# xorg-minimal-fonts-verify.awk -- the payload and fonts.dir checks in ONE gawk
# pass.
#
# WHY ONE PASS. 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. The shell side of this suite runs one find, one
# `sha256sum -c' down a pipe and this gawk; every payload and index result line
# below is produced here rather than by a shell loop over 45 files.
#
# INPUTS, each named by a -v so FILENAME can discriminate them:
#   REF   the reference manifest       "<size> <sha256> <path>"
#   SUMS  `sha256sum -c' over the manifest, one verdict per file:
#         "<path>: OK" / "<path>: FAILED" / "<path>: FAILED open or read".
#         It is -c rather than a plain digest run because the check list arrives
#         down a PIPE and so costs nothing on the argv: naming these 45 files by
#         absolute path builds a ~2.1 kB argv, which fits, but the pipeline is
#         the idiom the packager README settled on after a 10 kB one wedged a
#         guest outright (brintos/hardwarejs#15) and it cannot drift from the
#         manifest.
#   INV   find -printf '%p %y %s %m %l\n' over the font directory, its parent
#         and /usr/lib/X11 -- so the symlink, the co-tenant question and the
#         absence of fonts.alias are all answered from the same pass
#   FDIR  the on-target /usr/share/fonts/X11/misc/fonts.dir, read as data
#   FACES the expected face table   "<basename>|<family>|<style>|<pixelsize>",
#         MEASURED ON TARGET, with the literal family `!refused' marking the 37
#         faces fc-query produces no pattern for at all -- see that file's own
#         header, and the x86-64 control that shows the split is upstream
#         fontconfig's rather than this arm's
#   FCQ   `fc-query' over all 44 faces in one call -- STDOUT, same four fields
#   FCQE  the same call's STDERR, one "Can't query face ..." line per refusal.
#         Kept apart from FCQ on purpose: a refusal is an expected outcome for
#         37 of the 44 here, so it has to be counted, not discarded
#   FCL   `fc-list' under a fontconfig config scoped to this directory alone,
#         "<file>: <family>:style=<style>:pixelsize=<n>"
#   CONF  the image's /etc/fonts/fonts.conf, read to confirm the REAL
#         configuration covers this directory -- the one thing the scoped run
#         above deliberately cannot say
#
# WHAT IT IS A STATEMENT ABOUT. This recipe compiles nothing -- SRC_URI is
# `file://misc' and do_install copies 45 files poky carries in its own tree --
# so no red here is about the wasm32 compiler. It is about the install and
# packaging path: that all 45 files were packaged rather than a subset, that
# ${datadir} resolved to /usr/share, that nothing rewrote a byte, that the
# ${libdir}/X11/fonts symlink survived packaging as a symlink, and that no
# other recipe on this image claims the same directory.

function emit(id, ok, detail) {
    if (ok) printf "BST ok %s/%s\n", S, id
    else    printf "BST not ok %s/%s : %s\n", S, id, substr(detail, 1, 100)
}
function base(p,   n, a) { n = split(p, a, "/"); return a[n] }

BEGIN {
    S      = "xorg-minimal-fonts"
    DIR    = "/usr/share/fonts/X11/misc"
    XDIR   = "/usr/share/fonts/X11"
    LIBX   = "/usr/lib/X11"
    LINK   = "/usr/lib/X11/fonts"
    # THE TWO NUMBERS THE INDEX IS PINNED AGAINST, both measured on poky
    # b2c16f1e69 and both explained in the suite header: fonts.dir declares 409
    # entries and 365 of the files it names are NOT shipped by this package.
    WANT_DECL   = 409
    WANT_ABSENT = 365
    # FIFTEEN OF THE 409 XLFDs CONTAIN SPACES -- `-mutt-clearlyu alternate
    # glyphs-medium-r-normal--17-...' and its siblings. Measured; the first
    # version of this file required exactly two fields per line and reddened
    # on all fifteen. Pinned rather than merely tolerated, because "the index
    # has entries whose XLFD is not one word" is a property of the file worth
    # noticing if it changes.
    WANT_WIDE   = 15
    WANT_FILES  = 45
    WANT_FACES  = 44
    # THE SPLIT fc-query MAKES, measured on target and reproduced byte for byte
    # by fontconfig-native 2.15.0 on x86-64: seven faces answer, thirty-seven
    # are refused. Pinned in both directions.
    WANT_QUERYABLE = 7
    WANT_REFUSED   = 37
    # misc/ IS SHARED TOO. Each file another package puts here is named with the
    # owner dpkg's .list gives it, and left out of this package's counts -- the
    # 99-terminus-font scoping, since a co-tenant's payload is its own suite's
    # business. An unnamed arrival is still an extra-* red.
    MISCCOT["fonts.alias"] = "font-alias"
    MISCCOT["mtx.pcf"]     = "cmatrix"
    # DETERMINISTIC CASE ORDER. gawk leaves `for (k in arr)' unspecified, and a
    # transcript whose case order changes between runs is one nobody can diff
    # against the last one. The suite requires gawk, so this is available.
    PROCINFO["sorted_in"] = "@ind_str_asc"
}

FILENAME == REF  && /^#/               { next }
FILENAME == REF  && NF == 3            { WSZ[$3] = $1; WSHA[$3] = $2; NREF++; next }
FILENAME == SUMS && /: (OK|FAILED)/    { p = $0; sub(/: (OK|FAILED).*$/, "", p)
                                         GSUM[p] = ($0 ~ /: OK$/); NSUM++; next }
FILENAME == INV  && NF >= 4            { IT[$1] = $2; IS[$1] = $3; IM[$1] = $4
                                         IL[$1] = (NF >= 5 ? $5 : ""); NINV++; next }
FILENAME == FDIR && FNR == 1           { FCOUNT = $0; NFDIR++; next }
FILENAME == FDIR                       { NFDIR++
                                         if (NF < 2) { FBAD++; if (FBADL == "") FBADL = FNR ":" $0 }
                                         else { x = $0; sub(/^[ \t]*[^ \t]+[ \t]+/, "", x)
                                                # NOT substr($0, index($0, $2)): index() finds the
                                                # FIRST occurrence, and for `cursor.pcf.gz cursor'
                                                # that is inside field 1, so the XLFD came back as
                                                # the whole line. Measured -- it reddened exactly
                                                # the one entry this file most wants to pin.
                                                DECL[$1] = x; NDECL++
                                                if (NF > 2) NWIDE++ }
                                         next }
FILENAME == FACES && /^#/              { next }
FILENAME == FACES && NF > 0            { n = split($0, a, "|")
                                         if (n == 4) { EFAM[a[1]] = a[2]; ESTY[a[1]] = a[3]; EPX[a[1]] = a[4]; NFACE++ }
                                         next }
FILENAME == FCQ  && /\|/               { n = split($0, a, "|")
                                         if (n == 4) { QFAM[a[1]] = a[2]; QSTY[a[1]] = a[3]; QPX[a[1]] = a[4]; NFCQ++ }
                                         next }
FILENAME == FCQE && /Can't query face/ { p = $0; sub(/^.*of font file /, "", p)
                                         RFS[base(p)] = 1; NRFS++; next }
FILENAME == FCL  && /^\//              { p = $0; sub(/: .*$/, "", p)
                                         f = $0; sub(/^[^:]*: /, "", f)
                                         sub(/:.*$/, "", f)
                                         if (index(p, DIR "/") == 1 && !(base(p) in MISCCOT)) { LFAM[base(p)] = f; NFCL++ }
                                         next }
FILENAME == CONF && /<dir>/            { d = $0; sub(/^.*<dir>/, "", d); sub(/<\/dir>.*$/, "", d)
                                         CDIR[d] = 1; NCONF++; next }

END {
    # --- the four inputs were really read ---------------------------------
    # An empty input is the failure that hides best: every case below it would
    # be vacuously green. Each is asserted before anything is derived from it.
    emit("ref-rows",  NREF  == WANT_FILES, sprintf("manifest carries %d rows, want %d", NREF, WANT_FILES))
    emit("sum-rows",  NSUM  == WANT_FILES, sprintf("sha256sum -c returned %d verdicts, want %d", NSUM, WANT_FILES))
    emit("find-rows", NINV  >  0,          "find produced no rows -- the font directory is missing entirely")
    emit("fdir-rows", NFDIR >  0,          "fonts.dir produced no rows -- it did not ship or is empty")

    # --- the exact installed file list ------------------------------------
    # Counted from find rather than from the manifest, so an EXTRA file is a
    # red as well as a missing one. The names are pinned one case each below.
    nfound = 0
    for (p in IT) {
        if (p == DIR || p == XDIR || p == LIBX) continue
        if (index(p, DIR "/") != 1 || (base(p) in MISCCOT)) continue
        FOUND[p] = 1; nfound++
    }
    emit("file-count", nfound == WANT_FILES,
         sprintf("%s holds %d entries, want %d", DIR, nfound, WANT_FILES))

    for (p in FOUND)
        if (!(p in WSZ))
            emit("extra-" base(p), 0, sprintf("%s is installed but the manifest does not name it", p))

    # --- per-file: size, sha256, type and mode ----------------------------
    # One case per shipped file. 45 lines is what the ticket's "file names
    # pinned so a partial payload fails" asks for, and unlike terminus-font's
    # 238 collapsed console fonts these 45 files ARE the whole deliverable --
    # there is no second half for a reader to look at instead.
    for (p in WSZ) {
        id = "file-" base(p)
        if (!(p in IT))                 { emit(id, 0, "not installed"); continue }
        if (IT[p] != "f")               { emit(id, 0, sprintf("type is '%s', want a regular file", IT[p])); continue }
        if (IS[p] + 0 != WSZ[p] + 0)    { emit(id, 0, sprintf("size %s, want %s", IS[p], WSZ[p])); continue }
        if (IM[p] != "644")             { emit(id, 0, sprintf("mode %s, want 644", IM[p])); continue }
        if (!(p in GSUM))               { emit(id, 0, "sha256sum -c returned no verdict for it"); continue }
        if (!GSUM[p])                   { emit(id, 0, sprintf("sha256 differs from %s", substr(WSHA[p], 1, 16))); continue }
        emit(id, 1, "")
    }

    # --- the ${libdir}/X11/fonts symlink ----------------------------------
    # do_install makes it with `ln -sf ${datadir}/fonts/X11/ ...', so the target
    # carries a trailing slash and is absolute. It is asserted as a SYMLINK
    # because a packaging path that dereferenced it would produce a second copy
    # of the whole font tree and nothing else would notice.
    # THE FORM ON THE IMAGE IS RELATIVE, AND THAT IS NOT WHAT do_install WROTE.
    # The recipe runs `ln -sf ${datadir}/fonts/X11/ ${D}/${libdir}/X11/fonts',
    # an ABSOLUTE link, and xorg-minimal-fonts_1.0-r0_all.deb records exactly
    # that. The assembled rootfs -- both the staged tree and the .tar.zst it
    # comes from -- carries `../../share/fonts/X11' instead, so the conversion
    # happens during image assembly. Both spellings name the same directory, so
    # both are accepted and the case is about the link being a LINK to the font
    # tree; which form arrived is in the detail of a red.
    if (!(LINK in IT))            emit("libdir-symlink", 0, LINK " is absent")
    else if (IT[LINK] != "l")     emit("libdir-symlink", 0, sprintf("%s is type '%s', want a symlink", LINK, IT[LINK]))
    else if (IL[LINK] != XDIR && IL[LINK] != XDIR "/" &&
             IL[LINK] != "../../share/fonts/X11" && IL[LINK] != "../../share/fonts/X11/")
                                  emit("libdir-symlink", 0, sprintf("%s -> '%s', want '%s' or '../../share/fonts/X11'", LINK, IL[LINK], XDIR))
    else                          emit("libdir-symlink", 1, "")

    # --- fonts.alias IS here now, and the tripwire that said so has fired ---
    # This case used to assert the file's ABSENCE. It was written as an explicit
    # tripwire -- "if font-alias ever does land on the image its misc/fonts.alias
    # arrives in THIS directory, and the fc-* cases below would then be reporting
    # a different image than the one this port was verified on". BRINTOS-2310
    # ported font-alias and put it in packagegroup-brintos-extra, so it has.
    #
    # The tripwire is kept and turned round rather than deleted: a case removed to
    # make a suite green is what the partially-red rule forbids. What it now pins
    # is that the file is present AND that its arrival did not disturb the reading
    # below -- font-alias ships misc/fonts.alias, xorg-minimal-fonts ships the 44
    # faces and fonts.dir, the two sets are disjoint, and fontconfig reads neither
    # index: it scans the font files themselves with FreeType. So the fc-* cases
    # measure the same thing they measured on the 2343 image.
    #
    # Its OWNER is asserted, not just its presence. `misc/' is shared, so "the
    # file is there" would also pass if xorg-minimal-fonts had started shipping
    # one -- which is exactly what Debian Policy 11.8.5 forbids and what this
    # suite exists to notice.
    emit("fonts-alias-present", (DIR "/fonts.alias") in IT, DIR "/fonts.alias is absent -- BRINTOS-2310 put font-alias on this image and it ships that file")
    emit("no-fonts-scale", !((DIR "/fonts.scale") in IT), DIR "/fonts.scale is present -- something ran mkfontscale here")

    # --- /usr/share/fonts/X11 IS SHARED, and with whom is the assertion ----
    #
    # The package's FILES:${PN} claims ${datadir}/fonts/X11/ whole, which reads
    # as though this directory were its own. It is not, and this was measured
    # rather than assumed: ttf-pt-sans put eight faces in
    # /usr/share/fonts/X11/TTF before this port existed, and its own FILES
    # claims /usr/share. Nothing is shadowed -- the two subdirectories are
    # disjoint and neither package ships a file the other does -- so the
    # correct assertion is not "sole tenant" but "these tenants and no
    # others", the same scoping 99-terminus-font applies to the console-font
    # directory it shares with cmatrix.
    #
    # It is worth naming the neighbour rather than merely allowing it: the
    # largest of those eight, PTN77F.ttf at 584,812 bytes, is the face that
    # put this image over the file-backed mapping bracket
    # (linux-in-the-browser#165) and so is the direct reason the fontconfig
    # calls in this suite run under a scoped configuration.
    #
    # THE TENANT LIST GREW, AND EACH ARRIVAL HAS A NAMED OWNER. On the image
    # BRINTOS-2343 verified, TTF was the only entry beside misc/ and this case
    # pinned exactly that. BRINTOS-2310 added encodings, font-alias and font-util
    # to packagegroup-brintos-extra; read off the staged rootfs, with each
    # directory attributed to the package that owns it in dpkg's own .list file:
    #
    #     100dpi   75dpi   cyrillic   font-alias
    #     encodings                   encodings
    #     util                        font-util
    #     TTF                         ttf-pt-sans
    #
    # The assertion stays "these tenants and no others" -- the shape 2343 chose
    # deliberately over "sole tenant" -- with the roster widened to what is now
    # installed. A SEVENTH directory still trips it, which is the whole point of
    # the case; widening it to a bare count would have thrown that away.
    KNOWN["TTF"]       = "ttf-pt-sans"
    KNOWN["100dpi"]    = "font-alias"
    KNOWN["75dpi"]     = "font-alias"
    KNOWN["cyrillic"]  = "font-alias"
    KNOWN["encodings"] = "encodings"
    KNOWN["util"]      = "font-util"
    NKNOWN = 6

    ncot = 0; cot = ""; nbad = 0; badl = ""
    for (p in IT) {
        if (index(p, XDIR "/") != 1) continue
        if (p == DIR || index(p, DIR "/") == 1) continue
        b = base(p); ncot++
        if (!(b in KNOWN)) { nbad++; if (nbad <= 3) badl = badl " " b }
        else if (ncot <= 3) cot = cot " " b
    }
    emit("x11-dir-known-cotenants", nbad == 0 && ncot == NKNOWN,
         sprintf("%d entry(ies) under %s beside misc/, want exactly %d (TTF, 100dpi, 75dpi, cyrillic, encodings, util); unexpected:%s", ncot, XDIR, NKNOWN, badl))

    # --- fonts.dir: it parses, and it is self-consistent -------------------
    # "Parses" is three separate properties and they are asserted separately,
    # because a red in one localises differently from a red in the others:
    # the header is a bare decimal count; every remaining line is exactly two
    # whitespace-separated fields (a file name and an XLFD, neither of which
    # may contain a space); and the header agrees with the number of entries
    # that follow it.
    emit("fdir-header-numeric", FCOUNT ~ /^[0-9]+$/, sprintf("first line of fonts.dir is '%s', want a bare decimal count", substr(FCOUNT, 1, 40)))
    emit("fdir-fields", FBAD == 0, sprintf("%d fonts.dir line(s) carry no XLFD at all; first is %s", FBAD, FBADL))
    emit("fdir-multiword-xlfds", NWIDE == WANT_WIDE,
         sprintf("%d fonts.dir entries have an XLFD containing spaces, want %d", NWIDE, WANT_WIDE))
    emit("fdir-header-agrees", FCOUNT + 0 == NDECL, sprintf("fonts.dir header says %s, %d entry lines follow", FCOUNT, NDECL))
    emit("fdir-entry-count", NDECL == WANT_DECL, sprintf("fonts.dir declares %d entries, want %d", NDECL, WANT_DECL))

    # --- the index against what is actually installed ---------------------
    #
    # THE TICKET'S CLAUSE IS FALSE FOR THIS PAYLOAD AND THE SUITE PINS WHAT IS
    # TRUE INSTEAD. BRINTOS-2343 asks that the index's "declared entry count
    # matches the number of font files actually installed". It does not, and
    # cannot: this is the MINIMAL font set, whose fonts.dir is upstream's index
    # for the COMPLETE misc set. 409 entries are declared and 44 files ship, so
    # 365 declared names have no file. That is upstream's own state -- poky
    # carries both -- and a suite written to the clause as stated would be red
    # on a correct install.
    #
    # What IS assertable, and is what the clause was reaching for, is the
    # direction that matters: every file that ships is declared, so nothing is
    # installed that the index cannot name. Both numbers are pinned, so this
    # goes red if either the payload or the index moves.
    nmiss = 0; missl = ""
    for (p in FOUND) {
        b = base(p)
        if (b == "fonts.dir") continue
        if (!(b in DECL)) { nmiss++; if (nmiss <= 5) missl = missl " " b }
    }
    emit("fdir-covers-installed", nmiss == 0, sprintf("%d installed font(s) are not declared in fonts.dir:%s", nmiss, missl))

    nabs = 0
    for (b in DECL) if (!((DIR "/" b) in FOUND)) nabs++
    emit("fdir-declared-absent", nabs == WANT_ABSENT,
         sprintf("%d declared fonts are not shipped, want exactly %d -- the payload or the index has moved", nabs, WANT_ABSENT))

    # --- the entry the recipe's own LIC_FILES_CHKSUM names ------------------
    # cursor.pcf.gz is one of the two files the recipe checksums, and the one
    # every minimal X install expects to find. Its fonts.dir entry is also the
    # only one in the file that is not an XLFD: the name is the bare string
    # "cursor", which is what an X client asks for by name.
    emit("fdir-cursor-entry", DECL["cursor.pcf.gz"] == "cursor",
         sprintf("fonts.dir maps cursor.pcf.gz to '%s', want 'cursor'", DECL["cursor.pcf.gz"]))

    # ======================================================================
    # THE CONSUMER LAYER -- fontconfig, already on this image, can READ what
    # this package installed. A red here with the payload layer above green
    # says the bytes arrived and FreeType could not parse them; a red in both
    # says the bytes did not arrive. That split is the point of the two layers.
    # ======================================================================
    emit("faces-rows", NFACE == WANT_FACES, sprintf("expected-face table carries %d rows, want %d", NFACE, WANT_FACES))

    nq = 0; nr = 0
    for (b in EFAM) if (EFAM[b] == "!refused") nr++; else nq++
    emit("faces-split", nq == WANT_QUERYABLE && nr == WANT_REFUSED,
         sprintf("expected-face table says %d queryable / %d refused, want %d / %d", nq, nr, WANT_QUERYABLE, WANT_REFUSED))
    emit("fcq-queryable-count", NFCQ == WANT_QUERYABLE,
         sprintf("fc-query answered for %d face(s), want %d", NFCQ, WANT_QUERYABLE))
    emit("fcq-refused-count", NRFS == WANT_REFUSED,
         sprintf("fc-query refused %d face(s), want %d", NRFS, WANT_REFUSED))

    # --- per-face, and the refusals are pinned as firmly as the answers -----
    # The ticket asks that fc-query report family, style and pixel size FOR
    # EACH file. For 37 of the 44 it will not, on this arm or on x86-64, so
    # each of those is pinned as a REFUSAL instead: a face that starts
    # answering is as much a red as one that stops. That is the only form of
    # the clause that can be true of this payload, and it is per file rather
    # than a count precisely so that a single face changing sides is located.
    for (b in EFAM) {
        id = "fcq-" b
        if (EFAM[b] == "!refused") {
            if (b in QFAM)      { emit(id, 0, sprintf("fc-query now ANSWERS for it (%s/%s/%s) -- the table says it is refused", QFAM[b], QSTY[b], QPX[b])); continue }
            if (!(b in RFS))    { emit(id, 0, "fc-query neither answered for it nor refused it -- it was not read at all"); continue }
            emit(id, 1, ""); continue
        }
        if (b in RFS)                      { emit(id, 0, "fc-query REFUSED it -- the table says it answers"); continue }
        if (!(b in QFAM))                  { emit(id, 0, "fc-query reported nothing for it"); continue }
        if (QFAM[b] != EFAM[b])            { emit(id, 0, sprintf("family '%s', want '%s'", QFAM[b], EFAM[b])); continue }
        if (QSTY[b] != ESTY[b])            { emit(id, 0, sprintf("style '%s', want '%s'", QSTY[b], ESTY[b])); continue }
        if (QPX[b] + 0 != EPX[b] + 0)      { emit(id, 0, sprintf("pixelsize '%s', want '%s'", QPX[b], EPX[b])); continue }
        emit(id, 1, "")
    }
    for (b in QFAM)
        if (!(b in EFAM))
            emit("fcq-unexpected-" b, 0, "fc-query reported a face the expected-face table does not name")

    # --- fc-list enumerates them once the cache is built --------------------
    # fc-query reads a file it is handed; fc-list answers from the CACHE, so
    # this is the property the ticket actually names -- the faces are
    # discoverable by a program that was not told where they live. The seven
    # are the same seven, which is the cross-check worth having: the cache
    # cannot hold a face fontconfig could not query in the first place.
    emit("fc-list-faces", NFCL == WANT_QUERYABLE,
         sprintf("fc-list enumerates %d face(s) under %s, want %d", NFCL, DIR, WANT_QUERYABLE))

    FIXFAM = EFAM["6x13.pcf.gz"]; CURFAM = EFAM["cursor.pcf.gz"]
    nfix = 0; ncur = 0; nodd = 0; oddl = ""
    for (b in LFAM) {
        if      (LFAM[b] == FIXFAM) nfix++
        else if (LFAM[b] == CURFAM) ncur++
        else { nodd++; if (nodd <= 3) oddl = oddl " " b "=" LFAM[b] }
    }
    emit("fc-list-fixed", nfix == WANT_QUERYABLE - 1,
         sprintf("fc-list reports %d face(s) of family '%s', want %d", nfix, FIXFAM, WANT_QUERYABLE - 1))
    emit("fc-list-cursor", ncur == 1 && LFAM["cursor.pcf.gz"] == CURFAM,
         sprintf("fc-list reports %d face(s) of family '%s' and calls cursor.pcf.gz '%s'", ncur, CURFAM, LFAM["cursor.pcf.gz"]))
    emit("fc-list-no-other-family", nodd == 0,
         sprintf("%d face(s) under %s report neither '%s' nor '%s':%s", nodd, DIR, FIXFAM, CURFAM, oddl))

    # --- and the REAL configuration covers this directory -------------------
    # The fc-cache and fc-list above run under a config naming only this
    # directory, which is what keeps them off the image's other font
    # directories (see the suite header). That scoping is also the one thing
    # they cannot then attest: whether the image's own fontconfig would find
    # these files. This case answers it from /etc/fonts/fonts.conf, by
    # reading rather than by running -- a `<dir>' element that is this
    # directory or a prefix of it.
    ncov = 0; covl = ""
    for (d in CDIR) {
        if (d == DIR || index(DIR "/", d "/") == 1) { ncov++; if (covl == "") covl = d }
    }
    emit("fontconfig-scans-dir", NCONF > 0 && ncov > 0,
         sprintf("no <dir> in /etc/fonts/fonts.conf covers %s (%d <dir> element(s) read)", DIR, NCONF))
}