Collapse similar HTML elements

1 files changed, 115 insertions, 168 deletions

diff --git a/lib/document-to-html.js b/lib/document-to-html.js
index 8a7cec0..eef0f8e 100644
--- a/lib/document-to-html.js
+++ b/lib/document-to-html.js
@@ -4,9 +4,10 @@ var _ = require("underscore");
 var promises = require("./promises");
 var documents = require("./documents");
 var htmlPaths = require("./html-paths");
-var HtmlGenerator = require("./html-generator").HtmlGenerator;
 var results = require("./results");
 var images = require("./images");
+var Html = require("./html");
+var writers = require("./writers");
 
 exports.DocumentConverter = DocumentConverter;
 
@@ -42,48 +43,44 @@ function DocumentConversion(options) {
     
     var styleMap = options.styleMap || [];
     
-    function createHtmlGenerator() {
-        return new HtmlGenerator({
-            prettyPrint: options.prettyPrint,
-            outputFormat: options.outputFormat
-        });
-    }
-    
     function convertToHtml(document) {
         var messages = [];
-        var html = createHtmlGenerator();
         
-        return promises.nfcall(elementToHtml, document, html, messages).then(function() {
-            html.closeAll();
-            
-            return new results.Result(html.asString(), messages);
+        return promises.nfcall(elementToHtml, document, messages).then(function(html) {
+            var writer = writers.writer({
+                prettyPrint: options.prettyPrint,
+                outputFormat: options.outputFormat
+            });
+            Html.write(writer, Html.simplify(html));
+            return new results.Result(writer.asString(), messages);
         });
     }
     
-    function convertElements(elements, html, messages, callback) {
-        async.eachSeries(elements, function(element, callback) {
-            return elementToHtml(element, html, messages, callback);
+    function convertElements(elements, messages, callback) {
+        async.mapSeries(elements, function(element, callback) {
+            return elementToHtml(element, messages, callback);
         }, callback);
     }
 
-    function elementToHtml(element, html, messages, callback) {
+    function elementToHtml(element, messages, callback) {
         var handler = elementConverters[element.type];
         if (handler) {
-            handler(element, html, messages, callback);
+            handler(element, messages, callback);
         } else {
             callback();
         }
     }
     
-    function convertParagraph(element, html, messages, callback) {
-        html.satisfyPath(styleForParagraph(element, messages));
-        
-        if (!options.ignoreEmptyParagraphs) {
-            html.writeAll();
-        }
+    function convertParagraph(element, messages, callback) {
+        var style = styleForParagraph(element, messages);
         
         setImmediate(function() {
-            convertElements(element.children, html, messages, callback);
+            convertElements(element.children, messages, function(error, children) {
+                if (!options.ignoreEmptyParagraphs) {
+                    children = [Html.forceWrite, Html.fragment(children)];
+                }
+                callback(null, Html.pathToNode(style, children));
+            });
         });
     }
     
@@ -100,62 +97,63 @@ function DocumentConversion(options) {
         }
     }
     
-    function convertRun(run, html, messages, callback) {
-        var runHtml = createHtmlGenerator();
+    function convertRun(run, messages, callback) {
         if (run.styleId) {
             var style = findStyle(run);
-            if (style) {
-                runHtml.satisfyPath(style.to);
-            } else {
+            if (!style) {
                 messages.push(unrecognisedStyleWarning("run", run));
             }
         }
-        if (run.isBold) {
-            runHtml.open(htmlPaths.element("strong"));
-        }
-        if (run.isItalic) {
-            runHtml.open(htmlPaths.element("em"));
-        }
-        if (run.verticalAlignment === documents.verticalAlignment.superscript) {
-            runHtml.open(htmlPaths.element("sup"));
-        }
-        if (run.verticalAlignment === documents.verticalAlignment.subscript) {
-            runHtml.open(htmlPaths.element("sub"));
-        }
-        if (run.isUnderline) {
-            convertUnderline(runHtml);
-        }
-        if (run.isStrikethrough) {
-            convertStrikethrough(runHtml);
-        }
-        convertElements(run.children, runHtml, messages, function(err) {
-            if (err) {
-                callback(err);
-            } else {
-                runHtml.closeAll();
-                html.append(runHtml);
-                callback();
+        convertElements(run.children, messages, function(error, children) {
+            var result = Html.fragment(children);
+            if (run.isStrikethrough) {
+                result = convertStrikethrough(result);
+            }
+            if (run.isUnderline) {
+                result = convertUnderline(result);
+            }
+            if (run.verticalAlignment === documents.verticalAlignment.subscript) {
+                result = wrap(result, "sub");
+            }
+            if (run.verticalAlignment === documents.verticalAlignment.superscript) {
+                result = wrap(result, "sup");
+            }
+            if (run.isItalic) {
+                result = wrap(result, "em");
             }
+            if (run.isBold) {
+                result = wrap(result, "strong");
+            }
+            if (style) {
+                result = Html.pathToNode(style.to, [result]);
+            }
+            callback(null, result);
         });
     }
     
+    function wrap(node, tagName) {
+        return Html.elementWithTag(htmlPaths.element(tagName, {}, {fresh: false}), [node]);
+    }
+    
     function convertUnderline(runHtml) {
-        (options.convertUnderline || defaultConvertUnderline)(runHtml);
+        return (options.convertUnderline || defaultConvertUnderline)(runHtml);
     }
     
-    function convertStrikethrough(runHtml) {
+    function convertStrikethrough(result) {
         var style = findStyle({type: "strikethrough"});
         if (style) {
-            runHtml.appendPath(style.to);
+            return Html.pathToNode(style.to, [result]);
         } else {
-            runHtml.open(htmlPaths.element("s"));
+            return wrap(result, "s");
         }
     }
     
     function defaultConvertUnderline(runHtml) {
         var style = findStyle({type: "underline"});
         if (style) {
-            runHtml.appendPath(style.to);
+            return Html.pathToNode(style.to, [runHtml]);
+        } else {
+            return runHtml;
         }
     }
     
@@ -176,12 +174,14 @@ function DocumentConversion(options) {
     });
     
     function recoveringConvertImage(convertImage) {
-        return function(image, html, messages, callback) {
-            convertImage(image, html, messages, function(error) {
+        return function(image, messages, callback) {
+            convertImage(image, messages, function(error, result) {
                 if (error) {
                     messages.push(results.warning(error.message));
+                    callback(null, Html.fragment());
+                } else {
+                    callback(null, result);
                 }
-                callback();
             });
         };
     }
@@ -198,141 +198,88 @@ function DocumentConversion(options) {
         return idPrefix + "-" + suffix;
     }
     
-    function convertTable(element, html, messages, callback) {
-        html.closeAll();
-        wrapChildren(element, "table", html, messages, callback);
+    function convertTable(element, messages, callback) {
+        wrapChildren(element, "table", messages, callback);
     }
     
-    function convertTableRow(element, html, messages, callback) {
-        wrapChildren(element, "tr", html, messages, callback);
+    function convertTableRow(element, messages, callback) {
+        wrapChildren(element, "tr", messages, callback);
     }
     
-    function convertTableCell(element, html, messages, callback) {
-        html.open(htmlPaths.element("td"), {alwaysWrite: true});
-        
-        async.map(element.children, function(child, callback) {
-            var cellHtml = createHtmlGenerator();
-            elementToHtml(child, cellHtml, messages, function(error) {
-                if (error) {
-                    callback(error);
-                } else {
-                    cellHtml.closeAll();
-                    callback(null, cellHtml);
-                }
-            });
-        }, function(error, childrenHtml) {
-            if (error) {
-                callback(error);
-            } else {
-                childrenHtml.forEach(function(childHtml) {
-                    html.append(childHtml);
-                });
-                html.close();
-                callback();
-            }
-        });
+    function convertTableCell(element, messages, callback) {
+        wrapChildren(element, "td", messages, callback);
     }
     
-    function wrapChildren(element, wrapElementName, html, messages, callback) {
-        html.open(htmlPaths.element(wrapElementName));
-        convertElements(element.children, html, messages, function(err) {
-            if (err) {
-                callback(err);
-            } else {
-                html.close();
-                callback();
-            }
+    function wrapChildren(element, wrapElementName, messages, callback) {
+        convertElements(element.children, messages, function(error, children) {
+            callback(null, Html.element(wrapElementName, {}, [
+                Html.forceWrite,
+                Html.fragment(children)]));
         });
     }
 
     var elementConverters = {
-        "document": function(document, html, messages, callback) {
-            convertElements(document.children, html, messages, function(error) {
-                if (error) {
-                    callback(error);
-                } else {
-                    html.closeAll();
-                    html.open(htmlPaths.element("ol"));
-                    var notes = noteReferences.map(function(noteReference) {
-                        return document.notes.resolve(noteReference);
-                    });
-                    convertElements(notes, html, messages, function(error) {
-                        if (error) {
-                            callback(error);
-                        } else {
-                            html.close();
-                            callback();
-                        }
-                    });
-                }
+        "document": function(document, messages, callback) {
+            convertElements(document.children, messages, function(error, children) {
+                var notes = noteReferences.map(function(noteReference) {
+                    return document.notes.resolve(noteReference);
+                });
+                convertElements(notes, messages, function(error, notesNodes) {
+                    callback(null, Html.fragment([
+                        Html.fragment(children),
+                        Html.element("ol", {}, notesNodes)]));
+                });
             });
         },
         "paragraph": convertParagraph,
         "run": convertRun,
-        "text": function(element, html, messages, callback) {
-            html.text(element.value);
-            callback();
+        "text": function(element, messages, callback) {
+            callback(null, Html.text(element.value));
         },
-        "tab": function(element, html, messages, callback) {
-            html.text("\t");
-            callback();
+        "tab": function(element, messages, callback) {
+            callback(null, Html.text("\t"));
         },
-        "hyperlink": function(element, html, messages, callback) {
+        "hyperlink": function(element, messages, callback) {
             var href = element.anchor ? "#" + htmlId(element.anchor) : element.href;
 
-            html.open(htmlPaths.element("a", {href: href}));
-            convertElements(element.children, html, messages, function(err) {
-                if (err) {
-                    callback(err);
-                } else {
-                    html.close();
-                    callback();
-                }
+            convertElements(element.children, messages, function(error, children) {
+                callback(null, Html.element("a", {href: href}, children));
             });
         },
-        "bookmarkStart": function(element, html, messages, callback) {
-            html.open(htmlPaths.element("a", {
+        "bookmarkStart": function(element, messages, callback) {
+            callback(null, Html.element("a", {
                 id: htmlId(element.name)
-            }), {alwaysWrite: true});
-            html.close();
-            callback();
+            }, [Html.forceWrite]));
         },
-        "noteReference": function(element, html, messages, callback) {
-            html.open(htmlPaths.element("sup"));
-            html.open(htmlPaths.element("a", {
-                href: "#" + noteHtmlId(element),
-                id: noteRefHtmlId(element)
-            }));
+        "noteReference": function(element, messages, callback) {
             noteReferences.push(element);
-            html.text("[" + (noteNumber++) + "]");
-            html.close();
-            html.close();
-            callback();
+            
+            callback(null, Html.element("sup", {}, [
+                Html.element("a", {
+                    href: "#" + noteHtmlId(element),
+                    id: noteRefHtmlId(element)
+                }, [Html.text("[" + (noteNumber++) + "]")])
+            ]));
         },
-        "note": function(element, html, messages, callback) {
-            html.open(htmlPaths.element("li", {id: noteHtmlId(element)}));
-            var footnoteHtml = createHtmlGenerator();
-            convertElements(element.body, footnoteHtml, messages, function(err) {
-                if (err) {
-                    callback(err);
-                } else {
-                    footnoteHtml.text(" ");
-                    footnoteHtml.open(htmlPaths.element("a", {href: "#" + noteRefHtmlId(element)}));
-                    footnoteHtml.text("↑");
-                    footnoteHtml.closeAll();
-                    html.append(footnoteHtml);
-                    html.close();
-                    callback();
-                }
+        "note": function(element, messages, callback) {
+            convertElements(element.body, messages, function(error, children) {
+                callback(null, Html.element("li", {id: noteHtmlId(element)}, [
+                    Html.fragment(children),
+                    
+                    Html.elementWithTag(htmlPaths.element("p", {}, {fresh: false}), [
+                        Html.text(" "),
+                        Html.element("a", {href: "#" + noteRefHtmlId(element)}, [Html.text("↑")]),
+                        
+                    ])
+                ]));
             });
         },
         "image": recoveringConvertImage(options.convertImage || defaultConvertImage),
         "table": convertTable,
         "tableRow": convertTableRow,
         "tableCell": convertTableCell,
-        "lineBreak": function(element, html, messages, callback) {
-            html.selfClosing(htmlPaths.element("br"));
-            callback();
+        "lineBreak": function(element, messages, callback) {
+            callback(null, Html.selfClosingElement("br"));
         }
     };
     return {