# cantarell-fonts-verify.awk -- turn one find, one sha256sum, one dpkg file list
# and one fc-scan into the payload, inventory, packaging-footprint and
# fontconfig-derivation cases.
#
# WHY ONE AWK AND NOT A SHELL LOOP. 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. Everything below is derived from four
# files the .test already produced with four execs.
#
# INPUTS:
#   REF    the reference manifest, "<size> <sha256> <path>"
#   SUMS   `sha256sum` over the one payload file, "<sha256>  <path>"
#   INV    `find ... -printf '%p %y %s %m %l\n'` over the payload font directory
#          AND the two co-tenant font directories
#   PKGF   /var/lib/dpkg/info/cantarell-fonts.list -- one path per line
#   FACES  the native fontconfig control, "<file>|<family>|...|<variable>|<index>"
#   SCAN   this image's own fc-scan over the one installed file, same format
#
# THIS PORT GETS A DIRECTORY OF ITS OWN, WHICH NO FONT PORT BEFORE IT DID. The
# recipe's fontsdir defaults to ${datadir}/fonts/cantarell and the recipe does not
# override it, so /usr/share/fonts/cantarell is created by this package and holds
# nothing else. That removes the whole class of co-tenancy repair the last four
# font ports each had to do -- there is no neighbour to scope a census against,
# and no neighbour's suite to amend. What is asserted instead is the other
# direction: that the two directories this port does NOT write to are unchanged,
# so a packaging or fontconfig change that moved the fonts already here is a
# named red rather than a silent one.
#
# THE FILES VARIABLE NAMES A DIRECTORY THE BUILD NEVER CREATES, AND THAT IS
# ASSERTED RATHER THAN LEFT TO LOOK LIKE AN OMISSION. The recipe sets
#     FILES:${PN} = "${datadir}/fonts ${datadir}/fontconfig"
# as an ASSIGNMENT, and ${datadir}/fontconfig is over-declaration: nothing in the
# tarball installs there, on either the prebuilt or the from-source path, and
# there is no fontconfig fragment anywhere in cantarell-fonts-0.303.1. The
# directory nevertheless EXISTS on the image, owned by the `fontconfig` package
# (conf.avail/ and conf.d/ live under it), so a case asserting "the directory is
# absent" would be wrong and a case asserting "it is empty" would fail for a
# reason that has nothing to do with this port. pkgfiles-* asserts the only thing
# that is actually this port's business: the cantarell-fonts package's own file
# list contains exactly one path, and no path under ${datadir}/fontconfig.

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) { sub(/^.*\//, "", p); return p }

BEGIN {
    S = "cantarell-fonts"
    PROCINFO["sorted_in"] = "@ind_str_asc"
    FDIR   = "/usr/share/fonts/cantarell"
    COT1   = "/usr/share/fonts/ttf"        # 12 Liberation + 10 Vera
    COT2   = "/usr/share/fonts/truetype"   # 1 Inconsolata + 4 Gentium + 7 Droid
                                           #  + 14 Source Code Pro (BRINTOS-1455)
                                           #  + 12 ttf-liberation (BRINTOS-1721)
                                           #  + 20 Roboto (BRINTOS-1723)
                                           #  + 14 Ubuntu Font Family (BRINTOS-1724)
                                           #  + 6 Hunky (BRINTOS-1727)
                                           #  + 1 Abyssinica SIL (BRINTOS-1728)
    WANT_COT1 = 22
    WANT_COT2 = 137
    NFIELD = 12   # fields on an fcfaces/fc-scan line
    NFILE  = 1    # payload files this package installs
    NPAT   = 6    # fc-scan patterns the ONE file produces: 5 named instances + 1 variable
    OURFILE = "Cantarell-VF.otf"
}

FILENAME == REF && /^#/ { next }
FILENAME == REF && NF == 3 { WSIZE[$3] = $1; WSUM[$3] = $2; NREF++; next }

FILENAME == SUMS && NF >= 2 { p = $2; sub(/^\*/, "", p); GSUM[p] = $1; next }

FILENAME == INV && NF >= 4 { ITYPE[$1] = $2; ISIZE[$1] = $3; IMODE[$1] = $4; next }

FILENAME == PKGF && NF >= 1 { PKGPATH[$0] = 1; NPKG++; next }

FILENAME == FACES && /^#/ { next }
FILENAME == FACES && NF >= 1 { split($0, a, "|"); WFC[a[12] "|" a[11]] = $0; NFC++; next }

FILENAME == SCAN && NF >= 1 { split($0, a, "|"); GFC[a[12] "|" a[11]] = $0; NGFC++; next }

END {
    # THE INPUTS WERE REALLY READ. Every case below is a comparison, and a
    # comparison over two empty tables is vacuously true -- the failure that
    # reads as a full green.
    emit("ref-count",   NREF == NFILE, "the manifest carried " NREF " entries, want " NFILE)
    emit("faces-count", NFC  == NPAT,  "the fcfaces control carried " NFC " lines, want " NPAT)

    # ------------------------------------------------------------ the payload
    # Size, bytes, type and mode in one case. %y catches a symlink where a
    # regular file is meant and %m the mode, neither of which `test -f` sees; the
    # digest catches a file of the right length that is not the right file.
    #
    # MODE 644 IS A REAL ASSERTION, NOT BOILERPLATE. meson's install_data
    # defaults to 0644 and the upstream blob is 0644 in the tarball, so this row
    # says the install went through the class's own path rather than a copy that
    # preserved something else.
    for (p in WSIZE) {
        id = "file-" base(p)
        if (!(p in ITYPE))        { emit(id, 0, p " is absent"); continue }
        if (ITYPE[p] != "f")      { emit(id, 0, p " is type '" ITYPE[p] "', want a regular file"); continue }
        if (ISIZE[p] != WSIZE[p]) { emit(id, 0, p " is " ISIZE[p] " bytes, want " WSIZE[p]); continue }
        if (IMODE[p] != "644")    { emit(id, 0, p " is mode " IMODE[p] ", want 644"); continue }
        if (!(p in GSUM))         { emit(id, 0, p " was not digested"); continue }
        if (GSUM[p] != WSUM[p])   { emit(id, 0, p " sha256 " substr(GSUM[p],1,16) "..., want " substr(WSUM[p],1,16) "..."); continue }
        emit(id, 1)
    }

    # ---------------------------------------------------- the port's directory
    # THE DIRECTORY IS THIS PACKAGE'S ALONE, so unlike every font suite before it
    # this census is NOT scoped by basename -- it is entitled to say the whole
    # directory. A future font port that chose the same fontsdir would redden
    # this, which is the correct outcome: it would also be sharing a directory
    # with a suite that assumes it does not, and the two facts should surface
    # together rather than one silently.
    n = 0; extra = ""
    for (p in ITYPE) {
        if (p !~ ("^" FDIR "/")) continue
        if (ITYPE[p] != "f") { extra = extra " " base(p) "(type " ITYPE[p] ")"; continue }
        n++
        if (base(p) != OURFILE) extra = extra " " base(p)
    }
    emit("dir-count",     n == NFILE,  FDIR " holds " n " regular file(s), want " NFILE)
    emit("dir-no-extras", extra == "", FDIR " also holds:" extra)

    # ------------------------------------------- the package's own file list
    # WHAT THE PACKAGE CLAIMS, not what is on disk. file-* above would look
    # identical if a second package had installed the same bytes, and dir-count
    # would look identical if this package ALSO owned paths elsewhere -- which is
    # exactly what FILES:${PN} invites, since it names ${datadir}/fontconfig as
    # well. dpkg's own list is the only place the package's real footprint is
    # visible on target.
    #
    # dpkg lists the directories it created as well as the files, so the count
    # is of the FILE entries: the four parent directories (/usr, /usr/share,
    # /usr/share/fonts, /usr/share/fonts/cantarell) are legitimately in the list
    # and are not what this case is about.
    nf = 0; nfc = 0; stray = ""
    for (p in PKGPATH) {
        if (p ~ /^\/usr\/share\/fontconfig(\/|$)/) { nfc++; continue }
        if (!(p in ITYPE) || ITYPE[p] != "f") continue    # a directory entry
        nf++
        if (p != FDIR "/" OURFILE) stray = stray " " p
    }
    emit("pkgfiles-loaded",  NPKG > 0, "/var/lib/dpkg/info/cantarell-fonts.list was empty or unreadable")
    emit("pkgfiles-count",   nf == NFILE, "cantarell-fonts owns " nf " file(s), want " NFILE ";" stray)
    emit("pkgfiles-no-fontconfig", nfc == 0,
         "cantarell-fonts owns " nfc " path(s) under /usr/share/fontconfig -- FILES names ${datadir}/fontconfig but nothing in the tarball installs there")

    # ------------------------------------------------------- the co-tenants
    # THIS PORT MUST NOT TOUCH THE TWO DIRECTORIES THAT WERE ALREADY HERE. It
    # writes to neither, so these are the cases that say a fontconfig or
    # packaging change did not move the faces the image already scans on every
    # fc-cache. Both neighbourhoods assert their own payloads from the other
    # side, so a disagreement between the suites is visible rather than silent.
    #
    # WANT_COT2 MOVED 12 -> 26 FOR BRINTOS-1455, 26 -> 38 FOR BRINTOS-1721,
    # 38 -> 58 FOR BRINTOS-1723, 58 -> 72 FOR BRINTOS-1724 AND 72 -> 78 FOR
    # BRINTOS-1727, 78 -> 79 FOR BRINTOS-1728 AND 79 -> 137 FOR BRINTOS-1729,
    # AND THAT IS THIS CASE
    # WORKING. This is a WHOLE-DIRECTORY count, so it is meant to move
    # every time a font port lands in COT2; ttf-liberation added twelve faces,
    # ttf-roboto twenty (12 Roboto-* + 8 RobotoCondensed-*), the Ubuntu Font
    # Family fourteen (4 UbuntuMono-* + 10 Ubuntu-*). Each of the three was cut
    # from a base that predated the others and moved this constant for its own
    # faces alone, so the sides do NOT compose by taking any one of them: each
    # merge that lands later adds its own contribution to what is already there.
    #
    # HOW 72 WAS ESTABLISHED, AND WHAT IT IS NOT. It is NOT the sum in a conflict
    # marker, which is the thing that has been wrong before. It is also NOT a
    # count of the merged tree's staged rootfs: that build was run (2026-08-29,
    # hz1) and reached task 17128 of 17246 WITHOUT producing one, because nmap
    # and python3 fail do_compile on brintos/hwjs-cc#81's
    # vla/dynamic-alloca-address-escape-opaque-sink refusal since BRINTOS-902
    # landed shared libc -- so do_rootfs never ran and there was nothing to
    # count. See BRINTOS-1430 and BRINTOS-1767; nmap is not in
    # BRINTOS_HWJS81_EXCLUDED at all.
    #
    # What it IS: the union of four file sets, each measured on its own and
    # proved DISJOINT BY BASENAME (72 names, 72 distinct, no collision) --
    #
    #   26  the pre-existing tenants, live-counted in the staged rootfs of this
    #       branch's own pre-merge image build (Droid 7, Gentium 4, Inconsolata
    #       1, Source Code Pro 14)
    #   14  Ubuntu Font Family, from ttf-ubuntu-font-family-manifest.txt, read
    #       off the built .debs and byte-compared to the upstream zip
    #   20  Roboto, from ttf-roboto-manifest.txt (BRINTOS-1723, live-confirmed
    #       there at 58)
    #   12  ttf-liberation, from ttf-liberation-manifest.txt (BRINTOS-1721)
    #
    # 78 WAS DERIVED THE SAME WAY, AT THE MERGE, BY THE FIFTH PORT.
    # BRINTOS-1727 (ttf-hunkyfonts) adds six -- HunkySans-{Bold,BoldOblique,
    # Oblique,Roman}.ttf and HunkySerif-{Bold,Roman}.ttf -- and its own 26 is
    # the same pre-existing set counted above, live in the staged rootfs of its
    # pre-merge image build (which held 32). Not the sum in the conflict marker:
    # the union of ALL EIGHT truetype manifests in this merged tree was swept by
    # basename and came to 78 distinct names, with the per-manifest
    # contributions summing to 78 and NO collision between any two sets --
    # 1 Inconsolata, 4 Gentium, 7 Droid, 14 Source Code Pro, 12 ttf-liberation,
    # 20 Roboto, 14 Ubuntu, 6 Hunky.
    #
    # STILL NOT A LIVE COUNT OF A MERGED ROOTFS, for the reason above: nmap and
    # python3 still fail do_compile on this base, so the merged image still
    # cannot reach do_rootfs. Whoever lands the seventh port should re-derive
    # it the same way, and prefer a live count the moment the image builds
    # again.
    #
    # 79 IS THE SIXTH PORT'S NUMBER, RE-DERIVED THE SAME WAY AND NOT TAKEN FROM
    # THE CONFLICT MARKER. BRINTOS-1728 (ttf-abyssinica) adds ONE face,
    # AbyssinicaSIL-Regular.ttf. The union of every truetype-bearing manifest in
    # this merged tree was swept by basename again rather than 78 being taken on
    # trust: 78 distinct names on `main` alone -- 1 Inconsolata, 4 Gentium,
    # 7 Droid, 14 Source Code Pro, 12 ttf-liberation, 20 Roboto, 14 Ubuntu,
    # 6 Hunky -- and 79 with this port's face, with no collision between any two
    # sets.
    #
    # TWO MANIFESTS THAT REACHED `main` AFTER 78 WAS WRITTEN CONTRIBUTE NOTHING
    # HERE, and were checked rather than assumed: ttf-pt-sans installs eight
    # faces under /usr/share/fonts/X11/TTF, and ttf-liberation-sans-narrow four
    # under /usr/share/fonts/TTF. Neither directory is COT1 or COT2 -- and note
    # /usr/share/fonts/TTF is NOT COT1's /usr/share/fonts/ttf, which is why
    # WANT_COT1 is still 22 and why reading these two paths case-insensitively
    # would get both constants wrong.
    # These are TOTALS, not basename scopes -- deliberately, because cantarell
    # owns a directory of its own and is entitled to say the neighbours' two are
    # exactly as it found them. The price is one line of maintenance per font
    # port that lands in either, and source-code-pro-fonts is the first to
    # collect it: fourteen SourceCodePro-*.otf joined COT2. 99-source-code-pro-
    # fonts asserts the same twelve pre-existing faces from its own side, BY
    # NAME, so a displacement is still localised to the package that caused it
    # even though this case only reports a count.
    c1 = 0; c2 = 0
    for (p in ITYPE) {
        if (ITYPE[p] != "f") continue
        if (p ~ ("^" COT1 "/")) c1++
        if (p ~ ("^" COT2 "/")) c2++
    }
    emit("cotenant-ttf",      c1 == WANT_COT1, COT1 " holds " c1 " files, want " WANT_COT1 " (12 Liberation + 10 Vera)")
    emit("cotenant-truetype", c2 == WANT_COT2, COT2 " holds " c2 " files, want " WANT_COT2 " (1 Inconsolata + 4 Gentium + 7 Droid + 14 Source Code Pro + 12 ttf-liberation + 20 Roboto + 14 Ubuntu + 6 Hunky + 1 Abyssinica SIL + 58 TLWG)")

    # ------------------------------- fontconfig's own reading of the six patterns
    # ONE FILE, SIX PATTERNS. See the fcfaces control's header for why, and for
    # why the key is <index>|<variable> rather than the basename every other font
    # suite on this image uses.
    #
    # FIELD BY FIELD AGAINST THE NATIVE CONTROL, not as one string compare: a
    # whole-line case would name the pattern and nothing else, and these fields
    # fail for different reasons.
    emit("fcq-count", NGFC == NPAT, "fc-scan produced " NGFC " pattern(s) for the one file, want " NPAT)

    split("file family style slant weight spacing postscriptname fontformat scalable outline variable index", FN, " ")
    for (k in WFC) {
        id = "fcq-" k; gsub(/\|/, "-", id)
        if (!(k in GFC)) { emit(id, 0, "fc-scan produced no pattern at index/variable " k); continue }
        nw = split(WFC[k], w, "|"); ng = split(GFC[k], g, "|")
        if (nw != NFIELD || ng != NFIELD) {
            emit(id, 0, "field count " ng ", control " nw ", want " NFIELD); continue
        }
        bad = ""
        for (i = 1; i <= NFIELD; i++)
            if (w[i] != g[i]) bad = bad " " FN[i] "='" substr(g[i],1,24) "'(want '" substr(w[i],1,24) "')"
        emit(id, bad == "", k ":" bad)
    }

    # THE THREE DERIVATIONS THIS PORT TURNS ON, NAMED SEPARATELY. fcq-* above
    # already rejects any input these reject, so on their own they would be the
    # subsumed-guard mistake -- they are here because each localises a DIFFERENT
    # downstream consequence and a reviewer reading a red needs to know which.
    #
    #   spacing   decides whether this port could answer `fc-match monospace` and
    #             displace the generic ttf-inconsolata holds. It cannot: unset on
    #             all six. A single "unset" case is right here where 99-ttf-droid
    #             needed two, because that port shipped a mono face and this one
    #             does not.
    #   format    CFF on all six. 99-ttf-inconsolata asserts CFF for a STATIC
    #             `CFF ` font and 99-ttf-gentium/99-ttf-droid assert TrueType;
    #             this is the only place CFF2 is exercised, and fontconfig reports
    #             it under the same name.
    #   variable  EXACTLY ONE of the six patterns is the variable font itself.
    #             This is the case that says fontconfig enumerated the axis rather
    #             than treating the file as a single static face -- the whole
    #             reason this suite keys on index at all.
    nsp = 0; nfmt = 0; nvar = 0; spbad = ""; fmtbad = ""
    for (k in GFC) {
        split(GFC[k], g, "|")
        if (g[6] == "") nsp++; else spbad = spbad " " k "='" g[6] "'"
        if (g[8] == "CFF") nfmt++; else fmtbad = fmtbad " " k "='" g[8] "'"
        if (g[11] == "True") nvar++
    }
    emit("fcq-spacing-proportional", nsp == NPAT && NGFC == NPAT,
         "spacing must be unset on all " NPAT " patterns;" spbad)
    emit("fcq-format-cff", nfmt == NPAT && NGFC == NPAT,
         "fontformat must be CFF on all " NPAT " patterns;" fmtbad)
    emit("fcq-one-variable", nvar == 1 && NGFC == NPAT,
         "exactly one of the " NPAT " patterns must carry variable=True; " nvar " did")
}