@hyperlex/mammoth 1.4.9-beta

package/lib/docx/styles-reader.js

@@ -0,0 +1,70 @@
+exports.readStylesXml = readStylesXml;
+exports.Styles = Styles;
+exports.defaultStyles = new Styles({}, {});
+
+function Styles(paragraphStyles, characterStyles, tableStyles, numberingStyles) {
+    return {
+        findParagraphStyleById: function(styleId) {
+            return paragraphStyles[styleId];
+        },
+        findCharacterStyleById: function(styleId) {
+            return characterStyles[styleId];
+        },
+        findTableStyleById: function(styleId) {
+            return tableStyles[styleId];
+        },
+        findNumberingStyleById: function(styleId) {
+            return numberingStyles[styleId];
+        }
+    };
+}
+
+Styles.EMPTY = new Styles({}, {}, {}, {});
+
+function readStylesXml(root) {
+    var paragraphStyles = {};
+    var characterStyles = {};
+    var tableStyles = {};
+    var numberingStyles = {};
+
+    var styles = {
+        "paragraph": paragraphStyles,
+        "character": characterStyles,
+        "table": tableStyles
+    };
+
+    root.getElementsByTagName("w:style").forEach(function(styleElement) {
+        var style = readStyleElement(styleElement);
+        if (style.type === "numbering") {
+            numberingStyles[style.styleId] = readNumberingStyleElement(styleElement);
+        } else {
+            var styleSet = styles[style.type];
+            if (styleSet) {
+                styleSet[style.styleId] = style;
+            }
+        }
+    });
+
+    return new Styles(paragraphStyles, characterStyles, tableStyles, numberingStyles);
+}
+
+function readStyleElement(styleElement) {
+    var type = styleElement.attributes["w:type"];
+    var styleId = styleElement.attributes["w:styleId"];
+    var name = styleName(styleElement);
+    return {type: type, styleId: styleId, name: name};
+}
+
+function styleName(styleElement) {
+    var nameElement = styleElement.first("w:name");
+    return nameElement ? nameElement.attributes["w:val"] : null;
+}
+
+function readNumberingStyleElement(styleElement) {
+    var numId = styleElement
+        .firstOrEmpty("w:pPr")
+        .firstOrEmpty("w:numPr")
+        .firstOrEmpty("w:numId")
+        .attributes["w:val"];
+    return {numId: numId};
+}

package/lib/docx/uris.js

@@ -0,0 +1,21 @@
+exports.uriToZipEntryName = uriToZipEntryName;
+exports.replaceFragment = replaceFragment;
+
+function uriToZipEntryName(base, uri) {
+    if (uri.charAt(0) === "/") {
+        return uri.substr(1);
+    } else {
+        // In general, we should check first and second for trailing and leading slashes,
+        // but in our specific case this seems to be sufficient
+        return base + "/" + uri;
+    }
+}
+
+
+function replaceFragment(uri, fragment) {
+    var hashIndex = uri.indexOf("#");
+    if (hashIndex !== -1) {
+        uri = uri.substring(0, hashIndex);
+    }
+    return uri + "#" + fragment;
+}

package/lib/html/ast.js

@@ -0,0 +1,50 @@
+var htmlPaths = require("../styles/html-paths");
+
+
+function nonFreshElement(tagName, attributes, children) {
+    return elementWithTag(
+        htmlPaths.element(tagName, attributes, {fresh: false}),
+        children);
+}
+
+function freshElement(tagName, attributes, children) {
+    var tag = htmlPaths.element(tagName, attributes, {fresh: true});
+    return elementWithTag(tag, children);
+}
+
+function elementWithTag(tag, children) {
+    return {
+        type: "element",
+        tag: tag,
+        children: children || []
+    };
+}
+
+function text(value) {
+    return {
+        type: "text",
+        value: value
+    };
+}
+
+var forceWrite = {
+    type: "forceWrite"
+};
+
+exports.freshElement = freshElement;
+exports.nonFreshElement = nonFreshElement;
+exports.elementWithTag = elementWithTag;
+exports.text = text;
+exports.forceWrite = forceWrite;
+
+var voidTagNames = {
+    "br": true,
+    "hr": true,
+    "img": true
+};
+
+function isVoidElement(node) {
+    return (node.children.length === 0) && voidTagNames[node.tag.tagName];
+}
+
+exports.isVoidElement = isVoidElement;

package/lib/html/index.js

@@ -0,0 +1,41 @@
+var ast = require("./ast");
+
+exports.freshElement = ast.freshElement;
+exports.nonFreshElement = ast.nonFreshElement;
+exports.elementWithTag = ast.elementWithTag;
+exports.text = ast.text;
+exports.forceWrite = ast.forceWrite;
+
+exports.simplify = require("./simplify");
+
+function write(writer, nodes) {
+    nodes.forEach(function(node) {
+        writeNode(writer, node);
+    });
+}
+
+function writeNode(writer, node) {
+    toStrings[node.type](writer, node);
+}
+
+var toStrings = {
+    element: generateElementString,
+    text: generateTextString,
+    forceWrite: function() { }
+};
+
+function generateElementString(writer, node) {
+    if (ast.isVoidElement(node)) {
+        writer.selfClosing(node.tag.tagName, node.tag.attributes);
+    } else {
+        writer.open(node.tag.tagName, node.tag.attributes);
+        write(writer, node.children);
+        writer.close(node.tag.tagName);
+    }
+}
+
+function generateTextString(writer, node) {
+    writer.text(node.value);
+}
+
+exports.write = write;

package/lib/html/simplify.js

@@ -0,0 +1,88 @@
+var _ = require("underscore");
+
+var ast = require("./ast");
+
+function simplify(nodes) {
+    return collapse(removeEmpty(nodes));
+}
+
+function collapse(nodes) {
+    var children = [];
+    
+    nodes.map(collapseNode).forEach(function(child) {
+        appendChild(children, child);
+    });
+    return children;
+}
+
+function collapseNode(node) {
+    return collapsers[node.type](node);
+}
+
+var collapsers = {
+    element: collapseElement,
+    text: identity,
+    forceWrite: identity
+};
+
+function collapseElement(node) {
+    return ast.elementWithTag(node.tag, collapse(node.children));
+}
+
+function identity(value) {
+    return value;
+}
+
+function appendChild(children, child) {
+    var lastChild = children[children.length - 1];
+    if (child.type === "element" && !child.tag.fresh && lastChild && lastChild.type === "element" && child.tag.matchesElement(lastChild.tag)) {
+        if (child.tag.separator) {
+            appendChild(lastChild.children, ast.text(child.tag.separator));
+        }
+        child.children.forEach(function(grandChild) {
+            // Mutation is fine since simplifying elements create a copy of the children.
+            appendChild(lastChild.children, grandChild);
+        });
+    } else {
+        children.push(child);
+    }
+}
+
+function removeEmpty(nodes) {
+    return flatMap(nodes, function(node) {
+        return emptiers[node.type](node);
+    });
+}
+
+function flatMap(values, func) {
+    return _.flatten(_.map(values, func), true);
+}
+
+var emptiers = {
+    element: elementEmptier,
+    text: textEmptier,
+    forceWrite: neverEmpty
+};
+
+function neverEmpty(node) {
+    return [node];
+}
+
+function elementEmptier(element) {
+    var children = removeEmpty(element.children);
+    if (children.length === 0 && !ast.isVoidElement(element)) {
+        return [];
+    } else {
+        return [ast.elementWithTag(element.tag, children)];
+    }
+}
+
+function textEmptier(node) {
+    if (node.value.length === 0) {
+        return [];
+    } else {
+        return [node];
+    }
+}
+
+module.exports = simplify;

package/lib/images.js

@@ -0,0 +1,29 @@
+var _ = require("underscore");
+
+var promises = require("./promises");
+var Html = require("./html");
+
+exports.imgElement = imgElement;
+
+function imgElement(func) {
+    return function(element, messages) {
+        return promises.when(func(element)).then(function(result) {
+            var attributes = _.clone(result);
+            if (element.altText) {
+                attributes.alt = element.altText;
+            }
+            return [Html.freshElement("img", attributes)];
+        });
+    };
+}
+
+// Undocumented, but retained for backwards-compatibility with 0.3.x
+exports.inline = exports.imgElement;
+
+exports.dataUri = imgElement(function(element) {
+    return element.read("base64").then(function(imageBuffer) {
+        return {
+            src: "data:" + element.contentType + ";base64," + imageBuffer
+        };
+    });
+});

package/lib/index.js

@@ -0,0 +1,115 @@
+var _ = require("underscore");
+
+var docxReader = require("./docx/docx-reader");
+var docxStyleMap = require("./docx/style-map");
+var DocumentConverter = require("./document-to-html").DocumentConverter;
+var readStyle = require("./style-reader").readStyle;
+var readOptions = require("./options-reader").readOptions;
+var unzip = require("./unzip");
+var Result = require("./results").Result;
+var results = require("./results");
+
+exports.convertToHtml = convertToHtml;
+exports.convertToMarkdown = convertToMarkdown;
+exports.convert = convert;
+exports.extractRawText = extractRawText;
+exports.images = require("./images");
+exports.transforms = require("./transforms");
+exports.underline = require("./underline");
+exports.embedStyleMap = embedStyleMap;
+exports.readEmbeddedStyleMap = readEmbeddedStyleMap;
+
+function convertToHtml(input, options) {
+    return convert(input, options);
+}
+
+function convertToMarkdown(input, options) {
+    var markdownOptions = Object.create(options || {});
+    markdownOptions.outputFormat = "markdown";
+    return convert(input, markdownOptions);
+}
+
+function convert(input, options) {
+    options = readOptions(options);
+
+    return unzip.openZip(input)
+        .tap(function(docxFile) {
+            return docxStyleMap.readStyleMap(docxFile).then(function(styleMap) {
+                options.embeddedStyleMap = styleMap;
+            });
+        })
+        .then(function(docxFile) {
+            return docxReader.read(docxFile, input)
+                .then(function(documentResult) {
+                    return documentResult.map(options.transformDocument);
+                })
+                .then(function(documentResult) {
+                    return convertDocumentToHtml(documentResult, options);
+                });
+        });
+}
+
+function readEmbeddedStyleMap(input) {
+    return unzip.openZip(input)
+        .then(docxStyleMap.readStyleMap);
+}
+
+function convertDocumentToHtml(documentResult, options) {
+    var styleMapResult = parseStyleMap(options.readStyleMap());
+    var parsedOptions = _.extend({}, options, {
+        styleMap: styleMapResult.value
+    });
+    var documentConverter = new DocumentConverter(parsedOptions);
+
+    return documentResult.flatMapThen(function(document) {
+        return styleMapResult.flatMapThen(function(styleMap) {
+            return documentConverter.convertToHtml(document);
+        });
+    });
+}
+
+function parseStyleMap(styleMap) {
+    return Result.combine((styleMap || [])
+        .map(function(arg) {
+            return _.isObject(arg) ? results.success(arg) : readStyle(arg);
+        }))
+        .map(function(styleMap) {
+            return styleMap.filter(function(styleMapping) {
+                return !!styleMapping;
+            });
+        });
+}
+
+
+function extractRawText(input) {
+    return unzip.openZip(input)
+        .then(docxReader.read)
+        .then(function(documentResult) {
+            return documentResult.map(convertElementToRawText);
+        });
+}
+
+function convertElementToRawText(element) {
+    if (element.type === "text") {
+        return element.value;
+    } else {
+        var tail = element.type === "paragraph" ? "\n\n" : "";
+        return (element.children || []).map(convertElementToRawText).join("") + tail;
+    }
+}
+
+function embedStyleMap(input, styleMap) {
+    return unzip.openZip(input)
+        .tap(function(docxFile) {
+            return docxStyleMap.writeStyleMap(docxFile, styleMap);
+        })
+        .then(function(docxFile) {
+            return {
+                toBuffer: docxFile.toBuffer
+            };
+        });
+}
+
+exports.styleMapping = function() {
+    throw new Error('Use a raw string instead of mammoth.styleMapping e.g. "p[style-name=\'Title\'] => h1" instead of mammoth.styleMapping("p[style-name=\'Title\'] => h1")');
+};

package/lib/main.js

@@ -0,0 +1,63 @@
+/* global process */
+
+var fs = require("fs");
+var path = require("path");
+
+var mammoth = require("./");
+var promises = require("./promises");
+var images = require("./images");
+
+function main(argv) {
+    var docxPath = argv["docx-path"];
+    var outputPath = argv["output-path"];
+    var outputDir = argv.output_dir;
+    var outputFormat = argv.output_format;
+    var styleMapPath = argv.style_map;
+    
+    readStyleMap(styleMapPath).then(function(styleMap) {
+        var options = {
+            styleMap: styleMap,
+            outputFormat: outputFormat
+        };
+        
+        if (outputDir) {
+            var basename = path.basename(docxPath, ".docx");
+            outputPath = path.join(outputDir, basename + ".html");
+            var imageIndex = 0;
+            options.convertImage = images.imgElement(function(element) {
+                imageIndex++;
+                var extension = element.contentType.split("/")[1];
+                var filename = imageIndex + "." + extension;
+                
+                return element.read().then(function(imageBuffer) {
+                    var imagePath = path.join(outputDir, filename);
+                    return promises.nfcall(fs.writeFile, imagePath, imageBuffer);
+                }).then(function() {
+                    return {src: filename};
+                });
+            });
+        }
+        
+        return mammoth.convert({path: docxPath}, options)
+            .then(function(result) {
+                result.messages.forEach(function(message) {
+                    process.stderr.write(message.message);
+                    process.stderr.write("\n");
+                });
+                
+                var outputStream = outputPath ? fs.createWriteStream(outputPath) : process.stdout;
+                
+                outputStream.write(result.value);
+            });
+    }).done();
+}
+
+function readStyleMap(styleMapPath) {
+    if (styleMapPath) {
+        return promises.nfcall(fs.readFile, styleMapPath, "utf8");
+    } else {
+        return promises.resolve(null);
+    }
+}
+
+module.exports = main;

package/lib/options-reader.js

@@ -0,0 +1,98 @@
+exports.readOptions = readOptions;
+
+
+var _ = require("underscore");
+
+var defaultStyleMap = exports._defaultStyleMap = [
+    "p.Heading1 => h1:fresh",
+    "p.Heading2 => h2:fresh",
+    "p.Heading3 => h3:fresh",
+    "p.Heading4 => h4:fresh",
+    "p.Heading5 => h5:fresh",
+    "p.Heading6 => h6:fresh",
+    "p[style-name='Heading 1'] => h1:fresh",
+    "p[style-name='Heading 2'] => h2:fresh",
+    "p[style-name='Heading 3'] => h3:fresh",
+    "p[style-name='Heading 4'] => h4:fresh",
+    "p[style-name='Heading 5'] => h5:fresh",
+    "p[style-name='Heading 6'] => h6:fresh",
+    "p[style-name='heading 1'] => h1:fresh",
+    "p[style-name='heading 2'] => h2:fresh",
+    "p[style-name='heading 3'] => h3:fresh",
+    "p[style-name='heading 4'] => h4:fresh",
+    "p[style-name='heading 5'] => h5:fresh",
+    "p[style-name='heading 6'] => h6:fresh",
+    
+    "r[style-name='Strong'] => strong",
+    
+    "p[style-name='footnote text'] => p:fresh",
+    "r[style-name='footnote reference'] =>",
+    "p[style-name='endnote text'] => p:fresh",
+    "r[style-name='endnote reference'] =>",
+    "p[style-name='annotation text'] => p:fresh",
+    "r[style-name='annotation reference'] =>",
+    
+    // LibreOffice
+    "p[style-name='Footnote'] => p:fresh",
+    "r[style-name='Footnote anchor'] =>",
+    "p[style-name='Endnote'] => p:fresh",
+    "r[style-name='Endnote anchor'] =>",
+    
+    "p:unordered-list(1) => ul > li:fresh",
+    "p:unordered-list(2) => ul|ol > li > ul > li:fresh",
+    "p:unordered-list(3) => ul|ol > li > ul|ol > li > ul > li:fresh",
+    "p:unordered-list(4) => ul|ol > li > ul|ol > li > ul|ol > li > ul > li:fresh",
+    "p:unordered-list(5) => ul|ol > li > ul|ol > li > ul|ol > li > ul|ol > li > ul > li:fresh",
+    "p:ordered-list(1) => ol > li:fresh",
+    "p:ordered-list(2) => ul|ol > li > ol > li:fresh",
+    "p:ordered-list(3) => ul|ol > li > ul|ol > li > ol > li:fresh",
+    "p:ordered-list(4) => ul|ol > li > ul|ol > li > ul|ol > li > ol > li:fresh",
+    "p:ordered-list(5) => ul|ol > li > ul|ol > li > ul|ol > li > ul|ol > li > ol > li:fresh",
+    
+    "r[style-name='Hyperlink'] =>",
+    
+    "p[style-name='Normal'] => p:fresh"
+];
+
+var standardOptions = exports._standardOptions = {
+    transformDocument: identity,
+    includeDefaultStyleMap: true,
+    includeEmbeddedStyleMap: true
+};
+
+function readOptions(options) {
+    options = options || {};
+    return _.extend({}, standardOptions, options, {
+        customStyleMap: readStyleMap(options.styleMap),
+        readStyleMap: function() {
+            var styleMap = this.customStyleMap;
+            if (this.includeEmbeddedStyleMap) {
+                styleMap = styleMap.concat(readStyleMap(this.embeddedStyleMap));
+            }
+            if (this.includeDefaultStyleMap) {
+                styleMap = styleMap.concat(defaultStyleMap);
+            }
+            return styleMap;
+        }
+    });
+}
+
+function readStyleMap(styleMap) {
+    if (!styleMap) {
+        return [];
+    } else if (_.isString(styleMap)) {
+        return styleMap.split("\n")
+            .map(function(line) {
+                return line.trim();
+            })
+            .filter(function(line) {
+                return line !== "" && line.charAt(0) !== "#";
+            });
+    } else {
+        return styleMap;
+    }
+}
+
+function identity(value) {
+    return value;
+}

package/lib/promises.js

@@ -0,0 +1,42 @@
+var _ = require("underscore");
+var bluebird = require("bluebird/js/release/promise")();
+
+exports.defer = defer;
+exports.when = bluebird.resolve;
+exports.resolve = bluebird.resolve;
+exports.all = bluebird.all;
+exports.props = bluebird.props;
+exports.reject = bluebird.reject;
+exports.promisify = bluebird.promisify;
+exports.mapSeries = bluebird.mapSeries;
+exports.attempt = bluebird.attempt;
+
+exports.nfcall = function(func) {
+    var args = Array.prototype.slice.call(arguments, 1);
+    var promisedFunc = bluebird.promisify(func);
+    return promisedFunc.apply(null, args);
+};
+
+bluebird.prototype.fail = bluebird.prototype.caught;
+
+bluebird.prototype.also = function(func) {
+    return this.then(function(value) {
+        var returnValue = _.extend({}, value, func(value));
+        return bluebird.props(returnValue);
+    });
+};
+
+function defer() {
+    var resolve;
+    var reject;
+    var promise = new bluebird.Promise(function(resolveArg, rejectArg) {
+        resolve = resolveArg;
+        reject = rejectArg;
+    });
+
+    return {
+        resolve: resolve,
+        reject: reject,
+        promise: promise
+    };
+}

package/lib/results.js

@@ -0,0 +1,72 @@
+var _ = require("underscore");
+
+
+exports.Result = Result;
+exports.success = success;
+exports.warning = warning;
+exports.error = error;
+
+
+function Result(value, messages) {
+    this.value = value;
+    this.messages = messages || [];
+}
+
+Result.prototype.map = function(func) {
+    return new Result(func(this.value), this.messages);
+};
+
+Result.prototype.flatMap = function(func) {
+    var funcResult = func(this.value);
+    return new Result(funcResult.value, combineMessages([this, funcResult]));
+};
+
+Result.prototype.flatMapThen = function(func) {
+    var that = this;
+    return func(this.value).then(function(otherResult) {
+        return new Result(otherResult.value, combineMessages([that, otherResult]));
+    });
+};
+
+Result.combine = function(results) {
+    var values = _.flatten(_.pluck(results, "value"));
+    var messages = combineMessages(results);
+    return new Result(values, messages);
+};
+
+function success(value) {
+    return new Result(value, []);
+}
+
+function warning(message) {
+    return {
+        type: "warning",
+        message: message
+    };
+}
+
+function error(exception) {
+    return {
+        type: "error",
+        message: exception.message,
+        error: exception
+    };
+}
+
+function combineMessages(results) {
+    var messages = [];
+    _.flatten(_.pluck(results, "messages"), true).forEach(function(message) {
+        if (!containsMessage(messages, message)) {
+            messages.push(message);
+        }
+    });
+    return messages;
+}
+
+function containsMessage(messages, message) {
+    return _.find(messages, isSameMessage.bind(null, message)) !== undefined;
+}
+
+function isSameMessage(first, second) {
+    return first.type === second.type && first.message === second.message;
+}