From c08fadfaa93b22016c05cdb3905bd4973e91339b Mon Sep 17 00:00:00 2001
From: Ben Alpert <spicyjalapeno@gmail.com>
Date: Tue, 16 Jul 2013 22:00:54 -0700
Subject: [PATCH] Parse single superscripts and subscripts

Summary:
Add the ability to parse lone `^x` and `_y`, etc. This basically just
involves more checking of cases in the increasingly ugly `parseAtom` function.
Also, now we manually check for the cases of double superscripts and
subscripts.

Test Plan: Make sure the tests pass. Make sure things like `^x` and `_y` parse.

Reviewers: emily

Reviewed By: emily

Differential Revision: http://phabricator.khanacademy.org/D3095
---
 Parser.js           | 77 ++++++++++++++++++++++++++-------------------
 katex.js            |  4 +++
 test/katex-tests.js | 38 ++++++++++++++++++++--
 3 files changed, 84 insertions(+), 35 deletions(-)

diff --git a/Parser.js b/Parser.js
index a971f8ea2..6361e66aa 100644
--- a/Parser.js
+++ b/Parser.js
@@ -110,42 +110,55 @@ Parser.prototype.parseSubscript = function(pos) {
 Parser.prototype.parseAtom = function(pos) {
     // Parse the nucleus
     var nucleus = this.parseGroup(pos);
+    var nextPos = pos;
+    var nucleusNode;
+
     if (nucleus) {
-        // Now, we try to parse a subscript or a superscript. If one of those
-        // succeeds, we then try to parse the opposite one, and depending on
-        // whether that succeeds, we return the correct type.
-        var sup, sub;
-        if (sup = this.parseSuperscript(nucleus.position)) {
-            if (sub = this.parseSubscript(sup.position)) {
-                return new ParseResult(
-                    new ParseNode("supsub",
-                        {base: nucleus.result, sup: sup.result,
-                            sub: sub.result}),
-                    sub.position);
-            } else {
-                return new ParseResult(
-                    new ParseNode("sup",
-                        {base: nucleus.result, sup: sup.result}),
-                    sup.position);
+        nextPos = nucleus.position;
+        nucleusNode = nucleus.result;
+    }
+
+    var sup;
+    var sub;
+
+    // Now, we try to parse a subscript or a superscript (or both!), and
+    // depending on whether those succeed, we return the correct type.
+    while (true) {
+        var node;
+        if ((node = this.parseSuperscript(nextPos))) {
+            if (sup) {
+                throw "Parse error: Double superscript";
             }
-        } else if (sub = this.parseSubscript(nucleus.position)) {
-            if (sup = this.parseSuperscript(sub.position)) {
-                return new ParseResult(
-                    new ParseNode("supsub",
-                        {base: nucleus.result, sup: sup.result,
-                            sub: sub.result}),
-                    sup.position);
-            } else {
-                return new ParseResult(
-                    new ParseNode("sub",
-                        {base: nucleus.result, sub: sub.result}),
-                    sub.position);
-            }
-        } else {
-            return nucleus;
+            nextPos = node.position;
+            sup = node.result;
+            continue;
         }
+        if ((node = this.parseSubscript(nextPos))) {
+            if (sub) {
+                throw "Parse error: Double subscript";
+            }
+            nextPos = node.position;
+            sub = node.result;
+            continue;
+        }
+        break;
+    }
+
+    if (sup && sub) {
+        return new ParseResult(
+            new ParseNode("supsub", {base: nucleusNode, sup: sup,
+                    sub: sub}),
+            nextPos);
+    } else if (sup) {
+        return new ParseResult(
+            new ParseNode("sup", {base: nucleusNode, sup: sup}),
+            nextPos);
+    } else if (sub) {
+        return new ParseResult(
+            new ParseNode("sub", {base: nucleusNode, sub: sub}),
+            nextPos);
     } else {
-        return null;
+        return nucleus;
     }
 }
 
diff --git a/katex.js b/katex.js
index 48f54c2fc..c42c50104 100644
--- a/katex.js
+++ b/katex.js
@@ -27,6 +27,10 @@ var makeSpan = function(className, children) {
 };
 
 var buildGroup = function(style, color, group, prev) {
+    if (!group) {
+        return makeSpan();
+    }
+
     if (group.type === "mathord") {
         return makeSpan("mord" + color, [mathit(group.value)]);
     } else if (group.type === "textord") {
diff --git a/test/katex-tests.js b/test/katex-tests.js
index 213808745..f49691635 100644
--- a/test/katex-tests.js
+++ b/test/katex-tests.js
@@ -164,6 +164,24 @@ describe("A subscript and superscript parser", function() {
         }).not.toThrow();
     });
 
+    it("should not fail when there is no nucleus", function() {
+        expect(function() {
+            parseTree("^3");
+        }).not.toThrow();
+
+        expect(function() {
+            parseTree("_2");
+        }).not.toThrow();
+
+        expect(function() {
+            parseTree("^3_2");
+        }).not.toThrow();
+
+        expect(function() {
+            parseTree("_2^3");
+        }).not.toThrow();
+    });
+
     it("should produce sups for superscript", function() {
         var parse = parseTree("x^2")[0];
 
@@ -207,16 +225,30 @@ describe("A subscript and superscript parser", function() {
         expect(parseA).toEqual(parseB);
     });
 
-    it("should not parse x^x^x", function() {
+    it("should not parse double subscripts or superscripts", function() {
         expect(function() {
             parseTree("x^x^x");
         }).toThrow();
-    });
 
-    it("should not parse x_x_x", function() {
         expect(function() {
             parseTree("x_x_x");
         }).toThrow();
+
+        expect(function() {
+            parseTree("x_x^x_x");
+        }).toThrow();
+
+        expect(function() {
+            parseTree("x_x^x^x");
+        }).toThrow();
+
+        expect(function() {
+            parseTree("x^x_x_x");
+        }).toThrow();
+
+        expect(function() {
+            parseTree("x^x_x^x");
+        }).toThrow();
     });
 
     it("should work correctly with {}s", function() {