refactor: improve memory footprint of zipline-parser code

I miss the ART GC already...

Author: msfjarvis

zipline-parser/src/commonMain/kotlin/dev/msfjarvis/claw/parser/internal/CommentParsers.kt

@@ -31,11 +31,13 @@ private fun MutableList<Comment>.addSubtree(
   val comment = commentElement.toComment(parentComment)
   add(comment)
   val childContainer = if (subtree.`is`("div.comment")) subtree.parent() ?: subtree else subtree
-  childContainer
-    .children()
-    .filter { it.`is`("ol.comments") }
-    .flatMap { comments -> comments.children().filter { it.`is`("li.comments_subtree") } }
-    .forEach { child -> addSubtree(child, parentComment = comment.shortId, seen) }
+  for (childList in childContainer.children()) {
+    if (!childList.`is`("ol.comments")) continue
+    for (child in childList.children()) {
+      if (!child.`is`("li.comments_subtree")) continue
+      addSubtree(child, parentComment = comment.shortId, seen)
+    }
+  }
 }
 
 private fun Element.toComment(parentComment: String?): Comment {

zipline-parser/src/commonMain/kotlin/dev/msfjarvis/claw/parser/internal/PostDetailsParsers.kt

@@ -10,40 +10,30 @@ import com.fleeksoft.ksoup.Ksoup
 import dev.msfjarvis.claw.model.LobstersPostDetails
 
 private const val BASE_URL = "https://lobste.rs"
+private val commentCountRegex = "\\d+".toRegex()
+private const val STORY_SELECTOR = "ol.stories > li.story"
+private const val SUBMITTER_SELECTOR =
+  "ol.stories > li.story div.byline > a[href^=/~]:not([tabindex]):not([aria-hidden=true])"
 
 internal fun parsePostDetails(html: String): LobstersPostDetails {
   val document = Ksoup.parse(html, baseUri = BASE_URL)
+  val storyElement = document.select(STORY_SELECTOR)
+  val timestampElement = document.select("$STORY_SELECTOR div.byline > time")
+  val titleElement = document.select("$STORY_SELECTOR span.link.h-cite > a")
+  val commentsElement = document.select("$STORY_SELECTOR span.comments_label a")
+  val submitterElement = document.select(SUBMITTER_SELECTOR)
+  val tags = document.select("$STORY_SELECTOR span.tags > a").map { it.text() }
   return LobstersPostDetails(
-    shortId = document.select("ol.stories > li.story").attr("data-shortid"),
-    createdAt =
-      normalizeCreatedAt(
-        document.select("ol.stories > li.story div.byline > time").attr("data-at-unix")
-      ),
-    title = document.select("ol.stories > li.story span.link.h-cite > a").text(),
-    url = document.select("ol.stories > li.story span.link.h-cite > a").attr("abs:href"),
+    shortId = storyElement.attr("data-shortid"),
+    createdAt = normalizeCreatedAt(timestampElement.attr("data-at-unix")),
+    title = titleElement.text(),
+    url = titleElement.attr("abs:href"),
     description = document.select("div.story_content div.story_text").html(),
-    commentCount =
-      "\\d+"
-        .toRegex()
-        .find(document.select("ol.stories > li.story span.comments_label a").text())
-        ?.value
-        ?.toInt() ?: 0,
-    commentsUrl = document.select("ol.stories > li.story span.comments_label a").attr("abs:href"),
-    submitter =
-      document
-        .select(
-          "ol.stories > li.story div.byline > a[href^=/~]:not([tabindex]):not([aria-hidden=true])"
-        )
-        .text(),
-    tags = document.select("ol.stories > li.story span.tags > a").map { it.text() },
+    commentCount = commentCountRegex.find(commentsElement.text())?.value?.toInt() ?: 0,
+    commentsUrl = commentsElement.attr("abs:href"),
+    submitter = submitterElement.text(),
+    tags = tags,
     comments = parseComments(document),
-    userIsAuthor =
-      document
-        .select(
-          "ol.stories > li.story div.byline > a[href^=/~]:not([tabindex]):not([aria-hidden=true])"
-        )
-        .attr("class")
-        .split(' ')
-        .contains("user_is_author"),
+    userIsAuthor = submitterElement.attr("class").split(' ').contains("user_is_author"),
   )
 }

zipline-parser/src/commonMain/kotlin/dev/msfjarvis/claw/parser/internal/PostParsers.kt

@@ -11,6 +11,7 @@ import com.fleeksoft.ksoup.nodes.Element
 import dev.msfjarvis.claw.model.LobstersPost
 
 private const val BASE_URL = "https://lobste.rs"
+private val commentCountRegex = "\\d+".toRegex()
 
 internal fun parsePostsPage(html: String): List<LobstersPost> {
   return Ksoup.parse(html, baseUri = BASE_URL).select("li.story").map(::parsePost)
@@ -24,19 +25,20 @@ private fun parsePost(element: Element): LobstersPost {
       "> div.story_liner div.byline > a[href^=/~]:not([tabindex]):not([aria-hidden=true])"
     )
 
+  val timestampElement = element.select("> div.story_liner div.byline > time")
+  val descriptionElement = element.select("> div.story_liner a.description_present")
+  val tags = element.select("> div.story_liner span.tags > a").map(Element::text)
+
   return LobstersPost(
     shortId = element.attr("data-shortid"),
-    createdAt =
-      normalizeCreatedAt(
-        element.select("> div.story_liner div.byline > time").attr("data-at-unix")
-      ),
+    createdAt = normalizeCreatedAt(timestampElement.attr("data-at-unix")),
     title = titleElement.text(),
     url = titleElement.attr("abs:href"),
-    description = element.select("> div.story_liner a.description_present").attr("title"),
-    commentCount = "\\d+".toRegex().find(commentElement.text())?.value?.toInt() ?: 0,
+    description = descriptionElement.attr("title"),
+    commentCount = commentCountRegex.find(commentElement.text())?.value?.toInt() ?: 0,
     commentsUrl = commentElement.attr("abs:href"),
     submitter = submitterElement.text(),
     userIsAuthor = submitterElement.attr("class").split(' ').contains("user_is_author"),
-    tags = element.select("> div.story_liner span.tags > a").map(Element::text),
+    tags = tags,
   )
 }

zipline-parser/src/commonMain/kotlin/dev/msfjarvis/claw/parser/internal/TagParsers.kt

@@ -17,18 +17,15 @@ internal fun parseTagsPage(html: String): List<Tag> {
 }
 
 private fun parseTag(element: Element): Tag {
+  val tagElement = element.select("> a.tag")
+  val descriptionElement = element.select("> span:not(.byline)")
   return Tag(
-    tag = element.select("> a.tag").text(),
-    description = element.select("> span:not(.byline)").text(),
+    tag = tagElement.text(),
+    description = descriptionElement.text(),
     privileged = element.attr("data-privileged").toBoolean(),
-    active =
-      !element
-        .select("> span:not(.byline)")
-        .attr("class")
-        .split(Regex("\\s+"))
-        .contains("inactive_tag"),
+    active = !descriptionElement.hasClass("inactive_tag"),
     category = element.attr("data-category"),
-    isMedia = element.select("> a.tag").attr("class").split(Regex("\\s+")).contains("tag_is_media"),
+    isMedia = tagElement.hasClass("tag_is_media"),
     hotnessMod = element.attr("data-hotness-mod").toDoubleOrNull() ?: 0.0,
   )
 }