Merge pull request #169 from takker99:push-metadata

feat(websocket): Push all page metadata

browser/websocket/__snapshots__/findMetadata.test.ts.snap

@@ -0,0 +1,33 @@
+export const snapshot = {};
+
+snapshot[`findMetadata() 1`] = `
+[
+  [
+    "ふつうの",
+    "リンク2",
+    "hashtag",
+  ],
+  [
+    "/help-jp/外部リンク",
+  ],
+  [
+    "scrapbox",
+    "takker",
+  ],
+  "https://scrapbox.io/files/65f29c24974fd8002333b160.svg",
+  [
+    "65f29c24974fd8002333b160",
+    "65e7f82e03949c0024a367d0",
+    "65e7f4413bc95600258481fb",
+  ],
+  [
+    "助けてhelpfeel!!",
+  ],
+  [
+    "名前	[scrapbox.icon]",
+    "住所	[リンク2]を入れること",
+    "電話番号	#をつけてもリンクにならないよ",
+    "自分の強み	3個くらい列挙",
+  ],
+]
+`;

browser/websocket/findMetadata.test.ts

@@ -0,0 +1,35 @@
+import { findMetadata, getHelpfeels } from "./findMetadata.ts";
+import { assertEquals, assertSnapshot } from "../../deps/testing.ts";
+
+const text = `てすと
+[ふつうの]リンク
+　しかし\`これは[リンク]\`ではない
+
+code:code
+ コードブロック中の[リンク]や画像[https://scrapbox.io/files/65f29c0c9045b5002522c8bb.svg]は無視される
+
+
+   ? 助けてhelpfeel!!
+
+   table:infobox
+    名前	[scrapbox.icon]
+    住所	[リンク2]を入れること
+    電話番号	#をつけてもリンクにならないよ
+    自分の強み	3個くらい列挙
+
+#hashtag もつけるといいぞ？
+[/forum-jp]のようなリンクは対象外
+ [/help-jp/]もだめ
+ [/icons/なるほど.icon][takker.icon]
+[/help-jp/外部リンク]
+
+サムネを用意
+[https://scrapbox.io/files/65f29c24974fd8002333b160.svg]
+
+[https://scrapbox.io/files/65e7f4413bc95600258481fb.svg https://scrapbox.io/files/65e7f82e03949c0024a367d0.svg]`;
+
+Deno.test("findMetadata()", (t) => assertSnapshot(t, findMetadata(text)));
+Deno.test("getHelpfeels()", () =>
+  assertEquals(getHelpfeels(text.split("\n").map((text) => ({ text }))), [
+    "助けてhelpfeel!!",
+  ]));

browser/websocket/findMetadata.ts

@@ -0,0 +1,179 @@
+import { BaseLine, Node, parse } from "../../deps/scrapbox.ts";
+import { toTitleLc } from "../../title.ts";
+import { parseYoutube } from "../../parser/youtube.ts";
+
+/** テキストに含まれているメタデータを取り出す
+ *
+ * @param text Scrapboxのテキスト
+ * @return 順に、links, projectLinks, icons, image, files, helpfeels, infoboxDefinition
+ */
+export const findMetadata = (
+  text: string,
+): [
+  string[],
+  string[],
+  string[],
+  string | null,
+  string[],
+  string[],
+  string[],
+] => {
+  const blocks = parse(text, { hasTitle: true }).flatMap((block) => {
+    switch (block.type) {
+      case "codeBlock":
+      case "title":
+        return [];
+      case "line":
+      case "table":
+        return block;
+    }
+  });
+
+  /** 重複判定用map
+   *
+   * bracket link とhashtagを区別できるようにしている
+   * - bracket linkならtrue
+   *
+   * linkの形状はbracket linkを優先している
+   */
+  const linksLc = new Map<string, boolean>();
+  const links = [] as string[];
+  const projectLinksLc = new Set<string>();
+  const projectLinks = [] as string[];
+  const iconsLc = new Set<string>();
+  const icons = [] as string[];
+  let image: string | null = null;
+  const files = new Set<string>();
+  const helpfeels = new Set<string>();
+
+  const fileUrlPattern = new RegExp(
+    `${
+      location?.origin ?? "https://scrapbox.io"
+    }/files/([a-z0-9]{24})(?:|\\.[a-zA-Z0-9]+)(?:|\\?[^\\s]*)$`,
+  );
+
+  const lookup = (node: Node) => {
+    switch (node.type) {
+      case "hashTag":
+        if (linksLc.has(toTitleLc(node.href))) return;
+        linksLc.set(toTitleLc(node.href), false);
+        links.push(node.href);
+        return;
+      case "link":
+        switch (node.pathType) {
+          case "relative": {
+            const link = cutId(node.href);
+            if (linksLc.get(toTitleLc(link))) return;
+            linksLc.set(toTitleLc(link), true);
+            links.push(link);
+            return;
+          }
+          case "root": {
+            const link = cutId(node.href);
+            // ignore `/project` or `/project/`
+            if (/^\/[\w\d-]+\/?$/.test(link)) return;
+            if (projectLinksLc.has(toTitleLc(link))) return;
+            projectLinksLc.add(toTitleLc(link));
+            projectLinks.push(link);
+            return;
+          }
+          case "absolute": {
+            const props = parseYoutube(node.href);
+            if (props && props.pathType !== "list") {
+              image ??= `https://i.ytimg.com/vi/${props.videoId}/mqdefault.jpg`;
+              return;
+            }
+            const fileId = node.href.match(fileUrlPattern)?.[1];
+            if (fileId) files.add(fileId);
+            return;
+          }
+          default:
+            return;
+        }
+      case "icon":
+      case "strongIcon": {
+        if (node.pathType === "root") return;
+        if (iconsLc.has(toTitleLc(node.path))) return;
+        iconsLc.add(toTitleLc(node.path));
+        icons.push(node.path);
+        return;
+      }
+      case "image":
+      case "strongImage": {
+        image ??= node.src.endsWith("/thumb/1000")
+          ? node.src.replace(/\/thumb\/1000$/, "/raw")
+          : node.src;
+        {
+          const fileId = node.src.match(fileUrlPattern)?.[1];
+          if (fileId) files.add(fileId);
+        }
+        if (node.type === "image") {
+          const fileId = node.link.match(fileUrlPattern)?.[1];
+          if (fileId) files.add(fileId);
+        }
+        return;
+      }
+      case "helpfeel":
+        helpfeels.add(node.text);
+        return;
+      case "numberList":
+      case "strong":
+      case "quote":
+      case "decoration": {
+        for (const n of node.nodes) {
+          lookup(n);
+        }
+        return;
+      }
+      default:
+        return;
+    }
+  };
+
+  const infoboxDefinition = [] as string[];
+
+  for (const block of blocks) {
+    switch (block.type) {
+      case "line":
+        for (const node of block.nodes) {
+          lookup(node);
+        }
+        continue;
+      case "table": {
+        for (const row of block.cells) {
+          for (const nodes of row) {
+            for (const node of nodes) {
+              lookup(node);
+            }
+          }
+        }
+        if (!["infobox", "cosense"].includes(block.fileName)) continue;
+        infoboxDefinition.push(
+          ...block.cells.map((row) =>
+            row.map((cell) => cell.map((node) => node.raw).join("")).join("\t")
+              .trim()
+          ),
+        );
+        continue;
+      }
+    }
+  }
+
+  return [
+    links,
+    projectLinks,
+    icons,
+    image,
+    [...files],
+    [...helpfeels],
+    infoboxDefinition,
+  ];
+};
+
+const cutId = (link: string): string => link.replace(/#[a-f\d]{24,32}$/, "");
+
+/** テキストからHelpfeel記法のentryだけ取り出す */
+export const getHelpfeels = (lines: Pick<BaseLine, "text">[]): string[] =>
+  lines.flatMap(({ text }) =>
+    /^\s*\? .*$/.test(text) ? [text.trimStart().slice(2)] : []
+  );

browser/websocket/isSameArray.test.ts

@@ -0,0 +1,10 @@
+import { isSameArray } from "./isSameArray.ts";
+import { assert } from "../../deps/testing.ts";
+
+Deno.test("isSameArray()", () => {
+  assert(isSameArray([1, 2, 3], [1, 2, 3]));
+  assert(isSameArray([1, 2, 3], [3, 2, 1]));
+  assert(!isSameArray([1, 2, 3], [3, 2, 3]));
+  assert(!isSameArray([1, 2, 3], [1, 2]));
+  assert(isSameArray([], []));
+});

browser/websocket/isSameArray.ts

@@ -0,0 +1,2 @@
+export const isSameArray = <T>(a: T[], b: T[]): boolean =>
+  a.length === b.length && a.every((x) => b.includes(x));