Improve gitbook -> text

5 files changed, 171 insertions, 86 deletions

diff --git a/packages/gitbook-html/lib/glossary.js b/packages/gitbook-html/lib/glossary.js
index 9d3799b..89ff135 100755
--- a/packages/gitbook-html/lib/glossary.js
+++ b/packages/gitbook-html/lib/glossary.js
@@ -23,17 +23,4 @@ function parseGlossary(html) {
     return entries;
 }
 
-// Glossary -> HTML
-function glossaryToText(glossary) {
-    var bl = '\n';
-
-    var body = _.map(glossary, function(entry) {
-        return '<h2>' + entry.name + '</h2>' + bl + bl
-        + '<p>' + entry.description + '</p>';
-    }).join(bl+bl);
-
-    return '<h1>Glossary</h1>'+bl+bl+body;
-}
-
 module.exports = parseGlossary;
-module.exports.toText = glossaryToText;
diff --git a/packages/gitbook-html/lib/index.js b/packages/gitbook-html/lib/index.js
index 0e67c94..1c2b0e0 100755
--- a/packages/gitbook-html/lib/index.js
+++ b/packages/gitbook-html/lib/index.js
@@ -1,11 +1,12 @@
 var _ = require('lodash');
+var ToText = require('./totext');
 
 var htmlParser = {
-    summary: require("./summary"),
-    glossary: require("./glossary"),
-    langs: require("./langs"),
-    readme: require("./readme"),
-    page: require("./page")
+    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
@@ -19,15 +20,22 @@ function compose(toHTML, fn) {
 }
 
 // Create a GitBook parser
-function createParser(toHTML) {
-    return {
+function createParser(toHTML, toText) {
+    var parser = {
         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)
-    }
+    };
+
+    var _toText = new ToText(toText);
+    parser.summary.toText =_toText.summary;
+    parser.langs.toText =_toText.langs;
+    parser.glossary.toText =_toText.glossary;
+
+    return parser;
 }
 
-module.exports = htmlParser;
+module.exports = createParser(_.identity);
 module.exports.createParser = createParser;
diff --git a/packages/gitbook-html/lib/langs.js b/packages/gitbook-html/lib/langs.js
index 270a9f6..035091b 100755
--- a/packages/gitbook-html/lib/langs.js
+++ b/packages/gitbook-html/lib/langs.js
@@ -6,19 +6,5 @@ function parseLangs(content) {
     return parseSummary(content).parts[0].articles;
 }
 
-// Languages -> HTML
-function langsToText(langs) {
-    var bl = '\n';
-    var content = '<h1>Languages</h1>'+bl+bl;
-
-    content += '<ul>' + bl;
-    _.each(langs, function(lang) {
-        content = content + '    <li><a href="'+lang.path+'">'+lang.title+'</a></li>'+bl;
-    });
-    content += '</ul>' + bl;
-
-    return content;
-}
-
 module.exports = parseLangs;
-module.exports.toText = langsToText;
+
diff --git a/packages/gitbook-html/lib/summary.js b/packages/gitbook-html/lib/summary.js
index 607062e..9b6b688 100755
--- a/packages/gitbook-html/lib/summary.js
+++ b/packages/gitbook-html/lib/summary.js
@@ -24,7 +24,7 @@ function parseList($ul, $) {
 
         // Get text for the entry
         var $p = $li.children('p');
-        article.title = $p.text() ||  dom.textNode($li.get(0));
+        article.title = ($p.text() ||  dom.textNode($li.get(0))).trim();
 
         // Parse link
         var $a = $li.find(SELECTOR_LINK);
@@ -37,6 +37,7 @@ function parseList($ul, $) {
         var $sub = findList($li);
         article.articles = parseList($sub, $);
 
+        if (!article.title) return;
         articles.push(article);
     });
 
@@ -64,52 +65,4 @@ function parseSummary(html) {
     };
 }
 
-// Summary -> HTML
-function textPrefix(d) {
-    return Array(d*4).join(' ');
-}
-
-function articleToText(article, d) {
-    var prefix = textPrefix(d);
-    var content = prefix + '<li>';
-
-    if (article.path) {
-        content += '<a href="'+article.path+'">'+article.title+'</a>';
-    } else {
-        content += article.title;
-    }
-
-    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 content + BL;
-};
-
-
 module.exports = parseSummary;
-module.exports.toText = summaryToText;
diff --git a/packages/gitbook-html/lib/totext.js b/packages/gitbook-html/lib/totext.js
new file mode 100644
index 0000000..13b1cd8
--- /dev/null
+++ b/packages/gitbook-html/lib/totext.js
@@ -0,0 +1,151 @@
+var _ = require('lodash');
+
+function ToText(markup) {
+    _.extend(this, markup || {});
+    _.bindAll(this);
+};
+
+// Break line
+ToText.prototype.onBL = function() {
+    return '\n';
+};
+
+ToText.prototype.onText = function(text) {
+    return text;
+};
+
+// ---- TITLES
+
+ToText.prototype.onTitleStart = function(level) {
+    return '<h'+level+'>';
+};
+ToText.prototype.onTitleEnd = function(level) {
+    return '</h'+level+'>';
+};
+
+// ---- PARAGRAPHS / SECTIONS
+ToText.prototype.onParagraphStart = function() {
+    return '<p>';
+};
+ToText.prototype.onParagraphEnd = function() {
+    return '</p>';
+};
+
+
+ToText.prototype.onSection = function() {
+    return this.onBL();
+};
+
+// ---- LINKS
+ToText.prototype.onLinkStart = function(href) {
+    return '<a href="' + href + '">';
+};
+ToText.prototype.onLinkEnd = function(href) {
+    return '</a>';
+};
+
+// ---- LISTS
+ToText.prototype.onListItemStart = function(level) {
+    return this._spaces((level + 1) * 4) + '<li>';
+};
+ToText.prototype.onListItemEnd = function(level) {
+    return this._spaces((level + 1) * 4) + '</li>' + this.onBL();
+};
+ToText.prototype.onListStart = function(level) {
+    return this._spaces(level * 4) + '<ul>' + this.onBL();
+};
+ToText.prototype.onListEnd = function(level) {
+    return this._spaces(level * 4) + '</ul>' + this.onBL();
+};
+
+// ------ LANGS
+
+ToText.prototype.langs = function(languages) {
+    var content = '';
+    content += this.onTitleStart(1) + this.onText('Languages') + this.onTitleEnd(1);
+    content += this.onSection();
+
+    content += this._summaryArticles(languages);
+
+    return content;
+};
+
+// ------ GLOSSARY
+
+ToText.prototype.glossary = function(glossary) {
+    var content = '';
+    content += this.onTitleStart(1) + this.onText('Glossary') + this.onTitleEnd(1);
+    content += this.onSection();
+
+    _.each(glossary, function(entry) {
+        content += this.onTitleStart(2) + this.onText(entry.name) + this.onTitleEnd(2);
+        content += this.onParagraphStart();
+        content += this.onText(entry.description);
+        content += this.onParagraphEnd();
+        content += this.onSection();
+    }, this);
+
+    return content;
+};
+
+// ------ SUMMARY
+
+ToText.prototype._summaryArticle = function(article, level) {
+    var content = '';
+
+    content += this.onListItemStart(level);
+
+    if (article.path) content += this.onLinkStart(article.path)
+    content += this.onText(article.title)
+    if (article.path) content += this.onLinkEnd(article.path);
+    content += this.onBL();
+
+    if (article.articles && article.articles.length > 0) {
+        content += this._summaryArticles(article.articles, level + 1);
+    }
+
+    content += this.onListItemEnd(level);
+
+    return content;
+};
+ToText.prototype._summaryArticles = function(articles, level) {
+    var content = '';
+    level = level || 0;
+
+    content += this.onListStart(level);
+    _.each(articles, function(article) {
+        content += this._summaryArticle(article, level);
+    }, this);
+    content += this.onListEnd(level);
+
+    return content;
+};
+ToText.prototype._summaryPart = function(part) {
+    var content = '';
+
+    content += this._summaryArticles(part.articles);
+    content += this.onSection();
+
+    return content;
+};
+
+ToText.prototype.summary = function(summary) {
+    var content = '';
+    content += this.onTitleStart(1) + this.onText('Summary') + this.onTitleEnd(1);
+    content += this.onSection();
+
+    _.each(summary.parts, function(part) {
+        content += this._summaryPart(part);
+    }, this);
+
+    return content;
+};
+
+// ---- Utilities
+
+ToText.prototype._spaces =  function(n, s) {
+    return Array(n + 1).join(s || ' ');
+}
+
+module.exports = ToText;
+