From 4ad2de797f42983ed9a26cf09358fe66bb495601 Mon Sep 17 00:00:00 2001
From: Dan Stillman <dstillman@zotero.org>
Date: Fri, 25 Dec 2009 23:39:54 +0000
Subject: [PATCH] Restores proper copying support for notes, with separate
 output for text, rich text, and HTML-code-as-text (via Copy HTML pref)

New hidden prefs:

- export.quickCopy.compatibility.indentBlockquotes (default: true)

Word and TextEdit don't indent blockquotes on their own and need this, but OO gets it right, so this results in an extra indent

- export.quickCopy.compatibility.word (default: false)

Adds Word Normal style to paragraphs and adds double-spacing, but OO inserts the conditional style code as a document comment

- quickCopy.quoteBlockquotes.plainText (default: true)
- quickCopy.quoteBlockquotes.richText (default: true)

Adds quotes around blockquote paragraphs, since loss of attribution is serious (and who knows whether all apps that accept rich text even support indenting)

Not enabled for HTML code copying, since CSS can be used to render blockquotes as desired


Fixes #1397, toggle HTML tags on and off for quick copy of notes
---
 chrome/content/zotero/xpcom/quickCopy.js | 194 ++++++++++++++++++++---
 1 file changed, 174 insertions(+), 20 deletions(-)

diff --git a/chrome/content/zotero/xpcom/quickCopy.js b/chrome/content/zotero/xpcom/quickCopy.js
index 22daf86f20..767ebce3f0 100644
--- a/chrome/content/zotero/xpcom/quickCopy.js
+++ b/chrome/content/zotero/xpcom/quickCopy.js
@@ -165,31 +165,185 @@ Zotero.QuickCopy = new function() {
 			// If all notes, export full content
 			if (allNotes) {
 				var content = [];
-				for (var i=0; i<notes.length; i++) {
-					content.push(notes[i].getNote());
-				}
-				
 				default xml namespace = '';
-				var html = <div/>;
-				for (var i=0; i<content.length; i++) {
-					var p = <p>{content[i]}</p>;
-					p.@style = 'white-space: pre-wrap';
-					html.p += p;
+				XML.prettyPrinting = false;
+				XML.ignoreWhitespace = false;
+				var htmlXML = <div class="zotero-notes"/>;
+				for (var i=0; i<notes.length; i++) {
+					var noteContent = notes[i].getNote();
+					try {
+						var noteDiv = new XML('<div class="zotero-note">' + noteContent + '</div>');
+					}
+					catch (e) {
+						Zotero.debug(e);
+						Zotero.debug("Couldn't parse note as HTML -- try as CDATA", 2);
+						try {
+							var noteDiv = new XML('<div class="zotero-note"><pre><![CDATA[' + noteContent + ']]></pre></div>');
+						}
+						catch (e) {
+							Zotero.debug("Skipping note", 2);
+							continue;
+						}
+					}
+					
+					htmlXML.div += noteDiv;
 				}
 				
-				html = html.toXMLString();
+				// Raw HTML output
+				var html = htmlXML.toXMLString();
+				
+				var textXML = htmlXML.copy();
+				
+				//
+				// Text-only adjustments
+				//
+				
+				if (Zotero.Prefs.get('export.quickCopy.quoteBlockquotes.plainText')) {
+					// Add quotes around blockquote paragraphs
+					var nodes = textXML..blockquote;
+					for (var i in nodes) {
+						for (var j=0, len=nodes[i].p.length(); j<len; j++) {
+							var children = nodes[i].p[j].children();
+							children[0] = '\u201c' + children[0];
+							// Add closing double quotes to final paragraph
+							if (j == len - 1) {
+								children[children.length()-1] +=  '\u201d';
+							}
+						}
+					}
+				}
+				
+				// Replace tags with characters, like Thunderbird
+				//
+				// (It'd be nice to use the OutputFormatted mode of the native
+				// plaintext serializer, which appears to be what TB uses,
+				// but I wasn't able to get it to work from JS.)
+				var map = {
+					strong: "*",
+					em: "/"
+				}
+				for (var tag in map) {
+					var nodes = textXML.descendants(tag);
+					for (var i in nodes) {
+						var children = nodes[i].children();
+						children[0] = map[tag] + children[0];
+						children[children.length()-1] += map[tag];
+					}
+				}
+				
+				// Replace span styles with characters
+				var spanStyleMap = [
+					{
+						re: /text-decoration:\s*underline/,
+						repl: "_"
+					}
+				];
+				for each(var style in spanStyleMap) {
+					var nodes = textXML..span.(style.re.test(@style));
+					for (var i in nodes) {
+						var children = nodes[i].children();
+						children[0] = style.repl + children[0];
+						children[children.length()-1] += style.repl;
+					}
+				}
+				
+				//
+				// And add spaces for indents
+				//
+				// Placeholder for 4 spaces in final output
+				var ztab = "%%ZOTEROTAB%%";
+				var p = textXML..p;
+				for (var i in p) {
+					var children = p[i].children();
+					// TODO: Be a bit smarter and more flexible about this
+					if (p[i].@style.indexOf("padding-left: 30px") != -1) {
+						children[0] = ztab + children[0];
+					}
+					else if (p[i].@style.indexOf("padding-left: 60px") != -1) {
+						children[0] = ztab + ztab + children[0];
+					}
+					else if (p[i].@style.indexOf("padding-left: 90px") != -1) {
+						children[0] = ztab + ztab + ztab + children[0];
+					}
+					else if (p[i].@style.indexOf("padding-left: 120px") != -1) {
+						children[0] = ztab + ztab + ztab + ztab + children[0];
+					}
+				}
+				
+				var text = Zotero.Utilities.prototype.unescapeHTML(textXML.toXMLString());
+				text = text.replace(new RegExp(ztab, "g"), "    ");
+				
+				if (text.trim) {
+					text = text.trim();
+				}
+				// TODO: Remove once >=Fx3.5
+				else {
+					text = Zotero.Utilities.prototype.trim()
+				}
+				
+				//
+				// Adjustments for the HTML copied to the clipboard
+				//
+				
+				// Everything seems to like margin-left better than padding-left
+				var p = htmlXML..p;
+				for (var i in p) {
+					var children = p[i].children();
+					if (p[i].@style.toString().indexOf("padding-left") != -1) {
+						p[i].@style = p[i].@style.toString().replace("padding-left", "margin-left");
+					}
+				}
+				
+				// Add quotes around blockquote paragraphs
+				if (Zotero.Prefs.get('export.quickCopy.quoteBlockquotes.richText')) {
+					var nodes = htmlXML..blockquote;
+					for (var i in nodes) {
+						for (var j=0, len=nodes[i].p.length(); j<len; j++) {
+							var children = nodes[i].p[j].children();
+							children[0] = '\u201c' + children[0];
+							// Add closing double quotes to final paragraph
+							if (j == len - 1) {
+								children[children.length()-1] +=  '\u201d';
+							}
+						}
+					}
+				}
+				
+				// Word and TextEdit don't indent blockquotes on their own and need this
+				//
+				// OO gets it right, so this results in an extra indent
+				if (Zotero.Prefs.get('export.quickCopy.compatibility.indentBlockquotes')) {
+					var p = htmlXML..blockquote.p;
+					for (var i in p) {
+						var children = p[i].children();
+						p[i].@style = "margin-left: 30px";
+					}
+				}
+				
+				// Add Word Normal style to paragraphs and add double-spacing
+				//
+				// OO inserts the conditional style code as a document comment
+				if (Zotero.Prefs.get('export.quickCopy.compatibility.word')) {
+					var nodes = htmlXML..p;
+					for (var i in nodes) {
+						nodes[i].@class = "msoNormal";
+					}
+					var copyHTML = "<!--[if gte mso 0]>"
+									+ "<style>"
+									+ "p { margin-top:.1pt;margin-right:0in;margin-bottom:.1pt;margin-left:0in; line-height: 200%; }"
+									+ "li { margin-top:.1pt;margin-right:0in;margin-bottom:.1pt;margin-left:0in; line-height: 200%; }"
+									+ "blockquote p { margin-left: 11px; margin-right: 11px }"
+									+ "</style>"
+									+ "<![endif]-->\n"
+									+ htmlXML.toXMLString();
+				}
+				else {
+					var copyHTML = htmlXML.toXMLString();
+				}
 				
-				// Don't copy HTML, since we don't have rich-text notes and
-				// copying HTML on Windows just loses newlines (due to
-				// unsupported white-space: pre-wrap in Word and OO).
-				/*
 				var content = {
-					text: contentType == "html" ? html : content.join('\n\n\n'),
-					html: html
-				};
-				*/
-				var content = {
-					text: contentType == "html" ? html : content.join('\n\n\n')
+					text: contentType == "html" ? html : text,
+					html: copyHTML
 				};
 				
 				return content;