From 30ff60a44e45b6782a1db46063ca8c4bf0b49f94 Mon Sep 17 00:00:00 2001
From: Nat Wilson <natw@fortyninemaps.com>
Date: Sun, 29 Oct 2017 10:34:40 -0700
Subject: [PATCH] Insert a space when TJ string is offset by more than a
 threshold

That threshold is hard-coded as -100 (is this font specific?)
---
 pdf/contentstream/contentstream.go      | 13 +++++++++++--
 pdf/contentstream/contentstream_test.go | 25 +++++++++++++++++++++++++
 2 files changed, 36 insertions(+), 2 deletions(-)
 create mode 100644 pdf/contentstream/contentstream_test.go

diff --git a/pdf/contentstream/contentstream.go b/pdf/contentstream/contentstream.go
index f694451f..71fe51e8 100644
--- a/pdf/contentstream/contentstream.go
+++ b/pdf/contentstream/contentstream.go
@@ -131,8 +131,17 @@ func (this *ContentStreamParser) ExtractText() (string, error) {
 				return "", fmt.Errorf("Invalid parameter type, no array (%T)", op.Params[0])
 			}
 			for _, obj := range *paramList {
-				if strObj, ok := obj.(*PdfObjectString); ok {
-					txt += string(*strObj)
+				switch v := obj.(type) {
+				case *PdfObjectString:
+					txt += string(*v)
+				case *PdfObjectFloat:
+					if *v < -100 {
+						txt += " "
+					}
+				case *PdfObjectInteger:
+					if *v < -100 {
+						txt += " "
+					}
 				}
 			}
 		} else if inText && op.Operand == "Tj" {
diff --git a/pdf/contentstream/contentstream_test.go b/pdf/contentstream/contentstream_test.go
new file mode 100644
index 00000000..5237eb10
--- /dev/null
+++ b/pdf/contentstream/contentstream_test.go
@@ -0,0 +1,25 @@
+package contentstream
+
+import (
+	"testing"
+)
+
+func TestOperandTJSpacing(t *testing.T) {
+
+	content := `BT
+	[(are)-328(h)5(ypothesized)-328(to)-327(in\003uence)-328(the)-328(stability)-328(of)-328(the)-328(upstream)-327(glaciers,)-328(and)-328(thus)-328(of)-328(the)-328(entire)-327(ice)-328(sheet)]TJ
+	ET`
+	referenceText := "are hypothesized to in\003uence the stability of the upstream glaciers, and thus of the entire ice sheet"
+
+	cStreamParser := NewContentStreamParser(content)
+
+	text, err := cStreamParser.ExtractText()
+	if err != nil {
+		t.Error()
+	}
+
+	if text != referenceText {
+		t.Fail()
+	}
+
+}