| 1 | // Helper functions for HTML tag wrapping |
| 1 | 2 | function wrap(text, tag) { |
| 3 | if (!tag) { |
| 4 | throw new Error('Tag is required for wrapping'); |
| 5 | } |
| 6 | // text can be empty string, but not null/undefined |
| 7 | if (text === null || text === undefined) { |
| 8 | throw new Error('Text cannot be null or undefined'); |
| 9 | } |
| 2 | 10 | return `<${tag}>${text}</${tag}>`; |
| 3 | 11 | } |
| 4 | 12 | |
| 5 | | function isTag(text, tag) { |
| 6 | | return text.startsWith(`<${tag}>`); |
| 7 | | } |
| 13 | // Constants for better maintainability |
| 14 | const MARKDOWN_PATTERNS = { |
| 15 | STRONG: /__(.+?)__/g, |
| 16 | EMPHASIS: /_(.+?)_/g, |
| 17 | LIST_ITEM: /^\* /, |
| 18 | HEADER: /^#+/ |
| 19 | }; |
| 8 | 20 | |
| 9 | | function parser(markdown, delimiter, tag) { |
| 10 | | const pattern = new RegExp(`${delimiter}(.+)${delimiter}`); |
| 11 | | const replacement = `<${tag}>$1</${tag}>`; |
| 12 | | return markdown.replace(pattern, replacement); |
| 13 | | } |
| 21 | const HTML_TAGS = { |
| 22 | STRONG: 'strong', |
| 23 | EMPHASIS: 'em', |
| 24 | PARAGRAPH: 'p', |
| 25 | LIST: 'ul', |
| 26 | LIST_ITEM: 'li' |
| 27 | }; |
| 28 | |
| 29 | const MAX_HEADER_LEVEL = 6; |
| 14 | 30 | |
| 15 | | function parse__(markdown) { |
| 16 | | return parser(markdown, '__', 'strong'); |
| 31 | // Markdown inline element parsers |
| 32 | function parseStrong(markdown) { |
| 33 | const pattern = MARKDOWN_PATTERNS.STRONG; |
| 34 | const replacement = `<${HTML_TAGS.STRONG}>$1</${HTML_TAGS.STRONG}>`; |
| 35 | return markdown.replace(pattern, replacement); |
| 17 | 36 | } |
| 18 | 37 | |
| 19 | | function parse_(markdown) { |
| 20 | | return parser(markdown, '_', 'em'); |
| 38 | function parseEmphasis(markdown) { |
| 39 | const pattern = MARKDOWN_PATTERNS.EMPHASIS; |
| 40 | const replacement = `<${HTML_TAGS.EMPHASIS}>$1</${HTML_TAGS.EMPHASIS}>`; |
| 41 | return markdown.replace(pattern, replacement); |
| 21 | 42 | } |
| 22 | 43 | |
| 23 | | function parseText(markdown, list) { |
| 24 | | const parsedText = parse_(parse__(markdown)); |
| 25 | | if (list) { |
| 26 | | return parsedText; |
| 27 | | } else { |
| 28 | | return wrap(parsedText, 'p'); |
| 29 | | } |
| 44 | function parseText(markdown, isListItem = false) { |
| 45 | // Parse inline elements (__ for strong, _ for emphasis) |
| 46 | const parsedText = parseEmphasis(parseStrong(markdown)); |
| 47 | |
| 48 | // Wrap in paragraph tag unless it's a list item |
| 49 | return isListItem ? parsedText : wrap(parsedText, HTML_TAGS.PARAGRAPH); |
| 30 | 50 | } |
| 31 | 51 | |
| 32 | | function parseHeader(markdown, list) { |
| 33 | | let count = 0; |
| 34 | | for (let i = 0; i < markdown.length; i++) { |
| 35 | | if (markdown[i] === '#') { |
| 36 | | count += 1; |
| 52 | // Markdown block element parsers |
| 53 | function parseHeader(markdown, isCurrentlyInList) { |
| 54 | // Count leading # characters to determine header level |
| 55 | let headerLevel = 0; |
| 56 | for (const char of markdown) { |
| 57 | if (char === '#') { |
| 58 | headerLevel++; |
| 37 | 59 | } else { |
| 38 | 60 | break; |
| 39 | 61 | } |
| 40 | 62 | } |
| 41 | | if (count === 0 || count > 6) { |
| 42 | | return [null, list]; |
| 63 | |
| 64 | // Return null if not a valid header (no # or more than max allowed) |
| 65 | if (headerLevel === 0 || headerLevel > MAX_HEADER_LEVEL) { |
| 66 | return [null, isCurrentlyInList]; |
| 43 | 67 | } |
| 44 | | const headerTag = `h${count}`; |
| 45 | | const headerHtml = wrap(markdown.substring(count + 1), headerTag); |
| 46 | | if (list) { |
| 47 | | return [`</ul>${headerHtml}`, false]; |
| 48 | | } else { |
| 49 | | return [headerHtml, false]; |
| 68 | |
| 69 | // Extract header content (skip # and space) |
| 70 | const headerContent = markdown.substring(headerLevel + 1); |
| 71 | const headerTag = `h${headerLevel}`; |
| 72 | const headerHtml = wrap(headerContent, headerTag); |
| 73 | |
| 74 | // Close list if we were in one |
| 75 | if (isCurrentlyInList) { |
| 76 | return [`</${HTML_TAGS.LIST}>${headerHtml}`, false]; |
| 50 | 77 | } |
| 78 | |
| 79 | return [headerHtml, false]; |
| 51 | 80 | } |
| 52 | 81 | |
| 53 | | function parseLineItem(markdown, list) { |
| 54 | | if (markdown.startsWith('*')) { |
| 55 | | const innerHtml = wrap(parseText(markdown.substring(2), true), 'li'); |
| 56 | | if (list) { |
| 57 | | return [innerHtml, true]; |
| 58 | | } else { |
| 59 | | return [`<ul>${innerHtml}`, true]; |
| 60 | | } |
| 82 | function parseLineItem(markdown, isCurrentlyInList) { |
| 83 | // Check if line starts with list item marker |
| 84 | if (!MARKDOWN_PATTERNS.LIST_ITEM.test(markdown)) { |
| 85 | return [null, isCurrentlyInList]; |
| 86 | } |
| 87 | |
| 88 | // Extract list item content (skip '* ') |
| 89 | const itemContent = markdown.substring(2); |
| 90 | const itemHtml = wrap(parseText(itemContent, true), HTML_TAGS.LIST_ITEM); |
| 91 | |
| 92 | // Open list if we weren't already in one |
| 93 | if (!isCurrentlyInList) { |
| 94 | return [`<${HTML_TAGS.LIST}>${itemHtml}`, true]; |
| 61 | 95 | } |
| 62 | | return [null, list]; |
| 96 | |
| 97 | return [itemHtml, true]; |
| 63 | 98 | } |
| 64 | 99 | |
| 65 | | function parseParagraph(markdown, list) { |
| 66 | | if (!list) { |
| 67 | | return [parseText(markdown, false), false]; |
| 68 | | } else { |
| 69 | | return [`</ul>${parseText(markdown, false)}`, false]; |
| 100 | function parseParagraph(markdown, isCurrentlyInList) { |
| 101 | const paragraphHtml = parseText(markdown, false); |
| 102 | |
| 103 | // Close list if we were in one |
| 104 | if (isCurrentlyInList) { |
| 105 | return [`</${HTML_TAGS.LIST}>${paragraphHtml}`, false]; |
| 70 | 106 | } |
| 107 | |
| 108 | return [paragraphHtml, false]; |
| 71 | 109 | } |
| 72 | 110 | |
| 73 | | function parseLine(markdown, list) { |
| 74 | | let [result, inListAfter] = parseHeader(markdown, list); |
| 111 | // Main parsing functions |
| 112 | function parseLine(markdown, isCurrentlyInList) { |
| 113 | // Try parsing as header |
| 114 | let [result, newListState] = parseHeader(markdown, isCurrentlyInList); |
| 115 | |
| 116 | // If not a header, try parsing as list item |
| 75 | 117 | if (result === null) { |
| 76 | | [result, inListAfter] = parseLineItem(markdown, list); |
| 118 | [result, newListState] = parseLineItem(markdown, isCurrentlyInList); |
| 77 | 119 | } |
| 120 | |
| 121 | // If not a list item, parse as paragraph |
| 78 | 122 | if (result === null) { |
| 79 | | [result, inListAfter] = parseParagraph(markdown, list); |
| 123 | [result, newListState] = parseParagraph(markdown, isCurrentlyInList); |
| 80 | 124 | } |
| 125 | |
| 126 | // This should never happen with valid input |
| 81 | 127 | if (result === null) { |
| 82 | | throw new Error('Remove this line and implement the function'); |
| 128 | throw new Error('Invalid markdown line'); |
| 83 | 129 | } |
| 84 | | return [result, inListAfter]; |
| 130 | |
| 131 | return [result, newListState]; |
| 85 | 132 | } |
| 86 | 133 | |
| 87 | 134 | /** |
| 88 | | * @param {string} markdown |
| 89 | | * @returns {string} |
| 135 | * Parses markdown text and returns HTML |
| 136 | * @param {string} markdown - The markdown text to parse |
| 137 | * @returns {string} The corresponding HTML |
| 90 | 138 | */ |
| 91 | 139 | export function parse(markdown) { |
| 140 | // Validate input |
| 141 | if (typeof markdown !== 'string') { |
| 142 | throw new Error('Markdown input must be a string'); |
| 143 | } |
| 144 | |
| 92 | 145 | const lines = markdown.split('\n'); |
| 93 | | let result = ''; |
| 94 | | let list = false; |
| 95 | | for (let i = 0; i < lines.length; i++) { |
| 96 | | let [lineResult, newList] = parseLine(lines[i], list); |
| 97 | | result += lineResult; |
| 98 | | list = newList; |
| 146 | let htmlResult = ''; |
| 147 | let isInList = false; |
| 148 | |
| 149 | // Process each line of markdown |
| 150 | for (const line of lines) { |
| 151 | const [lineHtml, newInListState] = parseLine(line, isInList); |
| 152 | htmlResult += lineHtml; |
| 153 | isInList = newInListState; |
| 99 | 154 | } |
| 100 | | if (list) { |
| 101 | | return result + '</ul>'; |
| 102 | | } else { |
| 103 | | return result; |
| 155 | |
| 156 | // Close any unclosed list at the end |
| 157 | if (isInList) { |
| 158 | htmlResult += `</${HTML_TAGS.LIST}>`; |
| 104 | 159 | } |
| 160 | |
| 161 | return htmlResult; |
| 105 | 162 | } |