# BST-TIMEOUT: 900   # 272 cases; measured 111s on the BRINTOS-1997 batch image
#
# BRINTOS-2017 -- leptonica 1.82.0 on wasm32-hwjs.
#
# THIS PORT IS A LIBRARY WITH TEN CLI PROGRAMS IN FRONT OF IT, and the two halves
# are covered by two different things for a reason. The ten programs convert
# formats and report headers; they binarise nothing, erode nothing, scale nothing
# and measure no skew. So the processing layer -- which is the half an OCR engine
# is built on and the half this port exists for -- is driven from
# data/leptonica-probe, one process, 198 of the 272 cases below.
#
# UPSTREAM'S OWN REGRESSION SUITE IS NOT WHAT RUNS HERE, and that is a judgement
# about what it would measure rather than a limitation. prog/reg_wrapper.sh drives
# each of the 142 `*_reg' drivers as `foo_reg generate && foo_reg compare', and the
# tarball ships no golden directory -- so `generate' writes the reference that
# `compare' then reads back, seconds later, from the same binary. On target that
# asserts self-consistency and cannot catch a codec that is wrong the same way
# twice. It also wants prog/'s 284 fixture files, 16 MB. The port's bbappend keeps
# those drivers out of `make all' for the link cost; this suite covers the
# operations they cover, against expectations that did not come from a previous run
# of the code under test.
#
# WHERE EACH EXPECTATION COMES FROM, because they are not all the same kind:
#
#   * CLOSED FORM. The fixtures are built by an integer recipe (see
#     src/leptonica_probe.c and src/leptonica_mkfixture.py) chosen so the answers
#     are derivable on paper: 277 ON pixels; a threshold at 128 on a `4*x' ramp
#     keeps 32 columns of 64 rows = 2048; a 3x3 erosion of a 20x12 rectangle, a 6x6
#     and one isolated pixel leaves 18x10 + 4x4 + 0 = 196; three components with
#     bounding boxes the recipe states outright.
#   * ALGEBRA. Opening is idempotent and a subset of its input; four quarter turns
#     are the identity; a half turn twice is the identity. These hold for ANY
#     correct implementation and need no reference at all.
#   * AN INDEPENDENT ENCODER. data/leptonica-fixture.png is written by
#     src/leptonica_mkfixture.py from the PNG spec and zlib -- leptonica never
#     touched it -- and data/leptonica-fixture.rgb is the decode any correct reader
#     must produce. That is what makes `convertformat f.png -> PPM, compare the
#     payload' a claim about leptonica rather than a round trip with itself.
#   * MEASURED ON x86-64, and confined to the cases named `*-vs-x86'. The
#     interpolating operations -- pixScale's linear arm, Sauvola, the shear
#     rotation -- have no closed form worth writing down, so their values come from
#     an x86-64 build of the SAME source with the SAME codec set (giflib on,
#     openjpeg and webp off) and are pinned here. A divergence in one of those says
#     the arithmetic moved and says nothing about the rest of the suite.
#
# THE FORMAT LAYER IS ASSERTED FIRST AND SEPARATELY because it is where a
# cross-build goes wrong: every lossless format is gated on bit-identical round
# trip, and JPEG -- the only lossy one -- on a bounded per-channel maximum
# difference instead. PackBits is in twice on purpose: for RGB, leptonica's TIFF
# writer DOWNGRADES it to Deflate and says so, and the header code coming back as
# ZIP rather than PackBits pins that; on 1 bpp it is honoured and the code comes
# back PackBits.
#
# THE TWO DECODERS THAT LEAVE THROUGH longjmp ARE DRIVEN ON PURPOSE. libjpeg's
# error_exit and libpng's png_error never return, so leptonica installs a setjmp
# landing pad for each (src/jpegio.c, src/pngio.c) -- 19 sites, and the only
# setjmp in the tarball. A truncated file reaches the pad from inside the decoder,
# which is the cross-frame shape HwjsCoroutinize has to survive; the assertion is
# that pixRead returns NULL and the process is still running to say so.

. "$BST_LIB"
require fileinfo leptonica/present

tmp=${TMPDIR:-/tmp}/bst-leptonica.$$
rm -rf "$tmp"; mkdir -p "$tmp" || { bst_notok leptonica/tmpdir "cannot mkdir $tmp"; exit 0; }

FIXPNG="$BST_DATA/leptonica-fixture.png"
FIXRGB="$BST_DATA/leptonica-fixture.rgb"

size() { wc -c < "$1" 2>/dev/null | tr -d ' '; }

# usage <id> <want-rc> <needle> <cmd...> -- a syntax line and the status beside
# it. check_out cannot express this: every one of leptonica's ten programs prints
# its syntax and exits 1, and check_out requires exit 0.
usage() {
  _id=$1; _wrc=$2; _needle=$3; shift 3
  _out=$("$@" 2>&1); _rc=$?
  if [ "$_rc" != "$_wrc" ]; then bst_notok "$_id" "rc=$_rc want $_wrc"; return; fi
  case "$_out" in *"$_needle"*) bst_ok "$_id" ;;
                  *) bst_notok "$_id" "$(_bst_1line "$_out")" ;; esac
}

# magic <id> <file> <n> <hex> -- the first n bytes of a file, as lowercase hex.
magic() {
  got=$(dd if="$2" bs=1 count="$3" 2>/dev/null | od -An -tx1 | tr -d ' \n')
  if [ "$got" = "$4" ]; then bst_ok "$1"; else bst_notok "$1" "got $got want $4"; fi
}

# ----------------------------------------------------- 1. what the port ships
for t in convertfilestopdf convertfilestops convertformat convertsegfilestopdf \
         convertsegfilestops converttopdf converttops fileinfo imagetops \
         xtractprotos; do
  if command -v "$t" >/dev/null 2>&1; then bst_ok "leptonica/prog-$t"
  else bst_notok "leptonica/prog-$t" "not installed (upstream bin_PROGRAMS names ten)"; fi
done

# The 282 noinst programs upstream builds and throws away are kept out of `make
# all' by the port's patch. None of them may appear on the image; a re-appearance
# is a patch that stopped applying, which is otherwise silent.
for t in alltests_reg binarize_reg conncomp_reg skew_reg scale_reg rotate1_reg \
         ioformats_reg pageseg_reg; do
  if command -v "$t" >/dev/null 2>&1; then
    bst_notok "leptonica/noinst-$t" "$t is installed; it is a noinst regression driver"
  else bst_ok "leptonica/noinst-$t"; fi
done

check leptonica/staticlib  test -f /usr/lib/liblept.a
check leptonica/pkgconfig  test -f /usr/lib/pkgconfig/lept.pc
for h in allheaders.h alltypes.h pix.h environ.h imageio.h morph.h; do
  check "leptonica/header-$h" test -f "/usr/include/leptonica/$h"
done

# Shape, not absence: this arm converts to shared libraries per recipe
# (BRINTOS-1740), so a .so here would be a conversion rather than a breach.
bst_lib_shape leptonica/no-shared "/usr/lib/liblept.so*"

# The .pc a consumer reads. `-llept' rather than `-lleptonica' is upstream's
# leptonica_OUTPUT_NAME (configure.ac:269), and the header dir has the package
# name appended, which is why an #include is "leptonica/allheaders.h".
pc=$(cat /usr/lib/pkgconfig/lept.pc 2>/dev/null)
case "$pc" in *"Version: 1.82.0"*) bst_ok leptonica/pc-version ;;
              *) bst_notok leptonica/pc-version "$(_bst_1line "$pc")" ;; esac
case "$pc" in *"-llept"*) bst_ok leptonica/pc-llept ;;
              *) bst_notok leptonica/pc-llept "no -llept in Libs" ;; esac
case "$pc" in *"/leptonica"*) bst_ok leptonica/pc-incdir ;;
              *) bst_notok leptonica/pc-incdir "includedir is not .../leptonica" ;; esac

# ------------------------------------------ 2. the fixture leptonica did not write
check    leptonica/fixture-png      test -f "$FIXPNG"
check    leptonica/fixture-rgb      test -f "$FIXRGB"
check_eq leptonica/fixture-rgb-size 12288 "$(size "$FIXRGB")"

cp "$FIXPNG" "$tmp/f.png" 2>/dev/null
cp "$FIXRGB" "$tmp/f.rgb" 2>/dev/null
check leptonica/fixture-staged test -f "$tmp/f.png"

# --------------------------------------------------- 3. the CLI layer, end to end
# fileinfo prints its report on stderr and returns 1 with the wrong argc.
# EVERY ONE OF THE TEN PRINTS ITS SYNTAX AND EXITS 1 WITH NO ARGUMENTS, which is
# worth asserting for all ten together: it is the cheapest statement that each is
# the real program, that it started, that its argv parsing ran, and that it
# returned rather than trapping. Two of them -- imagetops and fileinfo -- route
# the message through leptonica's own ERROR_INT and so carry an `Error in' prefix
# the other eight do not.
usage leptonica/fileinfo-usage             1 "Syntax:  fileinfo filein"        fileinfo
usage leptonica/convertformat-usage        1 "Syntax: convertformat filein"    convertformat
usage leptonica/converttops-usage          1 "Syntax: converttops filein"      converttops
usage leptonica/converttopdf-usage         1 "Syntax: converttopdf dir"        converttopdf
usage leptonica/convertfilestopdf-usage    1 "Syntax: convertfilestopdf dirin" convertfilestopdf
usage leptonica/convertfilestops-usage     1 "Syntax: convertfilestops dirin"  convertfilestops
usage leptonica/convertsegfilestopdf-usage 1 "Syntax: convertsegfilestopdf"    convertsegfilestopdf
usage leptonica/convertsegfilestops-usage  1 "Syntax: convertsegfilestops"     convertsegfilestops
usage leptonica/imagetops-usage            1 "Syntax:  imagetops <filein>"     imagetops
usage leptonica/xtractprotos-usage         1 "xtractprotos [-prestring="       xtractprotos

info=$(fileinfo "$tmp/f.png" 2>&1); rc=$?
if [ $rc -eq 0 ]; then bst_ok leptonica/fileinfo-rc
else bst_notok leptonica/fileinfo-rc "rc=$rc"; fi
for pair in \
    "fmt:input image format type: png" \
    "hdr:w = 64, h = 64, bps = 8, spp = 3, iscmap = 0" \
    "full:w = 64, h = 64, d = 32, spp = 3, wpl = 64" \
    "nocmap:colormap does not exist" ; do
  id=${pair%%:*}; needle=${pair#*:}
  case "$info" in *"$needle"*) bst_ok "leptonica/fileinfo-$id" ;;
                  *) bst_notok "leptonica/fileinfo-$id" "no '$needle'" ;; esac
done

# THE DECODE ORACLE. The PPM leptonica writes carries a comment line, so the
# payload is taken from the end rather than at a fixed offset; 64*64*3 bytes of
# it must be exactly what mkfixture put in the PNG.
if convertformat "$tmp/f.png" "$tmp/o.pnm" >/dev/null 2>&1; then bst_ok leptonica/cf-png-pnm
else bst_notok leptonica/cf-png-pnm "convertformat png -> pnm failed"; fi
tail -c 12288 "$tmp/o.pnm" > "$tmp/o.rgb" 2>/dev/null
check leptonica/png-decode-vs-independent cmp "$tmp/o.rgb" "$tmp/f.rgb"

# BMP and GIF, each back out through PPM, so the comparison is against the same
# independent bytes rather than against leptonica's other encoder.
if convertformat "$tmp/f.png" "$tmp/o.bmp" >/dev/null 2>&1; then bst_ok leptonica/cf-png-bmp
else bst_notok leptonica/cf-png-bmp "convertformat png -> bmp failed"; fi
convertformat "$tmp/o.bmp" "$tmp/b.pnm" >/dev/null 2>&1
tail -c 12288 "$tmp/b.pnm" > "$tmp/b.rgb" 2>/dev/null
check leptonica/bmp-roundtrip-exact cmp "$tmp/b.rgb" "$tmp/f.rgb"

if convertformat "$tmp/f.png" "$tmp/o.gif" >/dev/null 2>&1; then bst_ok leptonica/cf-png-gif
else bst_notok leptonica/cf-png-gif "convertformat png -> gif failed"; fi
convertformat "$tmp/o.gif" "$tmp/g.pnm" >/dev/null 2>&1
tail -c 12288 "$tmp/g.pnm" > "$tmp/g.rgb" 2>/dev/null
check leptonica/gif-roundtrip-exact cmp "$tmp/g.rgb" "$tmp/f.rgb"

hdr=$(fileinfo "$tmp/o.bmp" 2>&1)
case "$hdr" in *"format type: bmp"*) bst_ok leptonica/hdr-bmp ;;
               *) bst_notok leptonica/hdr-bmp "$(_bst_1line "$hdr")" ;; esac
hdr=$(fileinfo "$tmp/o.gif" 2>&1)
case "$hdr" in *"format type: gif"*) bst_ok leptonica/hdr-gif ;;
               *) bst_notok leptonica/hdr-gif "$(_bst_1line "$hdr")" ;; esac
# The GIF comes back 4 bpp colormapped: sixteen colours is what the fixture has,
# so a wider bps here means the palette path widened rather than mapped.
case "$hdr" in *"bps = 4, spp = 1, iscmap = 1"*) bst_ok leptonica/hdr-gif-cmap ;;
               *) bst_notok leptonica/hdr-gif-cmap "$(_bst_1line "$hdr")" ;; esac

# A .tif EXTENSION MEANS TIFF_G4, WHICH IS BILEVEL, so an RGB input cannot use it
# -- and upstream's documented answer is to transcode to PNG and rename the
# output, not to fail. The named output does not appear; a .png does.
( cd "$tmp" && convertformat f.png ext.tif ) >/dev/null 2>&1
check leptonica/cf-ext-tif-fallback  test ! -f "$tmp/ext.tif"
check leptonica/cf-ext-tif-wrote-png test -f "$tmp/ext.png"
# Asked explicitly, TIFF is honoured and stays 8 bps RGB.
convertformat "$tmp/f.png" "$tmp/e.tif" TIFF >/dev/null 2>&1
hdr=$(fileinfo "$tmp/e.tif" 2>&1)
case "$hdr" in *"format type: tif"*) bst_ok leptonica/cf-explicit-tiff ;;
               *) bst_notok leptonica/cf-explicit-tiff "$(_bst_1line "$hdr")" ;; esac
case "$hdr" in *"bps = 8, spp = 3"*) bst_ok leptonica/cf-explicit-tiff-depth ;;
               *) bst_notok leptonica/cf-explicit-tiff-depth "$(_bst_1line "$hdr")" ;; esac

# PostScript and PDF: leptonica writes both and reads neither, so the assertion
# is the signature and a plausible size.
if converttops "$tmp/f.png" "$tmp/one.ps" >/dev/null 2>&1; then bst_ok leptonica/converttops-rc
else bst_notok leptonica/converttops-rc "converttops failed"; fi
magic leptonica/converttops-magic "$tmp/one.ps" 4 "25215053"
if [ "$(size "$tmp/one.ps")" -gt 500 ] 2>/dev/null; then bst_ok leptonica/converttops-size
else bst_notok leptonica/converttops-size "$(size "$tmp/one.ps") bytes"; fi

# imagetops takes a PostScript LEVEL, not a compression percentage, and refuses
# anything outside 1..3 by name -- which is worth pinning because the syntax line
# calls it a "compression level".
if imagetops "$tmp/f.png" 2 "$tmp/two.ps" >/dev/null 2>&1; then bst_ok leptonica/imagetops-rc
else bst_notok leptonica/imagetops-rc "imagetops level 2 failed"; fi
magic     leptonica/imagetops-magic "$tmp/two.ps" 4 "25215053"
usage leptonica/imagetops-level0 1 "valid levels are: 1, 2, 3" \
      imagetops "$tmp/f.png" 0 "$tmp/bad.ps"

mkdir -p "$tmp/d" && cp "$tmp/f.png" "$tmp/d/" 2>/dev/null
if converttopdf "$tmp/d" "$tmp/one.pdf" >/dev/null 2>&1; then bst_ok leptonica/converttopdf-rc
else bst_notok leptonica/converttopdf-rc "converttopdf failed"; fi
magic leptonica/converttopdf-magic "$tmp/one.pdf" 5 "255044462d"

if convertfilestopdf "$tmp/d" allfiles 300 1.0 0 0 none "$tmp/many.pdf" \
     >/dev/null 2>&1; then bst_ok leptonica/convertfilestopdf-rc
else bst_notok leptonica/convertfilestopdf-rc "convertfilestopdf failed"; fi
magic leptonica/convertfilestopdf-magic "$tmp/many.pdf" 5 "255044462d"

if convertfilestops "$tmp/d" png 300 "$tmp/many.ps" >/dev/null 2>&1; then
  bst_ok leptonica/convertfilestops-rc
else bst_notok leptonica/convertfilestops-rc "convertfilestops failed"; fi
magic leptonica/convertfilestops-magic "$tmp/many.ps" 4 "25215053"

# The two segmented-page converters are not driven any further than the usage
# block above. Each wants a mask directory, a boxa file and a page-numbering
# prefix this suite has no business synthesising, and a fixture invented to
# satisfy the argument list would assert that the fixture was well formed rather
# than anything about the port.

# --------------------------------------------------------------- 4. the probe
probe=$tmp/leptonica-probe
OUT=$tmp/probe.out
check leptonica/probe-payload test -f "$BST_DATA/leptonica-probe"

if cp "$BST_DATA/leptonica-probe" "$probe" 2>/dev/null && chmod +x "$probe" 2>/dev/null; then
  bst_ok leptonica/probe-stage
  "$probe" "$tmp" > "$OUT" 2>"$tmp/probe.err"; prc=$?
  if [ $prc -eq 0 ]; then bst_ok leptonica/probe-rc
  else bst_notok leptonica/probe-rc "rc=$prc $(_bst_1line "$(cat "$tmp/probe.err")")"; fi
else
  bst_notok leptonica/probe-stage "cannot stage $BST_DATA/leptonica-probe"
  : > "$OUT"
fi

# v <key> -- one probe value, or the empty string. Every assertion below goes
# through it, so a probe that printed nothing lands on the whole of section 4 as
# named `not ok' lines rather than contributing nothing to the count.
v() { awk -v k="$1" '$1=="LEPT" && $2==k {print $3; exit}' "$OUT"; }

# eq <case> <want> <key>
eq() { check_eq "leptonica/$1" "$2" "$(v "$3")"; }

# The probe reached its last line: everything above it ran.
eq probe-end ok end

# 4a. what the library says it is, and which codecs it was built against.
eq version leptonica-1.82.0 version
libs=$(awk '$1=="LEPT" && $2=="imagelibs" {$1="";$2="";print;exit}' "$OUT")
for want in libgif libjpeg libpng libtiff zlib; do
  case "$libs" in *"$want"*) bst_ok "leptonica/imagelib-$want" ;;
                  *) bst_notok "leptonica/imagelib-$want" "not in: $(_bst_1line "$libs")" ;; esac
done
# webp is off by the stock recipe's --without-libwebp and openjpeg by upstream's
# default PACKAGECONFIG. Both are on this image for other ports, so their absence
# here is the recipe's decision and not a missing dependency.
for no in libwebp jp2k; do
  case "$libs" in *"$no"*) bst_notok "leptonica/imagelib-no-$no" "$no is linked in: $(_bst_1line "$libs")" ;;
                  *) bst_ok "leptonica/imagelib-no-$no" ;; esac
done
# giflib and libtiff are pinned by their own ports on this image, so the version
# leptonica reports for each is a statement about the link, not about leptonica.
case "$libs" in *"libgif 5.2.2"*) bst_ok leptonica/imagelib-gif-version ;;
                *) bst_notok leptonica/imagelib-gif-version "$(_bst_1line "$libs")" ;; esac
case "$libs" in *"libtiff 4.6.0"*) bst_ok leptonica/imagelib-tiff-version ;;
                *) bst_notok leptonica/imagelib-tiff-version "$(_bst_1line "$libs")" ;; esac

# 4b. the fixtures are what the integer recipe says -- closed form throughout.
eq fixtures ok fixtures
eq rgb-w 64 rgb-w
eq rgb-h 64 rgb-h
eq rgb-d 32 rgb-d
eq rgb-p00-r 0   rgb-p00-r
eq rgb-p00-g 255 rgb-p00-g
eq rgb-p00-b 0   rgb-p00-b
eq rgb-p63-r 255 rgb-p63-r
eq rgb-p63-g 0   rgb-p63-g
eq rgb-p63-b 255 rgb-p63-b
eq gray-p0   0   gray-p0
eq gray-p63  252 gray-p63
eq bin-on    277 bin-on
eq text-on   39360 text-on

# 4c. every format the enabled PACKAGECONFIG admits.
#
# Format codes are IFF_* from src/imageio.h: 1 BMP, 2 JFIF_JPEG, 3 PNG, 4 TIFF,
# 5 TIFF_PACKBITS, 8 TIFF_G4, 9 TIFF_LZW, 10 TIFF_ZIP, 11 PNM, 13 GIF. They are
# what pixReadHeader made of a file it did not write, so they assert the reader's
# own sniffing rather than the writer's intent -- which is how the PackBits
# downgrade below becomes visible at all.
for spec in "png:3:3" "tiff:4:3" "tzip:10:3" "tlzw:9:3" \
            "gif:13:1" "bmp:1:3" "pnm:11:3" \
            "gpng:3:1" "gpnm:11:1" "bpng:3:1" "bg4:8:1" "bpbm:11:1" "bpb:5:1"; do
  tag=$(printf '%s' "$spec" | cut -d: -f1)
  wfmt=$(printf '%s' "$spec" | cut -d: -f2)
  wspp=$(printf '%s' "$spec" | cut -d: -f3)
  eq "rt-$tag-write" ok      "rt-$tag-write"
  eq "rt-$tag-read"  ok      "rt-$tag-read"
  eq "rt-$tag-fmt"   "$wfmt" "rt-$tag-hdr-fmt"
  eq "rt-$tag-w"     64      "rt-$tag-hdr-w"
  eq "rt-$tag-h"     64      "rt-$tag-hdr-h"
  eq "rt-$tag-spp"   "$wspp" "rt-$tag-hdr-spp"
  eq "rt-$tag-same"  1       "rt-$tag-same"
done

# The GIF path is the one that goes through a colormap, so it is the one whose
# cmap flag is worth asserting; every other format here comes back direct.
eq rt-gif-cmap 1 rt-gif-hdr-cmap
eq rt-png-cmap 0 rt-png-hdr-cmap

# PACKBITS ON RGB IS SILENTLY DEFLATE. pixWriteStreamTiffWA refuses PackBits for
# bpp > 1 -- "invalid compression type 5 for bpp > 1; using TIFF_ZIP" -- so the
# request is honoured as ZIP, and the header code coming back 10 rather than 5 is
# what says so. On 1 bpp (rt-bpb above) it is honoured and the code is 5.
eq rt-tpb-write ok rt-tpb-write
eq rt-tpb-read  ok rt-tpb-read
eq rt-tpb-fmt   10 rt-tpb-hdr-fmt
eq rt-tpb-same  1  rt-tpb-same

# The lossless formats are bit-identical, which the `-same' cases above already
# say; these pin the per-channel difference at zero so a decode that is equal by
# some other definition still lands on a named case.
for tag in png tiff tzip tlzw tpb gif bmp pnm; do
  eq "rt-$tag-maxdiff" 0 "rt-$tag-maxdiff"
done

# JPEG IS THE ONLY LOSSY FORMAT HERE AND IS BOUNDED RATHER THAN PINNED. The
# fixture's 16x16 blocks of flat colour are the worst case for a DCT at quality
# 75 -- every block boundary is a step -- so a difference is expected; what is
# asserted is that it stays inside the band an x86-64 build of the same source
# measured (26), with headroom for a different libjpeg-turbo build.
eq rt-jpeg-write ok rt-jpeg-write
eq rt-jpeg-read  ok rt-jpeg-read
eq rt-jpeg-fmt   2  rt-jpeg-hdr-fmt
eq rt-jpeg-w     64 rt-jpeg-hdr-w
eq rt-jpeg-h     64 rt-jpeg-hdr-h
eq rt-jpeg-spp   3  rt-jpeg-hdr-spp
eq rt-jpeg-same  0  rt-jpeg-same
jd=$(v rt-jpeg-maxdiff)
if [ -n "$jd" ] && [ "$jd" -ge 1 ] && [ "$jd" -le 40 ] 2>/dev/null; then
  bst_ok leptonica/rt-jpeg-maxdiff-band
else
  bst_notok leptonica/rt-jpeg-maxdiff-band "maxdiff=$jd, want 1..40 (x86-64 measured 26)"
fi

eq ps-write   ok         ps-write
eq ps-magic   25214164   ps-magic
eq pdf-write  ok         pdf-write
eq pdf-magic  255044462d pdf-magic

# 4d. the two decoders that must leave through longjmp.
eq corrupt-png  null corrupt-png
eq corrupt-jpeg null corrupt-jpeg

# 4e. binarisation.
# Closed form: the ramp is 4*x, so `< 128' is x < 32 -- 32 columns, 64 rows.
eq thresh128-on 2048 thresh128-on
eq otsu-rc 0 otsu-rc
# Otsu on a uniform ramp has no interesting maximum, so the threshold is reported
# and banded rather than pinned: what matters is that it lands inside the data's
# range and splits it, not exactly where.
ot=$(v otsu-thresh)
if [ -n "$ot" ] && [ "$ot" -gt 20 ] && [ "$ot" -lt 230 ] 2>/dev/null; then
  bst_ok leptonica/otsu-thresh-band
else
  bst_notok leptonica/otsu-thresh-band "thresh=$ot, want 20..230 (x86-64 measured 86)"
fi
on=$(v otsu-on)
if [ -n "$on" ] && [ "$on" -gt 0 ] && [ "$on" -lt 4096 ] 2>/dev/null; then
  bst_ok leptonica/otsu-on-band
else bst_notok leptonica/otsu-on-band "on=$on, want 1..4095 (x86-64 measured 1408)"; fi
eq sauvola-rc 0 sauvola-rc
eq sauvola-on-vs-x86 192 sauvola-on

# 4f. morphology with a 3x3 brick -- every count closed form (see the header).
eq erode-on  196 erode-on
eq dilate-on 381 dilate-on
eq open-on   276 open-on
eq close-on  277 close-on
# Algebra, true of any correct implementation and of no particular one.
eq open-idempotent 1 open-idempotent
eq open-subset     0 open-subset

# 4g. scaling.
eq scale2-w  128 scale2-w
eq scale2-h  128 scale2-h
eq scale15-w 96  scale15-w
eq scale15-h 96  scale15-h
eq sample05-w 32 sample05-w
eq sample05-h 32 sample05-h
eq areamap05-w 32 areamap05-w
eq areamap05-h 32 areamap05-h
eq smooth025-w 16 smooth025-w
eq smooth025-h 16 smooth025-h
# Sampling and area-mapping have closed forms on this ramp and are asserted as
# such: sampling takes column 2x, worth 8x, so a row sums to 8*(0+..+31) = 3968
# and the image to 32*3968 = 126976. Area mapping averages each pair, (8x + 8x+4)
# / 2 = 8x + 2, so a row is 3968 + 64 = 4032 and the image 129024.
eq sample05-sum  126976 sample05-sum
eq areamap05-sum 129024 areamap05-sum
# The linear-interpolation arm has no such form; these are the x86-64 values.
eq scale2-sum-vs-x86  2080 scale2-sum
eq scale15-sum-vs-x86 1167 scale15-sum

# 4h. rotation.
eq rot90-w 64 rot90-w
eq rot90-h 64 rot90-h
# A quarter turn maps source (0,63) to destination (0,0); block index there is
# k = 0 + 4*3 = 12, so (r,g,b) = (204, 51, 0). Closed form, not measured.
eq rot90-p00-r 204 rot90-p00-r
eq rot90-p00-g 51  rot90-p00-g
eq rot90-p00-b 0   rot90-p00-b
eq rot180-twice-same 1 rot180-twice-same
eq rot90-x4-same     1 rot90-x4-same
eq rotam-w 64 rotam-w
eq rotam-h 64 rotam-h
# Rotation is about the centre, so the centre pixel is the one interpolation
# cannot move: column 32 of the ramp is 4*32 = 128.
eq rotam-centre 128 rotam-centre
eq rotshear-w 64 rotshear-w
eq rotshear-h 64 rotshear-h
eq rotshear-on-vs-x86 272 rotshear-on

# 4i. skew. The bars go in level and come out tilted by exactly 2 degrees, so
# there is a right answer no previous run of this code supplied. leptonica
# reports the angle that would UNDO the skew, and its sweep step bounds how close
# it can get, so the assertion is a band around -2 rather than a value.
eq skew-fixture ok skew-fixture
eq skew-rc 0 skew-rc
ang=$(v skew-angle)
case "$ang" in
  -1.[6-9]*|-2.[0-3]*) bst_ok leptonica/skew-angle-band ;;
  *) bst_notok leptonica/skew-angle-band "angle=$ang, want about -2 (x86-64 measured -2.0938)" ;;
esac
# Below 3.0 confidence pixDeskewGeneral declines to rotate at all (skew.c:123),
# so the fixture's geometry was chosen to clear that floor with room to spare --
# and this case is what says it still does.
conf=$(v skew-conf)
conf_i=$(printf '%s' "$conf" | cut -d. -f1)
if [ -n "$conf_i" ] && [ "$conf_i" -ge 5 ] 2>/dev/null; then bst_ok leptonica/skew-conf-floor
else bst_notok leptonica/skew-conf-floor "conf=$conf, want >= 5 (x86-64 measured 12.5011)"; fi
# A page with no skew must read as zero, which is the control the measurement
# needs, and the deskewed page must read the same way -- that is the whole claim.
for pair in "flat:skew-flat-angle" "residual:deskew-residual"; do
  id=${pair%%:*}; key=${pair#*:}
  a=$(v "$key")
  case "$a" in
    0.0*|-0.0*|0.1*|-0.1*|0.2*|-0.2*) bst_ok "leptonica/skew-$id" ;;
    *) bst_notok "leptonica/skew-$id" "$key=$a, want within 0.3 of zero" ;;
  esac
done

# 4j. connected components -- closed form, boxes and all.
eq cc8-count 3 cc8-count
eq cc4-count 3 cc4-count
check_eq leptonica/cc8-box0 "58,4,1,1"   "$(v cc8-box0)"
check_eq leptonica/cc8-box1 "10,8,20,12" "$(v cc8-box1)"
check_eq leptonica/cc8-box2 "44,40,6,6"  "$(v cc8-box2)"
# A 3x3 opening removes the isolated pixel and nothing else, so the count drops
# by exactly one -- a relation between two operations rather than a fact about
# either, and the one case here that a consistently wrong morphology still fails.
eq cc8-after-open 2 cc8-after-open

rm -rf "$tmp"