From 7da60a0147116130a94274e4a20ae54dd7d59dea Mon Sep 17 00:00:00 2001
From: Akihiko Odaki <nekomanma@pixiv.co.jp>
Date: Sat, 31 Mar 2018 19:53:30 +0900
Subject: Store texts as HTML

---
 src/common/text/parse/core/syntax-highlighter.ts | 334 +++++++++++++++++++++++
 src/common/text/parse/elements/bold.ts           |  14 +
 src/common/text/parse/elements/code.ts           |  17 ++
 src/common/text/parse/elements/emoji.ts          |  14 +
 src/common/text/parse/elements/hashtag.ts        |  19 ++
 src/common/text/parse/elements/inline-code.ts    |  17 ++
 src/common/text/parse/elements/link.ts           |  19 ++
 src/common/text/parse/elements/mention.ts        |  17 ++
 src/common/text/parse/elements/quote.ts          |  14 +
 src/common/text/parse/elements/url.ts            |  14 +
 src/common/text/parse/index.ts                   |  72 +++++
 11 files changed, 551 insertions(+)
 create mode 100644 src/common/text/parse/core/syntax-highlighter.ts
 create mode 100644 src/common/text/parse/elements/bold.ts
 create mode 100644 src/common/text/parse/elements/code.ts
 create mode 100644 src/common/text/parse/elements/emoji.ts
 create mode 100644 src/common/text/parse/elements/hashtag.ts
 create mode 100644 src/common/text/parse/elements/inline-code.ts
 create mode 100644 src/common/text/parse/elements/link.ts
 create mode 100644 src/common/text/parse/elements/mention.ts
 create mode 100644 src/common/text/parse/elements/quote.ts
 create mode 100644 src/common/text/parse/elements/url.ts
 create mode 100644 src/common/text/parse/index.ts

(limited to 'src/common/text/parse')

diff --git a/src/common/text/parse/core/syntax-highlighter.ts b/src/common/text/parse/core/syntax-highlighter.ts
new file mode 100644
index 0000000000..c0396b1fc6
--- /dev/null
+++ b/src/common/text/parse/core/syntax-highlighter.ts
@@ -0,0 +1,334 @@
+function escape(text) {
+	return text
+		.replace(/>/g, '&gt;')
+		.replace(/</g, '&lt;');
+}
+
+// 文字数が多い順にソートします
+// そうしないと、「function」という文字列が与えられたときに「func」が先にマッチしてしまう可能性があるためです
+const _keywords = [
+	'true',
+	'false',
+	'null',
+	'nil',
+	'undefined',
+	'void',
+	'var',
+	'const',
+	'let',
+	'mut',
+	'dim',
+	'if',
+	'then',
+	'else',
+	'switch',
+	'match',
+	'case',
+	'default',
+	'for',
+	'each',
+	'in',
+	'while',
+	'loop',
+	'continue',
+	'break',
+	'do',
+	'goto',
+	'next',
+	'end',
+	'sub',
+	'throw',
+	'try',
+	'catch',
+	'finally',
+	'enum',
+	'delegate',
+	'function',
+	'func',
+	'fun',
+	'fn',
+	'return',
+	'yield',
+	'async',
+	'await',
+	'require',
+	'include',
+	'import',
+	'imports',
+	'export',
+	'exports',
+	'from',
+	'as',
+	'using',
+	'use',
+	'internal',
+	'module',
+	'namespace',
+	'where',
+	'select',
+	'struct',
+	'union',
+	'new',
+	'delete',
+	'this',
+	'super',
+	'base',
+	'class',
+	'interface',
+	'abstract',
+	'static',
+	'public',
+	'private',
+	'protected',
+	'virtual',
+	'partial',
+	'override',
+	'extends',
+	'implements',
+	'constructor'
+];
+
+const keywords = _keywords
+	.concat(_keywords.map(k => k[0].toUpperCase() + k.substr(1)))
+	.concat(_keywords.map(k => k.toUpperCase()))
+	.sort((a, b) => b.length - a.length);
+
+const symbols = [
+	'=',
+	'+',
+	'-',
+	'*',
+	'/',
+	'%',
+	'~',
+	'^',
+	'&',
+	'|',
+	'>',
+	'<',
+	'!',
+	'?'
+];
+
+const elements = [
+	// comment
+	code => {
+		if (code.substr(0, 2) != '//') return null;
+		const match = code.match(/^\/\/(.+?)(\n|$)/);
+		if (!match) return null;
+		const comment = match[0];
+		return {
+			html: `<span class="comment">${escape(comment)}</span>`,
+			next: comment.length
+		};
+	},
+
+	// block comment
+	code => {
+		const match = code.match(/^\/\*([\s\S]+?)\*\//);
+		if (!match) return null;
+		return {
+			html: `<span class="comment">${escape(match[0])}</span>`,
+			next: match[0].length
+		};
+	},
+
+	// string
+	code => {
+		if (!/^['"`]/.test(code)) return null;
+		const begin = code[0];
+		let str = begin;
+		let thisIsNotAString = false;
+		for (let i = 1; i < code.length; i++) {
+			const char = code[i];
+			if (char == '\\') {
+				str += char;
+				str += code[i + 1] || '';
+				i++;
+				continue;
+			} else if (char == begin) {
+				str += char;
+				break;
+			} else if (char == '\n' || i == (code.length - 1)) {
+				thisIsNotAString = true;
+				break;
+			} else {
+				str += char;
+			}
+		}
+		if (thisIsNotAString) {
+			return null;
+		} else {
+			return {
+				html: `<span class="string">${escape(str)}</span>`,
+				next: str.length
+			};
+		}
+	},
+
+	// regexp
+	code => {
+		if (code[0] != '/') return null;
+		let regexp = '';
+		let thisIsNotARegexp = false;
+		for (let i = 1; i < code.length; i++) {
+			const char = code[i];
+			if (char == '\\') {
+				regexp += char;
+				regexp += code[i + 1] || '';
+				i++;
+				continue;
+			} else if (char == '/') {
+				break;
+			} else if (char == '\n' || i == (code.length - 1)) {
+				thisIsNotARegexp = true;
+				break;
+			} else {
+				regexp += char;
+			}
+		}
+
+		if (thisIsNotARegexp) return null;
+		if (regexp == '') return null;
+		if (regexp[0] == ' ' && regexp[regexp.length - 1] == ' ') return null;
+
+		return {
+			html: `<span class="regexp">/${escape(regexp)}/</span>`,
+			next: regexp.length + 2
+		};
+	},
+
+	// label
+	code => {
+		if (code[0] != '@') return null;
+		const match = code.match(/^@([a-zA-Z_-]+?)\n/);
+		if (!match) return null;
+		const label = match[0];
+		return {
+			html: `<span class="label">${label}</span>`,
+			next: label.length
+		};
+	},
+
+	// number
+	(code, i, source) => {
+		const prev = source[i - 1];
+		if (prev && /[a-zA-Z]/.test(prev)) return null;
+		if (!/^[\-\+]?[0-9\.]+/.test(code)) return null;
+		const match = code.match(/^[\-\+]?[0-9\.]+/)[0];
+		if (match) {
+			return {
+				html: `<span class="number">${match}</span>`,
+				next: match.length
+			};
+		} else {
+			return null;
+		}
+	},
+
+	// nan
+	(code, i, source) => {
+		const prev = source[i - 1];
+		if (prev && /[a-zA-Z]/.test(prev)) return null;
+		if (code.substr(0, 3) == 'NaN') {
+			return {
+				html: `<span class="nan">NaN</span>`,
+				next: 3
+			};
+		} else {
+			return null;
+		}
+	},
+
+	// method
+	code => {
+		const match = code.match(/^([a-zA-Z_-]+?)\(/);
+		if (!match) return null;
+
+		if (match[1] == '-') return null;
+
+		return {
+			html: `<span class="method">${match[1]}</span>`,
+			next: match[1].length
+		};
+	},
+
+	// property
+	(code, i, source) => {
+		const prev = source[i - 1];
+		if (prev != '.') return null;
+
+		const match = code.match(/^[a-zA-Z0-9_-]+/);
+		if (!match) return null;
+
+		return {
+			html: `<span class="property">${match[0]}</span>`,
+			next: match[0].length
+		};
+	},
+
+	// keyword
+	(code, i, source) => {
+		const prev = source[i - 1];
+		if (prev && /[a-zA-Z]/.test(prev)) return null;
+
+		const match = keywords.filter(k => code.substr(0, k.length) == k)[0];
+		if (match) {
+			if (/^[a-zA-Z]/.test(code.substr(match.length))) return null;
+			return {
+				html: `<span class="keyword ${match}">${match}</span>`,
+				next: match.length
+			};
+		} else {
+			return null;
+		}
+	},
+
+	// symbol
+	code => {
+		const match = symbols.filter(s => code[0] == s)[0];
+		if (match) {
+			return {
+				html: `<span class="symbol">${match}</span>`,
+				next: 1
+			};
+		} else {
+			return null;
+		}
+	}
+];
+
+// specify lang is todo
+export default (source: string, lang?: string) => {
+	let code = source;
+	let html = '';
+
+	let i = 0;
+
+	function push(token) {
+		html += token.html;
+		code = code.substr(token.next);
+		i += token.next;
+	}
+
+	while (code != '') {
+		const parsed = elements.some(el => {
+			const e = el(code, i, source);
+			if (e) {
+				push(e);
+				return true;
+			} else {
+				return false;
+			}
+		});
+
+		if (!parsed) {
+			push({
+				html: escape(code[0]),
+				next: 1
+			});
+		}
+	}
+
+	return html;
+};
diff --git a/src/common/text/parse/elements/bold.ts b/src/common/text/parse/elements/bold.ts
new file mode 100644
index 0000000000..ce25764457
--- /dev/null
+++ b/src/common/text/parse/elements/bold.ts
@@ -0,0 +1,14 @@
+/**
+ * Bold
+ */
+
+module.exports = text => {
+	const match = text.match(/^\*\*(.+?)\*\*/);
+	if (!match) return null;
+	const bold = match[0];
+	return {
+		type: 'bold',
+		content: bold,
+		bold: bold.substr(2, bold.length - 4)
+	};
+};
diff --git a/src/common/text/parse/elements/code.ts b/src/common/text/parse/elements/code.ts
new file mode 100644
index 0000000000..4821e95fe2
--- /dev/null
+++ b/src/common/text/parse/elements/code.ts
@@ -0,0 +1,17 @@
+/**
+ * Code (block)
+ */
+
+import genHtml from '../core/syntax-highlighter';
+
+module.exports = text => {
+	const match = text.match(/^```([\s\S]+?)```/);
+	if (!match) return null;
+	const code = match[0];
+	return {
+		type: 'code',
+		content: code,
+		code: code.substr(3, code.length - 6).trim(),
+		html: genHtml(code.substr(3, code.length - 6).trim())
+	};
+};
diff --git a/src/common/text/parse/elements/emoji.ts b/src/common/text/parse/elements/emoji.ts
new file mode 100644
index 0000000000..e24231a223
--- /dev/null
+++ b/src/common/text/parse/elements/emoji.ts
@@ -0,0 +1,14 @@
+/**
+ * Emoji
+ */
+
+module.exports = text => {
+	const match = text.match(/^:[a-zA-Z0-9+-_]+:/);
+	if (!match) return null;
+	const emoji = match[0];
+	return {
+		type: 'emoji',
+		content: emoji,
+		emoji: emoji.substr(1, emoji.length - 2)
+	};
+};
diff --git a/src/common/text/parse/elements/hashtag.ts b/src/common/text/parse/elements/hashtag.ts
new file mode 100644
index 0000000000..ee57b140b8
--- /dev/null
+++ b/src/common/text/parse/elements/hashtag.ts
@@ -0,0 +1,19 @@
+/**
+ * Hashtag
+ */
+
+module.exports = (text, i) => {
+	if (!(/^\s#[^\s]+/.test(text) || (i == 0 && /^#[^\s]+/.test(text)))) return null;
+	const isHead = text[0] == '#';
+	const hashtag = text.match(/^\s?#[^\s]+/)[0];
+	const res: any[] = !isHead ? [{
+		type: 'text',
+		content: text[0]
+	}] : [];
+	res.push({
+		type: 'hashtag',
+		content: isHead ? hashtag : hashtag.substr(1),
+		hashtag: isHead ? hashtag.substr(1) : hashtag.substr(2)
+	});
+	return res;
+};
diff --git a/src/common/text/parse/elements/inline-code.ts b/src/common/text/parse/elements/inline-code.ts
new file mode 100644
index 0000000000..9f9ef51a2b
--- /dev/null
+++ b/src/common/text/parse/elements/inline-code.ts
@@ -0,0 +1,17 @@
+/**
+ * Code (inline)
+ */
+
+import genHtml from '../core/syntax-highlighter';
+
+module.exports = text => {
+	const match = text.match(/^`(.+?)`/);
+	if (!match) return null;
+	const code = match[0];
+	return {
+		type: 'inline-code',
+		content: code,
+		code: code.substr(1, code.length - 2).trim(),
+		html: genHtml(code.substr(1, code.length - 2).trim())
+	};
+};
diff --git a/src/common/text/parse/elements/link.ts b/src/common/text/parse/elements/link.ts
new file mode 100644
index 0000000000..35563ddc3d
--- /dev/null
+++ b/src/common/text/parse/elements/link.ts
@@ -0,0 +1,19 @@
+/**
+ * Link
+ */
+
+module.exports = text => {
+	const match = text.match(/^\??\[([^\[\]]+?)\]\((https?:\/\/[\w\/:%#@\$&\?!\(\)\[\]~\.=\+\-]+?)\)/);
+	if (!match) return null;
+	const silent = text[0] == '?';
+	const link = match[0];
+	const title = match[1];
+	const url = match[2];
+	return {
+		type: 'link',
+		content: link,
+		title: title,
+		url: url,
+		silent: silent
+	};
+};
diff --git a/src/common/text/parse/elements/mention.ts b/src/common/text/parse/elements/mention.ts
new file mode 100644
index 0000000000..2025dfdaad
--- /dev/null
+++ b/src/common/text/parse/elements/mention.ts
@@ -0,0 +1,17 @@
+/**
+ * Mention
+ */
+import parseAcct from '../../../../common/user/parse-acct';
+
+module.exports = text => {
+	const match = text.match(/^(?:@[a-zA-Z0-9\-]+){1,2}/);
+	if (!match) return null;
+	const mention = match[0];
+	const { username, host } = parseAcct(mention.substr(1));
+	return {
+		type: 'mention',
+		content: mention,
+		username,
+		host
+	};
+};
diff --git a/src/common/text/parse/elements/quote.ts b/src/common/text/parse/elements/quote.ts
new file mode 100644
index 0000000000..cc8cfffdc4
--- /dev/null
+++ b/src/common/text/parse/elements/quote.ts
@@ -0,0 +1,14 @@
+/**
+ * Quoted text
+ */
+
+module.exports = text => {
+	const match = text.match(/^"([\s\S]+?)\n"/);
+	if (!match) return null;
+	const quote = match[0];
+	return {
+		type: 'quote',
+		content: quote,
+		quote: quote.substr(1, quote.length - 2).trim(),
+	};
+};
diff --git a/src/common/text/parse/elements/url.ts b/src/common/text/parse/elements/url.ts
new file mode 100644
index 0000000000..1003aff9c3
--- /dev/null
+++ b/src/common/text/parse/elements/url.ts
@@ -0,0 +1,14 @@
+/**
+ * URL
+ */
+
+module.exports = text => {
+	const match = text.match(/^https?:\/\/[\w\/:%#@\$&\?!\(\)\[\]~\.=\+\-]+/);
+	if (!match) return null;
+	const url = match[0];
+	return {
+		type: 'url',
+		content: url,
+		url: url
+	};
+};
diff --git a/src/common/text/parse/index.ts b/src/common/text/parse/index.ts
new file mode 100644
index 0000000000..1e2398dc38
--- /dev/null
+++ b/src/common/text/parse/index.ts
@@ -0,0 +1,72 @@
+/**
+ * Misskey Text Analyzer
+ */
+
+const elements = [
+	require('./elements/bold'),
+	require('./elements/url'),
+	require('./elements/link'),
+	require('./elements/mention'),
+	require('./elements/hashtag'),
+	require('./elements/code'),
+	require('./elements/inline-code'),
+	require('./elements/quote'),
+	require('./elements/emoji')
+];
+
+export default (source: string) => {
+
+	if (source == '') {
+		return null;
+	}
+
+	const tokens = [];
+
+	function push(token) {
+		if (token != null) {
+			tokens.push(token);
+			source = source.substr(token.content.length);
+		}
+	}
+
+	let i = 0;
+
+	// パース
+	while (source != '') {
+		const parsed = elements.some(el => {
+			let _tokens = el(source, i);
+			if (_tokens) {
+				if (!Array.isArray(_tokens)) {
+					_tokens = [_tokens];
+				}
+				_tokens.forEach(push);
+				return true;
+			} else {
+				return false;
+			}
+		});
+
+		if (!parsed) {
+			push({
+				type: 'text',
+				content: source[0]
+			});
+		}
+
+		i++;
+	}
+
+	// テキストを纏める
+	tokens[0] = [tokens[0]];
+	return tokens.reduce((a, b) => {
+		if (a[a.length - 1].type == 'text' && b.type == 'text') {
+			const tail = a.pop();
+			return a.concat({
+				type: 'text',
+				content: tail.content + b.content
+			});
+		} else {
+			return a.concat(b);
+		}
+	});
+};
-- 
cgit v1.2.3-freya