. "$BST_LIB"

# GNU libunistring 1.2 -- Unicode string manipulation for UTF-8/16/32. The
# distro is static-only, so the payload is 21 headers (-dev) plus libunistring.a
# (-staticdev), and nothing on this image links it yet.
#
# THIS SUITE IS NOT PAYLOAD-ONLY, for the reason 76-onig.test established: a
# library nobody links has a perfect payload whether or not it works. Every
# static assertion below is equally true of an archive whose Unicode tables were
# never consulted, so the evidence that matters is the two probes --
# cross-linked with wasmcc-glibc against THIS libunistring.a and driven below,
# with every expected value taken from a native x86_64 build of the identical
# 1.2 source.
#
# THERE ARE TWO PROBES, AND THE SPLIT IS A FINDING RATHER THAN A TIDY-UP:
#
#   data/unistring-probe        24 cases -- the table-driven halves (decoding,
#                               classification, case, normalisation, width, the
#                               name database, the three break algorithms). All
#                               green on target; its tally is UNI-SUMMARY 24/24.
#   data/unistring-printf-probe  8 cases -- the unistdio (u*_asnprintf) half,
#                               which is where the platform defect lives. It
#                               ABORTS on target partway through, so only its
#                               first four cases are observable; see the block
#                               at the end of this file.
#
# They are separate binaries because u16_asnprintf("%s") is miscompiled on
# wasm32-hwjs (brintos/hwjs-cc#27): it returns an EMPTY result for short inputs
# and aborts with `free(): invalid size` for longer ones. While the unistdio
# cases lived in the main probe that abort destroyed the summary line and every
# case after it, which is exactly how a real defect in one corner can be made to
# look like a broken port everywhere. Split, the blast radius is the unistdio
# cases alone and the other 24 still report.
#
# Four things here would be got wrong by copying a neighbouring suite, all
# measured rather than assumed:
#
#   1. THERE IS NO .pc FILE AND NO onig-config-STYLE STUB. libunistring ships
#      neither, so the pkg-config assertions that 76-onig.test and 76-cmark.test
#      spend a dozen cases on have nothing to assert here. Their ABSENCE is
#      asserted instead, so that a future recipe change adding one is noticed
#      rather than silently unasserted.
#   2. THE MAIN libunistring PACKAGE IS EMPTY AND IS NOT IN THE IMAGE. Only
#      -dev and -staticdev are installed (two manifest rows, the libmxml/onig
#      shape), so there is no /usr/lib/libunistring.so and no binary to require.
#      `require` is therefore driven off the ARCHIVE, not off a tool.
#   3. THE c32 PATCH IS A BUILD-TIME FIX AND HAS NO RUNTIME TRIPWIRE. Unlike
#      onig's mixed-* cases, nothing this suite can run distinguishes a patched
#      from an unpatched libunistring: unpatched, do_compile fails and no image
#      exists to test. That is stated rather than faked with a case that looks
#      like a guard and is not.
#   4. FULL CASE MAPPING CHANGES LENGTH. u8_toupper("straße") is "STRASSE" --
#      seven characters from six, because U+00DF maps to two. A suite asserting
#      a length-preserving result would fail against a CORRECT port, and one
#      asserting only rc=0 would pass against a library that returned the input
#      untouched.

root=${BST_ROOT:-}
inc=$root/usr/include
lib=$root/usr/lib/libunistring.a

# ------------------------------------------------------------------ payload --
# The archive is the package: with no tool and no .so, its absence is what a
# `require` must key on.
# bst.sh's `require` keys on a TOOL in $PATH and this package installs none, so
# the guard is spelled out here against the archive instead. It ends the file
# with `exit 0`, exactly as `require` does -- the runner gives each test its own
# shell, so `return` would not end anything.
if [ -f "$lib" ]; then
  bst_ok libunistring/archive-present
else
  bst_skip libunistring/archive-present "no $lib — libunistring not installed"
  exit 0
fi

check libunistring/archive-nonempty test -s "$lib"

# The thirteen top-level headers, one case each: a partial -dev install is a
# real and quiet failure mode, and "some headers are there" would not catch it.
for h in unicase uniconv unictype unigbrk unilbrk unimetadata uniname \
         uninorm unistdio unistr unitypes uniwbrk uniwidth; do
  check libunistring/hdr-$h test -f "$inc/$h.h"
done
# ...and the eight private ones under unistring/, which unistr.h includes and a
# consumer never names directly -- so they are the ones most easily dropped.
for h in cdefs iconveh inline localcharset stdbool stdint version woe32dll; do
  check libunistring/hdr-unistring-$h test -f "$inc/unistring/$h.h"
done

# The version macro, by value. This is the one static fact the probe also
# reports at RUNTIME (UNI-version below compares the linked-in variable against
# this macro), so the pair catches a header/archive version skew.
check_eq  libunistring/version-macro 0x010200 \
  "$(awk '/define[ \t]+_LIBUNISTRING_VERSION/{print $3}' "$inc/unistring/version.h")"
check_out libunistring/hdr-guard "#define _UNISTRING_VERSION_H" cat "$inc/unistring/version.h"

# Note 1: neither exists, and asserting so is what makes a future addition
# visible rather than silently uncovered.
check libunistring/no-pkgconfig test ! -f "$root/usr/lib/pkgconfig/libunistring.pc"
check libunistring/no-shared    test ! -f "$root/usr/lib/libunistring.so"

# ------------------------------------------------------------------ runtime --
# data/unistring-probe is cross-linked against the shipped libunistring.a. Its
# link is itself evidence: the coroutinize pass reported `refused=0` over 425
# coroutinized functions, including the seven vasnprintf instantiations whose
# dynamic alloca was expected to be refused and was not (see the bbappend).
#
# Every expectation below is a literal line of native output. They are NOT
# relaxed to match target behaviour -- a divergence here is a platform finding
# and must go red, which is the whole reason the control was built first.
probe=${TMPDIR:-/tmp}/bst-unistring-probe.$$
if cp "$BST_DATA/unistring-probe" "$probe" 2>/dev/null && chmod +x "$probe" 2>/dev/null; then
  out=$("$probe" 2>&1)

  # The linked-in variable agrees with the header the -dev package shipped.
  check_out libunistring/run-version \
    "UNI-version OK [runtime=0x010200 header=0x010200]"          printf '%s\n' "$out"

  # unistr: validation both ways, then decode. The stride is asserted with the
  # code point because a decoder that returns the right character and the wrong
  # length desynchronises the caller on the NEXT character, not this one.
  check_out libunistring/run-check-good \
    "UNI-u8-check-good OK [r=null]"                              printf '%s\n' "$out"
  check_out libunistring/run-check-bad \
    "UNI-u8-check-bad OK [off=1]"                                printf '%s\n' "$out"
  check_out libunistring/run-mbtouc \
    "UNI-u8-mbtouc OK [U+0041/1 U+00E9/2 U+65E5/3 U+1F600/4]"    printf '%s\n' "$out"
  # Units vs characters -- the distinction the library exists for.
  check_out libunistring/run-len \
    "UNI-u8-len OK [units=10 chars=4]"                           printf '%s\n' "$out"
  # The astral character is what separates the three encodings: 4 units in
  # UTF-8, 1 in UTF-32, a surrogate PAIR in UTF-16. n16=5 is the assertion a
  # fixed-width UTF-16 converter fails.
  check_out libunistring/run-roundtrip \
    "UNI-u8-u32-u16-roundtrip OK [n32=4 n16=5 back=10]"          printf '%s\n' "$out"

  # unictype: classification, the general-category NAME, a binary property and a
  # script lookup -- four different compiled tables.
  check_out libunistring/run-class \
    "UNI-uc-class OK [alpha(e9)=1 alpha(3)=0 digit(3)=1]"        printf '%s\n' "$out"
  check_out libunistring/run-category \
    "UNI-uc-category OK [e9=Ll 41=Lu 20=Zs]"                     printf '%s\n' "$out"
  check_out libunistring/run-property \
    "UNI-uc-property OK [ws=1 idstart=1 idstart(sp)=0]"          printf '%s\n' "$out"
  check_out libunistring/run-script \
    "UNI-uc-script OK [script(03B1)=Greek]"                      printf '%s\n' "$out"

  # unicase: the simple per-character mapping, then Note 4 -- the full mapping
  # that CHANGES LENGTH, which is the half a per-character table gets wrong.
  check_out libunistring/run-case-simple \
    "UNI-uc-case-simple OK [toupper(E9)=U+00C9 tolower(C9)=U+00E9]" printf '%s\n' "$out"
  check_out libunistring/run-toupper-full \
    "UNI-u8-toupper-full OK [n=7 hex=53 54 52 41 53 53 45]"      printf '%s\n' "$out"
  check_out libunistring/run-casecmp \
    "UNI-u8-casecmp OK [rc=0 cmp=0]"                             printf '%s\n' "$out"

  # uninorm: both directions, because a normaliser that is the identity function
  # passes either one alone; then canonical equivalence, which is the operation
  # that makes normalisation worth shipping (raw_differs=1 proves the two inputs
  # really are distinct byte strings); then a compatibility fold.
  check_out libunistring/run-nfd  "UNI-u8-nfd OK [n=3 hex=65 CC 81]"  printf '%s\n' "$out"
  check_out libunistring/run-nfc  "UNI-u8-nfc OK [n=2 hex=C3 A9]"     printf '%s\n' "$out"
  check_out libunistring/run-normcmp \
    "UNI-u8-normcmp OK [rc=0 cmp=0 raw_differs=1]"               printf '%s\n' "$out"
  check_out libunistring/run-nfkc "UNI-u8-nfkc OK [n=2 hex=66 69]"    printf '%s\n' "$out"

  # uniwidth: wide and zero-width, not just ASCII. A width function returning 1
  # for everything is right about 'A' and wrong about its reason to exist.
  check_out libunistring/run-width \
    "UNI-uc-width OK [A=1 CJK=2 comb=0]"                         printf '%s\n' "$out"
  check_out libunistring/run-strwidth "UNI-u8-strwidth OK [w=4]" printf '%s\n' "$out"

  # uniname: the largest table in the library, both directions.
  check_out libunistring/run-uniname-fwd \
    "UNI-uniname-forward OK [name=LATIN CAPITAL LETTER A]"       printf '%s\n' "$out"
  check_out libunistring/run-uniname-rev \
    "UNI-uniname-reverse OK [cp=U+03B1]"                         printf '%s\n' "$out"

  # The three break algorithms -- separate tables and separate state machines.
  check_out libunistring/run-linebreaks \
    "UNI-u8-linebreaks OK [lb=11121]"                            printf '%s\n' "$out"
  check_out libunistring/run-wordbreaks \
    "UNI-u8-wordbreaks OK [wb=00110]"                            printf '%s\n' "$out"
  # base + combining mark is ONE cluster of 3 bytes: the case a byte- or even
  # character-oriented cursor gets wrong.
  check_out libunistring/run-grapheme \
    "UNI-u8-grapheme OK [adv=3]"                                 printf '%s\n' "$out"

  # The probe's own tally and exit status. 24/24 is what the identical source
  # scores against a native x86_64 build of the identical release. The unistdio
  # cases are NOT here -- they are a separate binary, driven below.
  check_out libunistring/run-summary "UNI-SUMMARY 24/24" printf '%s\n' "$out"
  check_rc  libunistring/run-exit0 0 "$probe"
  rm -f "$probe"
else
  bst_notok libunistring/run-stage "cannot stage $BST_DATA/unistring-probe"
fi

# ----------------------------------------------------------------- unistdio --
# data/unistring-printf-probe drives the u*_asnprintf half against the same
# shipped archive. It runs cheapest-risk first -- the narrow width, then the
# integer directives, then %s at each output width, which is where the defect
# is -- so the most is learned before the abort.
#
# ON TARGET IT ABORTS at the first %s-at-u16 case with `free(): invalid size`
# (brintos/hwjs-cc#27), so exactly four of its eight cases are observable. Those
# four are asserted BY VALUE below; the abort itself is asserted as its own
# case, and is left FAILING rather than skipped so it flips green on its own
# when the platform fix lands -- which is the regression signal this suite is
# buying. The three cases the abort masks (u16-s-long, u8-s, u32-s) are named in
# that failure rather than asserted separately: they are one defect, and
# spending three more reds on it would overstate a single platform bug.
#
# Measured against a native x86_64 build of the identical source, where all
# eight pass and the tally reads UNIP-SUMMARY 8/8.
pprobe=${TMPDIR:-/tmp}/bst-unistring-printf.$$
if cp "$BST_DATA/unistring-printf-probe" "$pprobe" 2>/dev/null && chmod +x "$pprobe" 2>/dev/null; then
  pout=$("$pprobe" 2>&1); prc=$?

  # %U is libunistring's own directive for a UTF-8 argument. Asserted by VALUE,
  # so an accepted-but-miscompiled lowering is caught as well as a refused one.
  check_out libunistring/run-p-u8-asnprintf \
    "UNIP-u8-asnprintf OK [n=16 hex=78 3D 34 32 20 73 3D C3 A9 E6 97 A5 20 65 6E 64]" \
    printf '%s\n' "$pout"
  # A width/precision directive: the vasnprintf buffer size is computed from
  # max_width_length and max_precision_length, so a constant-size case would not
  # exercise that path at all.
  check_out libunistring/run-p-u8-asnprintf-width \
    "UNIP-u8-asnprintf-width OK [n=14 out=[      3.1416]]"       printf '%s\n' "$pout"
  # %d at u16 and u32 -- the controls that make the failure below specific to
  # the %s directive rather than to the u16 instantiation as a whole.
  check_out libunistring/run-p-u16-asnprintf-int \
    "UNIP-u16-asnprintf-int OK [n=3]"                            printf '%s\n' "$pout"
  check_out libunistring/run-p-u32-asnprintf-int \
    "UNIP-u32-asnprintf-int OK [n=3]"                            printf '%s\n' "$pout"

  # THE PLATFORM DEFECT. Left red on purpose -- see the block comment above.
  case "$pout" in
    *"UNIP-u16-asnprintf-s OK"*)
      bst_ok libunistring/run-p-u16-s ;;
    *)
      bst_notok libunistring/run-p-u16-s "u16_asnprintf with a %s directive is miscompiled on wasm32-hwjs (brintos/hwjs-cc#27): empty result (n=0) for short inputs, 'free(): invalid size' abort for longer ones -- probe rc=$prc. The same directive at u8 and u32 width, and %d at u16 width, are correct. The abort also masks u16-s-long, u8-s and u32-s." ;;
  esac

  # The probe's own tally, which the abort prevents it from ever printing.
  case "$pout" in
    *"UNIP-SUMMARY 8/8"*)
      bst_ok libunistring/run-p-summary ;;
    *)
      bst_notok libunistring/run-p-summary "no 'UNIP-SUMMARY 8/8' -- the probe aborts before its tally (brintos/hwjs-cc#27)" ;;
  esac
  rm -f "$pprobe"
else
  bst_notok libunistring/run-p-stage "cannot stage $BST_DATA/unistring-printf-probe"
fi