|
| 1 | +#!/bin/bash |
| 2 | +# SPDX-License-Identifier: PMPL-1.0-or-later |
| 3 | +# Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk> |
| 4 | +# |
| 5 | +# Pipeline drift guard for nextgen-typing. |
| 6 | +# Ensures canonical pipeline docs and 6a2 machine-readable state stay aligned. |
| 7 | + |
| 8 | +set -euo pipefail |
| 9 | + |
| 10 | +REPO_ROOT="${1:-.}" |
| 11 | +ERRORS=0 |
| 12 | + |
| 13 | +README="$REPO_ROOT/README.adoc" |
| 14 | +ARCHITECTURE="$REPO_ROOT/docs/ARCHITECTURE.adoc" |
| 15 | +PIPELINE="$REPO_ROOT/docs/PIPELINE.adoc" |
| 16 | +STATE="$REPO_ROOT/.machine_readable/6a2/STATE.a2ml" |
| 17 | +META="$REPO_ROOT/.machine_readable/6a2/META.a2ml" |
| 18 | +ECOSYSTEM="$REPO_ROOT/.machine_readable/6a2/ECOSYSTEM.a2ml" |
| 19 | + |
| 20 | +CANON_CHAIN="katagoria → typell → typed-wasm → PanLL" |
| 21 | +CHAIN_REGEX='katagoria[[:space:]]*→[[:space:]]*typell[[:space:]]*→[[:space:]]*typed-wasm[[:space:]]*→[[:space:]]*PanLL' |
| 22 | +CHAIN_REGEX_WITH_OPTIONAL_ROLES='katagoria([[:space:]]*\([^)]*\))?[[:space:]]*→[[:space:]]*typell([[:space:]]*\([^)]*\))?[[:space:]]*→[[:space:]]*typed-wasm([[:space:]]*\([^)]*\))?[[:space:]]*→[[:space:]]*PanLL([[:space:]]*\([^)]*\))?' |
| 23 | +STALE_REGEX='Not yet created|not yet updated|Does not exist yet|🚧 Planned|has not been created yet' |
| 24 | + |
| 25 | +# ANSI colors |
| 26 | +RED='\033[0;31m' |
| 27 | +GREEN='\033[0;32m' |
| 28 | +BLUE='\033[0;34m' |
| 29 | +NC='\033[0m' |
| 30 | + |
| 31 | +log_info() { |
| 32 | + echo -e "${BLUE}INFO${NC}: $*" >&2 |
| 33 | +} |
| 34 | + |
| 35 | +log_pass() { |
| 36 | + echo -e "${GREEN}PASS${NC}: $*" >&2 |
| 37 | +} |
| 38 | + |
| 39 | +log_error() { |
| 40 | + echo -e "${RED}ERROR${NC}: $*" >&2 |
| 41 | + ERRORS=$((ERRORS + 1)) |
| 42 | +} |
| 43 | + |
| 44 | +require_file() { |
| 45 | + local file="$1" |
| 46 | + if [ -f "$file" ]; then |
| 47 | + log_pass "Found: ${file#$REPO_ROOT/}" |
| 48 | + else |
| 49 | + log_error "Missing required file: ${file#$REPO_ROOT/}" |
| 50 | + fi |
| 51 | +} |
| 52 | + |
| 53 | +extract_quoted_value() { |
| 54 | + local key="$1" |
| 55 | + local file="$2" |
| 56 | + grep -E "^[[:space:]]*$key[[:space:]]*=" "$file" | head -1 | sed -E 's/^[^"]*"([^"]+)".*/\1/' || true |
| 57 | +} |
| 58 | + |
| 59 | +check_contains_regex() { |
| 60 | + local file="$1" |
| 61 | + local regex="$2" |
| 62 | + local description="$3" |
| 63 | + if grep -Eq "$regex" "$file"; then |
| 64 | + log_pass "$description" |
| 65 | + else |
| 66 | + log_error "$description" |
| 67 | + fi |
| 68 | +} |
| 69 | + |
| 70 | +check_katagoria_active_row() { |
| 71 | + local file="$1" |
| 72 | + local row_regex="$2" |
| 73 | + local label="$3" |
| 74 | + if awk -v row_re="$row_regex" ' |
| 75 | + $0 ~ row_re { |
| 76 | + getline |
| 77 | + if ($0 ~ /^\|[[:space:]]*.*Active/) ok=1 |
| 78 | + } |
| 79 | + END { exit(ok ? 0 : 1) } |
| 80 | + ' "$file"; then |
| 81 | + log_pass "$label" |
| 82 | + else |
| 83 | + log_error "$label" |
| 84 | + fi |
| 85 | +} |
| 86 | + |
| 87 | +echo "" |
| 88 | +log_info "Pipeline drift guard: validating canonical files and cross-file invariants" |
| 89 | +echo "" |
| 90 | + |
| 91 | +require_file "$README" |
| 92 | +require_file "$ARCHITECTURE" |
| 93 | +require_file "$PIPELINE" |
| 94 | +require_file "$STATE" |
| 95 | +require_file "$META" |
| 96 | +require_file "$ECOSYSTEM" |
| 97 | + |
| 98 | +if [ "$ERRORS" -gt 0 ]; then |
| 99 | + echo "" |
| 100 | + echo -e "${RED}Drift guard failed due to missing files.${NC}" >&2 |
| 101 | + exit 1 |
| 102 | +fi |
| 103 | + |
| 104 | +echo "" |
| 105 | +log_info "Checking canonical chain and invariant language" |
| 106 | +echo "" |
| 107 | + |
| 108 | +check_contains_regex "$README" "$CHAIN_REGEX" "README contains canonical main chain" |
| 109 | +check_contains_regex "$ARCHITECTURE" "$CHAIN_REGEX" "ARCHITECTURE contains canonical main chain" |
| 110 | +check_contains_regex "$PIPELINE" "$CHAIN_REGEX" "PIPELINE contains canonical main chain" |
| 111 | +check_contains_regex "$STATE" "$CHAIN_REGEX_WITH_OPTIONAL_ROLES" "STATE purpose contains canonical main chain" |
| 112 | +check_contains_regex "$ECOSYSTEM" "$CHAIN_REGEX" "ECOSYSTEM purpose contains canonical main chain" |
| 113 | + |
| 114 | +PIPE_CHAIN="$(extract_quoted_value "chain" "$PIPELINE")" |
| 115 | +if [ "$PIPE_CHAIN" = "$CANON_CHAIN" ]; then |
| 116 | + log_pass "PIPELINE chain value exactly matches canonical chain" |
| 117 | +else |
| 118 | + log_error "PIPELINE chain mismatch: expected '$CANON_CHAIN', got '$PIPE_CHAIN'" |
| 119 | +fi |
| 120 | + |
| 121 | +check_contains_regex "$README" "open-ended" "README contains open-ended TypeLL framing" |
| 122 | +check_contains_regex "$ARCHITECTURE" "Key Invariant: TypeLL Is Open-Ended" "ARCHITECTURE contains TypeLL key invariant section" |
| 123 | +check_contains_regex "$PIPELINE" "open-ended progressive" "PIPELINE contains open-ended progressive invariant" |
| 124 | +check_contains_regex "$META" 'title = "TypeLL is open-ended progressive, not capped at 10 levels"' "META ADR-002 encodes open-ended TypeLL framing" |
| 125 | + |
| 126 | +check_katagoria_active_row "$README" '^[|][[:space:]]*katagoria[[:space:]]*$' "README marks katagoria row as Active" |
| 127 | +check_katagoria_active_row "$ARCHITECTURE" '^[|][[:space:]]*`katagoria`[[:space:]]*$' "ARCHITECTURE marks katagoria row as Active" |
| 128 | + |
| 129 | +echo "" |
| 130 | +log_info "Checking date synchronization across canonical state files" |
| 131 | +echo "" |
| 132 | + |
| 133 | +STATE_DATE="$(extract_quoted_value "last-updated" "$STATE")" |
| 134 | +META_DATE="$(extract_quoted_value "last-updated" "$META")" |
| 135 | +ECOSYSTEM_DATE="$(extract_quoted_value "last-updated" "$ECOSYSTEM")" |
| 136 | +PIPELINE_VERSION="$(extract_quoted_value "version" "$PIPELINE")" |
| 137 | +ARCH_DATE="$(grep -E '^[0-9]{4}-[0-9]{2}-[0-9]{2}$' "$ARCHITECTURE" | head -1 || true)" |
| 138 | + |
| 139 | +if [ -z "$STATE_DATE" ] || [ -z "$META_DATE" ] || [ -z "$ECOSYSTEM_DATE" ] || [ -z "$PIPELINE_VERSION" ] || [ -z "$ARCH_DATE" ]; then |
| 140 | + log_error "One or more date/version fields are missing (STATE/META/ECOSYSTEM last-updated, PIPELINE version, ARCHITECTURE header date)" |
| 141 | +else |
| 142 | + if [ "$META_DATE" = "$STATE_DATE" ] && [ "$ECOSYSTEM_DATE" = "$STATE_DATE" ] && [ "$PIPELINE_VERSION" = "$STATE_DATE" ] && [ "$ARCH_DATE" = "$STATE_DATE" ]; then |
| 143 | + log_pass "Dates synchronized: STATE/META/ECOSYSTEM last-updated == PIPELINE version == ARCHITECTURE header date ($STATE_DATE)" |
| 144 | + else |
| 145 | + log_error "Date drift detected: STATE=$STATE_DATE META=$META_DATE ECOSYSTEM=$ECOSYSTEM_DATE PIPELINE.version=$PIPELINE_VERSION ARCHITECTURE.date=$ARCH_DATE" |
| 146 | + fi |
| 147 | +fi |
| 148 | + |
| 149 | +echo "" |
| 150 | +log_info "Checking for known stale phrases" |
| 151 | +echo "" |
| 152 | + |
| 153 | +STALE_HITS="$(grep -nE "$STALE_REGEX" "$README" "$ARCHITECTURE" "$PIPELINE" "$STATE" "$META" "$ECOSYSTEM" || true)" |
| 154 | +if [ -n "$STALE_HITS" ]; then |
| 155 | + log_error "Stale status phrases detected in canonical files:" |
| 156 | + echo "$STALE_HITS" >&2 |
| 157 | +else |
| 158 | + log_pass "No stale status phrases found in canonical files" |
| 159 | +fi |
| 160 | + |
| 161 | +if grep -Fq 'docs/PIPELINE.adoc' "$README"; then |
| 162 | + log_pass "README links to docs/PIPELINE.adoc" |
| 163 | +else |
| 164 | + log_error "README is missing a link/reference to docs/PIPELINE.adoc" |
| 165 | +fi |
| 166 | + |
| 167 | +echo "" |
| 168 | +echo "═══════════════════════════════════════════════════════════════════════════════" >&2 |
| 169 | +echo "PIPELINE DRIFT GUARD SUMMARY" >&2 |
| 170 | +echo "═══════════════════════════════════════════════════════════════════════════════" >&2 |
| 171 | +if [ "$ERRORS" -eq 0 ]; then |
| 172 | + echo -e "${GREEN}PASS${NC}: no pipeline drift detected." >&2 |
| 173 | + exit 0 |
| 174 | +else |
| 175 | + echo -e "${RED}FAIL${NC}: $ERRORS issue(s) detected." >&2 |
| 176 | + exit 1 |
| 177 | +fi |
0 commit comments