WIP

scripts/check_sisyphus_set_relations.sh

@@ -0,0 +1,225 @@
+#!/bin/bash
+set -euo pipefail
+export LC_ALL=C
+
+MIRROR=https://ftp.altlinux.org/pub/distributions/ALTLinux
+REQUIRE_LIMIT=100
+
+SCRIPT_DIR=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)
+REPO_ROOT=$(cd -- "$SCRIPT_DIR/.." && pwd)
+WORK=$(mktemp -d "${TMPDIR:-/tmp}/arsv-set-check.XXXXXX")
+trap 'rm -rf "$WORK"' EXIT
+
+for command in apt-get apt-cache awk sort join comm paste cc; do
+    command -v "$command" >/dev/null || {
+        printf 'required command not found: %s\n' "$command" >&2
+        exit 2
+    }
+done
+
+build_comparator()
+{
+    : >"$WORK/rpmlib.h"
+    : >"$WORK/system.h"
+    cat >"$WORK/compare.c" <<'EOF'
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/types.h>
+
+int rpmsetcmp(const char *, const char *);
+
+int main(void)
+{
+    char *provided = NULL;
+    char *required = NULL;
+    size_t provided_size = 0;
+    size_t required_size = 0;
+    ssize_t provided_length;
+
+    while ((provided_length = getline(&provided, &provided_size, stdin)) >= 0) {
+        ssize_t required_length = getline(&required, &required_size, stdin);
+        if (provided_length < 1 || required_length < 1) {
+            free(provided);
+            free(required);
+            return 2;
+        }
+        if (provided[provided_length - 1] == '\n')
+            provided[provided_length - 1] = '\0';
+        if (required[required_length - 1] == '\n')
+            required[required_length - 1] = '\0';
+
+        printf("%d\n", rpmsetcmp(provided, required));
+    }
+    free(provided);
+    free(required);
+    return 0;
+}
+EOF
+
+    cc -O2 -std=gnu11 -D_GNU_SOURCE -I"$WORK" \
+        -include "$SCRIPT_DIR/newset_compat.h" \
+        "$REPO_ROOT/reimplement/newset.c" "$WORK/compare.c" \
+        -o "$WORK/compare"
+}
+
+fetch_metadata()
+{
+    mkdir -p "$WORK/apt/lists/partial" "$WORK/apt/archives/partial"
+    : >"$WORK/apt/apt.conf"
+    : >"$WORK/apt/status"
+    printf '%s\n' \
+        "rpm [alt] $MIRROR Sisyphus/x86_64 classic" \
+        "rpm [alt] $MIRROR Sisyphus/noarch classic" \
+        >"$WORK/apt/sources.list"
+
+    local options=(
+        -o 'Dir::Etc::main=-'
+        -o 'Dir::Etc::parts=-'
+        -o "Dir::Etc::sourcelist=$WORK/apt/sources.list"
+        -o 'Dir::Etc::sourceparts=-'
+        -o "Dir::State::lists=$WORK/apt/lists"
+        -o "Dir::State::status=$WORK/apt/status"
+        -o "Dir::Cache::archives=$WORK/apt/archives"
+        -o "Dir::Cache::pkgcache=$WORK/apt/pkgcache.bin"
+        -o "Dir::Cache::srcpkgcache=$WORK/apt/srcpkgcache.bin"
+    )
+
+    APT_CONFIG="$WORK/apt/apt.conf" apt-get "${options[@]}" -qq update
+    APT_CONFIG="$WORK/apt/apt.conf" apt-cache "${options[@]}" dumpavail \
+        >"$WORK/available.txt"
+}
+
+extract_relations()
+{
+    awk '
+    function emit(kind, text,    count, parts, i, item, capability, encoded) {
+        count = split(text, parts, /,[[:space:]]*/)
+        for (i = 1; i <= count; i++) {
+            item = parts[i]
+            if (kind == "P" && item !~ /[[:space:]]+\(?=[[:space:]]+set:[[:alnum:]]+\)?$/)
+                continue
+            if (kind == "R" && item !~ /[[:space:]]+\(?>=[[:space:]]+set:[[:alnum:]]+\)?$/)
+                continue
+
+            capability = item
+            sub(/[[:space:]]+\(?>?=[[:space:]]+set:[[:alnum:]]+\)?$/, "", capability)
+            encoded = item
+            sub(/^.*[[:space:]]+\(?>?=[[:space:]]+/, "", encoded)
+            sub(/\)$/, "", encoded)
+            printf "%s\t%s\t%s\t%s\n", kind, package, capability, encoded
+        }
+    }
+    function flush() {
+        if (field != "")
+            emit(field, value)
+        field = ""
+        value = ""
+    }
+    {
+        if ($0 ~ /^[[:space:]]/ && field != "") {
+            line = $0
+            sub(/^[[:space:]]+/, "", line)
+            value = value " " line
+            next
+        }
+
+        flush()
+        if ($0 ~ /^Package: /)
+            package = substr($0, 10)
+        else if ($0 ~ /^Provides: /) {
+            field = "P"
+            value = substr($0, 11)
+        } else if ($0 ~ /^(Depends|Pre-Depends): /) {
+            field = "R"
+            value = substr($0, index($0, ":") + 2)
+        } else if ($0 == "")
+            package = ""
+    }
+    END { flush() }
+    ' "$WORK/available.txt" >"$WORK/relations.tsv"
+}
+
+check_relations()
+{
+    local tab=$'\t' result
+
+    awk -F '\t' '$1 == "P" { print $3 "\t" $2 "\t" $4 }' \
+        "$WORK/relations.tsv" | LC_ALL=C sort -t "$tab" -k1,1 \
+        >"$WORK/providers.tsv"
+    awk -F '\t' -v limit="$REQUIRE_LIMIT" \
+        '$1 == "R" { print $3 "\t" $2 "\t" $4; if (++n == limit) exit }' \
+        "$WORK/relations.tsv" | LC_ALL=C sort -u \
+        >"$WORK/requirements.tsv"
+
+    join -t "$tab" "$WORK/providers.tsv" "$WORK/requirements.tsv" \
+        >"$WORK/pairs.tsv"
+    : >"$WORK/paired.tsv"
+    : >"$WORK/compatible.tsv"
+    : >"$WORK/decode-errors.tsv"
+
+    awk -F '\t' '{ print $3; print $5 }' "$WORK/pairs.tsv" \
+        | "$WORK/compare" >"$WORK/results.txt"
+    paste "$WORK/pairs.tsv" "$WORK/results.txt" >"$WORK/compared.tsv"
+
+    while IFS=$'\t' read -r capability provider provided consumer required result; do
+        printf '%s\t%s\t%s\n' "$capability" "$consumer" "$required" \
+            >>"$WORK/paired.tsv"
+        case $result in
+            0|1)
+                printf '%s\t%s\t%s\n' "$capability" "$consumer" "$required" \
+                    >>"$WORK/compatible.tsv"
+                ;;
+            -3|-4)
+                printf '%s\t%s\t%s\t%s\t%s\t%s\n' \
+                    "$consumer" "$provider" "$capability" "$provided" "$required" "$result" \
+                    >>"$WORK/decode-errors.tsv"
+                ;;
+        esac
+    done <"$WORK/compared.tsv"
+
+    for file in requirements paired compatible; do
+        LC_ALL=C sort -u "$WORK/$file.tsv" -o "$WORK/$file.tsv"
+    done
+    comm -23 "$WORK/requirements.tsv" "$WORK/paired.tsv" >"$WORK/no-provider.tsv"
+    comm -23 "$WORK/paired.tsv" "$WORK/compatible.tsv" >"$WORK/incompatible.tsv"
+}
+
+count_lines()
+{
+    awk 'END { print NR + 0 }' "$1"
+}
+
+build_comparator
+printf 'Updating Sisyphus metadata...\n'
+fetch_metadata
+extract_relations
+check_relations
+
+requirements=$(count_lines "$WORK/requirements.tsv")
+compatible=$(count_lines "$WORK/compatible.tsv")
+no_provider=$(count_lines "$WORK/no-provider.tsv")
+incompatible=$(count_lines "$WORK/incompatible.tsv")
+decode_errors=$(count_lines "$WORK/decode-errors.tsv")
+comparisons=$(count_lines "$WORK/pairs.tsv")
+
+printf '\nrequirements: %s\n' "$requirements"
+printf 'comparisons:  %s\n' "$comparisons"
+printf 'compatible:   %s\n' "$compatible"
+printf 'no provider:  %s\n' "$no_provider"
+printf 'incompatible: %s\n' "$incompatible"
+printf 'decode errors:%s\n' "$decode_errors"
+
+if ((no_provider)); then
+    printf '\nRequirements without a provider (first 10):\n'
+    sed -n '1,10p' "$WORK/no-provider.tsv"
+fi
+if ((incompatible)); then
+    printf '\nUnsatisfied set requirements (first 10):\n'
+    sed -n '1,10p' "$WORK/incompatible.tsv"
+fi
+
+((requirements > 0)) || {
+    printf 'no set requirements found\n' >&2
+    exit 2
+}
+((no_provider == 0 && incompatible == 0 && decode_errors == 0))

tests/test_check_sisyphus_set_relations.py

@@ -0,0 +1,132 @@
+from __future__ import annotations
+
+import os
+import subprocess
+from pathlib import Path
+
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+SCRIPT = REPO_ROOT / "scripts" / "check_sisyphus_set_relations.sh"
+PROVIDER_SET = "set:jg2ZwFjdTD2B0"
+REQUIRED_SET = "set:jiXMb"
+
+
+def run_with_metadata(
+    tmp_path: Path, metadata: str, *, locale: str | None = None
+) -> subprocess.CompletedProcess[str]:
+    fake_bin = tmp_path / "bin"
+    fake_bin.mkdir()
+    metadata_file = tmp_path / "available.txt"
+    metadata_file.write_text(metadata, encoding="utf-8")
+
+    (fake_bin / "apt-get").write_text(
+        """#!/bin/sh
+archive=
+for argument in "$@"; do
+    case $argument in
+        Dir::Cache::archives=*) archive=${argument#*=} ;;
+    esac
+done
+[ -n "$archive" ] && [ -d "$archive/partial" ]
+""",
+        encoding="utf-8",
+    )
+    (fake_bin / "apt-cache").write_text(
+        '#!/bin/sh\ncat "$FAKE_APT_DUMP"\n', encoding="utf-8"
+    )
+    (fake_bin / "apt-get").chmod(0o755)
+    (fake_bin / "apt-cache").chmod(0o755)
+
+    environment = os.environ.copy()
+    environment["PATH"] = f"{fake_bin}:{environment['PATH']}"
+    environment["FAKE_APT_DUMP"] = str(metadata_file)
+    if locale is not None:
+        environment["LC_ALL"] = locale
+    return subprocess.run(
+        ["bash", str(SCRIPT)],
+        cwd=REPO_ROOT,
+        text=True,
+        capture_output=True,
+        env=environment,
+    )
+
+
+def test_accepts_a_real_provider_subset_relation_and_continuation_line(tmp_path: Path) -> None:
+    metadata = f"""\
+Package: libexample
+Provides: libexample.so.1()(64bit) (= {PROVIDER_SET})
+
+Package: example
+Depends: plain-dependency,
+ libexample.so.1()(64bit) (>= {REQUIRED_SET})
+
+"""
+
+    completed = run_with_metadata(tmp_path, metadata)
+
+    assert completed.returncode == 0, completed.stderr
+    assert "requirements: 1" in completed.stdout
+    assert "compatible:   1" in completed.stdout
+    assert "incompatible: 0" in completed.stdout
+
+
+def test_rejects_a_provider_that_does_not_contain_the_required_set(tmp_path: Path) -> None:
+    metadata = f"""\
+Package: libexample
+Provides: libexample.so.1()(64bit) (= {REQUIRED_SET})
+
+Package: example
+Depends: libexample.so.1()(64bit) (>= {PROVIDER_SET})
+
+"""
+
+    completed = run_with_metadata(tmp_path, metadata)
+
+    assert completed.returncode == 1
+    assert "requirements: 1" in completed.stdout
+    assert "compatible:   0" in completed.stdout
+    assert "incompatible: 1" in completed.stdout
+    assert "example" in completed.stdout
+
+
+def test_uses_one_collation_order_for_sort_and_join(tmp_path: Path) -> None:
+    metadata = f"""\
+Package: userdb-provider
+Provides: /usr/lib64/courier-authlib/libauthuserdb.so.0()(64bit) (= {PROVIDER_SET})
+
+Package: auth-provider
+Provides: /usr/lib64/courier-authlib/libcourierauth.so.0()(64bit) (= {PROVIDER_SET})
+
+Package: auth-consumer
+Depends: /usr/lib64/courier-authlib/libcourierauth.so.0()(64bit) (>= {REQUIRED_SET})
+
+Package: common-consumer
+Depends: /usr/lib64/courier-authlib/libcourierauthcommon.so.0()(64bit) (>= {REQUIRED_SET})
+
+"""
+
+    completed = run_with_metadata(tmp_path, metadata, locale="ru_RU.utf8")
+
+    assert completed.returncode == 1
+    assert "is not sorted" not in completed.stderr
+    assert "requirements: 2" in completed.stdout
+    assert "compatible:   1" in completed.stdout
+    assert "no provider:  1" in completed.stdout
+
+
+def test_large_set_is_not_passed_as_a_command_line_argument(tmp_path: Path) -> None:
+    oversized_set = "set:" + "a" * 140_000
+    metadata = f"""\
+Package: libexample
+Provides: libexample.so.1()(64bit) (= {oversized_set})
+
+Package: example
+Depends: libexample.so.1()(64bit) (>= {REQUIRED_SET})
+
+"""
+
+    completed = run_with_metadata(tmp_path, metadata)
+
+    assert completed.returncode == 1
+    assert "Argument list too long" not in completed.stderr
+    assert "decode errors:1" in completed.stdout

tt

@@ -5,13 +5,7 @@
 
 
 2026-07-19T03:48:54Z	DEPENDENCY	SILLY	x86_64	mismatch	requires	libjpeg.so.62()(64bit)	>=	set:jfvZkgcUQqm1WcNFYdT6Zh	set:kgvLPFlp5EP4DOJVCPR1WC6	-
-2026-07-19T03:48:54Z	DEPENDENCY	SILLY	x86_64	match	requires	libpng16.so.16()(64bit)	>=	set:lhkC1fmC6EFQ5Klh4dezRCyvQFOXIZ3uwhXdUhBK49	set:lhkC1fmC6EFQ5Klh4dezRCyvQFOXIZ3uwhXdUhBK49	-
 
-2026-07-19T03:58:11Z	START	abuse	x86_64	processing	-	-	-	-	-	0.8-alt2@1308659965
-2026-07-19T03:58:33Z	DEPENDENCY	abuse	x86_64	match	requires	libSDL-1.2.so.0()(64bit)	>=	set:lgkZiCZcFvTDDX6Y1PaRgCdFuVAwizcZfaTezfxZFUxcDOoVbbKRxY6ahoMjPQkxX5fdZtjAKGEY4	set:lgkZiCZcFvTDDX6Y1PaRgCdFuVAwizcZfaTezfxZFUxcDOoVbbKRxY6ahoMjPQkxX5fdZtjAKGEY4	-
 2026-07-19T03:58:33Z	DEPENDENCY	abuse	x86_64	mismatch	requires	libSDL_mixer-1.2.so.0()(64bit)	>=	set:lgIdCOqw7JFqcojADYwCZ7pQY3CJ1LnAqVzI3Z0QXq0pLy1	set:kfMSoaHVyGBhV7CJ1YuVyx95uAdQPsq0b5zW6rUlsY3	-
-2026-07-19T03:58:33Z	SUMMARY	abuse	x86_64	mismatch	-	-	-	-	-	requires expected=2 generated=2 differences=1 extras=0; complete=1
 
 2026-07-19T04:08:20Z	DEPENDENCY	airstrike	x86_64	mismatch	requires	libSDL_mixer-1.2.so.0()(64bit)	>=	set:lirzppdVhrMiF4ChTZdaR8	set:khS6jpdPZrxpQlg2V9Bq4	-
-
-