|
| 1 | +#!/usr/bin/env bash |
| 2 | +# fix-haddock-links.sh |
| 3 | +# |
| 4 | +# Post-processes Haddock HTML generated by `cabal haddock-project` to fix |
| 5 | +# cross-package links. Without this, links to external dependencies are |
| 6 | +# broken 404s because the hosted site only contains docs for packages in |
| 7 | +# this repo. |
| 8 | +# |
| 9 | +# The script: |
| 10 | +# 1. Fetches the CHaP package index to classify packages |
| 11 | +# 2. Auto-discovers which packages have broken links |
| 12 | +# 3. Rewrites links to known doc sites or Hackage |
| 13 | +# 4. Validates all rewritten links with HTTP HEAD requests |
| 14 | +# 5. Replaces dead links (internal modules, missing docs) with annotated |
| 15 | +# plain text so there are zero broken clickable links |
| 16 | +# |
| 17 | +# Usage: ./scripts/fix-haddock-links.sh <website-directory> |
| 18 | + |
| 19 | +set -euo pipefail |
| 20 | + |
| 21 | +WEBSITE_DIR="${1:?Usage: $0 <website-directory>}" |
| 22 | + |
| 23 | +if [ ! -d "$WEBSITE_DIR" ]; then |
| 24 | + echo "Error: $WEBSITE_DIR is not a directory" >&2 |
| 25 | + exit 1 |
| 26 | +fi |
| 27 | + |
| 28 | +# ============================================================================ |
| 29 | +# Configuration: doc site mappings for CHaP packages |
| 30 | +# ============================================================================ |
| 31 | + |
| 32 | +LEDGER_DOCS="https://cardano-ledger.cardano.intersectmbo.org" |
| 33 | +BASE_DOCS="https://base.cardano.intersectmbo.org" |
| 34 | +PLUTUS_DOCS="https://plutus.cardano.intersectmbo.org/haddock/latest" |
| 35 | +CONSENSUS_DOCS="https://ouroboros-consensus.cardano.intersectmbo.org/haddocks" |
| 36 | +NETWORK_DOCS="https://ouroboros-network.cardano.intersectmbo.org" |
| 37 | +IO_SIM_DOCS="https://input-output-hk.github.io/io-sim" |
| 38 | + |
| 39 | +# Prefix rules: first match wins |
| 40 | +PREFIX_RULES=( |
| 41 | + "cardano-ledger-|$LEDGER_DOCS" |
| 42 | + "ouroboros-consensus|$CONSENSUS_DOCS" |
| 43 | + "ouroboros-network|$NETWORK_DOCS" |
| 44 | + "plutus-|$PLUTUS_DOCS" |
| 45 | + "io-classes|$IO_SIM_DOCS" |
| 46 | + "io-sim|$IO_SIM_DOCS" |
| 47 | +) |
| 48 | + |
| 49 | +# Exact overrides for CHaP packages whose names don't match a prefix rule |
| 50 | +declare -A EXACT_OVERRIDES=( |
| 51 | + # cardano-ledger repo |
| 52 | + [cardano-crypto-wrapper]="$LEDGER_DOCS" |
| 53 | + [cardano-data]="$LEDGER_DOCS" |
| 54 | + [cardano-protocol-tpraos]="$LEDGER_DOCS" |
| 55 | + [small-steps]="$LEDGER_DOCS" |
| 56 | + [byron-spec-chain]="$LEDGER_DOCS" |
| 57 | + [byron-spec-ledger]="$LEDGER_DOCS" |
| 58 | + [non-integral]="$LEDGER_DOCS" |
| 59 | + [vector-map]="$LEDGER_DOCS" |
| 60 | + # cardano-base repo |
| 61 | + [cardano-binary]="$BASE_DOCS" |
| 62 | + [cardano-crypto-class]="$BASE_DOCS" |
| 63 | + [cardano-crypto-praos]="$BASE_DOCS" |
| 64 | + [cardano-slotting]="$BASE_DOCS" |
| 65 | + [cardano-strict-containers]="$BASE_DOCS" |
| 66 | + [cardano-base]="$BASE_DOCS" |
| 67 | + [cardano-prelude]="$BASE_DOCS" |
| 68 | + [heapwords]="$BASE_DOCS" |
| 69 | + [measures]="$BASE_DOCS" |
| 70 | + # ouroboros-network repo |
| 71 | + [cardano-diffusion]="$NETWORK_DOCS" |
| 72 | + [network-mux]="$NETWORK_DOCS" |
| 73 | + [monoidal-synchronisation]="$NETWORK_DOCS" |
| 74 | + [typed-protocols]="$NETWORK_DOCS" |
| 75 | + [cardano-ping]="$NETWORK_DOCS" |
| 76 | + # io-sim repo |
| 77 | + [strict-checked-vars]="$IO_SIM_DOCS" |
| 78 | + [strict-sop-core]="$IO_SIM_DOCS" |
| 79 | + [strict-mvar]="$IO_SIM_DOCS" |
| 80 | +) |
| 81 | + |
| 82 | +# Resolve a package name to its doc site URL (or HACKAGE or NONE) |
| 83 | +resolve_url() { |
| 84 | + local pkg="$1" |
| 85 | + local is_chap="$2" # "yes" or "no" |
| 86 | + |
| 87 | + # Exact override? |
| 88 | + if [[ -v "EXACT_OVERRIDES[$pkg]" ]]; then |
| 89 | + echo "${EXACT_OVERRIDES[$pkg]}" |
| 90 | + return |
| 91 | + fi |
| 92 | + |
| 93 | + # Prefix rule? |
| 94 | + for rule in "${PREFIX_RULES[@]}"; do |
| 95 | + local prefix="${rule%%|*}" |
| 96 | + local url="${rule##*|}" |
| 97 | + if [[ "$pkg" == "${prefix}"* ]]; then |
| 98 | + echo "$url" |
| 99 | + return |
| 100 | + fi |
| 101 | + done |
| 102 | + |
| 103 | + # CHaP package with no known doc site |
| 104 | + if [[ "$is_chap" == "yes" ]]; then |
| 105 | + echo "NONE" |
| 106 | + return |
| 107 | + fi |
| 108 | + |
| 109 | + # Not on CHaP -> Hackage |
| 110 | + echo "HACKAGE" |
| 111 | +} |
| 112 | + |
| 113 | +# ============================================================================ |
| 114 | +# Phase 0: Create symlinks for local packages with versioned directory names |
| 115 | +# ============================================================================ |
| 116 | + |
| 117 | +echo "Phase 0: Creating symlinks for versioned directories..." |
| 118 | +symlink_count=0 |
| 119 | + |
| 120 | +# Build a set of existing short directory names |
| 121 | +declare -A SHORT_DIRS |
| 122 | +for dir in "$WEBSITE_DIR"/*/; do |
| 123 | + SHORT_DIRS["$(basename "$dir")"]=1 |
| 124 | +done |
| 125 | + |
| 126 | +# a) Symlink versioned-inplace dirs to short names (local cabal packages) |
| 127 | +# e.g. cardano-api-10.26.0.0-inplace -> cardano-api |
| 128 | +for dir in "$WEBSITE_DIR"/*/; do |
| 129 | + dirname="$(basename "$dir")" |
| 130 | + if [[ "$dirname" =~ ^(.+)-[0-9]+\.[0-9]+.*-inplace$ ]]; then |
| 131 | + short_name="${BASH_REMATCH[1]}" |
| 132 | + if [[ ! -e "$WEBSITE_DIR/$short_name" ]] && [[ ! "$dirname" =~ -inplace-.+ ]]; then |
| 133 | + ln -s "$dirname" "$WEBSITE_DIR/$short_name" |
| 134 | + symlink_count=$((symlink_count + 1)) |
| 135 | + fi |
| 136 | + fi |
| 137 | +done |
| 138 | + |
| 139 | +# b) Discover versioned-hash link targets that match existing short-name dirs |
| 140 | +# e.g. links to ../base-4.20.2.0-f074/ when ./website/base/ exists |
| 141 | +# Create symlinks so these links resolve locally |
| 142 | +VERSIONED_TARGETS=$(grep -rohP 'href="\.\./(\.\./)?\K[a-zA-Z][a-zA-Z0-9_.-]*(?=/)' "$WEBSITE_DIR" 2>/dev/null \ |
| 143 | + | grep -P '^.+-[0-9]+\.[0-9]+' | sort -u) |
| 144 | + |
| 145 | +while IFS= read -r target; do |
| 146 | + [ -z "$target" ] && continue |
| 147 | + # Already exists? |
| 148 | + [[ -e "$WEBSITE_DIR/$target" ]] && continue |
| 149 | + # Strip version+hash: "base-4.20.2.0-f074" -> "base" |
| 150 | + # Pattern: strip everything from the first "-DIGIT.DIGIT" onward |
| 151 | + if [[ "$target" =~ ^(.+)-[0-9]+\.[0-9] ]]; then |
| 152 | + short_name="${BASH_REMATCH[1]}" |
| 153 | + else |
| 154 | + short_name="$target" |
| 155 | + fi |
| 156 | + if [[ -v "SHORT_DIRS[$short_name]" ]]; then |
| 157 | + ln -s "$short_name" "$WEBSITE_DIR/$target" |
| 158 | + symlink_count=$((symlink_count + 1)) |
| 159 | + fi |
| 160 | +done <<< "$VERSIONED_TARGETS" |
| 161 | + |
| 162 | +echo " Created $symlink_count symlinks" |
| 163 | + |
| 164 | +# ============================================================================ |
| 165 | +# Phase 1: Fetch CHaP package list |
| 166 | +# ============================================================================ |
| 167 | + |
| 168 | +echo "Phase 1: Fetching CHaP package index..." |
| 169 | +CHAP_PKGS_FILE=$(mktemp) |
| 170 | +trap 'rm -f "$CHAP_PKGS_FILE"' EXIT |
| 171 | + |
| 172 | +curl -sL https://chap.intersectmbo.org/01-index.tar.gz \ |
| 173 | + | tar -tz \ |
| 174 | + | grep -oP '^[^/]+' \ |
| 175 | + | sort -u > "$CHAP_PKGS_FILE" |
| 176 | + |
| 177 | +chap_total=$(wc -l < "$CHAP_PKGS_FILE") |
| 178 | +echo " Fetched $chap_total CHaP packages" |
| 179 | + |
| 180 | +# Load into associative array for O(1) lookup |
| 181 | +declare -A CHAP_SET |
| 182 | +while IFS= read -r pkg; do |
| 183 | + CHAP_SET["$pkg"]=1 |
| 184 | +done < "$CHAP_PKGS_FILE" |
| 185 | + |
| 186 | +# ============================================================================ |
| 187 | +# Phase 2: Discover local packages |
| 188 | +# ============================================================================ |
| 189 | + |
| 190 | +echo "Phase 2: Discovering local packages..." |
| 191 | +declare -A LOCAL_SET |
| 192 | +for dir in "$WEBSITE_DIR"/*/; do |
| 193 | + [ -d "$dir" ] || continue |
| 194 | + name="$(basename "$dir")" |
| 195 | + LOCAL_SET["$name"]=1 |
| 196 | +done |
| 197 | +echo " Found ${#LOCAL_SET[@]} local directories" |
| 198 | + |
| 199 | +# ============================================================================ |
| 200 | +# Phase 3: Discover broken link targets |
| 201 | +# ============================================================================ |
| 202 | + |
| 203 | +echo "Phase 3: Discovering cross-package link targets..." |
| 204 | +DISCOVERED_PKGS=$(grep -rohP 'href="\.\./(\.\./)?\K[a-zA-Z][a-zA-Z0-9_.-]*(?=/)' "$WEBSITE_DIR" 2>/dev/null | sort -u) |
| 205 | +discovered_total=$(echo "$DISCOVERED_PKGS" | grep -c . || true) |
| 206 | +echo " Found $discovered_total unique link targets" |
| 207 | + |
| 208 | +# ============================================================================ |
| 209 | +# Phase 4 + 5: Classify, resolve, and apply rewrites |
| 210 | +# ============================================================================ |
| 211 | + |
| 212 | +echo "Phase 4: Classifying and rewriting links..." |
| 213 | + |
| 214 | +SED_ARGS=() |
| 215 | +UNMAPPED_CHAP=() |
| 216 | +REWRITTEN_HACKAGE=() |
| 217 | +REWRITTEN_DOCSITE=() |
| 218 | +skipped_count=0 |
| 219 | + |
| 220 | +while IFS= read -r pkg; do |
| 221 | + [ -z "$pkg" ] && continue |
| 222 | + |
| 223 | + # Skip local packages (directory already exists) |
| 224 | + if [[ -v "LOCAL_SET[$pkg]" ]]; then |
| 225 | + skipped_count=$((skipped_count + 1)) |
| 226 | + continue |
| 227 | + fi |
| 228 | + |
| 229 | + # Determine if CHaP package |
| 230 | + is_chap="no" |
| 231 | + if [[ -v "CHAP_SET[$pkg]" ]]; then |
| 232 | + is_chap="yes" |
| 233 | + fi |
| 234 | + |
| 235 | + url=$(resolve_url "$pkg" "$is_chap") |
| 236 | + |
| 237 | + if [[ "$url" == "NONE" ]]; then |
| 238 | + # CHaP package with no doc site — will be made unclickable |
| 239 | + UNMAPPED_CHAP+=("$pkg") |
| 240 | + continue |
| 241 | + elif [[ "$url" == "HACKAGE" ]]; then |
| 242 | + target="https://hackage.haskell.org/package/${pkg}/docs/" |
| 243 | + REWRITTEN_HACKAGE+=("$pkg") |
| 244 | + else |
| 245 | + target="${url}/${pkg}/" |
| 246 | + REWRITTEN_DOCSITE+=("$pkg") |
| 247 | + fi |
| 248 | + |
| 249 | + SED_ARGS+=(-e "s|href=\"\\.\\./${pkg}/|href=\"${target}|g") |
| 250 | + SED_ARGS+=(-e "s|href=\"\\.\\./\\.\\./${pkg}/|href=\"${target}|g") |
| 251 | +done <<< "$DISCOVERED_PKGS" |
| 252 | + |
| 253 | +echo " Local (skipped): $skipped_count" |
| 254 | +echo " Rewritten (doc site): ${#REWRITTEN_DOCSITE[@]}" |
| 255 | +echo " Rewritten (Hackage): ${#REWRITTEN_HACKAGE[@]}" |
| 256 | +echo " Unmapped CHaP: ${#UNMAPPED_CHAP[@]}" |
| 257 | + |
| 258 | +if [[ ${#UNMAPPED_CHAP[@]} -gt 0 ]]; then |
| 259 | + echo " Unmapped CHaP packages (will be made unclickable):" |
| 260 | + for pkg in "${UNMAPPED_CHAP[@]}"; do |
| 261 | + echo " - $pkg" |
| 262 | + done |
| 263 | +fi |
| 264 | + |
| 265 | +# Apply URL rewrites |
| 266 | +if [[ ${#SED_ARGS[@]} -gt 0 ]]; then |
| 267 | + echo "Phase 5: Applying link rewrites..." |
| 268 | + find "$WEBSITE_DIR" -name '*.html' -print0 | xargs -0 -P "$(nproc)" sed -i "${SED_ARGS[@]}" |
| 269 | + echo " Done" |
| 270 | +fi |
| 271 | + |
| 272 | +# ============================================================================ |
| 273 | +# Phase 5b: Make unmapped CHaP links unclickable |
| 274 | +# ============================================================================ |
| 275 | + |
| 276 | +if [[ ${#UNMAPPED_CHAP[@]} -gt 0 ]]; then |
| 277 | + echo "Phase 5b: Making unmapped CHaP links unclickable..." |
| 278 | + UNMAP_SED_ARGS=() |
| 279 | + for pkg in "${UNMAPPED_CHAP[@]}"; do |
| 280 | + # Replace <a href="../PKG/...">TEXT</a> with <span class="dead-link" title="...">TEXT</span> |
| 281 | + # We need to handle both ../ and ../../ prefixes |
| 282 | + UNMAP_SED_ARGS+=(-e "s|<a href=\"\\.\\./${pkg}/[^\"]*\"[^>]*>\\([^<]*\\)</a>|<span class=\"dead-link\" title=\"No hosted documentation available for ${pkg}\">\\1</span>|g") |
| 283 | + UNMAP_SED_ARGS+=(-e "s|<a href=\"\\.\\./\\.\\./${pkg}/[^\"]*\"[^>]*>\\([^<]*\\)</a>|<span class=\"dead-link\" title=\"No hosted documentation available for ${pkg}\">\\1</span>|g") |
| 284 | + done |
| 285 | + find "$WEBSITE_DIR" -name '*.html' -print0 | xargs -0 -P "$(nproc)" sed -i "${UNMAP_SED_ARGS[@]}" |
| 286 | + echo " Done" |
| 287 | +fi |
| 288 | + |
| 289 | +# ============================================================================ |
| 290 | +# Phase 6: Post-process validation |
| 291 | +# ============================================================================ |
| 292 | + |
| 293 | +echo "Phase 6: Validating rewritten links..." |
| 294 | + |
| 295 | +# Extract all unique rewritten external URLs (stripping #fragment) |
| 296 | +URLS_FILE=$(mktemp) |
| 297 | +trap 'rm -f "$CHAP_PKGS_FILE" "$URLS_FILE"' EXIT |
| 298 | + |
| 299 | +grep -rohP 'href="\Khttps://[^"#]+\.html' "$WEBSITE_DIR" 2>/dev/null | sort -u > "$URLS_FILE" |
| 300 | +url_count=$(wc -l < "$URLS_FILE") |
| 301 | +echo " Found $url_count unique external URLs to validate" |
| 302 | + |
| 303 | +# Validate each URL with HEAD request, collect 404s |
| 304 | +DEAD_URLS_FILE=$(mktemp) |
| 305 | +trap 'rm -f "$CHAP_PKGS_FILE" "$URLS_FILE" "$DEAD_URLS_FILE"' EXIT |
| 306 | + |
| 307 | +if [[ $url_count -gt 0 ]]; then |
| 308 | + # shellcheck disable=SC2016 # single quotes intentional: expansions evaluated by inner sh -c |
| 309 | + xargs -P 16 -I{} sh -c \ |
| 310 | + 'code=$(curl -sI -o /dev/null -w "%{http_code}" --connect-timeout 5 --max-time 10 "{}"); if [ "$code" != "200" ] && [ "$code" != "301" ] && [ "$code" != "302" ]; then echo "{}"; fi' \ |
| 311 | + < "$URLS_FILE" > "$DEAD_URLS_FILE" 2>/dev/null |
| 312 | +fi |
| 313 | + |
| 314 | +dead_count=$(wc -l < "$DEAD_URLS_FILE" | tr -d ' ') |
| 315 | +valid_count=$((url_count - dead_count)) |
| 316 | +echo " Valid: $valid_count" |
| 317 | +echo " Dead: $dead_count" |
| 318 | + |
| 319 | +# ============================================================================ |
| 320 | +# Phase 7: Replace dead links with annotated plain text |
| 321 | +# ============================================================================ |
| 322 | + |
| 323 | +if [[ $dead_count -gt 0 ]]; then |
| 324 | + echo "Phase 7: Replacing dead links with annotated text..." |
| 325 | + echo " Dead URLs:" |
| 326 | + |
| 327 | + DEAD_SED_ARGS=() |
| 328 | + while IFS= read -r dead_url; do |
| 329 | + [ -z "$dead_url" ] && continue |
| 330 | + # Extract package name from URL |
| 331 | + # Doc site pattern: https://SITE/PACKAGE/Module.html |
| 332 | + # Hackage pattern: https://hackage.haskell.org/package/PACKAGE/docs/Module.html |
| 333 | + if [[ "$dead_url" == *"hackage.haskell.org"* ]]; then |
| 334 | + pkg_name=$(echo "$dead_url" | grep -oP 'package/\K[^/]+') |
| 335 | + else |
| 336 | + # Take the second-to-last path component (PACKAGE from .../PACKAGE/Module.html) |
| 337 | + pkg_name=$(echo "$dead_url" | grep -oP '[^/]+(?=/[^/]+\.html$)') |
| 338 | + fi |
| 339 | + echo " - $dead_url ($pkg_name)" |
| 340 | + |
| 341 | + # Escape special chars for sed |
| 342 | + escaped_url=$(printf '%s' "$dead_url" | sed 's|[&/\]|\\&|g; s|\.|\\.|g') |
| 343 | + |
| 344 | + # Replace <a href="DEAD_URL...">TEXT</a> with dead-link span |
| 345 | + # The href may have a #fragment appended |
| 346 | + DEAD_SED_ARGS+=(-e "s|<a href=\"${escaped_url}[^\"]*\"[^>]*>\\([^<]*\\)</a>|<span class=\"dead-link\" title=\"From ${pkg_name} — documentation not available at the expected URL\">\\1</span>|g") |
| 347 | + done < "$DEAD_URLS_FILE" |
| 348 | + |
| 349 | + if [[ ${#DEAD_SED_ARGS[@]} -gt 0 ]]; then |
| 350 | + find "$WEBSITE_DIR" -name '*.html' -print0 | xargs -0 -P "$(nproc)" sed -i "${DEAD_SED_ARGS[@]}" |
| 351 | + fi |
| 352 | + |
| 353 | + # Inject CSS for dead-link styling into all HTML files |
| 354 | + echo " Injecting dead-link CSS..." |
| 355 | + find "$WEBSITE_DIR" -name '*.html' -print0 | xargs -0 -P "$(nproc)" sed -i \ |
| 356 | + 's|</head>|<style>.dead-link { border-bottom: 1px dotted \#888; color: \#888; cursor: help; }</style></head>|' |
| 357 | + |
| 358 | + echo " Done" |
| 359 | +fi |
| 360 | + |
| 361 | +# ============================================================================ |
| 362 | +# Summary |
| 363 | +# ============================================================================ |
| 364 | + |
| 365 | +echo "" |
| 366 | +echo "=== fix-haddock-links summary ===" |
| 367 | +echo " Packages discovered in HTML: $discovered_total" |
| 368 | +echo " Local (skipped): $skipped_count" |
| 369 | +echo " Rewritten to doc sites: ${#REWRITTEN_DOCSITE[@]}" |
| 370 | +echo " Rewritten to Hackage: ${#REWRITTEN_HACKAGE[@]}" |
| 371 | +echo " Unmapped CHaP (unclickable): ${#UNMAPPED_CHAP[@]}" |
| 372 | +echo " Links validated: $url_count" |
| 373 | +echo " Dead links (unclickable): $dead_count" |
| 374 | +echo "=================================" |
0 commit comments