changed for sg

writers/src/main/scala/edu/psu/sagnik/research/pdwriters/impl/CreateResultsBatch.scala

@@ -1,6 +1,9 @@
 package edu.psu.sagnik.research.pdwriters.impl
 
+import java.io.{ BufferedWriter, FileWriter }
+
 import edu.psu.sagnik.research.pdsimplify.model.PDDocumentSimple
+import edu.psu.sagnik.research.pdwriters.writers.json.CreateTextLinesJSON
 
 /**
  * Created by szr163 on 8/26/16.
@@ -40,21 +43,29 @@ object CreateResultsBatch {
 
     var doc = document
 
-    doc = getExtractionResult(doc, pageNum, chars.map(_.bb), Color.BLUE)
+    //doc = getExtractionResult(doc, pageNum, chars.map(_.bb), Color.BLUE)
 
-    doc = getExtractionResult(doc, pageNum, words.map(_.bb), Color.GREEN)
+    //doc = getExtractionResult(doc, pageNum, words.map(_.bb), Color.GREEN)
 
     doc = getExtractionResult(doc, pageNum, lines.map(_.bb), Color.RED)
 
-    doc = getExtractionResult(doc, pageNum, paragraphs.map(_.bb), Color.CYAN)
+    //doc = getExtractionResult(doc, pageNum, paragraphs.map(_.bb), Color.CYAN)
 
-    doc = getExtractionResult(doc, pageNum, rasters.map(_.bb), Color.MAGENTA)
+    //doc = getExtractionResult(doc, pageNum, rasters.map(_.bb), Color.MAGENTA)
 
-    doc = getExtractionResult(doc, pageNum, segments.map(_.bb), Color.PINK)
+    //doc = getExtractionResult(doc, pageNum, segments.map(_.bb), Color.PINK)
 
     doc
   }
 
+  def writeTextLinesJson(fileName: String, simplifiedDocument: PDDocumentSimple) = {
+    val file = new File(fileName)
+    val bw = new BufferedWriter(new FileWriter(file))
+    bw.write(CreateTextLinesJSON(simplifiedDocument))
+    println(s"JSON file written at: ${fileName}")
+    bw.close()
+  }
+
   def getExtractionResult(document: PDDocument, pageNum: Int, bbs: List[RectangleOTL], c: Color): PDDocument = {
     val page = document.getPage(pageNum)
     CreateMarkedPDF.rectMarkedContent(document, page, bbs, c)
@@ -71,7 +82,8 @@ object CreateResultsBatch {
   }
 
   def main(args: Array[String]): Unit = {
-    val DEFAULT_LOC = "/home/szr163/Downloads/f706gsd1.pdf"
+    val DEFAULT_LOC = "/media/sagnik/OS_Install/data/pdfsamples/nuance/nuancepdf/img04192017_0002.pdf"
+    //"/home/szr163/Downloads/f706gsd1.pdf"
     //"/Users/schoudhury/hassan/C10-2042.pdf"
     val pdLoc = if (args.length > 1) args(0) else DEFAULT_LOC
 

writers/src/main/scala/edu/psu/sagnik/research/pdwriters/impl/CreateResultsBatchFiles.scala

@@ -0,0 +1,72 @@
+package edu.psu.sagnik.research.pdwriters.impl
+
+import java.io.{ BufferedWriter, File, FileWriter }
+
+import edu.psu.sagnik.research.pdsimplify.impl.ProcessDocument
+import edu.psu.sagnik.research.pdsimplify.model.PDDocumentSimple
+import edu.psu.sagnik.research.pdwriters.writers.json.CreateTextLinesJSON
+import org.apache.pdfbox.pdmodel.PDDocument
+
+/**
+ * Created by sagnik on 7/7/17.
+ */
+
+object CreateResultsBatchFiles {
+
+  def writeTextLinesJson(fileName: String, simplifiedDocument: PDDocumentSimple) = {
+    val file = new File(fileName)
+    val bw = new BufferedWriter(new FileWriter(file))
+    bw.write(CreateTextLinesJSON(simplifiedDocument))
+    println(s"JSON file written at: ${fileName}")
+    bw.close()
+  }
+
+  import scala.util.matching.Regex
+  def recursiveListFiles(f: File, r: Regex): Array[File] = {
+    val these = f.listFiles
+    val good = these.filter(f => r.findFirstIn(f.getName).isDefined)
+    good ++ these.filter(_.isDirectory).flatMap(recursiveListFiles(_, r))
+  }
+
+  /*
+  def main(args: Array[String]): Unit = {
+    val DEFAULT_DIR = "/media/sagnik/OS_Install/data/pdfsamples/nuance/nuancepdf/"
+    //"/home/szr163/Downloads/f706gsd1.pdf"
+    //"/Users/schoudhury/hassan/C10-2042.pdf"
+    val pdDir = if (args.length > 1) args(0) else DEFAULT_DIR
+
+    val docs = recursiveListFiles(new File(pdDir), "pdf".r).map(_.getAbsolutePath)
+    docs.foreach { x =>
+      val document = PDDocument.load(new File(x))
+      val jsonLoc = s"${x.substring(0, x.length - 3)}json"
+      val simplifiedDocument = ProcessDocument(document)
+      writeTextLinesJson(jsonLoc, simplifiedDocument)
+      document.close()
+    }
+  }
+*/
+
+  def main(args: Array[String]): Unit = {
+    val DEFAULT_DIR = "/media/sagnik/OS_Install/data/pdfsamples/nuance/nuancepdf/"
+    //"/home/szr163/Downloads/f706gsd1.pdf"
+    //"/Users/schoudhury/hassan/C10-2042.pdf"
+    val pdDir = if (args.length > 1) args(0) else DEFAULT_DIR
+
+    val docs = recursiveListFiles(new File(pdDir), "pdf".r).map(_.getAbsolutePath)
+    docs.foreach { x =>
+      var document = PDDocument.load(new File(x))
+      val simplifiedDocument = ProcessDocument(document)
+      (0 until simplifiedDocument.pages.size).foreach(pageNum =>
+        document = CreateResultsBatch.createMarkedResult(
+          document,
+          simplifiedDocument,
+          pageNum = pageNum,
+          pdLoc = x
+        ))
+      document.save(s"${x.dropRight(4)}-marked.pdf")
+      println(s"[marked document written to]: ${x.dropRight(4)}-marked.pdf")
+      document.close()
+      document.close()
+    }
+  }
+}

writers/src/main/scala/edu/psu/sagnik/research/pdwriters/impl/ShowResults.scala

@@ -24,12 +24,13 @@ object ShowResults {
   def printExtractionResult(pdLoc: String, pageNum: Int, bbs: List[RectangleOTL], c: Color, qualifier: String) = {
     val document = PDDocument.load(new File(pdLoc))
     val page = document.getPage(pageNum)
-    CreateMarkedPNG(pdLoc, document, pageNum, page, bbs, c, qualifier)
+    CreateMarkedPDF(pdLoc, document, pageNum, page, bbs, c, qualifier)
     logger.fine(s"created ${qualifier.substring(0, qualifier.length - 1)} marked PDF")
   }
 
   def main(args: Array[String]): Unit = {
-    val DEFAULT_LOC = "src/test/resources/test1.pdf"
+    val DEFAULT_LOC =
+      "src/test/resources/test1.pdf"
     //"/Users/schoudhury/hassan/C10-2042.pdf"
     val DEFAULT_PAGE_NUM = 5
 

writers/src/main/scala/edu/psu/sagnik/research/pdwriters/writers/json/CreateJSON.scala

@@ -91,6 +91,7 @@ object CreateJSON {
                         ("content" -> pp.content)
                       )
                   })
+
             )
         }
 

writers/src/main/scala/edu/psu/sagnik/research/pdwriters/writers/json/CreateTextLinesJSON.scala

@@ -0,0 +1,64 @@
+package edu.psu.sagnik.research.pdwriters.writers.json
+
+import java.awt.geom.Point2D
+
+import edu.psu.sagnik.research.data.RectangleOTL
+import edu.psu.sagnik.research.pdsimplify.model.PDDocumentSimple
+import edu.psu.sagnik.research.pdsimplify.path.model.PDLine
+import org.json4s.JsonDSL._
+import org.json4s.JsonDSL._
+import org.json4s.jackson.JsonMethods._
+
+/**
+ * Created by schoudhury on 7/6/16.
+ */
+object CreateTextLinesJSON {
+
+  def precReduce(d: Float): Float = BigDecimal(d).setScale(2, BigDecimal.RoundingMode.HALF_UP).toFloat
+
+  def precReduce(point: Point2D.Float): List[Float] = List(precReduce(point.x), precReduce(point.y))
+
+  def precReduce(r: RectangleOTL): List[Float] = List(precReduce(r.xTopLeft), precReduce(r.yTopLeft), precReduce(r.widthRight), precReduce(r.heightDown))
+
+  case class PathStyle(
+    fill: Option[String],
+    fillRule: Option[String],
+    fillOpacity: Option[String],
+    stroke: Option[String],
+    strokeWidth: Option[String],
+    strokeLineCap: Option[String],
+    strokeLineJoin: Option[String],
+    strokeMiterLimit: Option[String],
+    strokeDashArray: Option[String],
+    strokeDashOffset: Option[String],
+    strokeOpacity: Option[String]
+  )
+
+  def apply(pDS: PDDocumentSimple): String = {
+    val jsonContent =
+      "pages" ->
+        pDS.pages.map {
+          p =>
+            (
+              ("pageNumber" -> p.pNum) ~
+              ("pageBB" -> precReduce(p.bb)) ~
+              ("textparagraphs" -> p.paragraphs.map {
+                pp =>
+                  (
+                    ("bb" -> precReduce(pp.bb)) ~
+                    ("textlines" -> pp.tLines.map {
+                      tl =>
+                        (
+                          ("bb" -> precReduce(tl.bb)) ~
+                          ("content" -> tl.content)
+                        )
+                    })
+                  )
+              })
+
+            )
+        }
+    pretty(render(jsonContent))
+  }
+
+}