diff options
Diffstat (limited to 'scalding/src/main')
8 files changed, 204 insertions, 36 deletions
diff --git a/scalding/src/main/resources/slug-blacklist.txt b/scalding/src/main/resources/slug-blacklist.txt new file mode 100644 index 0000000..7dc701f --- /dev/null +++ b/scalding/src/main/resources/slug-blacklist.txt @@ -0,0 +1,34 @@ +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 diff --git a/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala b/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala index ff8201a..5d1eaf5 100644 --- a/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala +++ b/scalding/src/main/scala/sandcrawler/CrossrefScorable.scala @@ -35,7 +35,7 @@ object CrossrefScorable { 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 = new ScorableFeatures(title=titles(0), doi=doi) + val sf : ScorableFeatures = ScorableFeatures.create(title=titles(0), doi=doi) new MapFeatures(sf.toSlug, sf.toString) } } else { diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala index 9a09e05..e510f75 100644 --- a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala +++ b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala @@ -24,10 +24,10 @@ 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) } @@ -37,7 +37,7 @@ class GrobidScorable extends Scorable with HBasePipeConversions { object GrobidScorable { 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 +45,7 @@ object GrobidScorable { case None => MapFeatures(Scorable.NoSlug, json) case Some(map) => { if (map contains "title") { - new ScorableFeatures(Scorable.getString(map, "title"), sha1=key).toMapFeatures + ScorableFeatures.create(title=Scorable.getString(map, "title"), 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..468b68e --- /dev/null +++ b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala @@ -0,0 +1,62 @@ + +package sandcrawler + +import cascading.flow.FlowDef +import cascading.pipe.Pipe +import cascading.tuple.Fields +import com.twitter.scalding._ +import com.twitter.scalding._ +import com.twitter.scalding.typed.TDsl._ +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/ScorableFeatures.scala b/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala index 8ed3369..0b9868a 100644 --- a/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala +++ b/scalding/src/main/scala/sandcrawler/ScorableFeatures.scala @@ -1,31 +1,35 @@ package sandcrawler +import java.io.InputStream + +import scala.io.Source import scala.util.parsing.json.JSONObject +object ScorableFeatures { + // TODO: Add exception handling. + val fileStream : InputStream = getClass.getResourceAsStream("/slug-blacklist.txt") + val SlugBlacklist : Set[String] = Source.fromInputStream(fileStream).getLines.toSet + fileStream.close + + // Static factory method + def create(title : String, year : Int = 0, doi : String = "", sha1 : String = "") : ScorableFeatures = { + new ScorableFeatures( + title=if (title == null) "" else title, + year=year, + doi=if (doi == null) "" else doi, + sha1=if (sha1 == null) "" else sha1) + } +} // Contains features needed to make slug and to score (in combination -// with a second ScorableFeatures). -class ScorableFeatures(title : String, year: Int = 0, doi : String = "", sha1: String = "") { - - 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") - - def toMap() : Map[String, Any] = { - Map("title" -> (if (title == null) "" else title), - "year" -> year, - "doi" -> (if (doi == null) "" else doi), - "sha1" -> (if (sha1 == null) "" else sha1)) - } +// with a second ScorableFeatures). Create with above static factory method. +class ScorableFeatures private(title : String, year: Int = 0, doi : String = "", sha1: String = "") { - override def toString() : String = { - JSONObject(toMap()).toString - } + def toMap() : Map[String, Any] = + Map("title" -> title, "year" -> year, "doi" -> doi, "sha1" -> sha1) + + override def toString() : String = + JSONObject(toMap).toString def toSlug() : String = { if (title == null) { @@ -34,11 +38,10 @@ class ScorableFeatures(title : String, year: Int = 0, doi : String = "", sha1: S val unaccented = StringUtilities.removeAccents(title) // Remove punctuation val slug = StringUtilities.removePunctuation((unaccented.toLowerCase())).replaceAll("\\s", "") - if (slug.isEmpty || slug == null || (slugBlacklist contains slug)) Scorable.NoSlug else slug + if (slug.isEmpty || slug == null || (ScorableFeatures.SlugBlacklist contains slug)) Scorable.NoSlug else slug } } - def toMapFeatures = { + def toMapFeatures : MapFeatures = MapFeatures(toSlug, toString) - } } |