From ef76200afac29b34962d125835bf7b99c60dd22b Mon Sep 17 00:00:00 2001
From: Daniel Shapiro <d.shapiro@berkeley.edu>
Date: Tue, 23 Apr 2019 21:32:06 -0700
Subject: [PATCH] 1.0.3

improvements to display of output
---
 build.sbt                                     |  2 +-
 src/main/scala/cladograms/Cladogram.scala     |  2 +-
 .../scala/cladograms/EnWikipediaClade.scala   | 21 +++++++++++--------
 3 files changed, 14 insertions(+), 11 deletions(-)

diff --git a/build.sbt b/build.sbt
index 29bd8da..b4ae614 100644
--- a/build.sbt
+++ b/build.sbt
@@ -1,6 +1,6 @@
 name := "wikiClade"
 
-version := "1.0.2"
+version := "1.0.3"
 
 scalaVersion := "2.12.8"
 
diff --git a/src/main/scala/cladograms/Cladogram.scala b/src/main/scala/cladograms/Cladogram.scala
index 14add62..60978eb 100644
--- a/src/main/scala/cladograms/Cladogram.scala
+++ b/src/main/scala/cladograms/Cladogram.scala
@@ -34,7 +34,7 @@ class Cladogram (val clade: Clade, var children: Set[Cladogram]) {
   def prominentDescendants(verbosity: Int): Set[Cladogram] = for {
     child <- children
     descendant =
-      if (child.children.size == 1 && !child.children.head.clade.shouldDisplay(verbosity))
+      if (child.children.size == 1 && !child.clade.shouldDisplay(verbosity))
         child.prominentDescendants(verbosity).head        //guaranteed to have exactly one element
       else child
   } yield descendant
diff --git a/src/main/scala/cladograms/EnWikipediaClade.scala b/src/main/scala/cladograms/EnWikipediaClade.scala
index 4df3f08..7c13361 100644
--- a/src/main/scala/cladograms/EnWikipediaClade.scala
+++ b/src/main/scala/cladograms/EnWikipediaClade.scala
@@ -5,22 +5,25 @@ import org.jsoup.nodes.{Document, Element}
 import org.jsoup.select.Elements
 
 import scala.util.{Failure, Success, Try}
-
 /**
   * Created by Daniel on 4/8/2019.
   */
 case class EnWikipediaClade(val name: String, val path: Option[String], val priorityOverride: Double = 100) extends Clade {
   val baseUrl = "https://en.wikipedia.org"
+  val ignorableCladeTypes = Set("Clade", "(unranked)")
+  val importantCladeTypes = Set("Kingdom", "Phylum", "Class", "Order", "Family", "Genus", "Species")
 
   lazy val meta: WikiCladeMetadata = getMeta
 
   def ancestors: List[Clade] = meta.ancestors
-  def priority: Double = Math.min(priorityOverride, meta.docPriority)
+  def priority: Double = Math.min(
+    Math.min(priorityOverride, meta.docPriority),
+    if (importantCladeTypes contains meta.cladeType) 20 else 100)
 
   override def shouldDisplay(verbosity: Int): Boolean = priority <= verbosity
 
   override def DOTDefinition: Option[String] = {
-    val cladeTypeStr = if (meta.cladeType.isEmpty) "" else meta.cladeType + " <br/>"
+    val cladeTypeStr = if (meta.cladeType.isEmpty) "" else s"""<FONT POINT-SIZE=\"10\">${meta.cladeType}</FONT><br/>"""
     Some(s""""$name" [label=<$cladeTypeStr<B>$name</B>>]""")
   }
 
@@ -37,7 +40,7 @@ case class EnWikipediaClade(val name: String, val path: Option[String], val prio
       else new EnWikipediaClade(details.name, Some(details.path))
     }
     val docPriority = priorityBasedOnDoc(docOpt)
-    WikiCladeMetadata(ancestors, cladeType, docPriority)
+    WikiCladeMetadata(ancestors, sanitizeCladeType(cladeType), docPriority)
   }
 
   private def getDoc: Option[Document] = path match {
@@ -92,15 +95,10 @@ case class EnWikipediaClade(val name: String, val path: Option[String], val prio
             val details = parseRow(row)
             if (details.path.nonEmpty) {
               val pagetry = Try(Jsoup.connect(baseUrl + details.path).get().select("title").text())
-//              val pagetry = doctry match {
-//                case Success(doc) => Try(doc.select("title").text())
-//                case Failure(e) => Failure(e)
-//              }
               pagetry match {
                 case Success(page) => if (knownPages contains page) {
                   iter(i + 1, started, knownPages, TaxonDetails(details.name, details.cladeType, "") :: taxList)
                 } else {
-                  //val pri = priorityBasedOnDoc(doctry.get)
                   iter(i + 1, started, knownPages + page, details :: taxList)
                 }
                 case Failure(_) =>
@@ -116,6 +114,11 @@ case class EnWikipediaClade(val name: String, val path: Option[String], val prio
     iter(0, false, Set(), List())
   }
 
+  private def sanitizeCladeType(cladeType: String): String = {
+    val cleaned = cladeType.replaceAll(":", "").trim
+    if (ignorableCladeTypes contains cleaned) "" else cleaned
+  }
+
   def priorityBasedOnDoc(docOpt: Option[Document]): Double = docOpt match {
     case Some(doc) => Math.min (99, Math.max (1, 100 - (15 * (Math.log (doc.text ().length) - 7) ) ) )
     case None => 99