Merge pull request #22 from remy/master

Add support for rounding to end of sentence
cgiffard · Sep 2, 2014 · 2eadf3e · 2eadf3e
2 parents e65fe59 + 5bf701a
commit 2eadf3e
Show file tree

Hide file tree

Showing 2 changed files with 64 additions and 8 deletions.
diff --git a/index.js b/index.js
@@ -3,13 +3,24 @@ var XRegexp = require('xregexp').XRegExp;
 (function (exportTo) {
     "use strict";
 
+    // Define our parse states
+    var PARSER_UNINITIALISED = 0,
+        PARSER_TAG_COMMENCED = 1,
+        PARSER_TAG_STRING = -1,
+        PARSER_TAG_STRING_SINGLE = -2,
+        PARSER_COMMENT = -3;
+
     // Nodes which should be considered implicitly self-closing
     // Taken from http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#void-elements
     var voidElements = [
         "area", "base", "br", "col", "command", "embed", "hr", "img", "input",
         "keygen", "link", "meta", "param", "source", "track", "wbr"
     ];
 
+    var sentenceTerminatorElements = [
+        "br", "p", "ol", "ul", "li", "div", "form", "article", "section", "body", "title", "h1", "h2", "h3", "h4", "h5", "h6", "header", "footer", "main", "dd", "dt", "pre", "figure", "figcapture", "td", "th"
+    ];
+
     var downsize = function (text, inputOptions, offset) {
         var stack = [],
             pointer = 0,
@@ -34,6 +45,14 @@ var XRegexp = require('xregexp').XRegExp;
             Number(options.characters);
         options.limit = isNaN(options.limit) ? Infinity : options.limit;
 
+        function peekTag() {
+            var peek = null;
+            text.slice(pointer).replace(/^(?:<[/]?([a-z]+)?)>/i, function (all, match, position) {
+                peek = match;
+            });
+            return peek;
+        }
+
         function isAtLimit() {
             var stackIndex = 0;
 
@@ -42,6 +61,23 @@ var XRegexp = require('xregexp').XRegExp;
                 return false;
             }
 
+            if (options.round) {
+                if (parseState !== PARSER_TAG_COMMENCED) {
+                    return false;
+                }
+
+                // If we are at a PARSER_TAG_COMMENCED then sniff the tag and check
+                // whether it's a sentence terminator
+                if (parseState === PARSER_TAG_COMMENCED) {
+                    var tag = peekTag();
+                    if (~sentenceTerminatorElements.indexOf(tag)) {
+                        return true;
+                    } else {
+                        return false;
+                    }
+                }
+            }
+
             // If we've got no special context to retain, do an early return.
             if (!options.keepContext) {
                 return true;
@@ -87,13 +123,6 @@ var XRegexp = require('xregexp').XRegExp;
             }
         }
 
-        // Define our parse states
-        var PARSER_UNINITIALISED = 0,
-            PARSER_TAG_COMMENCED = 1,
-            PARSER_TAG_STRING = -1,
-            PARSER_TAG_STRING_SINGLE = -2,
-            PARSER_COMMENT = -3;
-
         var exit = false;
         for (; pointer < text.length && !exit; pointer++) {
 
@@ -110,11 +139,11 @@ var XRegexp = require('xregexp').XRegExp;
 
                     if (parseState === PARSER_UNINITIALISED &&
                         text[pointer + 1].match(/[a-z0-9\-\_\/\!]/)) {
+                        parseState = PARSER_TAG_COMMENCED;
                         if (isAtLimit()) {
                             exit = true;
                             break;
                         }
-                        parseState = PARSER_TAG_COMMENCED;
                         tagBuffer += text[pointer];
                     }
 

diff --git a/test.js b/test.js
@@ -161,6 +161,33 @@ describe("Appending", function () {
 
 });
 
+describe("Rounding", function () {
+    it("should round a sentence up", function () {
+        downsize("<p>abcdefghij</p><p>klmnop</p><p>qrs</p>", {characters: 15, round: true})
+            .should.equal("<p>abcdefghij</p><p>klmnop</p>");
+    });
+
+    it("should handle sentences shorter than required", function () {
+        downsize("<p>here's some text.</p>", {words: 5, round: true})
+            .should.equal("<p>here's some text.</p>");
+    });
+
+    it("should round up to end of sentence, not just next tag", function () {
+        downsize("<p>here's <em>some</em> text.</p>", {characters: 2, round: true})
+            .should.equal("<p>here's <em>some</em> text.</p>");
+    });
+
+    it("should not have trailing empty tags", function () {
+        downsize("<p>characters</p><i>what</i>", {characters: 10, round: true})
+            .should.equal("<p>characters</p>");
+    });
+
+    it("should scoop up to end of text if no closing paragraph found", function () {
+        downsize("<p>characters</p><i>what</i>", {characters: 11, round: true})
+            .should.equal("<p>characters</p><i>what</i>");
+    });
+});
+
 describe("Performance", function () {
     var perfTestSeed = "";
     for (var i = 0; i < 1000000; i++) {