@depup/mammoth 1.12.0-depup.54

package/lib/docx/style-map.js

@@ -0,0 +1,75 @@
+var _ = require("underscore");
+
+var promises = require("../promises");
+var xml = require("../xml");
+
+exports.writeStyleMap = writeStyleMap;
+exports.readStyleMap = readStyleMap;
+
+
+var schema = "http://schemas.zwobble.org/mammoth/style-map";
+var styleMapPath = "mammoth/style-map";
+var styleMapAbsolutePath = "/" + styleMapPath;
+
+function writeStyleMap(docxFile, styleMap) {
+    docxFile.write(styleMapPath, styleMap);
+    return updateRelationships(docxFile).then(function() {
+        return updateContentTypes(docxFile);
+    });
+}
+
+function updateRelationships(docxFile) {
+    var path = "word/_rels/document.xml.rels";
+    var relationshipsUri = "http://schemas.openxmlformats.org/package/2006/relationships";
+    var relationshipElementName = "{" + relationshipsUri + "}Relationship";
+    return docxFile.read(path, "utf8")
+        .then(xml.readString)
+        .then(function(relationshipsContainer) {
+            var relationships = relationshipsContainer.children;
+            addOrUpdateElement(relationships, relationshipElementName, "Id", {
+                "Id": "rMammothStyleMap",
+                "Type": schema,
+                "Target": styleMapAbsolutePath
+            });
+            
+            var namespaces = {"": relationshipsUri};
+            return docxFile.write(path, xml.writeString(relationshipsContainer, namespaces));
+        });
+}
+
+function updateContentTypes(docxFile) {
+    var path = "[Content_Types].xml";
+    var contentTypesUri = "http://schemas.openxmlformats.org/package/2006/content-types";
+    var overrideName = "{" + contentTypesUri + "}Override";
+    return docxFile.read(path, "utf8")
+        .then(xml.readString)
+        .then(function(typesElement) {
+            var children = typesElement.children;
+            addOrUpdateElement(children, overrideName, "PartName", {
+                "PartName": styleMapAbsolutePath,
+                "ContentType": "text/prs.mammoth.style-map"
+            });
+            var namespaces = {"": contentTypesUri};
+            return docxFile.write(path, xml.writeString(typesElement, namespaces));
+        });
+}
+
+function addOrUpdateElement(elements, name, identifyingAttribute, attributes) {
+    var existingElement = _.find(elements, function(element) {
+        return element.name === name &&
+            element.attributes[identifyingAttribute] === attributes[identifyingAttribute];
+    });
+    if (existingElement) {
+        existingElement.attributes = attributes;
+    } else {
+        elements.push(xml.element(name, attributes));
+    }
+}
+
+function readStyleMap(docxFile) {
+    if (docxFile.exists(styleMapPath)) {
+        return docxFile.read(styleMapPath, "utf8");
+    } else {
+        return promises.resolve(null);
+    }
+}

package/lib/docx/styles-reader.js

@@ -0,0 +1,90 @@
+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,
+        "numbering": numberingStyles
+    };
+
+    root.getElementsByTagName("w:style").forEach(function(styleElement) {
+        var style = readStyleElement(styleElement);
+        var styleSet = styles[style.type];
+
+        // Per 17.7.4.17 style (Style Definition) of ECMA-376 4th edition Part 1:
+        //
+        // > If multiple style definitions each declare the same value for their
+        // > styleId, then the first such instance shall keep its current
+        // > identifier with all other instances being reassigned in any manner
+        // > desired.
+        //
+        // For the purpose of conversion, there's no point holding onto styles
+        // with reassigned style IDs, so we ignore such style definitions.
+
+        if (styleSet && styleSet[style.styleId] === undefined) {
+            styleSet[style.styleId] = style;
+        }
+    });
+
+    return new Styles(paragraphStyles, characterStyles, tableStyles, numberingStyles);
+}
+
+function readStyleElement(styleElement) {
+    var type = styleElement.attributes["w:type"];
+
+    if (type === "numbering") {
+        return readNumberingStyleElement(type, styleElement);
+    } else {
+        var styleId = readStyleId(styleElement);
+        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(type, styleElement) {
+    var styleId = readStyleId(styleElement);
+
+    var numId = styleElement
+        .firstOrEmpty("w:pPr")
+        .firstOrEmpty("w:numPr")
+        .firstOrEmpty("w:numId")
+        .attributes["w:val"];
+
+    return {type: type, numId: numId, styleId: styleId};
+}
+
+function readStyleId(styleElement) {
+    return styleElement.attributes["w:styleId"];
+}

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 +1,51 @@
+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,
+    "input": 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 +1,31 @@
+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 = {};
+            if (element.altText) {
+                attributes.alt = element.altText;
+            }
+            _.extend(attributes, result);
+
+            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.readAsBase64String().then(function(imageBuffer) {
+        return {
+            src: "data:" + element.contentType + ";base64," + imageBuffer
+        };
+    });
+});

package/lib/index.d.ts

@@ -0,0 +1,86 @@
+interface Mammoth {
+    convertToHtml: (input: Input, options?: Options) => Promise<Result>;
+    extractRawText: (input: Input) => Promise<Result>;
+    embedStyleMap: (input: Input, styleMap: string) => Promise<{
+        toArrayBuffer: () => ArrayBuffer,
+        toBuffer: () => Buffer,
+    }>;
+    images: Images;
+}
+
+type Input = NodeJsInput | BrowserInput;
+
+type NodeJsInput = PathInput | BufferInput;
+
+interface PathInput {
+    path: string;
+}
+
+interface BufferInput {
+    buffer: Buffer;
+}
+
+type BrowserInput = ArrayBufferInput;
+
+interface ArrayBufferInput {
+    arrayBuffer: ArrayBuffer;
+}
+
+interface Options {
+    styleMap?: string | Array<string>;
+    includeEmbeddedStyleMap?: boolean;
+    includeDefaultStyleMap?: boolean;
+    convertImage?: ImageConverter;
+    ignoreEmptyParagraphs?: boolean;
+    idPrefix?: string;
+    externalFileAccess?: boolean;
+    transformDocument?: (element: any) => any;
+}
+
+interface ImageConverter {
+    __mammothBrand: "ImageConverter";
+}
+
+interface Image {
+    contentType: string;
+    readAsArrayBuffer: () => Promise<ArrayBuffer>;
+    readAsBase64String: () => Promise<string>;
+    readAsBuffer: () => Promise<Buffer>;
+    read: ImageRead;
+}
+
+interface ImageRead {
+    (): Promise<Buffer>;
+    (encoding: string): Promise<string>;
+}
+
+interface ImageAttributes {
+    src: string;
+}
+
+interface Images {
+    dataUri: ImageConverter;
+    imgElement: (f: (image: Image) => Promise<ImageAttributes>) => ImageConverter;
+}
+
+interface Result {
+    value: string;
+    messages: Array<Message>;
+}
+
+type Message = Warning | Error;
+
+interface Warning {
+    type: "warning";
+    message: string;
+}
+
+interface Error {
+    type: "error";
+    message: string;
+    error: unknown;
+}
+
+declare const mammoth: Mammoth;
+
+export = mammoth;

package/lib/index.js

@@ -0,0 +1,111 @@
+var _ = require("underscore");
+
+var docxReader = require("./docx/docx-reader");
+var docxStyleMap = require("./docx/style-map");
+var DocumentConverter = require("./document-to-html").DocumentConverter;
+var convertElementToRawText = require("./raw-text").convertElementToRawText;
+var readStyle = require("./style-reader").readStyle;
+var readOptions = require("./options-reader").readOptions;
+var unzip = require("./unzip");
+var Result = require("./results").Result;
+
+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, options)
+                .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(readStyle))
+        .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 embedStyleMap(input, styleMap) {
+    return unzip.openZip(input)
+        .tap(function(docxFile) {
+            return docxStyleMap.writeStyleMap(docxFile, styleMap);
+        })
+        .then(function(docxFile) {
+            return docxFile.toArrayBuffer();
+        })
+        .then(function(arrayBuffer) {
+            return {
+                toArrayBuffer: function() {
+                    return arrayBuffer;
+                },
+                toBuffer: function() {
+                    return Buffer.from(arrayBuffer);
+                }
+            };
+        });
+}
+
+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;