@hyperlex/mammoth 1.4.9-beta

package/test/docx/style-map.tests.js

@@ -0,0 +1,112 @@
+var assert = require("assert");
+
+var JSZip = require("jszip");
+
+var zipfile = require("../../lib/zipfile");
+var styleMap = require("../../lib/docx/style-map");
+var test = require("../test")(module);
+
+test('reading embedded style map on document without embedded style map returns null', function() {
+    var zip = normalDocx();
+    
+    return styleMap.readStyleMap(zip).then(function(contents) {
+        assert.equal(contents, null);
+    });
+});
+
+test('embedded style map can be read after being written', function() {
+    var zip = normalDocx();
+    
+    return styleMap.writeStyleMap(zip, "p => h1").then(function() {
+        return styleMap.readStyleMap(zip).then(function(contents) {
+            assert.equal(contents, "p => h1");
+        });
+    });
+});
+
+test('embedded style map is written to separate file', function() {
+    var zip = normalDocx();
+    
+    return styleMap.writeStyleMap(zip, "p => h1").then(function() {
+        return zip.read("mammoth/style-map", "utf8").then(function(contents) {
+            assert.equal(contents, "p => h1");
+        });
+    });
+});
+
+test('embedded style map is referenced in relationships', function() {
+    var zip = normalDocx();
+    
+    return styleMap.writeStyleMap(zip, "p => h1").then(function() {
+        return zip.read("word/_rels/document.xml.rels", "utf8").then(function(contents) {
+            assert.equal(contents, expectedRelationshipsXml);
+        });
+    });
+});
+
+test('re-embedding style map replaces original', function() {
+    var zip = normalDocx();
+    
+    return styleMap.writeStyleMap(zip, "p => h1").then(function() {
+        return styleMap.writeStyleMap(zip, "p => h2");
+    }).then(function() {
+        return zip.read("word/_rels/document.xml.rels", "utf8").then(function(contents) {
+            assert.equal(contents, expectedRelationshipsXml);
+        });
+    }).then(function() {
+        return styleMap.readStyleMap(zip).then(function(contents) {
+            assert.equal(contents, "p => h2");
+        });
+    });
+});
+
+test('embedded style map has override content type in [Content_Types].xml', function() {
+    var zip = normalDocx();
+    
+    return styleMap.writeStyleMap(zip, "p => h1").then(function() {
+        return zip.read("[Content_Types].xml", "utf8").then(function(contents) {
+            assert.equal(contents, expectedContentTypesXml);
+        });
+    });
+});
+
+test('replacing style map keeps content type', function() {
+    var zip = normalDocx();
+    
+    return styleMap.writeStyleMap(zip, "p => h1").then(function() {
+        return styleMap.writeStyleMap(zip, "p => h2");
+    }).then(function() {
+        return zip.read("[Content_Types].xml", "utf8").then(function(contents) {
+            assert.equal(contents, expectedContentTypesXml);
+        });
+    });
+});
+
+var expectedRelationshipsXml = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' +
+    '<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">' +
+    '<Relationship Id="rId3" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/settings" Target="settings.xml"/>' +
+    '<Relationship Id="rMammothStyleMap" Type="http://schemas.zwobble.org/mammoth/style-map" Target="/mammoth/style-map"/>' +
+    '</Relationships>';
+
+var expectedContentTypesXml = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' +
+    '<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types">' +
+    '<Default Extension="png" ContentType="image/png"/>' +
+    '<Override PartName="/mammoth/style-map" ContentType="text/prs.mammoth.style-map"/>' +
+    '</Types>';
+
+function normalDocx() {
+    var zip = new JSZip();
+    var originalRelationshipsXml = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' +
+        '<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">' +
+        '<Relationship Id="rId3" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/settings" Target="settings.xml"/>' +
+        '</Relationships>';
+    var originalContentTypesXml = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' +
+        '<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types">' +
+        '<Default Extension="png" ContentType="image/png"/>' +
+        '</Types>';
+    zip.file("word/_rels/document.xml.rels", originalRelationshipsXml);
+    zip.file("[Content_Types].xml", originalContentTypesXml);
+    var buffer = zip.generate({type: "arraybuffer"});
+    return zipfile.openArrayBuffer(buffer);
+}
+

package/test/docx/styles-reader.tests.js

@@ -0,0 +1,133 @@
+var assert = require("assert");
+
+var readStylesXml = require("../../lib/docx/styles-reader").readStylesXml;
+var XmlElement = require("../../lib/xml").Element;
+var test = require("../test")(module);
+
+
+test('paragraph style is null if no style with that ID exists', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [])
+    );
+    assert.equal(styles.findParagraphStyleById("Heading1"), null);
+});
+
+test('paragraph style can be found by ID', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            paragraphStyleElement("Heading1", "Heading 1")
+        ])
+    );
+    assert.equal(styles.findParagraphStyleById("Heading1").styleId, "Heading1");
+});
+
+test('table style can be found by ID', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            tableStyleElement("TableNormal", "Normal Table")
+        ])
+    );
+    assert.equal(styles.findTableStyleById("TableNormal").styleId, "TableNormal");
+});
+
+test('character style can be found by ID', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            characterStyleElement("Heading1Char", "Heading 1 Char")
+        ])
+    );
+    assert.equal(styles.findCharacterStyleById("Heading1Char").styleId, "Heading1Char");
+});
+
+test('paragraph and character styles are distinct', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            paragraphStyleElement("Heading1", "Heading 1"),
+            characterStyleElement("Heading1Char", "Heading 1 Char")
+        ])
+    );
+    assert.equal(styles.findCharacterStyleById("Heading1"), null);
+    assert.equal(styles.findParagraphStyleById("Heading1Char"), null);
+});
+
+test('character and table styles are distinct', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            tableStyleElement("Heading1", "Heading 1")
+        ])
+    );
+    assert.equal(styles.findCharacterStyleById("Heading1"), null);
+});
+
+test('styles include names', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            paragraphStyleElement("Heading1", "Heading 1")
+        ])
+    );
+    assert.equal(styles.findParagraphStyleById("Heading1").name, "Heading 1");
+});
+
+test('style name is null if w:name element does not exist', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            styleWithoutWNameElement("paragraph", "Heading1"),
+            styleWithoutWNameElement("character", "Heading1Char")
+        ])
+    );
+    assert.equal(styles.findParagraphStyleById("Heading1").name, null);
+    assert.equal(styles.findCharacterStyleById("Heading1Char").name, null);
+});
+
+test('numbering style is null if no style with that ID exists', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [])
+    );
+    assert.equal(styles.findNumberingStyleById("List1"), null);
+});
+
+test('numbering style has null numId if style has no paragraph properties', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            new XmlElement("w:style", {"w:type": "numbering", "w:styleId": "List1"})
+        ])
+    );
+    assert.equal(styles.findNumberingStyleById("List1").numId, null);
+});
+
+test('numbering style has numId read from paragraph properties', function() {
+    var styles = readStylesXml(
+        new XmlElement("w:styles", {}, [
+            new XmlElement("w:style", {"w:type": "numbering", "w:styleId": "List1"}, [
+                new XmlElement("w:pPr", {}, [
+                    new XmlElement("w:numPr", {}, [
+                        new XmlElement("w:numId", {"w:val": "42"})
+                    ])
+                ])
+            ])
+        ])
+    );
+    assert.equal(styles.findNumberingStyleById("List1").numId, "42");
+});
+
+function paragraphStyleElement(id, name) {
+    return styleElement("paragraph", id, name);
+}
+
+function characterStyleElement(id, name) {
+    return styleElement("character", id, name);
+}
+
+function tableStyleElement(id, name) {
+    return styleElement("table", id, name);
+}
+
+function styleElement(type, id, name) {
+    return new XmlElement("w:style", {"w:type": type, "w:styleId": id}, [
+        new XmlElement("w:name", {"w:val": name}, [])
+    ]);
+}
+
+function styleWithoutWNameElement(type, id) {
+    return new XmlElement("w:style", {"w:type": type, "w:styleId": id}, []);
+}

package/test/docx/uris.tests.js

@@ -0,0 +1,22 @@
+var assert = require("assert");
+
+var zipfile = require("../../lib/docx/uris");
+var test = require("../test")(module);
+
+
+test("uriToZipEntryName", {
+    "when path does not have leading slash then path is resolved relative to base": function() {
+        assert.equal(
+            zipfile.uriToZipEntryName("one/two", "three/four"),
+            "one/two/three/four"
+        );
+    },
+
+    "when path has leading slash then base is ignored": function() {
+        assert.equal(
+            zipfile.uriToZipEntryName("one/two", "/three/four"),
+            "three/four"
+        );
+    }
+});
+

package/test/html/simplify.tests.js

@@ -0,0 +1,134 @@
+var assert = require("assert");
+
+var _ = require("underscore");
+
+var test = require("../test")(module);
+var html = require("../../lib/html");
+var htmlPaths = require("../../lib/styles/html-paths");
+
+var nonFreshElement = html.nonFreshElement;
+var text = html.text;
+
+test("empty text nodes are removed", function() {
+    assert.deepEqual(
+        simplifyNode(text("")),
+        []
+    );
+});
+
+test("elements with no children are removed", function() {
+    assert.deepEqual(
+        simplifyNode(nonFreshElement("p", {}, [])),
+        []
+    );
+});
+
+test("elements only containing empty nodes are removed", function() {
+    assert.deepEqual(
+        simplifyNode(nonFreshElement("p", {}, [text("")])),
+        []
+    );
+});
+
+test("empty children of element are removed", function() {
+    assert.deepEqual(
+        simplifyNode(nonFreshElement("p", {}, [text("Hello"), text("")])),
+        [nonFreshElement("p", {}, [text("Hello")])]
+    );
+});
+
+test("successive fresh elements are not collapsed", function() {
+    var path = htmlPaths.elements([
+        htmlPaths.element("p", {}, {fresh: true})
+    ]);
+    var original = concat(
+        pathToNodes(path, [text("Hello")]),
+        pathToNodes(path, [text(" there")])
+    );
+    
+    assert.deepEqual(
+        html.simplify(original),
+        original);
+});
+
+test("successive plain non-fresh elements are collapsed if they have the same tag name", function() {
+    var path = htmlPaths.elements([
+        htmlPaths.element("p", {}, {fresh: false})
+    ]);
+    assert.deepEqual(
+        html.simplify(concat(
+            pathToNodes(path, [text("Hello")]),
+            pathToNodes(path, [text(" there")])
+        )),
+        pathToNodes(path, [text("Hello"), text(" there")])
+    );
+});
+
+test("non-fresh can collapse into preceding fresh element", function() {
+    var freshPath = htmlPaths.elements([
+        htmlPaths.element("p", {}, {fresh: true})]);
+    var nonFreshPath = htmlPaths.elements([
+        htmlPaths.element("p", {}, {fresh: false})]);
+    assert.deepEqual(
+        html.simplify(concat(
+            pathToNodes(freshPath, [text("Hello")]),
+            pathToNodes(nonFreshPath, [text(" there")])
+        )),
+        pathToNodes(freshPath, [text("Hello"), text(" there")])
+    );
+});
+
+test("children of collapsed element can collapse with children of another collapsed element", function() {
+    assert.deepEqual(
+        html.simplify([
+            nonFreshElement("blockquote", {}, [nonFreshElement("p", {}, [text("Hello")])]),
+            nonFreshElement("blockquote", {}, [nonFreshElement("p", {}, [text("there")])])
+        ]),
+        [nonFreshElement("blockquote", {}, [nonFreshElement("p", {}, [text("Hello"), text("there")])])]
+    );
+});
+
+test("empty elements are removed before collapsing", function() {
+    var freshPath = htmlPaths.elements([
+        htmlPaths.element("p", {}, {fresh: true})]);
+    var nonFreshPath = htmlPaths.elements([
+        htmlPaths.element("p", {}, {fresh: false})]);
+    assert.deepEqual(
+        html.simplify(concat(
+            pathToNodes(nonFreshPath, [text("Hello")]),
+            pathToNodes(freshPath, []),
+            pathToNodes(nonFreshPath, [text(" there")])
+        )),
+        pathToNodes(nonFreshPath, [text("Hello"), text(" there")])
+    );
+});
+
+test("when separator is present then separator is prepended to collapsed element", function() {
+    var unseparatedPath = htmlPaths.elements([
+        htmlPaths.element("pre", {}, {fresh: false})
+    ]);
+    var separatedPath = htmlPaths.elements([
+        htmlPaths.element("pre", {}, {fresh: false, separator: "\n"})
+    ]);
+    assert.deepEqual(
+        html.simplify(concat(
+            pathToNodes(unseparatedPath, [text("Hello")]),
+            pathToNodes(separatedPath, [text(" the"), text("re")])
+        )),
+        pathToNodes(unseparatedPath, [text("Hello"), text("\n"), text(" the"), text("re")])
+    );
+});
+
+function simplifyNode(node) {
+    return html.simplify([node]);
+}
+
+function concat() {
+    return _.flatten(arguments, true);
+}
+
+function pathToNodes(path, nodes) {
+    return path.wrap(function() {
+        return nodes;
+    });
+}

package/test/html/write.tests.js

@@ -0,0 +1,42 @@
+var assert = require("assert");
+
+var test = require("../test")(module);
+var html = require("../../lib/html");
+var writers = require("../../lib/writers");
+
+
+test("text is HTML escaped", function() {
+    assert.equal(
+        generateString(html.text("<>&")),
+        "&lt;&gt;&amp;");
+});
+
+test("double quotes outside of attributes are not escaped", function() {
+    assert.equal(
+        generateString(html.text('"')),
+        '"');
+});
+
+test("element attributes are HTML escaped", function() {
+    assert.equal(
+        generateString(html.freshElement("p", {"x": "<"})),
+        '<p x="&lt;"></p>');
+});
+
+test("double quotes inside attributes are escaped", function() {
+    assert.equal(
+        generateString(html.freshElement("p", {"x": '"'})),
+        '<p x="&quot;"></p>');
+});
+
+test("element children are written", function() {
+    assert.equal(
+        generateString(html.freshElement("p", {}, [html.text("Hello")])),
+        '<p>Hello</p>');
+});
+
+function generateString(node) {
+    var writer = writers.writer();
+    html.write(writer, [node]);
+    return writer.asString();
+}

package/test/images.tests.js

@@ -0,0 +1,34 @@
+var assert = require("assert");
+
+var hamjest = require("hamjest");
+var assertThat = hamjest.assertThat;
+var contains = hamjest.contains;
+var hasProperties = hamjest.hasProperties;
+
+var mammoth = require("../");
+var documents = require("../lib/documents");
+var promises = require("../lib/promises");
+
+var test = require("./test")(module);
+
+
+test('mammoth.images.inline() should be an alias of mammoth.images.imgElement()', function() {
+    assert.ok(mammoth.images.inline === mammoth.images.imgElement);
+});
+
+
+test('mammoth.images.dataUri() encodes images in base64', function() {
+    var imageBuffer = new Buffer("abc");
+    var image = new documents.Image({
+        readImage: function(encoding) {
+            return promises.when(imageBuffer.toString(encoding));
+        },
+        contentType: "image/jpeg"
+    });
+    
+    return mammoth.images.dataUri(image).then(function(result) {
+        assertThat(result, contains(
+            hasProperties({tag: hasProperties({attributes: {"src": "data:image/jpeg;base64,YWJj"}})})
+        ));
+    });
+});

package/test/main.tests.js

@@ -0,0 +1,89 @@
+var assert = require("assert");
+var child_process = require("child_process"); // eslint-disable-line camelcase
+var path = require("path");
+var fs = require("fs");
+
+var temp = require('temp').track();
+
+var promises = require("../lib/promises");
+var test = require("./test")(module);
+var testPath = require("./testing").testPath;
+
+test("HTML is printed to stdout if output file is not set", function() {
+    return runMammoth(testPath("single-paragraph.docx")).then(function(result) {
+        assert.equal(result.stderrOutput, "");
+        assert.equal(result.output, "<p>Walking on imported air</p>");
+    });
+});
+
+test("HTML is written to file if output file is set", function() {
+    return createTempDir().then(function(tempDir) {
+        var outputPath = path.join(tempDir, "output.html");
+        return runMammoth(testPath("single-paragraph.docx"), outputPath).then(function(result) {
+            assert.equal(result.stderrOutput, "");
+            assert.equal(result.output, "");
+            assert.equal(fs.readFileSync(outputPath, "utf8"), "<p>Walking on imported air</p>");
+        });
+    });
+});
+
+var imageBase64 = "iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=";
+
+test("inline images are included in output if writing to single file", function() {
+    return runMammoth(testPath("tiny-picture.docx")).then(function(result) {
+        assert.equal(result.stderrOutput, "");
+        assert.equal(result.output, '<p><img src="data:image/png;base64,' + imageBase64 + '" /></p>');
+    });
+});
+
+test("images are written to separate files if output dir is set", function() {
+    return createTempDir().then(function(tempDir) {
+        var outputPath = path.join(tempDir, "tiny-picture.html");
+        var imagePath = path.join(tempDir, "1.png");
+        return runMammoth(testPath("tiny-picture.docx"), "--output-dir", tempDir).then(function(result) {
+            assert.equal(result.stderrOutput, "");
+            assert.equal(result.output, "");
+            assert.equal(fs.readFileSync(outputPath, "utf8"), '<p><img src="1.png" /></p>');
+            assert.equal(fs.readFileSync(imagePath, "base64"), imageBase64);
+        });
+    });
+});
+
+test("style map is used if set", function() {
+    return createTempDir().then(function(tempDir) {
+        var styleMapPath = path.join(tempDir, "style-map");
+        fs.writeFileSync(styleMapPath, "p => span:fresh");
+        return runMammoth(testPath("single-paragraph.docx"), "--style-map", styleMapPath).then(function(result) {
+            assert.equal(result.stderrOutput, "");
+            assert.equal(result.output, "<span>Walking on imported air</span>");
+        });
+    });
+});
+
+test("--output-format=markdown option generate markdown output", function() {
+    return runMammoth(testPath("single-paragraph.docx"), "--output-format=markdown").then(function(result) {
+        assert.equal(result.stderrOutput, "");
+        assert.equal(result.output, "Walking on imported air\n\n");
+    });
+});
+
+
+function runMammoth() {
+    var args = Array.prototype.slice.call(arguments, 0);
+    var deferred = promises.defer();
+    
+    var processArgs = ["node", "bin/mammoth"].concat(args);
+    // TODO: proper escaping of args
+    var command = processArgs.join(" ");
+    child_process.exec(command, function(error, stdout, stderr) { // eslint-disable-line camelcase
+        console.log(stderr); // eslint-disable-line no-console
+        assert.equal(error, null);
+        deferred.resolve({output: stdout, stderrOutput: stderr});
+    });
+    
+    return deferred.promise;
+}
+
+function createTempDir() {
+    return promises.nfcall(temp.mkdir, null);
+}