Improve summary parser

6 files changed, 103 insertions, 47 deletions

diff --git a/packages/gitbook-html/lib/dom.js b/packages/gitbook-html/lib/dom.js
index 2c2eaf7..df4de90 100644
--- a/packages/gitbook-html/lib/dom.js
+++ b/packages/gitbook-html/lib/dom.js
@@ -3,12 +3,18 @@ var cheerio = require('cheerio');
 
 // Parse an HTML string and return its content
 function parse(html) {
-    var $ = cheerio.load('<div>'+html+'</div>');
+    var $ = cheerio.load(html);
     var $el = $('html, body').first();
 
     return $el.length > 0? $el : $;
 }
 
+// Return main element
+function root($) {
+    var $el = $('html, body, > div').first();
+    return $el.length > 0? $el : $.root();
+}
+
 // Return text node of an element
 function textNode($el) {
     return _.reduce($el.children, function(text, e) {
@@ -19,5 +25,6 @@ function textNode($el) {
 
 module.exports = {
     parse: parse,
-    textNode: textNode
+    textNode: textNode,
+    root: root
 };
diff --git a/packages/gitbook-html/lib/glossary.js b/packages/gitbook-html/lib/glossary.js
index 648ba1a..9d3799b 100755
--- a/packages/gitbook-html/lib/glossary.js
+++ b/packages/gitbook-html/lib/glossary.js
@@ -7,9 +7,10 @@ function parseGlossary(html) {
 
     var entries = [];
 
-    $("h2").each(function() {
+    $('h2').each(function() {
         var $heading = $(this);
-        var $p = $heading.next();
+        var $next = $heading.next()
+        var $p =  $next.is('p')? $next.first() : $next.find('p').first();
 
         var entry = {};
 
diff --git a/packages/gitbook-html/lib/index.js b/packages/gitbook-html/lib/index.js
index a7c478c..0e67c94 100755
--- a/packages/gitbook-html/lib/index.js
+++ b/packages/gitbook-html/lib/index.js
@@ -1,8 +1,33 @@
+var _ = require('lodash');
 
-module.exports = {
+var htmlParser = {
     summary: require("./summary"),
     glossary: require("./glossary"),
     langs: require("./langs"),
     readme: require("./readme"),
     page: require("./page")
 };
+
+// Compose a function with a transform function for the first args
+function compose(toHTML, fn) {
+    return function() {
+        var args = _.toArray(arguments);
+        args[0] = toHTML(args[0]);
+
+        return fn.apply(null, args);
+    }
+}
+
+// Create a GitBook parser
+function createParser(toHTML) {
+    return {
+        summary: compose(toHTML, htmlParser.summary),
+        glossary: compose(toHTML, htmlParser.glossary),
+        langs: compose(toHTML, htmlParser.langs),
+        readme: compose(toHTML, htmlParser.readme),
+        page: compose(toHTML, htmlParser.page)
+    }
+}
+
+module.exports = htmlParser;
+module.exports.createParser = createParser;
diff --git a/packages/gitbook-html/lib/langs.js b/packages/gitbook-html/lib/langs.js
index 1042dcb..270a9f6 100755
--- a/packages/gitbook-html/lib/langs.js
+++ b/packages/gitbook-html/lib/langs.js
@@ -1,9 +1,9 @@
 var _ = require('lodash');
-var parseEntries = require('./summary').entries;
+var parseSummary = require('./summary');
 
 // HTML -> Languages
 function parseLangs(content) {
-    return parseEntries(content);
+    return parseSummary(content).parts[0].articles;
 }
 
 // Languages -> HTML
diff --git a/packages/gitbook-html/lib/page.js b/packages/gitbook-html/lib/page.js
index e687050..6f056fc 100755
--- a/packages/gitbook-html/lib/page.js
+++ b/packages/gitbook-html/lib/page.js
@@ -2,8 +2,10 @@ var Q = require('q');
 var _ = require('lodash');
 
 // HTML -> HTML
-function parsePage(src) {
-    return src;
+function parsePage(html) {
+    return {
+        content: html
+    };
 }
 
 module.exports = parsePage;
diff --git a/packages/gitbook-html/lib/summary.js b/packages/gitbook-html/lib/summary.js
index 1e2d63d..e71d6b5 100755
--- a/packages/gitbook-html/lib/summary.js
+++ b/packages/gitbook-html/lib/summary.js
@@ -1,14 +1,17 @@
 var _ = require('lodash');
 var dom = require('./dom');
 
+var SELECTOR_LIST = '.olist > ol, ol, ul';
+var SELECTOR_LINK = 'a, p > a';
+
+var BL = '\n';
 
 // parse a ul list and return list of chapters recursvely
 function parseList($ul, $) {
     var articles = [];
 
-    $ul.children('>li').each(function() {
+    $ul.children('li').each(function() {
         var article = {};
-
         var $li = $(this);
 
         // Get text for the entry
@@ -16,14 +19,14 @@ function parseList($ul, $) {
         article.title = $p.text() ||  dom.textNode($li.get(0));
 
         // Parse link
-        var $a = $li.find('> a, > p > a');
+        var $a = $li.children(SELECTOR_LINK);
         if ($a.length > 0) {
             article.title = $a.first().text();
             article.path = $a.attr('href').replace(/\\/g, '/').replace(/^\/+/, '')
         }
 
         // Sub articles
-        var $sub = $li.children('> .olist > ol, > ol, > ul');
+        var $sub = $li.children(SELECTOR_LIST).first();
         article.articles = parseList($sub, $);
 
         articles.push(article);
@@ -32,55 +35,73 @@ function parseList($ul, $) {
     return articles;
 }
 
-// Return a list of entries in a div
-function parseEntries (html) {
+// HTML -> Summary
+function parseSummary(html) {
     var $ = dom.parse(html);
-    var chapters = parseList($("> ol, > ul").first(), $);
-    return chapters;
-}
+    var $root = dom.root($);
 
-// HTML -> Summary
-function parseSummary(src) {
-    var chapters = parseEntries(src);
+    var $lists = $root.children(SELECTOR_LIST);
+    var parts = [];
+
+    $lists.each(function() {
+        var $list = $(this);
+
+        parts.push({
+            articles: parseList($(SELECTOR_LIST).first(), $)
+        });
+    });
 
     return {
-        chapters: chapters
+        parts: parts
     };
 }
 
 // Summary -> HTML
-function summaryToText(summary) {
-    var bl = '\n';
-
-    var _base = function(article) {
-        if (article.path) {
-            return '<a href="'+article.path+'">'+article.title+'</a>';
-        } else {
-            return article.title;
-        }
-    };
+function textPrefix(d) {
+    return Array(d*4).join(' ');
+}
 
-    var convertArticle = function(article, d) {
-        var content = Array(d+2).join(' ') + '<li>' + _base(article);
+function articleToText(article, d) {
+    var prefix = textPrefix(d);
+    var content = prefix + '<li>';
 
-        if (article.articles.length > 0) {
-            content += convertArticles(article.articles, d);
-        }
-        return content + '</li>' + bl;
-    };
+    if (article.path) {
+        content += '<a href="'+article.path+'">'+article.title+'</a>';
+    } else {
+        content += article.title;
+    }
 
-    var convertArticles = function(articles, d) {
-        var content = '<ul>' + bl;
-        _.each(articles, function(_article) {
-            content += convertArticle(_article, d + 1);
-        });
-        return content + '<ul>' + bl;
+    if (article.articles.length > 0) {
+        content += BL + articlesToText(article.articles, d) + prefix;
     }
+    content +=  '</li>' + BL;
+
+    return content;
+}
+
+function articlesToText(articles, d) {
+    var prefix = textPrefix(d);
+    var content = prefix + '<ul>' + BL;
+    _.each(articles, function(_article) {
+        content += articleToText(_article, d + 1);
+    });
+    return content + '</ul>' + BL;
+}
+
+function partsToText(part) {
+    return articlesToText(part.articles, 0) + BL + BL;
+}
+
+function summaryToText(summary) {
+    var content = '<h1>Summary</h1>' + BL;
+
+    _.each(summary.parts, function(part) {
+        content += partsToText(part);
+    });
 
-    return '<h1>Summary</h1>'+ bl+bl + convertArticles(summary.chapters, 0) + bl;
+    return content + BL;
 };
 
 
 module.exports = parseSummary;
-module.exports.entries = parseEntries;
 module.exports.toText = summaryToText;