From a4bae749953e6bb3a2fcaa8fa885cf2c97f64b47 Mon Sep 17 00:00:00 2001
From: Alex Kocharin <alex@kocharin.ru>
Date: Mon, 13 Oct 2014 22:02:42 +0400
Subject: [PATCH] Add rule to replace quotes with typographic ones

---
 lib/rules_typographer/smartquotes.js     | 110 +++++++++++++++++++++++
 lib/typographer.js                       |   1 +
 test/fixtures/remarkable/smartquotes.txt |  71 +++++++++++++++
 3 files changed, 182 insertions(+)
 create mode 100644 lib/rules_typographer/smartquotes.js
 create mode 100644 test/fixtures/remarkable/smartquotes.txt

diff --git a/lib/rules_typographer/smartquotes.js b/lib/rules_typographer/smartquotes.js
new file mode 100644
index 0000000..6d87310
--- /dev/null
+++ b/lib/rules_typographer/smartquotes.js
@@ -0,0 +1,110 @@
+// Convert straight quotation marks to typographic ones
+//
+'use strict';
+
+
+var quoteReg = /&quot;|'/g;
+var punctReg = /[-\s()\[\]]/;
+var apostrophe = '’';
+
+// This function returns true if the character at `pos`
+// could be inside a word.
+function isLetter(str, pos) {
+  if (pos < 0 || pos >= str.length) { return false; }
+  return !punctReg.test(str[pos]);
+}
+
+
+function addQuote(obj, tokenId, posId, str) {
+  if (!obj[tokenId]) { obj[tokenId] = {}; }
+  obj[tokenId][posId] = str;
+}
+
+
+module.exports = function smartquotes(typographer, state) {
+  var i, token, text, t, pos, max, thisLevel, lastSpace, nextSpace, item, canOpen, canClose, j, isSingle, fn, chars,
+      options = typographer.options,
+      replace = {},
+      tokens = state.tokens,
+      stack = [];
+
+  for (i = 0; i < tokens.length; i++) {
+    token = tokens[i];
+    thisLevel = tokens[i].level;
+
+    for (j = stack.length - 1; j >= 0; j--) {
+      if (stack[j].level <= thisLevel) { break; }
+    }
+    stack.length = j + 1;
+
+    if (token.type === 'text') {
+      text = token.content;
+      pos = 0;
+      max = text.length;
+
+      while (pos < max) {
+        quoteReg.lastIndex = pos;
+        t = quoteReg.exec(text);
+        if (!t) { break; }
+
+        lastSpace = !isLetter(text, t.index - 1);
+        pos = t.index + t[0].length;
+        isSingle = t[0] === "'";
+        nextSpace = !isLetter(text, pos);
+
+        if (!nextSpace && !lastSpace) {
+          // middle word
+          if (isSingle) {
+            addQuote(replace, i, t.index, apostrophe);
+          }
+          continue;
+        }
+
+        canOpen = !nextSpace;
+        canClose = !lastSpace;
+
+        if (canClose) {
+          // this could be a closing quote, rewind the stack to get a match
+          for (j = stack.length - 1; j >= 0; j--) {
+            item = stack[j];
+            if (stack[j].level < thisLevel) { break; }
+            if (item.single === isSingle && stack[j].level === thisLevel) {
+              item = stack[j];
+              chars = isSingle ? options.singleQuotes : options.doubleQuotes;
+              if (chars) {
+                addQuote(replace, item.token, item.start, chars[0]);
+                addQuote(replace, i, t.index, chars[1]);
+              }
+              stack.length = j;
+              canOpen = false; // should be "continue OUTER;", but eslint refuses labels :(
+              break;
+            }
+          }
+        }
+
+        if (canOpen) {
+          stack.push({
+            token: i,
+            start: t.index,
+            end: pos,
+            single: isSingle,
+            level: thisLevel
+          });
+        } else if (canClose && isSingle) {
+          addQuote(replace, i, t.index, apostrophe);
+        }
+      }
+    }
+  }
+
+  fn = function(str, pos) {
+    if (!replace[i][pos]) { return str; }
+    return replace[i][pos];
+  };
+
+  for (i = 0; i < tokens.length; i++) {
+    if (!replace[i]) { continue; }
+    quoteReg.lastIndex = 0;
+    tokens[i].content = tokens[i].content.replace(quoteReg, fn);
+  }
+};
diff --git a/lib/typographer.js b/lib/typographer.js
index 6167e4c..db53161 100644
--- a/lib/typographer.js
+++ b/lib/typographer.js
@@ -16,6 +16,7 @@ var rules = [];
 
 
 rules.push(require('./rules_typographer/replace'));
+rules.push(require('./rules_typographer/smartquotes'));
 
 
 function Typographer() {
diff --git a/test/fixtures/remarkable/smartquotes.txt b/test/fixtures/remarkable/smartquotes.txt
new file mode 100644
index 0000000..b3c7c77
--- /dev/null
+++ b/test/fixtures/remarkable/smartquotes.txt
@@ -0,0 +1,71 @@
+
+Should parse nested quotes:
+
+.
+"foo 'bar' baz"
+.
+<p>“foo ‘bar’ baz”</p>
+.
+
+.
+'foo 'bar' baz'
+.
+<p>‘foo ‘bar’ baz’</p>
+.
+
+Should not overlap quotes:
+
+.
+'foo "bar' baz"
+.
+<p>‘foo &quot;bar’ baz&quot;</p>
+.
+
+Should match quotes on the same level:
+
+.
+"foo *bar* baz"
+.
+<p>“foo <em>bar</em> baz”</p>
+.
+
+Should not match quotes on different levels:
+
+.
+*"foo* bar"
+
+"foo *bar"*
+.
+<p><em>&quot;foo</em> bar&quot;</p>
+<p>&quot;foo <em>bar&quot;</em></p>
+.
+
+.
+*"foo* bar *baz"*
+.
+<p><em>&quot;foo</em> bar <em>baz&quot;</em></p>
+.
+
+Should try and find matching quote in this case:
+
+.
+"foo "bar 'baz"
+.
+<p>&quot;foo “bar 'baz”</p>
+.
+
+Should render an apostrophe as a rsquo:
+
+.
+This isn't and can't be the best approach to implement this...
+.
+<p>This isn’t and can’t be the best approach to implement this…</p>
+.
+
+Apostrophe could end the word, that's why original smartypants replaces all of them as rsquo:
+
+.
+users' stuff
+.
+<p>users’ stuff</p>
+.