# source-code-pro-fonts-verify.awk -- turn one find, one sha256sum, the package's
# own 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>" -- FIFTEEN rows
#   SUMS   `sha256sum` over the fifteen payload files, "<sha256>  <path>"
#   INV    `find ... -printf '%p %y %s %m %l\n'` over the payload font directory,
#          /etc/fonts/conf.d, and /usr/share/fonts/ttf (the other font directory)
#   PKGF   /var/lib/dpkg/info/source-code-pro-fonts.list, one path per line
#   FACES  the native fontconfig control, "<basename>|<family>|..."
#   SCAN   this image's own fc-scan over the fourteen installed faces, same format
#
# FIFTEEN PATHS, AND THE FIFTEENTH IS THE WHOLE STORY OF THIS PORT. Every font
# port on this image before it installed fonts and nothing else; this recipe also
# installs /etc/fonts/conf.d/44-source-code-pro-fonts-fontconfig.conf, which is
# what moves the `monospace` generic. It is asserted here by DIGEST, at the file
# layer, so a fragment that arrived truncated or edited reds one row with a name
# that says so -- rather than surfacing three layers down as an fc-match answer
# nobody can localise.
#
# THERE IS NO DOC DIRECTORY IN THIS SUITE, AND THAT IS THE RECIPE. The zip's
# LICENSE.txt is never installed: the recipe's do_install writes the .otf files
# and the fragment, and FILES:${PN} is an ASSIGNMENT naming ${sysconfdir}/fonts
# and ${datadir}/fonts only. So ${PN}-doc is empty and dpkg builds none, which
# pkg-no-doc-package in the .test asserts. The licence is not lost -- name IDs
# 13/14 inside each face carry the OFL notice and its URL.

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 = "source-code-pro-fonts"
    PROCINFO["sorted_in"] = "@ind_str_asc"
    TTFDIR = "/usr/share/fonts/truetype"
    CONFDIR = "/etc/fonts/conf.d"
    COTDIR = "/usr/share/fonts/ttf"
    CONF   = CONFDIR "/44-source-code-pro-fonts-fontconfig.conf"
    # The 22 files liberation-fonts (12) and ttf-bitstream-vera (10) install into
    # COTDIR. This port never writes there; see the cotenant cases below.
    WANT_COT = 22
    NFIELD = 11   # fields on an fcfaces/fc-scan line
    NFACE  = 14   # font faces
    NREFROW = 15  # manifest rows: the fourteen faces plus the fragment

    # This package's own fourteen basenames. Everything scoped below is scoped
    # against THIS list rather than against "not ours" -- the distinction that
    # keeps the next Debian-layout font from reddening this suite, and the
    # mistake 99-ttf-inconsolata had to have repaired for it when ttf-gentium
    # moved in beside it.
    split("SourceCodePro-Black.otf SourceCodePro-BlackIt.otf SourceCodePro-Bold.otf " \
          "SourceCodePro-BoldIt.otf SourceCodePro-ExtraLight.otf SourceCodePro-ExtraLightIt.otf " \
          "SourceCodePro-It.otf SourceCodePro-Light.otf SourceCodePro-LightIt.otf " \
          "SourceCodePro-Medium.otf SourceCodePro-MediumIt.otf SourceCodePro-Regular.otf " \
          "SourceCodePro-Semibold.otf SourceCodePro-SemiboldIt.otf", _o, " ")
    for (i in _o) OURS[_o[i]] = 1

    # THE FOUR RIBBI FACES, WHICH ARE THE ONES WHOSE NAME IDs 16/17 EQUAL 1/2.
    # fontconfig records both readings and comma-joins them, so these four carry
    # ONE family value and the other ten carry TWO. See fcq-family-typographic.
    split("SourceCodePro-Regular.otf SourceCodePro-Bold.otf SourceCodePro-It.otf " \
          "SourceCodePro-BoldIt.otf", _r, " ")
    for (i in _r) RIBBI[_r[i]] = 1

    # The twelve faces already in TTFDIR when this port arrived. Named, not
    # counted: a count would go stale the next time a font lands in this
    # directory, and naming them is what says WHICH neighbour a displacement
    # took. ttf-inconsolata (1), ttf-gentium (4), ttf-droid (7).
    split("Inconsolata.ttf GenAI102.ttf GenAR102.ttf GenI102.ttf GenR102.ttf " \
          "DroidSans-Bold.ttf DroidSans.ttf DroidSansMono.ttf DroidSerif-Bold.ttf " \
          "DroidSerif-BoldItalic.ttf DroidSerif-Italic.ttf DroidSerif-Regular.ttf", _c, " ")
    for (i in _c) COTEN[_c[i]] = 1
}

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 { PKG[$0] = 1; NPKG++; next }

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

FILENAME == SCAN && NF >= 1 { split($0, a, "|"); GFC[a[1]] = $0; 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 == NREFROW, "the manifest carried " NREF " entries, want " NREFROW)
    emit("faces-count", NFC  == NFACE,   "the fcfaces control carried " NFC " lines, want " NFACE)

    # ------------------------------------------------------------ per file
    # Size, bytes, type and mode in one case per file. %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 HERE, NOT BOILERPLATE. The .otf files are 0644
    # inside the zip already, but LICENSE.txt beside them is 0755 -- so the
    # archive is NOT uniformly moded and it is the recipe's `install -m 0644`
    # that guarantees these. A do_install that stopped passing -m would show up
    # on these rows.
    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 payload directory
    # THE DIRECTORY IS SHARED FOUR WAYS NOW -- Inconsolata, the four Gentium
    # faces, the seven Droid ones, and these fourteen. So this case is scoped to
    # OUR OWN basename shape: a suite is entitled to say its own payload is
    # intact and nothing more, and the co-tenant census belongs to whoever is
    # counting.
    #
    # THE SHAPE IS `SourceCodePro-*.otf`, WHICH IS ALSO WHAT THE zip'S TTF/ AND
    # WOFF/ DIRECTORIES WOULD NOT MATCH. The recipe's do_install replaces
    # ttf.inc's `*.tt[cf]` glob with an `*.otf` one; a revision that restored the
    # inc's version would install fourteen SourceCodePro-*.ttf beside these, and
    # that is a different basename shape which no case here would see. So this
    # case is deliberately anchored on the extension too.
    extra = ""
    for (p in ITYPE) {
        if (ITYPE[p] != "f") continue
        if (p !~ ("^" TTFDIR "/")) continue
        b = base(p)
        if (b in OURS) continue
        if (b ~ /^SourceCodePro.*\.(otf|ttf)$/) extra = extra " " b
    }
    emit("ttf-no-extras", extra == "", TTFDIR " also holds SourceCodePro-shaped file(s):" extra)

    # ...AND THE FRAGMENT DIRECTORY THE SAME WAY. /etc/fonts/conf.d is owned by
    # the fontconfig package, which fills it with symlinks into conf.avail; this
    # port adds ONE REGULAR FILE to it. A second fragment of ours -- a recipe
    # revision, a stale copy from an earlier build -- would silently double the
    # alias rules, and outranking is decided by filename order, so which one won
    # would depend on the name.
    cextra = ""
    for (p in ITYPE) {
        if (p !~ ("^" CONFDIR "/")) continue
        b = base(p)
        if (p == CONF) continue
        if (b ~ /source-code-pro/) cextra = cextra " " b
    }
    emit("conf-no-extras", cextra == "", CONFDIR " also holds source-code-pro fragment(s):" cextra)

    # THE FRAGMENT IS A REGULAR FILE, NOT A SYMLINK. Every other entry in
    # conf.d on this image is a symlink into /usr/share/fontconfig/conf.avail,
    # placed there by the fontconfig package. This one is installed directly by
    # the recipe, so `%y` reading `l` here would mean something quite different
    # had happened -- and file-44-... above would still be green, because it
    # follows the link.
    emit("conf-regular-file", (CONF in ITYPE) && ITYPE[CONF] == "f",
         CONF " is type '" ((CONF in ITYPE) ? ITYPE[CONF] : "absent") "', want a regular file")

    # ------------------------------------------------------- the co-tenants
    # THIS PORT MUST ADD TO ITS DIRECTORY, NOT REPLACE IT. The recipe's
    # do_install is a `find -exec install` into a shared directory, so the
    # failure this guards is real rather than theoretical: a do_install that
    # cleaned its target first, or a packaging change that made one package claim
    # the whole directory, would take three neighbours with it and every file-*
    # case above would stay green.
    #
    # NAMED, NOT COUNTED. All three neighbours assert their own payload from the
    # other side, so a disagreement between the suites is visible rather than
    # silent -- and a name says WHICH package was displaced where a total only
    # says that one was.
    miss = ""
    for (b in COTEN) if (!((TTFDIR "/" b) in ITYPE)) miss = miss " " b
    emit("cotenant-truetype", miss == "",
         TTFDIR " is missing co-tenant face(s):" miss " -- this port displaced a neighbour")

    # AND THE OTHER FONT DIRECTORY IS UNTOUCHED. The 22 files liberation-fonts
    # and ttf-bitstream-vera own live in a directory this port never writes to,
    # so this is the case that says a fontconfig or packaging change did not move
    # the fonts that were already here.
    cot = 0
    for (p in ITYPE) if (p ~ ("^" COTDIR "/") && ITYPE[p] == "f") cot++
    emit("cotenant-ttfdir", cot == WANT_COT,
         COTDIR " holds " cot " files, want " WANT_COT " (12 Liberation + 10 Vera)")

    # ------------------------------------------- the package's real footprint
    # THE FILE LIST IS THE ONLY PLACE THE PACKAGE'S OWN CLAIM IS VISIBLE. The
    # file-* cases would look identical if some other package had installed the
    # same bytes; this says source-code-pro-fonts claims exactly the fifteen
    # paths the manifest names and nothing else. FILES:${PN} is
    # `${sysconfdir}/fonts ${datadir}/fonts` -- two DIRECTORY trees, not a file
    # list -- so a recipe or class that dropped anything else under either would
    # be swept into this package silently. That is what this case refuses.
    #
    # dpkg's list carries the parent directories too, so only the regular-file
    # entries are compared; the directories are the packaging's business.
    emit("pkgfiles-loaded", NPKG > 0, "the package file list was empty or absent")
    pmiss = ""; pextra = ""
    for (p in WSIZE) if (!(p in PKG)) pmiss = pmiss " " p
    for (p in PKG) {
        if (p in WSIZE) continue
        if ((p in ITYPE) && ITYPE[p] != "f") continue   # a directory entry
        if (!(p in ITYPE)) continue                     # not walked by our find
        pextra = pextra " " p
    }
    emit("pkgfiles-exact", pmiss == "" && pextra == "",
         "package claims differ from the manifest -- missing:" pmiss " extra:" pextra)

    # ------------------------------------- fontconfig's own reading of the faces
    # FIELD BY FIELD AGAINST THE NATIVE CONTROL, not as one string compare: a
    # whole-line case would name the file and nothing else, and these fields fail
    # for different reasons. See the control file's header for what each is.
    split("file family style slant weight spacing postscriptname fontformat scalable outline charset", FN, " ")
    for (b in WFC) {
        if (!(b in GFC)) { emit("fcq-" b, 0, "fc-scan produced no line for " b); continue }
        nw = split(WFC[b], w, "|"); ng = split(GFC[b], g, "|")
        if (nw != NFIELD || ng != NFIELD) {
            emit("fcq-" b, 0, "field count " ng ", control " nw ", want " NFIELD); continue
        }
        bad = ""
        for (i = 2; i <= NFIELD; i++)
            if (w[i] != g[i]) bad = bad " " FN[i] "='" substr(g[i],1,24) "'(want '" substr(w[i],1,24) "')"
        emit("fcq-" b, bad == "", b ":" 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.
    #
    # Keyed on the LITERAL basenames rather than on `for (b in GFC)`: an empty
    # control table would make the loop body unreachable and silently drop these
    # cases, which is the vacuous-green shape the count guards above refuse.
    nsp = 0; nfmt = 0; ntyp = 0; spbad = ""; fmtbad = ""; typbad = ""
    for (i in _o) {
        b = _o[i]
        if (!(b in GFC)) { spbad  = spbad  " " b "=(no fc-scan line)"
                           fmtbad = fmtbad " " b "=(no fc-scan line)"
                           typbad = typbad " " b "=(no fc-scan line)"; continue }
        split(GFC[b], g, "|")

        # SPACING IS 100 ON ALL FOURTEEN, WHICH IS WHAT MAKES THIS PORT ABLE TO
        # ANSWER `monospace` AT ALL. 99-ttf-droid's equivalent splits 6-and-1 and
        # 99-ttf-gentium's is unset on all four; this one is the first where
        # every face carries FC_MONO. It is the precondition for the generic
        # handover the .test's fc-match-monospace case asserts -- fontconfig
        # would not put a proportional face at the head of monospace's list
        # however loudly a fragment preferred it.
        if (g[6] == "100") nsp++; else spbad = spbad " " b "='" g[6] "'"

        # fontformat=CFF is the counterpart of 99-ttf-droid's TrueType assertion:
        # between them the two suites pin both of freetype's outline drivers on
        # this image from fontconfig's side.
        if (g[8] == "CFF") nfmt++; else fmtbad = fmtbad " " b "='" g[8] "'"

        # THE TYPOGRAPHIC-NAME SPLIT, ASSERTED AS THE 10-AND-4 IT IS. fontconfig
        # comma-joins ID16 then ID1; the four RIBBI faces have them equal and so
        # carry ONE value, and the other ten carry TWO. A single "all fourteen
        # have two" case would be WRONG here and a single "all fourteen have one"
        # equally so -- and getting it wrong in either direction is exactly the
        # failure that would let ten faces collapse onto one style.
        nval = split(g[2], _fv, ",")
        want = (b in RIBBI) ? 1 : 2
        if (nval == want) ntyp++
        else typbad = typbad " " b "=" nval "(want " want ")"
    }
    emit("fcq-spacing-mono", nsp == NFACE,
         "spacing must be 100 (FC_MONO) on all " NFACE " faces;" spbad)
    emit("fcq-format-cff", nfmt == NFACE,
         "fontformat must be CFF on all " NFACE " faces;" fmtbad)
    emit("fcq-family-typographic", ntyp == NFACE,
         "family value count must be 2 except on the four RIBBI faces;" typbad)
}