# BST-TIMEOUT: 300   # reads ~114 kB across four files; the cost here is execs, not bytes
#
# mailcap 2.1.54 -- /etc/mailcap, the RFC 1524 table that says which program
# opens which MIME type, and the mime.types extension-to-type map that goes with
# it. There is no consumer on this image yet (mime-support's run-mailcap is
# BRINTOS-1190 and had not landed when this was written), so the deliverable is
# the data, and this suite asserts that the data PARSES and RESOLVES rather than
# that the files exist.
#
# WHAT IS ACTUALLY IN THE PACKAGE, MEASURED, NOT ASSUMED. /etc/mailcap is 272
# bytes and SIX entries, all of them /usr/bin/xdg-open, with exactly one named
# field between them (copiousoutput on text/html) and NO continuation lines.
# That is upstream's own file -- byte-identical to an x86-64 install of the same
# release -- not a truncated payload. It follows that the shipped table cannot
# exercise a mailcap parser's two divergence points, so those are covered
# against data/mailcap-rfc1524.fixture instead and the real table is then read
# by a parser shown to handle them.
#
# THE TABLE IS TESTED BY ITS REAL INSTALLED PATH, NOT THROUGH THE SYMLINK.
# mailcap declares ALTERNATIVE mime.types on /etc/mime.types with no explicit
# priority, so update-alternatives installs the file as /etc/mime.types.mailcap
# and leaves /etc/mime.types a symlink to it -- and mime-support declares the
# same alternative at priority 90 against this one's class default of 10, so
# when that package lands the symlink moves to ITS table and this one stays
# where it is. Everything below reads /etc/mime.types.mailcap; the symlink gets
# one case of its own, whose expectation is DERIVED from what is installed.
#
# WRITTEN TO EXEC AS LITTLE AS POSSIBLE. On this arm a process's memory is not
# returned when it exits and /bin/sh is a 16 MB wasm module, so a suite's cost
# is dominated by how many programs it starts. Five execs cover the whole file.
. "$BST_LIB"

S=mailcap

# THE CANARY, which is not an assertion about mailcap at all.
#
# brintos_local_verify.py loses a suite's FIRST BST result line before the host
# parser ever sees it -- BRINTOS-1149, a defect in our own host-side tooling and
# not in the runner, the guest or any package. The harness reports the deficit
# (`declared N case(s), N-1 reached the console`) but 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` therefore loses the one line that mattered,
# silently -- and for this suite that first case would be `present-mailcap`, i.e.
# "the package's own table is on the image", which is the last red anyone would
# want eaten.
#
# So a case that asserts nothing stands in that position and the eaten line costs
# nothing. Nothing was moved out of the slot; a nothing-case was added in front
# of it, and every real case below is still counted. Same idiom and same reason
# as 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 1149 is fixed.
bst_ok $S/suite-canary

D=$BST_DATA
tmp=${TMPDIR:-/tmp}
W=$tmp/bst-mailcap.$$

require gawk      $S/gawk-present
require sha256sum $S/sha256sum-present
require find      $S/find-present

MAILCAP=/etc/mailcap
MTREAL=/etc/mime.types.mailcap
MTLINK=/etc/mime.types
NGINX=/etc/nginx/mime.types
FIXTURE=$D/mailcap-rfc1524.fixture

# --------------------------------------------------------------- payload shape
# One find over every path this suite has an opinion about, including the two
# that belong to mime-support: their presence is what selects the expectation
# for the alternatives case below, so it has to be measured, not assumed.
find "$MAILCAP" "$MTLINK" "$MTREAL" "$NGINX" \
     /etc/mime.types.mime-support /usr/bin/run-mailcap /usr/lib/mime/mailcap \
     -maxdepth 0 -printf '%p %y %s %l\n' 2>/dev/null > "$W.inv"

sha256sum "$MAILCAP" "$MTREAL" "$NGINX" "$MTLINK" 2>/dev/null > "$W.sums"

for pair in "mailcap:$MAILCAP" "mime-types:$MTREAL" "nginx:$NGINX" "fixture:$FIXTURE"; do
  nm=${pair%%:*}; f=${pair#*:}
  if [ -f "$f" ]; then bst_ok "$S/present-$nm"
  else bst_notok "$S/present-$nm" "$f is not a regular file"; fi
done

if [ ! -f "$MAILCAP" ] || [ ! -f "$MTREAL" ] || [ ! -f "$NGINX" ] || [ ! -f "$FIXTURE" ]; then
  bst_notok $S/parse-prereq "a payload file is missing; the parse batches cannot run"
  return 2>/dev/null || exit 0
fi

# ------------------------------------------------------- RFC 1524: shipped table
gawk -v PFX=MC \
     -v QUERIES='audio/mpeg,image/png,application/pdf,application/msword,application/postscript,text/html,text/plain,application/json,video/mp4' \
     -f "$D/mailcap-parse.awk" "$MAILCAP" > "$W.mc"
. "$W.mc"

check_eq $S/mailcap-entries   6 "$MC_ENTRIES"
check_eq $S/mailcap-comments  3 "$MC_COMMENTS"
check_eq $S/mailcap-blanks    4 "$MC_BLANKS"
# Every field-1 is a well-formed type/subtype and every named field is one RFC
# 1524 defines. A table that parses into entries but carries a field the spec
# has no name for is broken in a way a count of entries cannot see.
check_eq $S/mailcap-badtype   0 "$MC_BADTYPE"
check_eq $S/mailcap-badfield  0 "$MC_BADFIELD"
check_eq $S/mailcap-dangling  0 "$MC_DANGLING"
check_eq $S/mailcap-types \
  'audio/*,image/*,application/msword,application/pdf,application/postscript,text/html' \
  "$MC_TYPES"
# The shipped table has no continuations. Asserted rather than left implicit:
# if upstream ever adds one, the fixture-backed paths below start carrying the
# real table too and that is worth knowing.
check_eq $S/mailcap-nocont    0 "$MC_CONT"

# The fifth type in that list is written 'application/postscript ;' in the file,
# with a space before the separator. It appears there without the space because
# the parser trims each field -- which is the whole of that case; a separate
# assertion for it would only be the same statement said twice.

# ------------------------------------------------ RFC 1524: resolving a spread
# The entry INDEX is asserted alongside the command, so a table that resolves to
# the right program via the wrong entry still fails.
check_eq $S/resolve-audio-mpeg           '1:/usr/bin/xdg-open %s' "$MC_R_audio_mpeg"
check_eq $S/resolve-image-png            '2:/usr/bin/xdg-open %s' "$MC_R_image_png"
check_eq $S/resolve-application-msword   '3:/usr/bin/xdg-open %s' "$MC_R_application_msword"
check_eq $S/resolve-application-pdf      '4:/usr/bin/xdg-open %s' "$MC_R_application_pdf"
check_eq $S/resolve-application-ps       '5:/usr/bin/xdg-open %s' "$MC_R_application_postscript"
check_eq $S/resolve-text-html            '6:/usr/bin/xdg-open %s' "$MC_R_text_html"
check_eq $S/resolve-text-html-flag       'copiousoutput'          "$MC_RF_text_html"
# The negatives are the half that matters: there is no text/* and no
# application/* wildcard in this table, so a parser that treats 'audio/*' as
# matching anything passes every case above and fails these three.
check_eq $S/resolve-text-plain-none      'none' "$MC_R_text_plain"
check_eq $S/resolve-application-json-none 'none' "$MC_R_application_json"
check_eq $S/resolve-video-mp4-none       'none' "$MC_R_video_mp4"

# ------------------------------------- RFC 1524: continuation and \; escaping
gawk -v PFX=FX -f "$D/mailcap-parse.awk" "$FIXTURE" > "$W.fx"
. "$W.fx"

check_eq $S/fixture-entries  4 "$FX_ENTRIES"
check_eq $S/fixture-badtype  0 "$FX_BADTYPE"
check_eq $S/fixture-badfield 0 "$FX_BADFIELD"
check_eq $S/fixture-cont     2 "$FX_CONT"
# Three physical lines joined into one view-command field, with the backslash
# and the newline removed and nothing else.
check_eq $S/cont-physlines   3 "$FX_PHYS_1"
check_eq $S/cont-view        '/bin/bst-view %s --flag --tail' "$FX_VIEW_1"
check_eq $S/cont-flag        'needsterminal' "$FX_FLAGS_1"
check_eq $S/cont-description 'joined across three lines' "$FX_NV_1_description"
# The escaped '\;' must not split the field: a naive split(';') reports four
# fields and a view command truncated at '%s'.
check_eq $S/escape-nfields   3 "$FX_NF_2"
check_eq $S/escape-view      '/bin/bst-view %s ; /bin/bst-second %s' "$FX_VIEW_2"
check_eq $S/escape-flag      'copiousoutput' "$FX_FLAGS_2"
# Continuations that begin a new field: the leading blanks belong to the
# separator, not to the value, so the trim is what makes these readable.
check_eq $S/named-physlines  4 "$FX_PHYS_3"
check_eq $S/named-order      'test,nametemplate,description' "$FX_NAMED_3"
check_eq $S/named-test       'test -n "$DISPLAY"' "$FX_NV_3_test"
check_eq $S/named-nametmpl   '%s.bst' "$FX_NV_3_nametemplate"
check_eq $S/flags-both       'needsterminal,copiousoutput' "$FX_FLAGS_4"

# ------------------------------------------- mime.types, nginx table, integrity
gawk -v LOOKUPS='pdf,html,png,tar,json,mp3,ogg,ics,wasm,bst-no-such-ext' \
     -v RLOOKUPS='application/pdf,text/html,application/wasm,image/png,application/json' \
     -v CORE="$MAILCAP,$MTLINK,$MTREAL,$NGINX" \
     -f "$D/mailcap-mimetypes.awk" "$MTREAL" "$NGINX" "$W.sums" "$W.inv" > "$W.mt"
. "$W.mt"

# The inventory pass saw all four paths. This is the case that names the cause
# when the find above produces nothing -- which is what a find without -printf
# does, silently, leaving every size and type assertion below reading ''.
check_eq $S/inventory-core 4 "$MT_INV_CORE"

# These five numbers are upstream's own make-check (python3 test.py < mime.types)
# re-derived on target by the same rules. Pinning them is what makes a truncated
# or half-installed table fail: the file would still parse.
check_eq $S/mimetypes-types      2151 "$MT_TYPES"
check_eq $S/mimetypes-extensions 1393 "$MT_EXTS"
check_eq $S/mimetypes-dup-ext    0    "$MT_DUPEXT"
check_eq $S/mimetypes-dup-type   0    "$MT_DUPTYPE"
check_eq $S/mimetypes-bad-top    0    "$MT_BADTOP"
check_eq $S/mimetypes-toplevel \
  'application:1604,audio:171,chemical:1,font:6,image:89,message:24,model:41,multipart:16,text:95,video:102,x-conference:1,x-epoc:1' \
  "$MT_TOPLEVEL"

check_eq $S/ext-pdf   'application/pdf'   "$MT_E_pdf"
check_eq $S/ext-html  'text/html'         "$MT_E_html"
check_eq $S/ext-png   'image/png'         "$MT_E_png"
check_eq $S/ext-tar   'application/x-tar' "$MT_E_tar"
check_eq $S/ext-json  'application/json'  "$MT_E_json"
check_eq $S/ext-mp3   'audio/mpeg'        "$MT_E_mp3"
check_eq $S/ext-ogg   'audio/ogg'         "$MT_E_ogg"
check_eq $S/ext-ics   'text/calendar'     "$MT_E_ics"
check_eq $S/ext-wasm  'application/wasm'  "$MT_E_wasm"
check_eq $S/ext-absent 'none'             "$MT_E_bst_no_such_ext"

# The reverse direction, only where the table is unambiguous -- and text/html is
# carried as the case that shows the qualifier is real rather than decorative:
# it maps to two extensions (html, htm) and the suite says so instead of
# picking one.
check_eq $S/type-application-pdf  'pdf'         "$MT_T_application_pdf"
check_eq $S/type-image-png        'png'         "$MT_T_image_png"
check_eq $S/type-application-json 'json'        "$MT_T_application_json"
check_eq $S/type-application-wasm 'wasm'        "$MT_T_application_wasm"
check_eq $S/type-text-html-ambig  'ambiguous:2' "$MT_T_text_html"

# /etc/nginx/mime.types is GENERATED at build time from mime.types by
# generate-nginx-mimetypes.sh, on the build host, so it is the one file here
# that is not simply copied out of the tarball. Re-deriving it on target and
# comparing line for line is the assertion that the generation ran and produced
# the right thing for this arm.
check_eq $S/nginx-derived-lines 1078 "$MT_NGX_DERIV"
check_eq $S/nginx-file-lines    1080 "$MT_NGX_LINES"
check_eq $S/nginx-derivation    0    "$MT_NGX_DIFF"

# ------------------------------------------------ sizes and content, vs x86-64
# The three sha256 sums are of an INDEPENDENT x86-64 install of mailcap 2.1.54 --
# tarball unpacked, its own Makefile install run -- not of the .deb this image
# built, which would have made the comparison tautological. mailcap does not
# inherit allarch, so the wasm32 build genuinely ran its own install and its own
# nginx-table generation; that the bytes agree is a result, not a tautology.
check_eq $S/size-mailcap     272   "$MT_SZ__etc_mailcap"
check_eq $S/size-mime-types  74869 "$MT_SZ__etc_mime_types_mailcap"
check_eq $S/size-nginx       38661 "$MT_SZ__etc_nginx_mime_types"
check_eq $S/sha-mailcap \
  '9e701bfd119a566cd01950a1fa4049e3d0caf8cad684228ff04afb232aa911b8' "$MT_SHA__etc_mailcap"
check_eq $S/sha-mime-types \
  '6d89583307fd2941a32039b9d8f392195c6e9a2fc89c681e693a06a714b3a23d' "$MT_SHA__etc_mime_types_mailcap"
check_eq $S/sha-nginx \
  '845b3cf44e3d178299f218f6fffc4067a989d3f88895dc03bd931c7c9e7230c4' "$MT_SHA__etc_nginx_mime_types"

# --------------------------------------------------------- the alternatives link
# Both halves of the collision are asserted, and the EXPECTATION IS DERIVED from
# what is installed rather than written down: mime-support wins at priority 90
# when it is there, mailcap holds the link at the class default of 10 when it is
# not. Either way this package's own table is the file checked above, so the
# suite stays true whichever package landed first.
check_eq $S/alt-real-is-file 'f' "$MT_TY__etc_mime_types_mailcap"
check_eq $S/alt-link-is-link 'l' "$MT_TY__etc_mime_types"
if [ -f /etc/mime.types.mime-support ]; then
  alt_want=mime.types.mime-support
  alt_why='mime-support installed, ALTERNATIVE_PRIORITY 90'
else
  alt_want=mime.types.mailcap
  alt_why='mailcap alone, ALTERNATIVE_PRIORITY 10 (the update-alternatives default)'
fi
check_eq $S/alt-winner "$alt_want" "$MT_LN__etc_mime_types"
# And what the link RESOLVES to, not just what it says: sha256sum follows the
# symlink, so this is the same table read the way anything on the image would
# read it. It is only a statement about mailcap while mailcap holds the link.
if [ "$alt_want" = mime.types.mailcap ]; then
  check_eq $S/alt-link-resolves \
    '6d89583307fd2941a32039b9d8f392195c6e9a2fc89c681e693a06a714b3a23d' "$MT_SHA__etc_mime_types"
else
  bst_skip $S/alt-link-resolves "$alt_why -- /etc/mime.types is not this package's table"
fi

# ------------------------------------------------------------------ run-mailcap
# The ticket asks for this WHERE mime-support's run-mailcap is present, which is
# the only assertion here that drives the tables with the tool that exists to
# consume them: run-mailcap maps the EXTENSION through mime.types, then the TYPE
# through /etc/mailcap, and prints the command it would run. Both cases below
# therefore span both files at once.
#
# WHICH mime.types it reads is whoever holds the alternatives link -- ours today,
# mime-support's once BRINTOS-1190 lands. The /etc/mailcap half is always this
# package's, because mime-support installs its own table to /usr/lib/mime/mailcap
# and never to /etc/mailcap, so there is no collision on that side.
#
# It is a skip rather than a silent omission while the tool is absent: a skip is
# a coverage prompt the sweep can see. The branch is not untested for being
# skipped -- it was developed and run against these exact tables on an x86-64
# host that does carry run-mailcap, which is how the first draft's use of an
# extensionless filename (resolving to application/octet-stream, rc=3) was found.
#
# BRINTOS-1190 LANDED, SO run-mailcap IS NOW ON THE IMAGE -- AND IT RUNS.
# The skip below became these two cases. They were briefly RED: when 1190 first
# built, run-mailcap could not start at all on this arm because perl died in
# glibc's allocator, 20/20 measured -- linux-in-the-browser#154, the platform
# rather than either package. It measures 0/20 now, on a base that did not move
# (a60034d50dbb); see the #154 block in 99-mime-support.test for the before/after
# table and for what did change. Both cases are GREEN, and every data assertion
# above passed unchanged throughout -- neither this table nor mime-support's was
# ever implicated.
#
# THE CLASSIFIER IS THE POINT OF THIS BLOCK, and it is kept now that the reds it
# was written for have gone: #154 is still open upstream, nothing here fixed it
# deliberately, and a return would come back through this exact call. Left as a
# bare check_out, both reds carried `perl: warning: Setting locale failed` in
# their detail field: that preamble -- SEVEN lines, not the four this comment
# used to claim -- is on EVERY perl run on this image including the ones that
# succeed, and it filled the 100 characters a later reader attributes by, so a
# platform block read as a locale bug. Stripping it and naming the issue is what
# would make a returning red actionable. Same idiom, same wide pattern set as
# 36-libperlio-gzip-perl and 99-mime-support; the set is wide by measurement,
# not guesswork -- update-mime produced four different glibc spellings in 20 runs.
#
# BRINTOS-WORKAROUND: linux-in-the-browser#154 remove -- when #154 is CLOSED
# upstream these two become the ordinary check_out assertions they were written
# as, and this whole classifier goes.
if command -v run-mailcap >/dev/null 2>&1; then
  : > "$W.probe.pdf"; : > "$W.probe.html"
  for pair in "pdf:$W.probe.pdf:/usr/bin/xdg-open" "copiousoutput:$W.probe.html:.html |"; do
    cs=${pair%%:*}; rest=${pair#*:}; cf=${rest%%:*}; cw=${rest#*:}
    co=$(run-mailcap --norun --action=view "$cf" 2>&1); cr=$?
    cd_=$(printf '%s\n' "$co" | awk '/^perl: warning/{s=1;next} /^[ \t]/&&s{next} {s=0;print}')
    [ -n "$cd_" ] || cd_=$co
    case "$cd_" in
      *mremap_chunk*|*munmap_chunk*|*"corrupted size"*|*"corrupted double-linked list"*|\
      *"double free or corruption"*|*"free(): "*|*"malloc(): "*|*"realloc(): "*)
        bst_notok "$S/run-mailcap-$cs" \
          "rc=$cr $(_bst_1line "$cd_") -- linux-in-the-browser#154, perl aborts in glibc allocator; this table is not implicated" ;;
      *)
        case "$co" in
          *"$cw"*) [ "$cr" -eq 0 ] && bst_ok "$S/run-mailcap-$cs" \
                     || bst_notok "$S/run-mailcap-$cs" "rc=$cr $(_bst_1line "$cd_")" ;;
          *) bst_notok "$S/run-mailcap-$cs" "rc=$cr want:'$cw' got:'$(_bst_1line "$cd_")'" ;;
        esac ;;
    esac
  done
else
  bst_skip $S/run-mailcap "run-mailcap not installed on this image (mime-support: BRINTOS-1190)"
fi

rm -f "$W.inv" "$W.sums" "$W.mc" "$W.fx" "$W.mt" "$W.probe.pdf" "$W.probe.html"