import { Injectable } from '@angular/core'; import { LineNumberingMode } from 'app/site/motions/models/view-motion'; /** * Shape of line number objects */ interface LineNumberObject { lineNumber: number; marginBottom?: number; } /** * Converts HTML strings to pdfmake compatible document definition. * * TODO: Bring back upstream to pdfmake, so other projects may benefit from this converter and * to exclude complex code from OpenSlides. * Everything OpenSlides specific, such as line numbering and change recommendations, * should be excluded from this and handled elsewhere. * * @example * ``` * const dd = htmlToPdfService.convertHtml('

Hello World!

'); * ``` */ @Injectable({ providedIn: 'root' }) export class HtmlToPdfService { /** * holds the desired line number mode */ private lineNumberingMode: LineNumberingMode; /** * Space between list elements */ private LI_MARGIN_BOTTOM = 8; /** * Normal line height for paragraphs */ private LINE_HEIGHT = 1.25; /** * space between paragraphs */ private P_MARGIN_BOTTOM = 4.0; /** * Space above H */ private H_MARGIN_TOP = 10.0; /** * Conversion of HTML tags into pdfmake directives */ private elementStyles = { // should be the same for most HTML code b: ['font-weight:bold'], strong: ['font-weight:bold'], u: ['text-decoration:underline'], em: ['font-style:italic'], i: ['font-style:italic'], h1: ['font-size:14', 'font-weight:bold'], h2: ['font-size:12', 'font-weight:bold'], h3: ['font-size:10', 'font-weight:bold'], h4: ['font-size:10', 'font-style:italic'], h5: ['font-size:10'], h6: ['font-size:10'], a: ['color:blue', 'text-decoration:underline'], strike: ['text-decoration:line-through'], // Pretty specific stuff that might be excluded for other projects than OpenSlides del: ['color:red', 'text-decoration:line-through'], ins: ['color:green', 'text-decoration:underline'] }; /** * Treatment of required CSS-Classes * Checking CSS is not possible */ private classStyles = { delete: ['color:red', 'text-decoration:line-through'], insert: ['color:green', 'text-decoration:underline'], paragraphcontext: ['color:grey'] }; /** * Constructor */ public constructor() {} /** * Determine the ideal top margin for a given node * * @param nodeName the node to parse * @returns the margin tip as number */ private getMarginTop(nodeName: string): number { switch (nodeName) { case 'h1': case 'h2': case 'h3': case 'h4': case 'h5': case 'h6': { return this.H_MARGIN_TOP; } default: { return 0; } } } /** * Determine the ideal margin for a given node * * @param nodeName the node to parse * @returns the margin bottom as number */ private getMarginBottom(nodeName: string): number { switch (nodeName) { case 'h1': case 'h2': case 'h3': case 'h4': case 'h5': case 'h6': { return this.P_MARGIN_BOTTOM; } case 'li': { return this.LI_MARGIN_BOTTOM; } default: { return this.P_MARGIN_BOTTOM; } } } /** * Takes an HTML string, converts to HTML using a DOM parser and recursivly parses * the content into pdfmake compatible doc definition * * @param htmlText the html text to translate as string * @param lnMode determines the line numbering * @returns pdfmake doc definition as object */ public convertHtml(htmlText: string, lnMode?: LineNumberingMode): object { const docDef = []; this.lineNumberingMode = lnMode || LineNumberingMode.None; // Cleanup of dirty html would happen here // Create a HTML DOM tree out of html string const parser = new DOMParser(); const parsedHtml = parser.parseFromString(htmlText, 'text/html'); // Since the spread operator did not work for HTMLCollection, use Array.from const htmlArray = Array.from(parsedHtml.body.childNodes) as Element[]; // Parse the children of the current HTML element for (const child of htmlArray) { const parsedElement = this.parseElement(child); docDef.push(parsedElement); } return docDef; } /** * Converts a single HTML element to pdfmake, calls itself recursively for child html elements * * @param element can be an HTML element (

) or plain text ("Hello World") * @param currentParagraph usually holds the parent element, to allow nested structures * @param styles holds the style attributes of HTML elements (`

...`) * @returns the doc def to the given element in consideration to the given paragraph and styles */ public parseElement(element: Element, styles?: string[]): any { const nodeName = element.nodeName.toLowerCase(); let classes = []; let newParagraph: any; // extract explicit style information styles = styles || []; // to leave out plain text elements if (element.getAttribute) { const nodeStyle = element.getAttribute('style'); const nodeClass = element.getAttribute('class'); // add styles like `color:#ff00ff` content into styles array if (nodeStyle) { styles = nodeStyle .split(';') .map(style => style.replace(/\s/g, '')) .concat(styles); } // Handle CSS classes if (nodeClass) { classes = nodeClass.toLowerCase().split(' '); for (const cssClass of classes) { if (this.classStyles[cssClass]) { this.classStyles[cssClass].forEach(style => { styles.push(style); }); } } } } switch (nodeName) { case 'h1': case 'h2': case 'h3': case 'h4': case 'h5': case 'h6': case 'li': case 'p': case 'div': { const children = this.parseChildren(element, styles); // this introduces a bug with rendering sub-lists in PDF if ( this.lineNumberingMode === LineNumberingMode.Outside && !this.isInsideAList(element) && !classes.includes('insert') ) { newParagraph = this.create('stack'); newParagraph.stack = children; } else { newParagraph = this.create('text'); newParagraph.text = children; } newParagraph.margin = [0, 0, 0, 0]; // determine the "normal" top and button margins newParagraph.margin[1] = this.getMarginTop(nodeName); newParagraph.margin[3] = this.getMarginBottom(nodeName); if (this.lineNumberingMode === LineNumberingMode.Outside) { // that is usually the case for inserted change which should appear // under a set of line numbers with correct alignment if (classes.includes('insert')) { newParagraph.margin[0] = 20; newParagraph.margin[3] = this.P_MARGIN_BOTTOM; } } // stop enumeration if the list was inserted if (classes.includes('os-split-before')) { newParagraph.listType = 'none'; } // if the list ends (usually due to a new insert cr) prevent margins if (classes.includes('os-split-after')) { newParagraph.margin[3] = 0; } newParagraph.lineHeight = this.LINE_HEIGHT; newParagraph = { ...newParagraph, ...this.computeStyle(styles), ...this.computeStyle(this.elementStyles[nodeName]) }; break; } case 'a': case 'b': case 'strong': case 'u': case 'em': case 'i': case 'ins': case 'del': case 'strike': { const children = this.parseChildren(element, styles.concat(this.elementStyles[nodeName])); newParagraph = this.create('text'); newParagraph.text = children; break; } case 'span': { // Line numbering feature, will prevent compatibility to most other projects if (element.getAttribute('data-line-number') && !this.isInsideAList(element)) { if (this.lineNumberingMode === LineNumberingMode.Inside) { // TODO: algorithm for "inline" line numbers is not yet implemented } else if (this.lineNumberingMode === LineNumberingMode.Outside) { const currentLineNumber = element.getAttribute('data-line-number'); newParagraph = { columns: [ // the line number column this.getLineNumberObject({ lineNumber: +currentLineNumber }), { text: [] } ] }; } } else { const children = this.parseChildren(element, styles); newParagraph = { ...this.create('text'), ...this.computeStyle(styles) }; newParagraph.text = children; } break; } case 'br': { if (this.lineNumberingMode === LineNumberingMode.None && classes.includes('os-line-break')) { break; } else { newParagraph = this.create('text'); // yep thats all newParagraph.text = '\n'; newParagraph.lineHeight = this.LINE_HEIGHT; } break; } case 'ul': case 'ol': { const list = this.create(nodeName); // keep the numbers of the ol list if (nodeName === 'ol') { const start = element.getAttribute('start'); if (start) { list.start = +start; } } // in case of line numbers and only of the list is not nested in another list. if (this.lineNumberingMode === LineNumberingMode.Outside && !this.isInsideAList(element)) { const lines = this.extractLineNumbers(element); const cleanedChildDom = this.cleanLineNumbers(element); const cleanedChildren = this.parseChildren(cleanedChildDom, styles); if (lines.length > 0) { const listCol = { columns: [ { width: 20, stack: [] } ], margin: [0, 0, 0, 0] }; // if this is a "fake list" lower put it close to the element above if (this.isFakeList(element)) { listCol.margin[3] = -this.LI_MARGIN_BOTTOM; } for (const line of lines) { listCol.columns[0].stack.push(this.getLineNumberObject(line)); } list[nodeName] = cleanedChildren; listCol.columns.push(list); newParagraph = listCol; } else { // that is usually the case for "inserted" lists during change recomendations list.margin = [20, 0, 0, 0]; newParagraph = list; newParagraph[nodeName] = cleanedChildren; } } else { const children = this.parseChildren(element, styles); newParagraph = list; newParagraph[nodeName] = children; } break; } default: { newParagraph = { ...this.create('text', element.textContent.replace(/\n/g, '')), ...this.computeStyle(styles) }; break; } } return newParagraph; } /** * Helper routine to parse an elements children and return the children as parsed pdfmake doc string * * @param element the parent element to parse * @param currentParagraph the context of the element * @param styles the styles array, usually just to parse back into the `parseElement` function * @returns an array of parsed children */ private parseChildren(element: Element, styles?: Array): Element[] { const childNodes = Array.from(element.childNodes) as Element[]; const paragraph = []; if (childNodes.length > 0) { for (const child of childNodes) { // skip empty child nodes if (!(child.nodeName === '#text' && child.textContent.trim() === '')) { const parsedElement = this.parseElement(child, styles); const firstChild = element.firstChild as Element; if ( // add the line number column this.lineNumberingMode === LineNumberingMode.Outside && child && child.classList && child.classList.contains('os-line-number') ) { paragraph.push(parsedElement); } else if ( // if the first child of the parsed element is line number this.lineNumberingMode === LineNumberingMode.Outside && firstChild && firstChild.classList && firstChild.classList.contains('os-line-number') ) { const currentLine = paragraph.pop(); // push the parsed element into the "text" array currentLine.columns[1].text.push(parsedElement); paragraph.push(currentLine); } else { paragraph.push(parsedElement); } } } } return paragraph; } /** * Helper function to make a line-number object * * @param line and object in the shape: { lineNumber: X } * @returns line number as pdfmake-object */ private getLineNumberObject(line: LineNumberObject): object { return { width: 20, text: [ { // Add a blank with the normal font size here, so in rare cases the text // is rendered on the next page and the line number on the previous page. text: ' ', fontSize: 10, decoration: '' }, { text: line.lineNumber, color: 'gray', fontSize: 8 } ], marginBottom: line.marginBottom, lineHeight: this.LINE_HEIGHT }; } /** * Cleans the elements children from line-number spans * * @param element a html dom tree * @returns a DOM element without line number spans */ private cleanLineNumbers(element: Element): Element { const elementCopy = element.cloneNode(true) as Element; const children = elementCopy.childNodes; // using for-of did not work as expected for (let i = 0; i < children.length; i++) { if (this.getLineNumber(children[i] as Element)) { children[i].remove(); } if (children[i].childNodes.length > 0) { const cleanChildren = this.cleanLineNumbers(children[i] as Element); elementCopy.replaceChild(cleanChildren, children[i]); } } return elementCopy; } /** * Helper function to extract line numbers from child elements * * TODO: Cleanup * * @param element element to check for containing line numbers (usually a list) * @returns a list with the line numbers */ private extractLineNumbers(element: Element): LineNumberObject[] { let foundLineNumbers = []; const lineNumber = this.getLineNumber(element); if (lineNumber) { foundLineNumbers.push({ lineNumber: lineNumber }); } else if (element.nodeName === 'BR') { // Check if there is a new line, but it does not get a line number. // If so, insert a dummy line, so the line numbers stays aligned with // the text. if (!this.getLineNumber(element.nextSibling as Element)) { foundLineNumbers.push({ lineNumber: '' }); } } else { const children = Array.from(element.childNodes) as Element[]; let childrenLength = children.length; let childrenLineNumbers = []; for (let i = 0; i < children.length; i++) { childrenLineNumbers = childrenLineNumbers.concat(this.extractLineNumbers(children[i])); if (children.length < childrenLength) { i -= childrenLength - children.length; childrenLength = children.length; } } // If this is an list item, add some space to the lineNumbers: if (childrenLineNumbers.length && element.nodeName === 'LI') { childrenLineNumbers[childrenLineNumbers.length - 1].marginBottom = this.LI_MARGIN_BOTTOM; } foundLineNumbers = foundLineNumbers.concat(childrenLineNumbers); } return foundLineNumbers; } /** * Recursive helper function to determine if the element is inside a list * * @param element the current html node * @returns wether the element is inside a list or not */ private isInsideAList(element: Element): boolean { let parent = element.parentNode; while (parent !== null) { if (parent.nodeName === 'UL' || parent.nodeName === 'OL') { return true; } parent = parent.parentNode; } return false; } /** * Checks if a given UL or LI list (as element) is a "fake list" * Fake lists in fact lists by should appear like the parent list * would seamlessly continue. * This usually happens when a user makes change recommendations in * lists * * @param element the list to check, can be UL or LI * returns wether the list is fake or not */ private isFakeList(element: Element): boolean { if (element.firstElementChild && element.classList.contains('os-split-after')) { // either first child has split-before or last child has split-after const firstChild = element.firstElementChild; const lastChild = element.childNodes[element.childNodes.length - 1] as Element; const splitBefore = firstChild.nodeName === 'LI' && firstChild.classList.contains('os-split-before'); const splitAfter = lastChild.nodeName === 'LI' && lastChild.classList.contains('os-split-after'); return splitBefore || splitAfter; } return false; } /** * Helper function to safer extract a line number from an element * * @param element * @returns the line number of the element */ private getLineNumber(element: Element): number { if ( element && element.nodeName === 'SPAN' && element.getAttribute('class') && element.getAttribute('class').indexOf('os-line-number') > -1 ) { return +element.getAttribute('data-line-number'); } } /** * Extracts the style information from the given array * * @param styles an array of inline css styles (i.e. `style="margin: 10px"`) * @returns an object with style pdfmake compatible style information */ private computeStyle(styles: string[]): any { const styleObject: any = {}; if (styles && styles.length > 0) { for (const style of styles) { const styleDefinition = style .trim() .toLowerCase() .split(':'); const key = styleDefinition[0]; const value = styleDefinition[1]; if (styleDefinition.length === 2) { switch (key) { case 'padding-left': { styleObject.margin = [+value, 0, 0, 0]; break; } case 'font-size': { styleObject.fontSize = +value; break; } case 'text-align': { switch (value) { case 'right': case 'center': case 'justify': { styleObject.alignment = value; break; } } break; } case 'font-weight': { switch (value) { case 'bold': { styleObject.bold = true; break; } } break; } case 'text-decoration': { switch (value) { case 'underline': { styleObject.decoration = 'underline'; break; } case 'line-through': { styleObject.decoration = 'lineThrough'; break; } } break; } case 'font-style': { switch (value) { case 'italic': { styleObject.italics = true; break; } } break; } case 'color': { styleObject.color = this.parseColor(value); break; } case 'background-color': { styleObject.background = this.parseColor(value); break; } } } } } return styleObject; } /** * Returns the color in a hex format (e.g. #12ff00). * Also tries to convert RGB colors into hex values * * @param color color as string representation * @returns color as hex values for pdfmake */ private parseColor(color: string): string { const haxRegex = new RegExp('^#([0-9a-f]{3}|[0-9a-f]{6})$'); // e.g. `#fff` or `#ff0048` const rgbRegex = new RegExp('^rgb\\((\\d+),\\s*(\\d+),\\s*(\\d+)\\)$'); // e.g. rgb(0,255,34) or rgb(22, 0, 0) const nameRegex = new RegExp('^[a-z]+$'); if (haxRegex.test(color)) { return color; } else if (rgbRegex.test(color)) { const decimalColors = rgbRegex.exec(color).slice(1); for (let i = 0; i < 3; i++) { let decimalValue = +decimalColors[i]; if (decimalValue > 255) { decimalValue = 255; } let hexString = '0' + decimalValue.toString(16); hexString = hexString.slice(-2); decimalColors[i] = hexString; } return '#' + decimalColors.join(''); } else if (nameRegex.test(color)) { return color; } else { console.error('Could not parse color "' + color + '"'); return color; } } /** * Helper function to create valid doc definitions container elements for pdfmake * * @param name should be a pdfMake container element, like 'text' or 'stack' * @param content */ private create(name: string, content?: any): any { const container = {}; const docDef = content ? content : []; container[name] = docDef; return container; } }