|
| 1 | +//> using scala 3.6.3 |
| 2 | +//> using dep org.virtuslab::scala-yaml::0.3.1 |
| 3 | +//> using dep co.fs2::fs2-io::3.12.2 |
| 4 | +//> using dep com.typesafe:config:1.4.5 |
| 5 | + |
| 6 | +import cats.effect.{IO, IOApp} |
| 7 | +import cats.syntax.all.* |
| 8 | +import fs2.io.file.{Files, Path} |
| 9 | +import org.virtuslab.yaml.* |
| 10 | +import com.typesafe.config.{Config, ConfigFactory} |
| 11 | +import java.time.LocalDate |
| 12 | +import scala.jdk.CollectionConverters.* |
| 13 | + |
| 14 | +case class ScheduleItem( |
| 15 | + time: String, |
| 16 | + title: String, |
| 17 | + speakers: Option[List[String]], |
| 18 | + summary: Option[String] |
| 19 | +) derives YamlCodec |
| 20 | + |
| 21 | +case class Sponsor( |
| 22 | + name: String, |
| 23 | + logo: String, |
| 24 | + link: String, |
| 25 | + `type`: String, |
| 26 | + height: Option[Int] |
| 27 | +) derives YamlCodec |
| 28 | + |
| 29 | +case class Meta(meetup: Option[String]) derives YamlCodec |
| 30 | + |
| 31 | +case class EventConfig( |
| 32 | + title: String, |
| 33 | + short_title: Option[String], |
| 34 | + date_string: String, |
| 35 | + location: String, |
| 36 | + description: String, |
| 37 | + poster_hero: Option[String], |
| 38 | + poster_thumb: Option[String], |
| 39 | + schedule: Option[List[ScheduleItem]], |
| 40 | + sponsors: Option[List[Sponsor]], |
| 41 | + meta: Option[Meta] |
| 42 | +) derives YamlCodec |
| 43 | + |
| 44 | +case class Event(conf: EventConfig, content: String, originalYaml: String) { |
| 45 | + |
| 46 | + def loadSpeakerDirectory(): Map[String, String] = |
| 47 | + try { |
| 48 | + val config = ConfigFactory.parseFile( |
| 49 | + Path("src/blog/directory.conf").toNioPath.toFile |
| 50 | + ) |
| 51 | + val speakerNames = config |
| 52 | + .root() |
| 53 | + .keySet() |
| 54 | + .asScala |
| 55 | + .toList |
| 56 | + .map { key => |
| 57 | + key -> config.getConfig(key).getString("name") |
| 58 | + } |
| 59 | + .toMap |
| 60 | + speakerNames |
| 61 | + } catch { |
| 62 | + case _: Exception => Map.empty[String, String] |
| 63 | + } |
| 64 | + |
| 65 | + def cleanOtherLinks(markdown: String): String = { |
| 66 | + var cleaned = markdown |
| 67 | + |
| 68 | + // https://typelevel.org/blog/YYYY/MM/DD/post-name.html -> post-name.md |
| 69 | + val typelevelBlogPattern = |
| 70 | + """https://typelevel\.org/blog/\d{4}/\d{2}/\d{2}/([^)\s]+)\.html""".r |
| 71 | + cleaned = typelevelBlogPattern.replaceAllIn(cleaned, "$1.md") |
| 72 | + |
| 73 | + // /blog/YYYY/MM/DD/post-name.html -> post-name.md |
| 74 | + val relativeBlogPattern = |
| 75 | + """(?<![a-z])/blog/\d{4}/\d{2}/\d{2}/([^)\s]+)\.html""".r |
| 76 | + cleaned = relativeBlogPattern.replaceAllIn(cleaned, "$1.md") |
| 77 | + |
| 78 | + val siteUrlPattern = """\{\{\s*site\.url\s*\}\}""".r |
| 79 | + cleaned = siteUrlPattern.replaceAllIn(cleaned, "") |
| 80 | + |
| 81 | + // Replace {{ page.meta.meetup }} with just the URL |
| 82 | + val meetupPattern = """\{\{\s*page\.meta\.meetup\s*\}\}""".r |
| 83 | + cleaned = conf.meta.flatMap(_.meetup) match { |
| 84 | + case Some(meetupUrl) => meetupPattern.replaceAllIn(cleaned, meetupUrl) |
| 85 | + case None => meetupPattern.replaceAllIn(cleaned, "") |
| 86 | + } |
| 87 | + |
| 88 | + // Replace .html extensions with .md in relative links (but not absolute URLs starting with http) |
| 89 | + val htmlToMdPattern = """(?<!https?://[^\s)]*)(\\.html)""".r |
| 90 | + cleaned = htmlToMdPattern.replaceAllIn(cleaned, ".md") |
| 91 | + |
| 92 | + cleaned = cleaned.replace("/projects", "/projects/README.md") |
| 93 | + cleaned = cleaned.replace("/conduct.html", "/code-of-conduct/README.md") |
| 94 | + cleaned = |
| 95 | + cleaned.replace("/code-of-conduct.html", "/code-of-conduct/README.md") |
| 96 | + |
| 97 | + cleaned |
| 98 | + } |
| 99 | + |
| 100 | + def buildHoconMetadata(date: String, eventDate: String, eventLocation: String, tags: List[String]): String = |
| 101 | + s"""|{% |
| 102 | + | laika.html.template: event.template.html |
| 103 | + | date: "$date" |
| 104 | + | event-date: "$eventDate" |
| 105 | + | event-location: "$eventLocation" |
| 106 | + | tags: ${tags.mkString("[", ", ", "]")} |
| 107 | + |%}""".stripMargin |
| 108 | + |
| 109 | + def generateScheduleMarkdown(): String = conf.schedule |
| 110 | + .map { scheduleItems => |
| 111 | + val tableRows = scheduleItems |
| 112 | + .map { item => |
| 113 | + val timeColumn = item.time |
| 114 | + |
| 115 | + val talkColumn = if (item.speakers.isEmpty) { |
| 116 | + item.title |
| 117 | + } else { |
| 118 | + item.speakers |
| 119 | + .map { speakers => |
| 120 | + val speakerDirectory = loadSpeakerDirectory() |
| 121 | + val speakerNames = speakers |
| 122 | + .map(s => speakerDirectory.getOrElse(s, s)) |
| 123 | + .mkString(", ") |
| 124 | + item.summary match |
| 125 | + case Some(value) => |
| 126 | + s"@:style(schedule-title)${item.title}@:@ @:style(schedule-byline)${speakerNames}@:@ ${value}" |
| 127 | + case None => s"**${item.title}**<br/>${speakerNames}" |
| 128 | + } |
| 129 | + .getOrElse(item.title) |
| 130 | + } |
| 131 | + |
| 132 | + s"| ${timeColumn} | ${talkColumn} |" |
| 133 | + } |
| 134 | + .mkString("\n") |
| 135 | + |
| 136 | + s"""|| Time | Talk | |
| 137 | + ||------|------| |
| 138 | + |$tableRows""".stripMargin |
| 139 | + } |
| 140 | + .getOrElse("") |
| 141 | + |
| 142 | + def generateSponsorsHtml(): String = { |
| 143 | + conf.sponsors |
| 144 | + .map { sponsors => |
| 145 | + val sponsorsByType = sponsors.groupBy(_.`type`) |
| 146 | + |
| 147 | + val sections = List("platinum", "gold", "silver").flatMap { |
| 148 | + sponsorType => |
| 149 | + sponsorsByType.get(sponsorType).map { typeSponsors => |
| 150 | + val sponsorCells = typeSponsors |
| 151 | + .map { sponsor => |
| 152 | + s"@:style(bulma-cell bulma-has-text-centered)[@:image(${sponsor.logo}) { alt: ${sponsor.name}, title: ${sponsor.name}, style: legacy-event-sponsor }](${sponsor.link})@:@" |
| 153 | + } |
| 154 | + .mkString("\n") |
| 155 | + |
| 156 | + s"""|### ${sponsorType.capitalize} |
| 157 | + |@:style(bulma-grid bulma-is-col-min-12) |
| 158 | + |$sponsorCells |
| 159 | + |@:@""".stripMargin |
| 160 | + } |
| 161 | + } |
| 162 | + |
| 163 | + sections.mkString("\n\n") |
| 164 | + } |
| 165 | + .getOrElse("") |
| 166 | + } |
| 167 | + |
| 168 | + def toLaika(date: String, stage: Int): String = { |
| 169 | + val tags = Option.when(conf.title.contains("Summit"))("summits").toList ::: "events" :: Nil |
| 170 | + val metadata = buildHoconMetadata(date, conf.date_string, conf.location, tags) |
| 171 | + val title = s"# ${conf.title}" |
| 172 | + val image = |
| 173 | + conf.poster_hero.map(img => s"").getOrElse("") |
| 174 | + |
| 175 | + stage match { |
| 176 | + case 1 => |
| 177 | + // Stage 1: Just move to new location, keep original format |
| 178 | + s"---\n$originalYaml---\n\n$content\n" |
| 179 | + |
| 180 | + case 2 => |
| 181 | + // Stage 2: HOCON metadata + title, no content changes |
| 182 | + s"$metadata\n\n$title\n\n$image\n\n$content\n" |
| 183 | + |
| 184 | + case 3 => |
| 185 | + // Stage 3: Stage 2 + link cleaning |
| 186 | + val transformedContent = cleanOtherLinks(content) |
| 187 | + s"$metadata\n\n$title\n\n$image\n\n$transformedContent\n" |
| 188 | + |
| 189 | + case _ => |
| 190 | + // Stage 4+: Use original content and replace Jekyll includes with generated HTML |
| 191 | + val transformedContent = cleanOtherLinks(content) |
| 192 | + |
| 193 | + // Replace Jekyll includes with generated HTML |
| 194 | + var processedContent = transformedContent |
| 195 | + |
| 196 | + // Remove schedule assign and replace schedule include |
| 197 | + val scheduleAssignPattern = |
| 198 | + """\{\%\s*assign\s+schedule\s*=\s*page\.schedule\s*%\}\s*""".r |
| 199 | + processedContent = |
| 200 | + scheduleAssignPattern.replaceAllIn(processedContent, "") |
| 201 | + |
| 202 | + val schedulePattern = """\{\%\s*include\s+schedule\.html\s*%\}""".r |
| 203 | + val scheduleReplacement = |
| 204 | + if (conf.schedule.isDefined) generateScheduleMarkdown() else "" |
| 205 | + processedContent = |
| 206 | + schedulePattern.replaceAllIn(processedContent, scheduleReplacement) |
| 207 | + |
| 208 | + // Replace sponsors include |
| 209 | + val sponsorsPattern = """\{\%\s*include\s+sponsors\.html\s*%\}""".r |
| 210 | + val sponsorsReplacement = |
| 211 | + if (conf.sponsors.isDefined) generateSponsorsHtml() else "" |
| 212 | + processedContent = |
| 213 | + sponsorsPattern.replaceAllIn(processedContent, sponsorsReplacement) |
| 214 | + |
| 215 | + // Remove venue_map includes (not supported) |
| 216 | + val venueMapPattern = """\{\%\s*include\s+venue_map\.html\s*%\}""".r |
| 217 | + processedContent = venueMapPattern.replaceAllIn(processedContent, "") |
| 218 | + |
| 219 | + s"$metadata\n\n$title\n\n$image\n\n$processedContent\n" |
| 220 | + } |
| 221 | + } |
| 222 | +} |
| 223 | + |
| 224 | +object EventParser { |
| 225 | + def parse(path: Path, content: String): Either[Throwable, Event] = { |
| 226 | + // Normalize Windows line endings to Unix |
| 227 | + val normalized = content.replace("\r\n", "\n") |
| 228 | + val parts = normalized.split("---\n", 3) |
| 229 | + if (parts.length < 3) { |
| 230 | + val fn = path.fileName |
| 231 | + Left(new Exception(s"Invalid event '$fn': no YAML front matter found")) |
| 232 | + } else { |
| 233 | + val yamlContent = parts(1) |
| 234 | + val markdownContent = parts(2).trim |
| 235 | + yamlContent |
| 236 | + .as[EventConfig] |
| 237 | + .map(conf => Event(conf, markdownContent, yamlContent)) |
| 238 | + } |
| 239 | + } |
| 240 | +} |
| 241 | + |
| 242 | +object MigrateEvents extends IOApp { |
| 243 | + val oldEventsDir = Path("collections/_events") |
| 244 | + val newBlogDir = Path("src/blog") |
| 245 | + |
| 246 | + def getDateAndName(path: Path): Either[Throwable, (String, String)] = { |
| 247 | + val filename = path.fileName.toString |
| 248 | + val datePattern = """(\d{4}-\d{2}-\d{2})-(.+)""".r |
| 249 | + filename match { |
| 250 | + case datePattern(date, rest) => |
| 251 | + Right((date, filename)) // Keep full filename |
| 252 | + case _ => |
| 253 | + Left(new Exception(s"Filename doesn't match pattern: $filename")) |
| 254 | + } |
| 255 | + } |
| 256 | + |
| 257 | + def readEvent(path: Path): IO[String] = Files[IO] |
| 258 | + .readAll(path) |
| 259 | + .through(fs2.text.utf8.decode) |
| 260 | + .compile |
| 261 | + .string |
| 262 | + |
| 263 | + def writeEvent(path: Path, content: String): IO[Unit] = fs2.Stream |
| 264 | + .emit(content) |
| 265 | + .through(fs2.text.utf8.encode) |
| 266 | + .through(Files[IO].writeAll(path)) |
| 267 | + .compile |
| 268 | + .drain |
| 269 | + |
| 270 | + def migrateEvent(sourcePath: Path, stage: Int): IO[String] = for { |
| 271 | + (date, fullFilename) <- IO.fromEither(getDateAndName(sourcePath)) |
| 272 | + content <- readEvent(sourcePath) |
| 273 | + event <- IO.fromEither(EventParser.parse(sourcePath, content)) |
| 274 | + laikaContent = event.toLaika(date, stage) |
| 275 | + destPath = newBlogDir / fullFilename |
| 276 | + _ <- writeEvent(destPath, laikaContent) |
| 277 | + } yield fullFilename |
| 278 | + |
| 279 | + def migrateAllEvents(stage: Int): IO[Long] = Files[IO] |
| 280 | + .list(oldEventsDir) |
| 281 | + .filter(_.fileName.toString.matches("""^\d{4}-\d{2}-\d{2}-.+\.md$""")) |
| 282 | + .evalMap(path => migrateEvent(path, stage)) |
| 283 | + .evalMap(fullFilename => IO.println(s"Migrated: $fullFilename")) |
| 284 | + .compile |
| 285 | + .count |
| 286 | + |
| 287 | + def run(args: List[String]): IO[cats.effect.ExitCode] = { |
| 288 | + val stage = args.headOption.flatMap(_.toIntOption).getOrElse(4) |
| 289 | + IO.println(s"Running migration with stage $stage") *> |
| 290 | + migrateAllEvents(stage) |
| 291 | + .flatMap(c => IO.println(s"Migrated $c events")) |
| 292 | + .as(cats.effect.ExitCode.Success) |
| 293 | + } |
| 294 | +} |
0 commit comments