mammoth/lib/docx/styles-reader.js

Convert Word documents from docx to simple HTML and Markdown

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"];
}