@hyperlex/mammoth 1.4.9-beta

package/lib/style-reader.js

@@ -0,0 +1,321 @@
+var _ = require("underscore");
+var lop = require("lop");
+
+var documentMatchers = require("./styles/document-matchers");
+var htmlPaths = require("./styles/html-paths");
+var tokenise = require("./styles/parser/tokeniser").tokenise;
+var results = require("./results");
+
+exports.readHtmlPath = readHtmlPath;
+exports.readDocumentMatcher = readDocumentMatcher;
+exports.readStyle = readStyle;
+
+
+function readStyle(string) {
+    return parseString(styleRule, string);
+}
+
+function createStyleRule() {
+    return lop.rules.sequence(
+        lop.rules.sequence.capture(documentMatcherRule()),
+        lop.rules.tokenOfType("whitespace"),
+        lop.rules.tokenOfType("arrow"),
+        lop.rules.sequence.capture(lop.rules.optional(lop.rules.sequence(
+            lop.rules.tokenOfType("whitespace"),
+            lop.rules.sequence.capture(htmlPathRule())
+        ).head())),
+        lop.rules.tokenOfType("end")
+    ).map(function(documentMatcher, htmlPath) {
+        return {
+            from: documentMatcher,
+            to: htmlPath.valueOrElse(htmlPaths.empty)
+        };
+    });
+}
+
+function readDocumentMatcher(string) {
+    return parseString(documentMatcherRule(), string);
+}
+
+function documentMatcherRule() {
+    var sequence = lop.rules.sequence;
+    
+    var identifierToConstant = function(identifier, constant) {
+        return lop.rules.then(
+            lop.rules.token("identifier", identifier),
+            function() {
+                return constant;
+            }
+        );
+    };
+    
+    var paragraphRule = identifierToConstant("p", documentMatchers.paragraph);
+    var runRule = identifierToConstant("r", documentMatchers.run);
+    
+    var elementTypeRule = lop.rules.firstOf("p or r or table",
+        paragraphRule,
+        runRule
+    );
+    
+    var styleIdRule = lop.rules.then(
+        classRule,
+        function(styleId) {
+            return {styleId: styleId};
+        }
+    );
+
+    var styleNameMatcherRule = lop.rules.firstOf("style name matcher",
+        lop.rules.then(
+            lop.rules.sequence(
+                lop.rules.tokenOfType("equals"),
+                lop.rules.sequence.cut(),
+                lop.rules.sequence.capture(stringRule)
+            ).head(),
+            function(styleName) {
+                return {styleName: documentMatchers.equalTo(styleName)};
+            }
+        ),
+        lop.rules.then(
+            lop.rules.sequence(
+                lop.rules.tokenOfType("startsWith"),
+                lop.rules.sequence.cut(),
+                lop.rules.sequence.capture(stringRule)
+            ).head(),
+            function(styleName) {
+                return {styleName: documentMatchers.startsWith(styleName)};
+            }
+        )
+    );
+    
+    var styleNameRule = lop.rules.sequence(
+        lop.rules.tokenOfType("open-square-bracket"),
+        lop.rules.sequence.cut(),
+        lop.rules.token("identifier", "style-name"),
+        lop.rules.sequence.capture(styleNameMatcherRule),
+        lop.rules.tokenOfType("close-square-bracket")
+    ).head();
+
+    
+    var listTypeRule = lop.rules.firstOf("list type",
+        identifierToConstant("ordered-list", {isOrdered: true}),
+        identifierToConstant("unordered-list", {isOrdered: false})
+    );
+    var listRule = sequence(
+        lop.rules.tokenOfType("colon"),
+        sequence.capture(listTypeRule),
+        sequence.cut(),
+        lop.rules.tokenOfType("open-paren"),
+        sequence.capture(integerRule),
+        lop.rules.tokenOfType("close-paren")
+    ).map(function(listType, levelNumber) {
+        return {
+            list: {
+                isOrdered: listType.isOrdered,
+                levelIndex: levelNumber - 1
+            }
+        };
+    });
+
+    function createMatcherSuffixesRule(rules) {
+        var matcherSuffix = lop.rules.firstOf.apply(
+            lop.rules.firstOf,
+            ["matcher suffix"].concat(rules)
+        );
+        var matcherSuffixes = lop.rules.zeroOrMore(matcherSuffix);
+        return lop.rules.then(matcherSuffixes, function(suffixes) {
+            var matcherOptions = {};
+            suffixes.forEach(function(suffix) {
+                _.extend(matcherOptions, suffix);
+            });
+            return matcherOptions;
+        });
+    }
+    
+    var paragraphOrRun = sequence(
+        sequence.capture(elementTypeRule),
+        sequence.capture(createMatcherSuffixesRule([
+            styleIdRule,
+            styleNameRule,
+            listRule
+        ]))
+    ).map(function(createMatcher, matcherOptions) {
+        return createMatcher(matcherOptions);
+    });
+    
+    var table = sequence(
+        lop.rules.token("identifier", "table"),
+        sequence.capture(createMatcherSuffixesRule([
+            styleIdRule,
+            styleNameRule
+        ]))
+    ).map(function(options) {
+        return documentMatchers.table(options);
+    });
+
+    var bold = identifierToConstant("b", documentMatchers.bold);
+    var italic = identifierToConstant("i", documentMatchers.italic);
+    var underline = identifierToConstant("u", documentMatchers.underline);
+    var strikethrough = identifierToConstant("strike", documentMatchers.strikethrough);
+    var smallCaps = identifierToConstant("small-caps", documentMatchers.smallCaps);
+    var commentReference = identifierToConstant("comment-reference", documentMatchers.commentReference);
+    
+    var breakMatcher = sequence(
+        lop.rules.token("identifier", "br"),
+        sequence.cut(),
+        lop.rules.tokenOfType("open-square-bracket"),
+        lop.rules.token("identifier", "type"),
+        lop.rules.tokenOfType("equals"),
+        sequence.capture(stringRule),
+        lop.rules.tokenOfType("close-square-bracket")
+    ).map(function(breakType) {
+        switch (breakType) {
+        case "line":
+            return documentMatchers.lineBreak;
+        case "page":
+            return documentMatchers.pageBreak;
+        case "column":
+            return documentMatchers.columnBreak;
+        default:
+            // TODO: handle unknown document matchers
+        }
+    });
+
+    return lop.rules.firstOf("element type",
+        paragraphOrRun,
+        table,
+        bold,
+        italic,
+        underline,
+        strikethrough,
+        smallCaps,
+        commentReference,
+        breakMatcher
+    );
+}
+
+function readHtmlPath(string) {
+    return parseString(htmlPathRule(), string);
+}
+
+function htmlPathRule() {
+    var capture = lop.rules.sequence.capture;
+    var whitespaceRule = lop.rules.tokenOfType("whitespace");
+    var freshRule = lop.rules.then(
+        lop.rules.optional(lop.rules.sequence(
+            lop.rules.tokenOfType("colon"),
+            lop.rules.token("identifier", "fresh")
+        )),
+        function(option) {
+            return option.map(function() {
+                return true;
+            }).valueOrElse(false);
+        }
+    );
+    
+    var separatorRule = lop.rules.then(
+        lop.rules.optional(lop.rules.sequence(
+            lop.rules.tokenOfType("colon"),
+            lop.rules.token("identifier", "separator"),
+            lop.rules.tokenOfType("open-paren"),
+            capture(stringRule),
+            lop.rules.tokenOfType("close-paren")
+        ).head()),
+        function(option) {
+            return option.valueOrElse("");
+        }
+    );
+
+    var tagNamesRule = lop.rules.oneOrMoreWithSeparator(
+        identifierRule,
+        lop.rules.tokenOfType("choice")
+    );
+    
+    var styleElementRule = lop.rules.sequence(
+        capture(tagNamesRule),
+        capture(lop.rules.zeroOrMore(classRule)),
+        capture(freshRule),
+        capture(separatorRule)
+    ).map(function(tagName, classNames, fresh, separator) {
+        var attributes = {};
+        var options = {};
+        if (classNames.length > 0) {
+            attributes["class"] = classNames.join(" ");
+        }
+        if (fresh) {
+            options.fresh = true;
+        }
+        if (separator) {
+            options.separator = separator;
+        }
+        return htmlPaths.element(tagName, attributes, options);
+    });
+    
+    return lop.rules.firstOf("html path",
+        lop.rules.then(lop.rules.tokenOfType("bang"), function() {
+            return htmlPaths.ignore;
+        }),
+        lop.rules.then(
+            lop.rules.zeroOrMoreWithSeparator(
+                styleElementRule,
+                lop.rules.sequence(
+                    whitespaceRule,
+                    lop.rules.tokenOfType("gt"),
+                    whitespaceRule
+                )
+            ),
+            htmlPaths.elements
+        )
+    );
+}
+    
+var identifierRule = lop.rules.then(
+    lop.rules.tokenOfType("identifier"),
+    decodeEscapeSequences
+);
+var integerRule = lop.rules.tokenOfType("integer");
+
+var stringRule = lop.rules.then(
+    lop.rules.tokenOfType("string"),
+    decodeEscapeSequences
+);
+
+var escapeSequences = {
+    "n": "\n",
+    "r": "\r",
+    "t": "\t"
+};
+
+function decodeEscapeSequences(value) {
+    return value.replace(/\\(.)/g, function(match, code) {
+        return escapeSequences[code] || code;
+    });
+}
+
+var classRule = lop.rules.sequence(
+    lop.rules.tokenOfType("dot"),
+    lop.rules.sequence.cut(),
+    lop.rules.sequence.capture(identifierRule)
+).head();
+
+function parseString(rule, string) {
+    var tokens = tokenise(string);
+    var parser = lop.Parser();
+    var parseResult = parser.parseTokens(rule, tokens);
+    if (parseResult.isSuccess()) {
+        return results.success(parseResult.value());
+    } else {
+        return new results.Result(null, [results.warning(describeFailure(string, parseResult))]);
+    }
+}
+
+function describeFailure(input, parseResult) {
+    return "Did not understand this style mapping, so ignored it: " + input + "\n" +
+        parseResult.errors().map(describeError).join("\n");
+}
+
+function describeError(error) {
+    return "Error was at character number " + error.characterNumber() + ": " +
+        "Expected " + error.expected + " but got " + error.actual;
+}
+
+var styleRule = createStyleRule();

package/lib/styles/document-matchers.js

@@ -0,0 +1,74 @@
+exports.paragraph = paragraph;
+exports.run = run;
+exports.table = table;
+exports.bold = new Matcher("bold");
+exports.italic = new Matcher("italic");
+exports.underline = new Matcher("underline");
+exports.strikethrough = new Matcher("strikethrough");
+exports.smallCaps = new Matcher("smallCaps");
+exports.commentReference = new Matcher("commentReference");
+exports.lineBreak = new Matcher("break", {breakType: "line"});
+exports.pageBreak = new Matcher("break", {breakType: "page"});
+exports.columnBreak = new Matcher("break", {breakType: "column"});
+exports.equalTo = equalTo;
+exports.startsWith = startsWith;
+
+
+function paragraph(options) {
+    return new Matcher("paragraph", options);
+}
+
+function run(options) {
+    return new Matcher("run", options);
+}
+
+function table(options) {
+    return new Matcher("table", options);
+}
+
+function Matcher(elementType, options) {
+    options = options || {};
+    this._elementType = elementType;
+    this._styleId = options.styleId;
+    this._styleName = options.styleName;
+    if (options.list) {
+        this._listIndex = options.list.levelIndex;
+        this._listIsOrdered = options.list.isOrdered;
+    }
+}
+
+Matcher.prototype.matches = function(element) {
+    return element.type === this._elementType &&
+        (this._styleId === undefined || element.styleId === this._styleId) &&
+        (this._styleName === undefined || (element.styleName && this._styleName.operator(this._styleName.operand, element.styleName))) &&
+        (this._listIndex === undefined || isList(element, this._listIndex, this._listIsOrdered)) &&
+        (this._breakType === undefined || this._breakType === element.breakType);
+};
+
+function isList(element, levelIndex, isOrdered) {
+    return element.numbering &&
+        element.numbering.level == levelIndex &&
+        element.numbering.isOrdered == isOrdered;
+}
+
+function equalTo(value) {
+    return {
+        operator: operatorEqualTo,
+        operand: value
+    };
+}
+
+function startsWith(value) {
+    return {
+        operator: operatorStartsWith,
+        operand: value
+    };
+}
+
+function operatorEqualTo(first, second) {
+    return first.toUpperCase() === second.toUpperCase();
+}
+
+function operatorStartsWith(first, second) {
+    return second.toUpperCase().indexOf(first.toUpperCase()) === 0;
+}

package/lib/styles/html-paths.js

@@ -0,0 +1,81 @@
+var _ = require("underscore");
+
+var html = require("../html");
+
+exports.topLevelElement = topLevelElement;
+exports.elements = elements;
+exports.element = element;
+
+function topLevelElement(tagName, attributes, calculateAttributes) {
+    return elements([element(tagName, attributes, {fresh: true}, calculateAttributes)]);
+}
+
+function elements(elementStyles) {
+    return new HtmlPath(elementStyles.map(function(elementStyle) {
+        if (_.isString(elementStyle)) {
+            return element(elementStyle);
+        } else {
+            return elementStyle;
+        }
+    }));
+}
+
+function HtmlPath(elements) {
+    this._elements = elements;
+}
+
+HtmlPath.prototype.wrap = function wrap(children, element) {
+    var result = children();
+    for (var index = this._elements.length - 1; index >= 0; index--) {
+        result = this._elements[index].wrapNodes(result, element);
+    }
+    return result;
+};
+
+function element(tagName, attributes, options, calculateAttributes) {
+    options = options || {};
+    return new Element(tagName, attributes, options, calculateAttributes);
+}
+
+function Element(tagName, attributes, options, calculateAttributes) {
+    var tagNames = {};
+    if (_.isArray(tagName)) {
+        tagName.forEach(function(tagName) {
+            tagNames[tagName] = true;
+        });
+        tagName = tagName[0];
+    } else {
+        tagNames[tagName] = true;
+    }
+
+    this.tagName = tagName;
+    this.tagNames = tagNames;
+    this.attributes = attributes || {};
+    this.calculateAttributes = calculateAttributes;
+    this.fresh = options.fresh;
+    this.separator = options.separator;
+}
+
+Element.prototype.matchesElement = function(element) {
+    return this.tagNames[element.tagName] && _.isEqual(this.attributes || {}, element.attributes || {});
+};
+
+Element.prototype.wrap = function wrap(generateNodes, element) {
+    return this.wrapNodes(generateNodes(), element);
+};
+
+Element.prototype.wrapNodes = function wrapNodes(nodes, element) {
+    var attributes = this.attributes;
+    if (this.calculateAttributes) {
+        attributes = this.calculateAttributes(element);
+    }
+    var tag = _.extend({}, this, {attributes: attributes});
+    return [html.elementWithTag(tag, nodes)];
+};
+
+exports.empty = elements([]);
+exports.ignore = {
+    wrap: function() {
+        return [];
+    }
+};

package/lib/styles/parser/tokeniser.js

@@ -0,0 +1,30 @@
+var lop = require("lop");
+var RegexTokeniser = lop.RegexTokeniser;
+
+exports.tokenise = tokenise;
+
+var stringPrefix = "'((?:\\\\.|[^'])*)";
+
+function tokenise(string) {
+    var identifierCharacter = "(?:[a-zA-Z\\-_]|\\\\.)";
+    var tokeniser = new RegexTokeniser([
+        {name: "identifier", regex: new RegExp("(" + identifierCharacter + "(?:" + identifierCharacter + "|[0-9])*)")},
+        {name: "dot", regex: /\./},
+        {name: "colon", regex: /:/},
+        {name: "gt", regex: />/},
+        {name: "whitespace", regex: /\s+/},
+        {name: "arrow", regex: /=>/},
+        {name: "equals", regex: /=/},
+        {name: "startsWith", regex: /\^=/},
+        {name: "open-paren", regex: /\(/},
+        {name: "close-paren", regex: /\)/},
+        {name: "open-square-bracket", regex: /\[/},
+        {name: "close-square-bracket", regex: /\]/},
+        {name: "string", regex: new RegExp(stringPrefix + "'")},
+        {name: "unterminated-string", regex: new RegExp(stringPrefix)},
+        {name: "integer", regex: /([0-9]+)/},
+        {name: "choice", regex: /\|/},
+        {name: "bang", regex: /(!)/}
+    ]);
+    return tokeniser.tokenise(string);
+}

package/lib/transforms.js

@@ -0,0 +1,61 @@
+var _ = require("underscore");
+
+exports.paragraph = paragraph;
+exports.run = run;
+exports._elements = elements;
+exports.getDescendantsOfType = getDescendantsOfType;
+exports.getDescendants = getDescendants;
+
+function paragraph(transform) {
+    return elementsOfType("paragraph", transform);
+}
+
+function run(transform) {
+    return elementsOfType("run", transform);
+}
+
+function elementsOfType(elementType, transform) {
+    return elements(function(element) {
+        if (element.type === elementType) {
+            return transform(element);
+        } else {
+            return element;
+        }
+    });
+}
+
+function elements(transform) {
+    return function transformElement(element) {
+        if (element.children) {
+            var children = _.map(element.children, transformElement);
+            element = _.extend(element, {children: children});
+        }
+        return transform(element);
+    };
+}
+
+
+function getDescendantsOfType(element, type) {
+    return getDescendants(element).filter(function(descendant) {
+        return descendant.type === type;
+    });
+}
+
+function getDescendants(element) {
+    var descendants = [];
+
+    visitDescendants(element, function(descendant) {
+        descendants.push(descendant);
+    });
+
+    return descendants;
+}
+
+function visitDescendants(element, visit) {
+    if (element.children) {
+        element.children.forEach(function(child) {
+            visitDescendants(child, visit);
+            visit(child);
+        });
+    }
+}

package/lib/underline.js

@@ -0,0 +1,11 @@
+var htmlPaths = require("./styles/html-paths");
+var Html = require("./html");
+
+
+exports.element = element;
+
+function element(name) {
+    return function(html) {
+        return Html.elementWithTag(htmlPaths.element(name), [html]);
+    };
+}

package/lib/unzip.js

@@ -0,0 +1,22 @@
+exports.openZip = openZip;
+
+var fs = require("fs");
+
+var promises = require("./promises");
+var zipfile = require("./zipfile");
+
+exports.openZip = openZip;
+
+var readFile = promises.promisify(fs.readFile);
+
+function openZip(options) {
+    if (options.path) {
+        return readFile(options.path).then(zipfile.openArrayBuffer);
+    } else if (options.buffer) {
+        return promises.resolve(zipfile.openArrayBuffer(options.buffer));
+    } else if (options.file) {
+        return promises.resolve(options.file);
+    } else {
+        return promises.reject(new Error("Could not find file in options"));
+    }
+}