diff options
Diffstat (limited to 'scalding')
19 files changed, 1060 insertions, 117 deletions
diff --git a/scalding/build.sbt b/scalding/build.sbt index d477399..01f55ca 100644 --- a/scalding/build.sbt +++ b/scalding/build.sbt @@ -42,7 +42,7 @@ lazy val root = (project in file(".")). libraryDependencies += "org.apache.hadoop" % "hadoop-client" % hadoopVersion, libraryDependencies += "org.apache.hadoop" % "hadoop-mapreduce-client-jobclient" % hadoopVersion classifier "tests", libraryDependencies += "org.apache.hbase" % "hbase-common" % hbaseVersion, - libraryDependencies += "parallelai" % "parallelai.spyglass" % "2.11_0.17.2_cdh5.3.1", + libraryDependencies += "parallelai" % "parallelai.spyglass" % "2.11_0.17.2_cdh5.3.1-p1", // cargo-culted from twitter/scalding's build.sbt // hint via https://stackoverflow.com/questions/23280494/sbt-assembly-error-deduplicate-different-file-contents-found-in-the-following#23280952 diff --git a/scalding/src/main/resources/slug-blacklist.txt b/scalding/src/main/resources/slug-blacklist.txt index 7dc701f..ad3dc1d 100644 --- a/scalding/src/main/resources/slug-blacklist.txt +++ b/scalding/src/main/resources/slug-blacklist.txt @@ -1,34 +1,458 @@ abbreviations +abbreviationsandacronyms +aboutauthors +abouttheauthor +abouttheauthors +aboutthecover +abouttheeditors +abreviations abstract +abstractnotsubmittedforonlinepublication +abstracts +abstractsofaapaposterandpodiumpresentations +abstractsofcommunications +abstractsofthesesfromthescandinaviancountries +abstractwithdrawn +acknowledgement acknowledgements +acknowledgementsvii +acknowledgementtoreferees +acknowledgment +acknowledgmentofreferees +acknowledgments +addendum +additionalresources +address +advertisersindex +affect +affiliation +agenda +agradecimientos +aimsandscope +annexa +annualacknowledgementofmanuscriptreviewers +appendices +appendix1 +appendixa +appendixb +appointmentsandstaffchanges +approximation +apresentacao article +articles +articlesofsignificantinterestselectedfromthisissuebytheeditors +associationnews +ataglance +atribute +authorguidelines +authorindex +authorindexforvolume81 authorreply +authors authorsreply +authorsresponse +avantpropos +award +awardsappointmentsannouncements +backcover +background +bibliografia +bibliography +bigdata +blankpage +blood +boardoftrustees +bookofabstracts bookreview bookreviews +bookreviewsandnotices +bookreviewssection +booksreceived +buchbesprechungen +bulletin +calendar +calendarofevents +calendarofmeetings +callforarticles +callforpapers casereport +casereports +chairmansopeningremarks +changes +chaos +chapter1 +chapter10 +chapter1introduction +chapter7 +chapteri +chapterone +chapteroneintroduction +chaptertwo +citation +classes +classified +classifieds +collaborateurs +comment +commentaries commentary commentaryon commenton +comments commentto +committee +communication +communications +communicationstotheeditor +communiquedepresse +community +components +comptesrendus +computerscience +conclusion +conclusions +conferencereport +congratulations +congresscalendar +conservation +content contents +context +continuingeducation +continuingmedicaleducation +contributors +copyright +copyrightform +copyrightnotice +correction +corrections correspondence +corrigenda +corrigendum +councilminutes +cover +coverimage +currentresearch +curriculumvitae +danksagung +dearreaders +decisionmaking dedication +dedicatoria +definition +description +discussion +distribution +documents +ear +editorial editorialadvisoryboard +editorialboard +editorialcomment +editorialcomments +editorialconsultants +editoriale +editorialeditorial +editorialinformation +editorialintroduction +editorialnote +editorials +editorialsoftwaresurveysection +editorialstaff +editorialstatement +editorinchief +editors +editorschoice +editorscomment +editorscorner +editorscorrespondence +editorsintroduction +editorsnote +editorspicks +editorspreface +education +einfuhrung +einleitung +electrophoresis +employment +endnotes +entrevista +entscheidungsverzeichnis +epilogue +equipment +errata +erratum +essay +essays +executivesummary +exercises +extendedabstracts +feature +features +fichatecnica +figure3 +finalexam +finalreport focus +foreward +foreword +forthcomingarticles +forthcomingevents +fortherecord +forum +frequentlyaskedquestions +fromtheeditor +fromtheeditors +fromthepresident +frontmatter +furtherreadings +genealogy +generaldiscussion +generalinformation +generalintroduction +germany +glosario +glossary +glossaryofterms +guesteditorial +guideforauthors +guidelinesforcontributors +health +heartfailure +highlights +highlightsfromthisissue +history +home +homework hypothesis +iii +importantnotice +impressum inbrief +index +indexofauthors +indexofauthorsandtitles +indice +indicegeneral +informationforauthors +informationtoauthors +inhalt +inhaltsverzeichnis +inmemoriam +inreply +insidethisissue +institutenews +instructionsforauthors +instructionstoauthors +interview +inthestudy +inthisissue +introduccion introduction +introductiongenerale introductiontotheissue +introductorycomments +inventions +invitedcommentary +issuesandevents +jobdescription +keywords +languageteaching +lecture +letter +letterfromtheeditor +letters +letterstotheeditor lettertotheeditor +lettertotheeditors +linearalgebra +linearregression +links +listedestableaux listofabbreviations +listofcontributors +listoffigures +listofparticipants +listofpublications +listofreferees +listofreviewers +listoftables +literacy +literature +literaturecited +literaturrundschau +litteraturverzeichniss +livresrecus +lucina +lungcancer +magazin +maintenance +materials +materialsafetydatasheet +materialsandmethods +medicinalchemistry +meetingabstracts +meetings +meetingsandconferences +meetingsofinterest +membershipapplication +memoranda +memorandum +messagefromgeneralcochairs +messagefromthechairs +messagefromtheprogramchairs +messagefromtheprogramcochairs +metaanalysis +missionstatement +motivation +mrsnews +name +newbooks +newlyelectedmembersofthecollege +newproducts +news +newsandnotes +newsandreviews +newsandviews +newsviews +noii note +notes +notesandcomments +notesandnews +notesforcontributors +notesoncontributors +notice +noticeboard +notitle +notitleavailable +obituaries +obituary +online +openaccess +oralabstracts +oralpresentations +organizingcommittee +originalarticle +originalarticles +other +outline overview +panorama +papers +paperstoappearinforthcomingissues +partone +personalandmiscellaneous +perspective +perspectives +place +positionsavailable +poster +posterpresentations +postscript preface +preliminarymaterial +presentacio +presentacion +presentation +pressrelease +print +printing +proceedings +profile +programcommittee +projectmanagement +publication +publichealth +publishersnote +question +radiology +readersforum +recensions +recentpublications +redaktorensforord references +referenciasbibliograficas +regression +rehabilitation +rejoinder +remerciements +reply +replybyauthors +researchresearchers +resenas +resources +response +responsetothelettertotheeditor results +resume +resumen +resumes +resumo review reviewarticle +revieweracknowledgement +revieweracknowledgement2013 +reviewers +reviewessay +reviews +reviewsanddescriptionsoftablesandbooks +rezension +safety +section +security +selectedbibliography +shortcommunication +shorternotices +socialengineering +sommaire +sommario +specialsection +specifications +subjectindex +subscriptions +suggestedreadings +sumario +summaries +summariesofkeyjournalarticles summary +summaryofproceedings +summer +sun +supplementarymaterial +symposium +symptom +synthese +tabledesmatieres +tableofcontents +tableofcontentsandprologue +technicalreport +theauthors +thebasics +theeditorsdesk +thefirstauthorreplies +thelancet +theoreticalbackground +thetimes +theworldbank +theyearinreview +thismonthin +timemanagement +titeleiinhaltsverzeichnis title -name +titlepage +titlepagei +tocorrespondents +totheeditor +unitedkingdom +unitednations +unitedstates +upcomingevents +vorwort +website +welcome +whatshappening +whatsnew +workscited +yourquestionsanswered +zusammenfassung diff --git a/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala b/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala index 5d1eaf5..039fa85 100644 --- a/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala +++ b/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala @@ -2,6 +2,7 @@ package sandcrawler import scala.math import scala.util.parsing.json.JSON +import scala.util.parsing.json.JSONArray import scala.util.parsing.json.JSONObject import cascading.flow.FlowDef @@ -19,29 +20,100 @@ class CrossrefScorable extends Scorable with HBasePipeConversions { def getFeaturesPipe(args : Args)(implicit mode : Mode, flowDef : FlowDef) : TypedPipe[MapFeatures] = { getSource(args).read .toTypedPipe[String](new Fields("line")) + .filter { CrossrefScorable.keepRecord(_) } .map { CrossrefScorable.jsonToMapFeatures(_) } } } object CrossrefScorable { + + val ContentTypeWhitelist: Set[String] = Set( + "book", + "book-chapter", + "dataset", + "dissertation", + "journal-article", + "letter", + "monograph", + "posted-content", + "pre-print", + "proceedings-article", + "report", + "working-paper") + + def keepRecord(json : String) : Boolean = { + Scorable.jsonToMap(json) match { + case None => false + case Some(map) => { + mapToTitle(map) match { + case None => false + case Some(title) => title.length <= Scorable.MaxTitleLength + } + } + } + } + + // Returns None if title is null, empty, or too long. + def mapToTitle(map : Map[String, Any]) : Option[String] = { + if (map contains "title") { + val titles = map("title").asInstanceOf[List[String]] + if (titles.isEmpty || titles == null) { + None + } else { + val title = titles(0) + if (title == null || title.isEmpty || title.length > Scorable.MaxTitleLength) None else Some(title) + } + } else { + None + } + } + + def mapToAuthorList(map : Map[String, Any]) : List[String] = { + if (map contains "author") { + val objArray = map("author").asInstanceOf[List[Any]].map(e => e.asInstanceOf[Map[String,Any]]) + // TODO(bnewbold): combine given and family names? + objArray + .filter(e => e contains "family") + .map(e => e.get("family").get.asInstanceOf[String]) + } else { + List() + } + } + + def mapToYear(map : Map[String, Any]) : Option[Int] = { + map.get("created") match { + case None => None + case Some(created) => { + Some(created.asInstanceOf[Map[String,Any]] + .get("date-parts") + .get + .asInstanceOf[List[Any]](0) + .asInstanceOf[List[Any]](0) + .asInstanceOf[Double] + .toInt) + } + } + } + def jsonToMapFeatures(json : String) : MapFeatures = { Scorable.jsonToMap(json) match { case None => MapFeatures(Scorable.NoSlug, json) - case Some(map) => { - if ((map contains "title") && (map contains "DOI")) { - val titles = map("title").asInstanceOf[List[String]] - val doi = Scorable.getString(map, "DOI") - if (titles.isEmpty || titles == null || doi.isEmpty || doi == null) { - new MapFeatures(Scorable.NoSlug, json) - } else { - // bnewbold: not checking that titles(0) is non-null/non-empty; case would be, in JSON, "title": [ null ] - val sf : ScorableFeatures = ScorableFeatures.create(title=titles(0), doi=doi) - new MapFeatures(sf.toSlug, sf.toString) + case Some(map) => + mapToTitle(map) match { + case None => MapFeatures(Scorable.NoSlug, json) + case Some(title) => { + val doi = Scorable.getString(map, "DOI") + val authors: List[String] = mapToAuthorList(map) + val year: Int = mapToYear(map).getOrElse(0) + val contentType: String = map.get("type").map(e => e.asInstanceOf[String]).getOrElse("MISSING-CONTENT-TYPE") + if (doi.isEmpty || doi == null || authors.length == 0 || !(ContentTypeWhitelist contains contentType)) { + MapFeatures(Scorable.NoSlug, json) + } else { + val sf : ScorableFeatures = ScorableFeatures.create(title=title, authors=authors, doi=doi.toLowerCase(), year=year) + MapFeatures(sf.toSlug, sf.toString) + } } - } else { - new MapFeatures(Scorable.NoSlug, json) } - } } } } diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala index d7a1eea..c55cb40 100644 --- a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala +++ b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala @@ -24,20 +24,46 @@ class GrobidScorable extends Scorable with HBasePipeConversions { getSource(args) .read // Can't just "fromBytesWritable" because we have multiple types? - .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "tei_json", "status_code")) - .filter { case (_, tei_json, status_code) => tei_json != null && status_code != null } - .map { case (key, tei_json, status_code) => - (Bytes.toString(key.copyBytes()), Bytes.toString(tei_json.copyBytes()), Bytes.toLong(status_code.copyBytes())) + .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "metadata", "status_code")) + .filter { case (_, metadata, status_code) => metadata != null && status_code != null } + .map { case (key, metadata, status_code) => + (Bytes.toString(key.copyBytes()), Bytes.toString(metadata.copyBytes()), Bytes.toLong(status_code.copyBytes())) } // TODO: Should I combine next two stages for efficiency? .collect { case (key, json, StatusOK) => (key, json) } + .filter { case (key, json) => GrobidScorable.keepRecord(json) } .map { entry : (String, String) => GrobidScorable.jsonToMapFeatures(entry._1, entry._2) } } } object GrobidScorable { + def keepRecord(json : String) : Boolean = { + Scorable.jsonToMap(json) match { + case None => false + case Some(map) => { + if (map contains "title") { + val title = Scorable.getString(map, "title") + title != null && title.length <= Scorable.MaxTitleLength + } else { + false + } + } + } + } + + def mapToAuthorList(map : Map[String, Any]) : List[String] = { + if (map contains "authors") { + val objArray = map("authors").asInstanceOf[List[Any]].map(e => e.asInstanceOf[Map[String,Any]]) + objArray + .filter(e => e contains "name") + .map(e => e.get("name").get.asInstanceOf[String]) + } else { + List() + } + } + def getHBaseSource(table : String, host : String) : HBaseSource = { - HBaseBuilder.build(table, host, List("grobid0:tei_json", "grobid0:status_code"), SourceMode.SCAN_ALL) + HBaseBuilder.build(table, host, List("grobid0:metadata", "grobid0:status_code"), SourceMode.SCAN_ALL) } def jsonToMapFeatures(key : String, json : String) : MapFeatures = { @@ -45,7 +71,9 @@ object GrobidScorable { case None => MapFeatures(Scorable.NoSlug, json) case Some(map) => { if (map contains "title") { - ScorableFeatures.create(title=Scorable.getString(map, "title"), sha1=key).toMapFeatures + val authors: List[String] = mapToAuthorList(map) + val title = Scorable.getString(map, "title") + ScorableFeatures.create(title=title, authors=authors, sha1=key).toMapFeatures } else { MapFeatures(Scorable.NoSlug, json) } diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala new file mode 100644 index 0000000..f4e84fe --- /dev/null +++ b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala @@ -0,0 +1,60 @@ + +package sandcrawler + +import cascading.flow.FlowDef +import cascading.pipe.Pipe +import cascading.tuple.Fields +import com.twitter.scalding._ +import com.twitter.scalding.typed.TDsl._ +import org.apache.hadoop.hbase.io.ImmutableBytesWritable +import org.apache.hadoop.hbase.util.Bytes +import parallelai.spyglass.base.JobBase +import parallelai.spyglass.hbase.HBaseConstants.SourceMode +import parallelai.spyglass.hbase.HBasePipeConversions +import parallelai.spyglass.hbase.HBaseSource + +class GrobidScorableDumpJob(args: Args) extends JobBase(args) { + + val grobidHbaseRows = Stat("hbase-rows-scanned", "hbase-grobid-dump") + val filteredGrobidRows = Stat("grobid-rows-filtered", "hbase-grobid-dump") + val parsedGrobidRows = Stat("grobid-rows-parsed", "hbase-grobid-dump") + val validGrobidRows = Stat("grobid-rows-valid-slug", "hbase-grobid-dump") + + val pipe = GrobidScorable.getHBaseSource(args("hbase-table"), args("zookeeper-hosts")) + .read + // Can't just "fromBytesWritable" because we have multiple types? + .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "metadata", "status_code")) + .filter { case (_, metadata, status_code) => + grobidHbaseRows.inc + metadata != null && status_code != null + } + .map { case (key, metadata, status_code) => + (Bytes.toString(key.copyBytes()), Bytes.toString(metadata.copyBytes()), Bytes.toLong(status_code.copyBytes())) + } + // TODO: Should I combine next two stages for efficiency? + .collect { case (key, json, 200) => + filteredGrobidRows.inc + (key, json) + } + .map { entry : (String, String) => + parsedGrobidRows.inc + GrobidScorable.jsonToMapFeatures(entry._1, entry._2) + } + .filter { entry => Scorable.isValidSlug(entry.slug) } + .map { entry => + validGrobidRows.inc + entry + } + // XXX: this groupBy after the map? + .groupBy { case MapFeatures(slug, json) => slug } + .map { tuple => + val (slug : String, features : MapFeatures) = tuple + (slug, ReduceFeatures(features.json)) + } + + pipe + .map { case (slug, features) => + (slug, features.json) + } + .write(TypedTsv[(String, String)](args("output"))) +} diff --git a/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala new file mode 100644 index 0000000..20cc7a1 --- /dev/null +++ b/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala @@ -0,0 +1,37 @@ +package sandcrawler + +import java.util.Properties + +import cascading.property.AppProps +import cascading.tuple.Fields +import com.twitter.scalding._ +import parallelai.spyglass.base.JobBase +import parallelai.spyglass.hbase.HBaseConstants.SourceMode +import parallelai.spyglass.hbase.HBasePipeConversions +import parallelai.spyglass.hbase.HBaseSource + +class HBaseColCountJob(args: Args) extends JobBase(args) with HBasePipeConversions { + + val output = args("output") + + HBaseColCountJob.getHBaseSource( + args("hbase-table"), + args("zookeeper-hosts"), + args("column")) + .read + .debug + .groupAll { _.size('count) } + .write(Tsv(output)) +} + +object HBaseColCountJob { + + // eg, "wbgrp-journal-extract-0-qa",7 "mtrcs-zk1.us.archive.org:2181" + def getHBaseSource(hbaseTable: String, zookeeperHosts: String, col: String) : HBaseSource = { + HBaseBuilder.build( + hbaseTable, + zookeeperHosts, + List(col), + SourceMode.SCAN_ALL) + } +} diff --git a/scalding/src/main/scala/sandcrawler/HBaseStatusCodeCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseStatusCodeCountJob.scala new file mode 100644 index 0000000..4d9880f --- /dev/null +++ b/scalding/src/main/scala/sandcrawler/HBaseStatusCodeCountJob.scala @@ -0,0 +1,32 @@ +package sandcrawler + +import java.util.Properties + +import cascading.property.AppProps +import cascading.tuple.Fields +import com.twitter.scalding._ +import com.twitter.scalding.typed.TDsl._ +import org.apache.hadoop.hbase.io.ImmutableBytesWritable +import org.apache.hadoop.hbase.util.Bytes +import parallelai.spyglass.base.JobBase +import parallelai.spyglass.hbase.HBaseConstants.SourceMode +import parallelai.spyglass.hbase.HBasePipeConversions +import parallelai.spyglass.hbase.HBaseSource + +class HBaseStatusCodeCountJob(args: Args) extends JobBase(args) with HBasePipeConversions { + + val source = HBaseCountJob.getHBaseSource( + args("hbase-table"), + args("zookeeper-hosts"), + "grobid0:status_code") + + val statusPipe : TypedPipe[Long] = source + .read + .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable)]('key, 'status_code) + .map { case (key, raw_code) => Bytes.toLong(raw_code.copyBytes()) } + + statusPipe.groupBy { identity } + .size + .debug + .write(TypedTsv[(Long,Long)](args("output"))) +} diff --git a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala index fd0b4e2..f79d672 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala @@ -18,15 +18,15 @@ class HBaseStatusCountJob(args: Args) extends JobBase(args) with HBasePipeConver val source = HBaseCountJob.getHBaseSource( args("hbase-table"), args("zookeeper-hosts"), - "grobid0:status_code") + "grobid0:status") - val statusPipe : TypedPipe[Long] = source + val statusPipe : TypedPipe[String] = source .read - .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable)]('key, 'status_code) - .map { case (key, raw_code) => Bytes.toLong(raw_code.copyBytes()) } + .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable)]('key, 'status) + .map { case (key, raw_status) => Bytes.toString(raw_status.copyBytes()) } statusPipe.groupBy { identity } .size .debug - .write(TypedTsv[(Long,Long)](args("output"))) + .write(TypedTsv[(String,Long)](args("output"))) } diff --git a/scalding/src/main/scala/sandcrawler/Scorable.scala b/scalding/src/main/scala/sandcrawler/Scorable.scala index 9b9c633..5aac032 100644 --- a/scalding/src/main/scala/sandcrawler/Scorable.scala +++ b/scalding/src/main/scala/sandcrawler/Scorable.scala @@ -30,6 +30,7 @@ abstract class Scorable { } object Scorable { + val MaxTitleLength = 1023 val NoSlug = "NO SLUG" // Used for slug if title is empty or unparseable def isValidSlug(slug : String) : Boolean = { diff --git a/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala b/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala index 0b9868a..241db79 100644 --- a/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala +++ b/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala @@ -3,6 +3,7 @@ package sandcrawler import java.io.InputStream import scala.io.Source +import scala.util.parsing.json.JSONArray import scala.util.parsing.json.JSONObject object ScorableFeatures { @@ -10,11 +11,13 @@ object ScorableFeatures { val fileStream : InputStream = getClass.getResourceAsStream("/slug-blacklist.txt") val SlugBlacklist : Set[String] = Source.fromInputStream(fileStream).getLines.toSet fileStream.close + val MinSlugLength = 8 // Static factory method - def create(title : String, year : Int = 0, doi : String = "", sha1 : String = "") : ScorableFeatures = { + def create(title : String, authors : List[Any] = List(), year : Int = 0, doi : String = "", sha1 : String = "") : ScorableFeatures = { new ScorableFeatures( title=if (title == null) "" else title, + authors=if (authors == null) List() else authors.map(a => if (a == null) "" else a), year=year, doi=if (doi == null) "" else doi, sha1=if (sha1 == null) "" else sha1) @@ -23,13 +26,14 @@ object ScorableFeatures { // Contains features needed to make slug and to score (in combination // with a second ScorableFeatures). Create with above static factory method. -class ScorableFeatures private(title : String, year: Int = 0, doi : String = "", sha1: String = "") { +class ScorableFeatures private(title : String, authors : List[Any] = List(), year: Int = 0, doi : String = "", sha1: String = "") { def toMap() : Map[String, Any] = - Map("title" -> title, "year" -> year, "doi" -> doi, "sha1" -> sha1) + Map("title" -> title, "authors" -> JSONArray(authors), "year" -> year, "doi" -> doi, "sha1" -> sha1) - override def toString() : String = + override def toString() : String = { JSONObject(toMap).toString + } def toSlug() : String = { if (title == null) { @@ -38,7 +42,10 @@ class ScorableFeatures private(title : String, year: Int = 0, doi : String = "", val unaccented = StringUtilities.removeAccents(title) // Remove punctuation val slug = StringUtilities.removePunctuation((unaccented.toLowerCase())).replaceAll("\\s", "") - if (slug.isEmpty || slug == null || (ScorableFeatures.SlugBlacklist contains slug)) Scorable.NoSlug else slug + if (slug.isEmpty + || slug == null + || (ScorableFeatures.SlugBlacklist contains slug) + || (slug.length < ScorableFeatures.MinSlugLength)) Scorable.NoSlug else slug } } diff --git a/scalding/src/main/scala/sandcrawler/ScoreJob.scala b/scalding/src/main/scala/sandcrawler/ScoreJob.scala index 75d45e9..107f504 100644 --- a/scalding/src/main/scala/sandcrawler/ScoreJob.scala +++ b/scalding/src/main/scala/sandcrawler/ScoreJob.scala @@ -2,56 +2,63 @@ package sandcrawler import cascading.pipe.Pipe import com.twitter.scalding.Args +import com.twitter.scalding.Stat import com.twitter.scalding.TypedPipe import com.twitter.scalding.TypedTsv import parallelai.spyglass.base.JobBase class ScoreJob(args: Args) extends JobBase(args) { - // TODO: Instantiate any subclass of Scorable specified in args. - val sc1 : Scorable = new GrobidScorable() - val sc2 : Scorable = new CrossrefScorable() - val pipe1 : TypedPipe[(String, ReduceFeatures)] = sc1.getInputPipe(args) - val pipe2 : TypedPipe[(String, ReduceFeatures)] = sc2.getInputPipe(args) - pipe1.join(pipe2).map { entry => - val (slug : String, (features1 : ReduceFeatures, features2 : ReduceFeatures)) = entry - new ReduceOutput( - slug, - Scorable.computeSimilarity(features1, features2), - features1.json, - features2.json) - } - //TypedTsv doesn't work over case classes. - .map { entry => (entry.slug, entry.score, entry.json1, entry.json2) } - .write(TypedTsv[(String, Int, String, String)](args("output"))) -} - -/* -// Ugly hack to get non-String information into ScoreJob above. -object ScoreJob { - var scorable1 : Option[Scorable] = None - var scorable2 : Option[Scorable] = None - - def setScorable1(s : Scorable) { - scorable1 = Some(s) - } + val grobidRowCount = Stat("grobid-rows-filtered", "sandcrawler") + val crossrefRowCount = Stat("crossref-rows-filtered", "sandcrawler") + val joinedRowCount = Stat("joined-rows", "sandcrawler") + /* TODO: + val uniqueDoiCount = Stat("unique-doi-count", "sandcrawler") + val uniqueSha1Count = Stat("unique-sha1-count", "sandcrawler") + */ - def getScorable1() : Scorable = { - scorable1 match { - case Some(s) => s - case None => null + val grobidScorable : Scorable = new GrobidScorable() + val crossrefScorable : Scorable = new CrossrefScorable() + val grobidPipe : TypedPipe[(String, ReduceFeatures)] = grobidScorable + .getInputPipe(args) + .map { r => + grobidRowCount.inc + r } - } + val crossrefPipe : TypedPipe[(String, ReduceFeatures)] = crossrefScorable + .getInputPipe(args) + .map { r => + crossrefRowCount.inc + r + } + + val joinedPipe = grobidPipe + .addTrap(TypedTsv(args("output") + ".trapped")) + .join(crossrefPipe) - def setScorable2(s: Scorable) { - scorable2 = Some(s) - } + /* TODO: + // Reduces to count unique SHA1 and DOI + joinedPipe + .map { case (_, (grobidFeatures, _)) => grobidFeatures.sha } + .distinct + .map { _ => uniqueSha1Count.inc } + joinedPipe + .map { case (_, (_, crossrefFeatures)) => crossrefFeatures.doi } + .distinct + .map { _ => uniqueDoiCount.inc } + */ - def getScorable2() : Scorable = { - scorable2 match { - case Some(s) => s - case None => null + // TypedTsv doesn't work over case classes. + joinedPipe + .map { case (slug, (grobidFeatures, crossrefFeatures)) => + joinedRowCount.inc + //val (slug : String, (grobidFeatures: ReduceFeatures, crossrefFeatures: ReduceFeatures)) = entry + new ReduceOutput( + slug, + Scorable.computeSimilarity(grobidFeatures, crossrefFeatures), + grobidFeatures.json, + crossrefFeatures.json) } - } + .map { entry => (entry.slug, entry.score, entry.json1, entry.json2) } + .write(TypedTsv[(String, Int, String, String)](args("output"))) } - */ diff --git a/scalding/src/main/scala/sandcrawler/StringUtilities.scala b/scalding/src/main/scala/sandcrawler/StringUtilities.scala index 2745875..e03b60d 100644 --- a/scalding/src/main/scala/sandcrawler/StringUtilities.scala +++ b/scalding/src/main/scala/sandcrawler/StringUtilities.scala @@ -36,7 +36,7 @@ object StringUtilities { // Source: https://stackoverflow.com/a/30076541/631051 def removePunctuation(s: String) : String = { - s.replaceAll("""[\p{Punct}]""", "") + s.replaceAll("""[\p{Punct}’·“”‘’“”«»「」]""", "") } // Adapted from: https://stackoverflow.com/a/16018452/631051 diff --git a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala index 1789d1a..f598cae 100644 --- a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala +++ b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala @@ -64,12 +64,18 @@ class CrossrefScorableTest extends FlatSpec with Matchers { "issn-type" : [ { "value" : "0987-7983", "type" : "print" } ], "subject" : [ "Pediatrics, Perinatology, and Child Health" ] } -""" +""".replace("<<DOI>>", "10.123/aBc") // scalastyle:on - val CrossrefStringWithTitle = CrossrefString.replace("<<TITLE>>", "Some Title") + val CrossrefStringWithGoodTitle = CrossrefString.replace("<<TITLE>>", "Some Title") + val CrossrefStringWithMaximumTitle = CrossrefString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength) + val CrossrefStringWithExcessiveTitle = CrossrefString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength + "0") + val CrossrefStringWithNullTitle = CrossrefString.replace("\"<<TITLE>>\"", "null") val CrossrefStringWithEmptyTitle = CrossrefString.replace("<<TITLE>>", "") val CrossrefStringWithoutTitle = CrossrefString.replace("title", "nottitle") val MalformedCrossrefString = CrossrefString.replace("}", "") + val CrossrefStringWithNoAuthors = CrossrefString.replace("<<TITLE>>", "Some Valid Title").replace("author", "no-author") + val CrossrefStringWrongType = CrossrefString.replace("<<TITLE>>", "Some Valid Title").replace("journal-article", "other") + val CrossrefStringNoType = CrossrefString.replace("<<TITLE>>", "Some Valid Title").replace("type", "not-type") // Unit tests "CrossrefScorable.jsonToMapFeatures()" should "handle invalid JSON" in { @@ -82,19 +88,64 @@ class CrossrefScorableTest extends FlatSpec with Matchers { result.slug shouldBe Scorable.NoSlug } + it should "handle null title" in { + val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithNullTitle) + result.slug shouldBe Scorable.NoSlug + } + it should "handle empty title" in { val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithEmptyTitle) result.slug shouldBe Scorable.NoSlug } + it should "handle missing authors" in { + val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithNoAuthors) + result.slug shouldBe Scorable.NoSlug + } + it should "handle valid input" in { - val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithTitle) + val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithGoodTitle) result.slug shouldBe "sometitle" Scorable.jsonToMap(result.json) match { case None => fail() case Some(map) => { map("title").asInstanceOf[String] shouldBe "Some Title" + map("doi").asInstanceOf[String] shouldBe "10.123/abc" + // TODO: full name? not just a string? + map("authors").asInstanceOf[List[String]] shouldBe List("Gaier") + map("year").asInstanceOf[Double].toInt shouldBe 2002 } } } + + "CrossrefScorable.keepRecord()" should "return true for valid JSON with title" in { + CrossrefScorable.keepRecord(CrossrefStringWithGoodTitle) shouldBe true + } + + it should "return true for valid JSON with a title of maximum permitted length" in { + CrossrefScorable.keepRecord(CrossrefStringWithMaximumTitle) shouldBe true + } + + it should "return false for valid JSON with excessively long title" in { + CrossrefScorable.keepRecord(CrossrefStringWithExcessiveTitle) shouldBe false + } + + it should "return false for valid JSON with null title" in { + CrossrefScorable.keepRecord(CrossrefStringWithNullTitle) shouldBe false + } + + it should "return false for valid JSON with no title" in { + CrossrefScorable.keepRecord(CrossrefStringWithoutTitle) shouldBe false + } + + it should "return false for invalid JSON" in { + CrossrefScorable.keepRecord(CrossrefStringWithoutTitle) shouldBe false + } + + it should "handle content types" in { + val resultWrong = CrossrefScorable.jsonToMapFeatures(CrossrefStringWrongType) + resultWrong.slug shouldBe Scorable.NoSlug + val resultMissing = CrossrefScorable.jsonToMapFeatures(CrossrefStringNoType) + resultMissing.slug shouldBe Scorable.NoSlug + } } diff --git a/scalding/src/test/scala/sandcrawler/GrobidScorableDumpJobTest.scala b/scalding/src/test/scala/sandcrawler/GrobidScorableDumpJobTest.scala new file mode 100644 index 0000000..bf9343b --- /dev/null +++ b/scalding/src/test/scala/sandcrawler/GrobidScorableDumpJobTest.scala @@ -0,0 +1,124 @@ + +package sandcrawler + +import cascading.tuple.Fields +import cascading.tuple.Tuple +import com.twitter.scalding.JobTest +import com.twitter.scalding.TextLine +import com.twitter.scalding.TupleConversions +import com.twitter.scalding.TypedTsv +import org.apache.hadoop.hbase.io.ImmutableBytesWritable +import org.apache.hadoop.hbase.util.Bytes +import org.scalatest._ +import parallelai.spyglass.hbase.HBaseConstants.SourceMode + +class GrobidScorableDumpJobTest extends FlatSpec with Matchers { + //scalastyle:off + val JsonString = """ +{ + "title": "<<TITLE>>", + "authors": [ + {"name": "Brewster Kahle"}, + {"name": "J Doe"} + ], + "journal": { + "name": "Dummy Example File. Journal of Fake News. pp. 1-2. ISSN 1234-5678", + "eissn": null, + "issn": null, + "issue": null, + "publisher": null, + "volume": null + }, + "date": "2000", + "doi": null, + "citations": [ + { "authors": [{"name": "A Seaperson"}], + "date": "2001", + "id": "b0", + "index": 0, + "issue": null, + "journal": "Letters in the Alphabet", + "publisher": null, + "title": "Everything is Wonderful", + "url": null, + "volume": "20"}, + { "authors": [], + "date": "2011-03-28", + "id": "b1", + "index": 1, + "issue": null, + "journal": "The Dictionary", + "publisher": null, + "title": "All about Facts", + "url": null, + "volume": "14"} + ], + "abstract": "Everything you ever wanted to know about nothing", + "body": "Introduction \nEverything starts somewhere, as somebody [1] once said. \n\n In Depth \n Meat \nYou know, for kids. \n Potatos \nQED.", + "acknowledgement": null, + "annex": null +} +""" + // scalastyle:on + val JsonStringWithTitle = JsonString.replace("<<TITLE>>", "Dummy Example File") + val JsonStringWithoutTitle = JsonString.replace("title", "nottitle") + val MalformedJsonString = JsonString.replace("}", "") + + // Pipeline tests + val output = "/tmp/testOutput" + val input = "/tmp/testInput" + val (testTable, testHost) = ("test-table", "dummy-host:2181") + + val Sha1Strings : List[String] = List( + "sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q", // good + "sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU", // good + "sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT", // good + "sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56", // bad status + "sha1:93187A85273589347598473894839443", // malformed + "sha1:024937534094897039547e9824382943") // bad status + + val JsonStrings : List[String] = List( + JsonString.replace("<<TITLE>>", "Title 1: The Classic"), + JsonString.replace("<<TITLE>>", "Title 2: TNG"), + JsonString.replace("<<TITLE>>", "Title 3: The Sequel"), + // This will have bad status. + JsonString.replace("<<TITLE>>", "Title 1: The Classic"), + MalformedJsonString, + // This will have bad status. + JsonString.replace("<<TITLE>>", "Title 2: Not TNG") + ) + + // bnewbold: status codes aren't strings, they are uint64 + val Ok : Long = 200 + val Bad : Long = 400 + val StatusCodes = List(Ok, Ok, Ok, Bad, Ok, Bad) + + val SampleDataHead : List[Tuple] = (Sha1Strings, JsonStrings, StatusCodes) + .zipped + .toList + .map { case (sha, json, status) => List(Bytes.toBytes(sha), Bytes.toBytes(json), Bytes.toBytes(status)) } + .map { l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*) } + + // scalastyle:off null + // Add example of lines without GROBID data + val SampleData = SampleDataHead :+ new Tuple( + new ImmutableBytesWritable(Bytes.toBytes("sha1:35985C3YNNEGH5WAG5ZAA88888888888")), null, null) + // scalastyle:on null + + JobTest("sandcrawler.GrobidScorableDumpJob") + .arg("test", "") + .arg("app.conf.path", "app.conf") + .arg("output", output) + .arg("hbase-table", testTable) + .arg("zookeeper-hosts", testHost) + .arg("debug", "true") + .source[Tuple](GrobidScorable.getHBaseSource(testTable, testHost), SampleData) + .sink[(String, String)](TypedTsv[(String, String)](output)) { + outputBuffer => + "The pipeline" should "return correct-length list" in { + outputBuffer should have length 3 + } + } + .run + .finish +} diff --git a/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala b/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala index 661824b..119cf90 100644 --- a/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala +++ b/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala @@ -57,7 +57,10 @@ class GrobidScorableTest extends FlatSpec with Matchers { "annex": null } """ - val GrobidStringWithTitle = GrobidString.replace("<<TITLE>>", "Dummy Example File") + val GrobidStringWithGoodTitle = GrobidString.replace("<<TITLE>>", "Dummy Example File") + val GrobidStringWithMaximumTitle = GrobidString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength) + val GrobidStringWithExcessiveTitle = GrobidString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength + "0") + val GrobidStringWithNullTitle = GrobidString.replace("\"<<TITLE>>\"", "null") val GrobidStringWithoutTitle = GrobidString.replace("title", "nottitle") val MalformedGrobidString = GrobidString.replace("}", "") val Key = "Dummy Key" @@ -69,20 +72,50 @@ class GrobidScorableTest extends FlatSpec with Matchers { result.slug shouldBe Scorable.NoSlug } + it should "handle null title" in { + val result = GrobidScorable.jsonToMapFeatures(Key, GrobidStringWithNullTitle) + result.slug shouldBe Scorable.NoSlug + } + it should "handle missing title" in { val result = GrobidScorable.jsonToMapFeatures(Key, GrobidStringWithoutTitle) result.slug shouldBe Scorable.NoSlug } it should "handle valid input" in { - val result = GrobidScorable.jsonToMapFeatures(Key, GrobidStringWithTitle) + val result = GrobidScorable.jsonToMapFeatures(Key, GrobidStringWithGoodTitle) result.slug shouldBe "dummyexamplefile" Scorable.jsonToMap(result.json) match { case None => fail() case Some(map) => { map should contain key "title" map("title").asInstanceOf[String] shouldBe "Dummy Example File" + map("authors").asInstanceOf[List[String]] shouldBe List("Brewster Kahle", "J Doe") } } } + + "GrobidScorable.keepRecord()" should "return true for valid JSON with title" in { + GrobidScorable.keepRecord(GrobidStringWithGoodTitle) shouldBe true + } + + it should "return true for valid JSON with a title of maximum permitted length" in { + GrobidScorable.keepRecord(GrobidStringWithMaximumTitle) shouldBe true + } + + it should "return false for valid JSON with excessively long title" in { + GrobidScorable.keepRecord(GrobidStringWithExcessiveTitle) shouldBe false + } + + it should "return false for valid JSON with null title" in { + GrobidScorable.keepRecord(GrobidStringWithNullTitle) shouldBe false + } + + it should "return false for valid JSON with no title" in { + GrobidScorable.keepRecord(GrobidStringWithoutTitle) shouldBe false + } + + it should "return false for invalid JSON" in { + GrobidScorable.keepRecord(GrobidStringWithoutTitle) shouldBe false + } } diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala new file mode 100644 index 0000000..d2cf9de --- /dev/null +++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala @@ -0,0 +1,71 @@ +package sandcrawler + +import cascading.tuple.Fields +import cascading.tuple.Tuple +import com.twitter.scalding.JobTest +import com.twitter.scalding.Tsv +import com.twitter.scalding.TupleConversions +import com.twitter.scalding.TypedTsv +import org.apache.hadoop.hbase.io.ImmutableBytesWritable +import org.apache.hadoop.hbase.util.Bytes +import org.junit.runner.RunWith +import org.scalatest.FunSpec +import org.scalatest.junit.JUnitRunner +import org.slf4j.LoggerFactory +import parallelai.spyglass.hbase.HBaseConstants.SourceMode +import parallelai.spyglass.hbase.HBaseSource +import scala._ + +@RunWith(classOf[JUnitRunner]) +class HBaseStatusCodeCountTest extends FunSpec with TupleConversions { + + val output = "/tmp/testOutput" + val (testTable, testHost) = ("test-table", "dummy-host:2181") + + val log = LoggerFactory.getLogger(this.getClass.getName) + + val statusType1 : Long = 200 + val statusType2 : Long = 404 + val statusType1Bytes = Bytes.toBytes(statusType1) + val statusType2Bytes = Bytes.toBytes(statusType2) + + // TODO(bnewbold): now to express a null (empty value) in this list? + val sampleData : List[List[Array[Byte]]] = List( + ("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q", statusType1Bytes), + ("sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU", statusType1Bytes), + ("sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT", statusType2Bytes), + ("sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56", statusType2Bytes), + ("sha1:885C3YNNEGH5WAG5ZAAXWA8BNXJWT6CZ", statusType2Bytes), + ("sha1:00904C3YNNEGH5WAG5ZA9XWAEBNXJWT6", statusType2Bytes), + ("sha1:249C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ", statusType1Bytes), + ("sha1:095893C3YNNEGH5WAG5ZAAXWAEBNXJWT", statusType2Bytes)) + .map(pair => List(Bytes.toBytes(pair._1), pair._2)) + + val statusType1Count = sampleData.count(lst => lst(1) == statusType1Bytes) + val statusType2Count = sampleData.count(lst => lst(1) == statusType2Bytes) + + JobTest("sandcrawler.HBaseStatusCodeCountJob") + .arg("test", "") + .arg("app.conf.path", "app.conf") + .arg("output", output) + .arg("hbase-table", testTable) + .arg("zookeeper-hosts", testHost) + .arg("debug", "true") + .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "grobid0:status_code"), + sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*))) + .sink[Tuple](TypedTsv[(Long, Long)](output)) { + outputBuffer => + it("should return a correct number of elements.") { + assert(outputBuffer.size === 2) + } + + // Convert List[Tuple] to Map[Long, Long]. + val counts = outputBuffer.map(t => (t.getLong(0), t.getLong(1))).toMap + it("should have the appropriate number of each status type") { + assert(counts(statusType1) == statusType1Count) + assert(counts(statusType2) == statusType2Count) + } + } + .run + .finish +} diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala index 3291670..7e91af3 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala @@ -24,10 +24,8 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions { val log = LoggerFactory.getLogger(this.getClass.getName) - val statusType1 : Long = 200 - val statusType2 : Long = 404 - val statusType1Bytes = Bytes.toBytes(statusType1) - val statusType2Bytes = Bytes.toBytes(statusType2) + val statusType1Bytes = Bytes.toBytes("""{"status": "success"}""") + val statusType2Bytes = Bytes.toBytes("""{"status": "partial"}""") // TODO(bnewbold): now to express a null (empty value) in this list? val sampleData : List[List[Array[Byte]]] = List( @@ -51,20 +49,13 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions { .arg("hbase-table", testTable) .arg("zookeeper-hosts", testHost) .arg("debug", "true") - .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "grobid0:status_code"), + .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "grobid0:status"), sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*))) - .sink[Tuple](TypedTsv[(Long, Long)](output)) { + .sink[Tuple](TypedTsv[(String, Long)](output)) { outputBuffer => it("should return a 2-element list.") { assert(outputBuffer.size === 2) } - - // Convert List[Tuple] to Map[Long, Long]. - val counts = outputBuffer.map(t => (t.getLong(0), t.getLong(1))).toMap - it("should have the appropriate number of each status type") { - assert(counts(statusType1) == statusType1Count) - assert(counts(statusType2) == statusType2Count) - } } .run .finish diff --git a/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala b/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala index 5a22ef8..474f69a 100644 --- a/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala +++ b/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala @@ -9,22 +9,6 @@ import org.scalatest._ // scalastyle:off null class ScorableFeaturesTest extends FlatSpec with Matchers { - // TODO: Remove this when we're convinced that our file-reading code - // works. (I'm already convinced. --Ellen) - "read slugs" should "work" in { - val SlugBlacklist = Set( "abbreviations", "abstract", "acknowledgements", - "article", "authorreply", "authorsreply", "bookreview", "bookreviews", - "casereport", "commentary", "commentaryon", "commenton", "commentto", - "contents", "correspondence", "dedication", "editorialadvisoryboard", - "focus", "hypothesis", "inbrief", "introduction", "introductiontotheissue", - "lettertotheeditor", "listofabbreviations", "note", "overview", "preface", - "references", "results", "review", "reviewarticle", "summary", "title", - "name") - - ScorableFeatures.SlugBlacklist.size shouldBe SlugBlacklist.size - for (s <- ScorableFeatures.SlugBlacklist) SlugBlacklist should contain (s) - } - private def titleToSlug(s : String) : String = { ScorableFeatures.create(title = s).toSlug } @@ -52,7 +36,7 @@ class ScorableFeaturesTest extends FlatSpec with Matchers { it should "strip punctuation" in { titleToSlug("HELLO!:the:re") shouldBe "hellothere" - titleToSlug("a:b:c") shouldBe "abc" + titleToSlug("a:b:cdefgh") shouldBe "abcdefgh" titleToSlug( "If you're happy and you know it, clap your hands!") shouldBe "ifyourehappyandyouknowitclapyourhands" titleToSlug(":;\"\'") shouldBe Scorable.NoSlug @@ -65,14 +49,19 @@ class ScorableFeaturesTest extends FlatSpec with Matchers { } it should "strip special characters" in { - titleToSlug(":;!',|\"\'`.#?!-@*/\\=+~%$^{}()[]<>-_") shouldBe Scorable.NoSlug + titleToSlug(":;!',|\"\'`.#?!-@*/\\=+~%$^{}()[]<>-_’·“”‘’“”«»「」") shouldBe Scorable.NoSlug // TODO: titleToSlug("©™₨№…") shouldBe Scorable.NoSlug // TODO: titleToSlug("πµΣσ") shouldBe Scorable.NoSlug } it should "remove whitespace" in { titleToSlug("foo bar : baz ::") shouldBe "foobarbaz" - titleToSlug("\na\t:b:c") shouldBe "abc" + titleToSlug("\na\t:b:cdefghi") shouldBe "abcdefghi" titleToSlug("\n \t \r ") shouldBe Scorable.NoSlug } + + it should "skip very short slugs" in { + titleToSlug("short") shouldBe Scorable.NoSlug + titleToSlug("a longer, more in depth title") shouldBe "alongermoreindepthtitle" + } } diff --git a/scalding/src/test/scala/sandcrawler/ScoreJobTest.scala b/scalding/src/test/scala/sandcrawler/ScoreJobTest.scala index 00e4659..32fb16c 100644 --- a/scalding/src/test/scala/sandcrawler/ScoreJobTest.scala +++ b/scalding/src/test/scala/sandcrawler/ScoreJobTest.scala @@ -117,6 +117,8 @@ class ScoreJobTest extends FlatSpec with Matchers { } """ // scalastyle:on + val TooLongOfTitle = "X" * Scorable.MaxTitleLength + "Y" // arbitrary long string + val TooShortOfTitle = "X" * (ScorableFeatures.MinSlugLength - 1) val CrossrefStringWithTitle = CrossrefString.replace("<<TITLE>>", "SomeTitle") val CrossrefStringWithoutTitle = CrossrefString.replace("title", "nottitle") val MalformedCrossrefString = CrossrefString.replace("}", "") @@ -124,7 +126,9 @@ class ScoreJobTest extends FlatSpec with Matchers { CrossrefString.replace("<<TITLE>>", "Title 2: TNG").replace("<<DOI>>", "DOI-0"), CrossrefString.replace("<<TITLE>>", "Title 1: TNG 2A").replace("<<DOI>>", "DOI-0.5"), CrossrefString.replace("<<TITLE>>", "Title 1: TNG 3").replace("<<DOI>>", "DOI-0.75"), - CrossrefString.replace("<<TITLE>>", "Title 2: Rebooted").replace("<<DOI>>", "DOI-1")) + CrossrefString.replace("<<TITLE>>", "Title 2: Rebooted").replace("<<DOI>>", "DOI-1"), + CrossrefString.replace("<<TITLE>>", TooLongOfTitle).replace("<<DOI>>", "DOI-1"), + CrossrefString.replace("<<TITLE>>", TooShortOfTitle).replace("<<DOI>>", "DOI-1")) // Pipeline tests val output = "/tmp/testOutput" @@ -137,23 +141,28 @@ class ScoreJobTest extends FlatSpec with Matchers { "sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT", "sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56", "sha1:93187A85273589347598473894839443", - "sha1:024937534094897039547e9824382943") + "sha1:024937534094897039547e9824382943", + "sha1:93229759932857982837892347893892", + "sha1:83229759932857982837892347893892") val JsonStrings : List[String] = List( - JsonString.replace("<<TITLE>>", "Title 1"), + JsonString.replace("<<TITLE>>", "Title 1: The Original"), JsonString.replace("<<TITLE>>", "Title 2: TNG"), JsonString.replace("<<TITLE>>", "Title 3: The Sequel"), // This will have bad status. - JsonString.replace("<<TITLE>>", "Title 1"), + JsonString.replace("<<TITLE>>", "Title 1: The Original"), MalformedJsonString, // This will have bad status. - JsonString.replace("<<TITLE>>", "Title 2") + JsonString.replace("<<TITLE>>", "Title 2: Not TNG"), + // These are in both sources but have bad titles + JsonString.replace("<<TITLE>>", TooLongOfTitle), + JsonString.replace("<<TITLE>>", TooShortOfTitle) ) // bnewbold: status codes aren't strings, they are uint64 val Ok : Long = 200 val Bad : Long = 400 - val StatusCodes = List(Ok, Ok, Ok, Bad, Ok, Bad) + val StatusCodes = List(Ok, Ok, Ok, Bad, Ok, Bad, Ok, Ok) val SampleDataHead : List[Tuple] = (Sha1Strings, JsonStrings, StatusCodes) .zipped @@ -181,17 +190,24 @@ class ScoreJobTest extends FlatSpec with Matchers { 0 -> CrossrefStrings(0), 1 -> CrossrefStrings(1), 2 -> CrossrefStrings(2), - 3 -> CrossrefStrings(3))) + 3 -> CrossrefStrings(3), + 4 -> CrossrefStrings(4), + 4 -> CrossrefStrings(5))) + .sink[(String, ReduceFeatures)](TypedTsv[(String, ReduceFeatures)](output + ".trapped")) { _ => () } .sink[(String, Int, String, String)](TypedTsv[(String, Int, String, String)](output)) { // Grobid titles and slugs (in parentheses): // Title 1 (title1) // Title 2: TNG (title2tng) // Title 3: The Sequel (title3thesequel) + // <too long of a title> + // <too short of a title> // crossref titles and slugs (in parentheses): // Title 2: TNG (title2tng) // Title 1: TNG 2A (title1tng2a) // Title 1: TNG 3 (title1tng3) // Title 2: Rebooted (title2rebooted) + // <too long of a title> + // <too short of a title> // XXX: Join should have 3 "title1" slugs and 1 "title2tng" slug outputBuffer => "The pipeline" should "return a 1-element list" in { |