather than a regression.
#
# AND SEPARATELY, idn2(1) ABORTS on this arm -- `free(): invalid size` on any
# input string, while the library underneath is clean. Nine CLI cases are red for
# that reason and each one says so in its own detail field, naming
# brintos/hwjs-cc#46; the full analysis and what rules libidn2 out is in the CLI
# block below. Read that before treating any of those reds as a fault in this
# port.
#
# ---------------------------------------------------------------------------
# SO THE LIBRARY IS COVERED BY A CROSS-LINKED PROBE, WHICH IS ALSO THE POINT
#
# libidn2's own callers are C programs -- curl, wget and libpsl all call
# idn2_lookup_u8 or the _8z forms, none of which touches a locale -- and this
# ticket ships libidn2 as an ENABLER for exactly those. data/idn2-probe is
# src/libidn2_probe.c cross-linked with wasmcc-glibc against THE VERY ARCHIVE
# THAT SHIPS, so it answers the question a library port has to answer: can a
# consumer link this .a and get right answers? It is the BRINTOS-217 instrument
# (oniguruma compiled and packaged perfectly, then returned wrong offsets to
# every caller), and a payload-shape-only suite would go green on that.
#
# The probe MEASURES and does not judge: 122 lines of `IDN2-<case> <rc-name>
# <value>`, scored against data/libidn2-native.txt -- the byte-identical output of
# the same source built natively against the same libidn2 2.3.7 and libunistring
# 1.2 tarballs (src/libidn2_mknative.sh, which also runs upstream's own `make
# check` against that build: 15/15 PASS). Upstream's suite is sound and is not the
# oracle only because it cannot run here -- its drivers are in-tree test programs
# no package installs, IdnaTestV2.txt is not installed either, and there is no
# compiler on the image to build them with.
#
# Nothing the probe prints carries a pointer width, a sizeof, an address, a locale
# or a clock, so no value is allowed to differ between the arms -- with the single
# exception of its first line, the pointer-width banner, which is stripped before
# the comparison and asserted on its own.

ORACLE="$BST_DATA/libidn2-native.txt"
PROBE="$BST_DATA/idn2-probe"
TMP=${TMPDIR:-/tmp}

# --- payload shape ----------------------------------------------------------
check idn2/lib-archive test -f /usr/lib/libidn2.a
check idn2/header     test -f /usr/include/idn2.h
check idn2/pkgconfig  test -f /usr/lib/pkgconfig/libidn2.pc
check idn2/cli        test -x /usr/bin/idn2

# Static-only arm's standing invariant. `set --` so the glob is evaluated once and
# its non-expansion (no match) is the passing case.
set -- /usr/lib/libidn2.so*
if [ "$1" = '/usr/lib/libidn2.so*' ]; then bst_ok idn2/no-shared
else bst_notok idn2/no-shared "shared object present: $*"; fi

# --- the CLI ----------------------------------------------------------------
#
# THESE CASES ARE RED ON PURPOSE AND EACH SAYS WHY. idn2(1) aborts with
# `free(): invalid size` on ANY input string on wasm32-hwjs -- so the ten cases
# below that give it one are reds attributed to that defect, not to libidn2.
# They are NOT downgraded to skip: the tool is present and broken, which is
# exactly what `not ok` is for, and turning it green would hide the one thing
# this suite found that the port itself did not.
#
# What is known about the defect (BRINTOS-544 Investigation Log 5, measured on
# this image):
#   - `idn2 --version` and `--help` are clean, rc=0. They return before
#     process_input(); every input path dies there, before printing, so the
#     computed value never reaches even a line-buffered tty.
#   - glibc's own iconv_open/iconv/iconv_close for UTF-8 <- ANSI_X3.4-1968 is
#     clean on this image.
#   - the library is clean: 129/129 of the probe cases below pass, including
#     idn2_to_ascii_lz() -- the very call process_input() makes -- returning the
#     right answer and freeing cleanly, in a probe linked against the same two
#     archives.
#   - so the fault is in the shipped binary's own module, not in libidn2, not in
#     libunistring's conversion, and not in glibc's iconv. It reproduces 100% in
#     /usr/bin/idn2 and not at all in a differently-linked probe of the same
#     calls, which is a heap corruption sensitive to module layout -- a
#     miscompile signature. Filed as brintos/hwjs-cc#46. `free(): invalid size`
#     from a gnulib malloca/freea path is the same shape as brintos/hwjs-cc#27
#     (u16_asnprintf %s) and the coroutinize pass's own ALLOCSIZE-STRIP note on
#     libunistring_mmalloca (brintos/hwjs-cc#12), but SAME ROOT CAUSE IS NOT
#     ESTABLISHED here.
#
# One helper runs every CLI case so the abort is recognised and attributed
# uniformly -- and so that when the defect is fixed these cases start asserting
# the real behaviour again with no edit.
IDN2_ABORT='free(): invalid size'
IDN2_ATTR='idn2(1) aborts with free(): invalid size in process_input() on wasm32-hwjs; the LIBRARY is clean (129/129 probe cases, incl. idn2_to_ascii_lz). Not libidn2 and not this port -- brintos/hwjs-cc#46 https://brintos.io/brintos/hwjs-cc/issues/46 ; see 78-libidn2.test header'

# cli <id> <want-rc> <want-substring> <args...>
cli() {
  _cid=$1; _crc=$2; _cw=$3; shift 3
  _co=$(idn2 "$@" 2>&1); _cr=$?
  case "$_co" in
    *"$IDN2_ABORT"*)
      bst_notok "$_cid" "$IDN2_ATTR (rc=$_cr)" ; return ;;
  esac
  if bst_execfail "$_co"; then bst_unrun "$_cid" "$_co"; return; fi
  case "$_co" in *"$_cw"*) _ch=1 ;; *) _ch=0 ;; esac
  if [ "$_cr" -eq "$_crc" ] && [ "$_ch" -eq 1 ]; then bst_ok "$_cid"
  else bst_notok "$_cid" "rc=$_cr want=$_crc want:'$_cw' got:'$(_bst_1line "$_co")'"; fi
}

cli idn2/cli-version      0 "idn2 (Libidn2) 2.3.7"   --version
cli idn2/cli-ascii        0 "example.com"            example.com
cli idn2/cli-alabel       0 "xn--rksmrgs-5wao1o.se"  xn--rksmrgs-5wao1o.se
cli idn2/cli-decode-ascii 0 "example.com"            -d example.com

# stdin rather than argv -- a different path into the same process_input()
sout=$(printf 'example.com\n' | idn2 2>&1); src=$?
case "$sout" in
  *"$IDN2_ABORT"*) bst_notok idn2/cli-stdin "$IDN2_ATTR (rc=$src)" ;;
  *example.com*)   [ "$src" -eq 0 ] && bst_ok idn2/cli-stdin \
                     || bst_notok idn2/cli-stdin "rc=$src" ;;
  *)               bst_notok idn2/cli-stdin "rc=$src got:'$(_bst_1line "$sout")'" ;;
esac

# A 64-octet label is one over the IDNA limit; the CLI reports it and exits 1.
long=aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
cli idn2/cli-rc-toolong  1 ""                                          "$long.com"
cli idn2/cli-msg-toolong 1 "domain label longer than 63 characters"     "$long.com"
cli idn2/cli-rc-hyphen   1 "string start/ends with forbidden hyphen"    -- -bad.com

# THE IMAGE'S LOCALE BOUNDARY, asserted rather than discovered. See the header:
# these three describe the image, not libidn2, and they change the day a UTF-8
# locale ships. The UTF-8 bytes are written as octal escapes so this file stays
# pure ASCII in transport -- "raksmorgas.se" with a-diaeresis, o-diaeresis and
# a-ring.
#
# locale-charset is the one that survives the abort today: idn2 --debug prints
# Charset: to STDERR, unbuffered, before process_input() dies -- so it is scored
# on the charset alone and the abort is attributed separately.
dbg=$(idn2 --debug example.com 2>&1); dbgrc=$?
case "$dbg" in
  *"Charset: ANSI_X3.4-1968"*) bst_ok idn2/locale-charset ;;
  *"$IDN2_ABORT"*)             bst_notok idn2/locale-charset "$IDN2_ATTR (rc=$dbgrc)" ;;
  *)                           bst_notok idn2/locale-charset "rc=$dbgrc got:'$(_bst_1line "$dbg")'" ;;
esac

uarg=$(printf 'r\303\244ksm\303\266rg\303\245s.se')
cli idn2/nonascii-refused 1 ""                                     "$uarg"
cli idn2/nonascii-message 1 "could not convert string to UTF-8"    "$uarg"

# --- the library RUNNING, via the cross-linked probe -------------------------
if [ ! -f "$ORACLE" ]; then
  bst_notok idn2/oracle "native control $ORACLE not installed"
  exit 0
fi
bst_ok idn2/oracle

body=$TMP/bst-idn2-out.$$
ctl=$TMP/bst-idn2-ctl.$$
probe=$TMP/bst-idn2-probe.$$
trap 'rm -f "$body" "$ctl" "$probe"' EXIT

if ! cp "$PROBE" "$probe" 2>/dev/null || ! chmod +x "$probe" 2>/dev/null; then
  bst_notok idn2/probe-stage "cannot stage $PROBE"
  exit 0
fi
bst_ok idn2/probe-stage

# Run ONCE and keep both the output and the status: on this runtime a multi-MB
# wasm module's start-up dominates the whole suite's cost, so a second run to
# re-read the exit code would roughly double it.
raw=$("$probe" 2>&1); prc=$?

if bst_execfail "$raw"; then
  bst_unrun idn2/probe-exit "$raw"
  exit 0
fi
check_eq idn2/probe-exit 0 "$prc"

# The banner is the one value that legitimately differs between the arms, so it
# is asserted here and stripped from everything below.
banner=$(printf '%s\n' "$raw" | sed -n 1p)
check_eq idn2/probe-pointer-width "IDN2-PROBE libidn2 probe, 32-bit pointers" "$banner"
printf '%s\n' "$raw" | sed 1d > "$body"

grep -v '^#' "$ORACLE" > "$ctl"

# Completeness first: a per-line comparison that silently ran out of lines would
# report every remaining case as a mismatch against an empty string, or worse
# stop early and read as coverage. The count and the whole-file compare bound
# that before any individual case is scored.
check_eq idn2/native-line-count "$(wc -l < "$ctl" | tr -d ' ')" "$(wc -l < "$body" | tr -d ' ')"
if cmp -s "$ctl" "$body"; then bst_ok idn2/native-byte-identical
else bst_notok idn2/native-byte-identical "probe output differs from the native control -- see the per-case results below"; fi

# One case per measurement, read in lockstep off two descriptors so the loop
# forks nothing: check_eq is a shell function and the two files are already here.
# The key is compared as well as the value, so a misalignment reports itself
# instead of scoring case N's value against case N+1's expectation.
exec 3< "$ctl"
exec 4< "$body"
while read -r want <&3; do
  read -r got <&4 || got=""
  id=${want%% *}
  gid=${got%% *}
  case "$id" in IDN2-*) ;; *) continue ;; esac
  if [ "$id" != "$gid" ]; then
    bst_notok "idn2/${id#IDN2-}" "line misaligned: expected key $id, got '${gid:-<eof>}'"
    continue
  fi
  check_eq "idn2/${id#IDN2-}" "$want" "$got"
done
exec 3<&-
exec 4<&-