From c9c929e2d6ca39a61fa1dc6f636fe8eb884bb488 Mon Sep 17 00:00:00 2001
From: matt-farmer <solversurfer@icloud.com>
Date: Mon, 16 Apr 2018 14:21:23 +1000
Subject: [PATCH] Maintain paragraphing, remove escaped chars

Maintains the original paragraph strucures
Handles escaped chars e.g emojis
---
 ...sample.csv => writing-response-sample.csv} |  0
 nap-writing-print/html-sanitizer.go           | 19 +++++++++++++++++--
 nap-writing-print/script-writer.go            | 16 +++++++++++++---
 3 files changed, 30 insertions(+), 5 deletions(-)
 rename app/nap-writing-print/in/{sample.csv => writing-response-sample.csv} (100%)

diff --git a/app/nap-writing-print/in/sample.csv b/app/nap-writing-print/in/writing-response-sample.csv
similarity index 100%
rename from app/nap-writing-print/in/sample.csv
rename to app/nap-writing-print/in/writing-response-sample.csv
diff --git a/nap-writing-print/html-sanitizer.go b/nap-writing-print/html-sanitizer.go
index 7934865..f7c3e79 100644
--- a/nap-writing-print/html-sanitizer.go
+++ b/nap-writing-print/html-sanitizer.go
@@ -4,6 +4,8 @@ package nap_writing_print
 
 import (
 	"context"
+	"html"
+	"strings"
 
 	"github.com/microcosm-cc/bluemonday"
 )
@@ -25,8 +27,21 @@ func createHtmlSanitizer(ctx context.Context, in <-chan map[string]string) (
 			rmap := rMap
 
 			p := bluemonday.StrictPolicy()
-			html := p.Sanitize(rmap["Item Response"])
-			rmap["Item Response"] = html
+			// keep paragraph markers to preserve layout in pdf printing
+			p.AllowElements("p")
+			sanitizedHtml := p.Sanitize(rmap["Item Response"])
+			// now unescape the html for speech marks, apostrophe's etc. for printing
+			unescapedHtml := html.UnescapeString(sanitizedHtml)
+
+			// remove unnecessary characters
+			// remove non-breaking spaces and line feeds & backticks
+			noBrHtml := strings.Replace(unescapedHtml, "&nbsp;", " ", -1)
+			noLfHtml := strings.Replace(noBrHtml, "\\n", "", -1)
+
+			// remove end-of para markers, not needed for pdf print
+			noEndParaHtml := strings.Replace(noLfHtml, "</p>", "", -1)
+
+			rmap["Item Response"] = noEndParaHtml
 
 			select {
 			case out <- rmap:
diff --git a/nap-writing-print/script-writer.go b/nap-writing-print/script-writer.go
index 2cd6ba2..3b84893 100644
--- a/nap-writing-print/script-writer.go
+++ b/nap-writing-print/script-writer.go
@@ -35,13 +35,13 @@ func createScriptWriter(ctx context.Context, in <-chan map[string]string) (<-cha
 			doc.SetColor("black").
 				SetFont("Helvetica-Bold", 24).
 				SetXY(3.5, 2.7).DrawText(rmap["Anonymised Id"])
-			//
-			// draw the column of text
-			var colText = strings.Replace(rmap["Item Response"], "\\n", " ", -1)
+
+			colText := splitParas(rmap["Item Response"])
 			doc.SetColor("black").
 				SetFont("Helvetica", 12).
 				// DrawUnitGrid().
 				DrawTextInBox(3.5, 4, 12, 28, "LT", colText)
+
 			//
 			// save the files
 			doc.SaveFile(outputFile1)
@@ -57,3 +57,13 @@ func createScriptWriter(ctx context.Context, in <-chan map[string]string) (<-cha
 	return errc, nil
 
 }
+
+//
+// use the paragraph markers that have been left in the text
+// to insert spacing characters.
+//
+func splitParas(fullText string) string {
+
+	return strings.Replace(fullText, "<p>", "\n\n", -1)
+
+}