Merge pull request #85 from xpadev-net/refactor/input-parser

[改善] 入力のパース処理を改善

src/@types/index.ts

@@ -10,6 +10,7 @@ export * from "./format.legacy";
 export * from "./format.owner";
 export * from "./format.v1";
 export * from "./IComment";
+export * from "./input-parser";
 export * from "./IPlugins";
 export * from "./options";
 export * from "./renderer";

src/@types/input-parser.ts

@@ -0,0 +1,6 @@
+import type { FormattedComment } from "@/@types/format.formatted";
+
+export interface InputParser {
+  key: string[];
+  parse: (input: unknown) => FormattedComment[];
+}

src/input/empty.ts

@@ -0,0 +1,8 @@
+import type { FormattedComment, InputParser } from "@/@types";
+
+export const EmptyParser: InputParser = {
+  key: ["empty"],
+  parse: (): FormattedComment[] => {
+    return [];
+  },
+};

src/input/formatted.ts

@@ -0,0 +1,11 @@
+import { array, parse } from "valibot";
+
+import type { FormattedComment, InputParser } from "@/@types";
+import { ZFormattedComment } from "@/@types";
+
+export const FormattedParser: InputParser = {
+  key: ["formatted", "niconicome"],
+  parse: (input: unknown): FormattedComment[] => {
+    return parse(array(ZFormattedComment), input);
+  },
+};

src/input/index.ts

@@ -0,0 +1,19 @@
+import type { InputParser } from "@/@types";
+
+import { EmptyParser } from "./empty";
+import { FormattedParser } from "./formatted";
+import { LegacyParser } from "./legacy";
+import { LegacyOwnerParser } from "./legacyOwner";
+import { OwnerParser } from "./owner";
+import { V1Parser } from "./v1";
+import { XmlDocumentParser } from "./xmlDocument";
+
+export const parsers: InputParser[] = [
+  EmptyParser,
+  FormattedParser,
+  LegacyParser,
+  LegacyOwnerParser,
+  OwnerParser,
+  V1Parser,
+  XmlDocumentParser,
+];

src/input/legacy.ts

@@ -0,0 +1,61 @@
+import { array, parse, safeParse } from "valibot";
+
+import type { InputParser } from "@/@types";
+import {
+  type FormattedComment,
+  type RawApiResponse,
+  ZApiChat,
+  ZRawApiResponse,
+} from "@/@types";
+
+export const LegacyParser: InputParser = {
+  key: ["legacy"],
+  parse: (input) => {
+    return fromLegacy(parse(array(ZRawApiResponse), input));
+  },
+};
+
+/**
+ * ニコニコ公式のlegacy apiから帰ってきたデータ処理する
+ * @param data legacy apiから帰ってきたデータ
+ * @returns 変換後のデータ
+ */
+const fromLegacy = (data: RawApiResponse[]): FormattedComment[] => {
+  const data_: FormattedComment[] = [],
+    userList: string[] = [];
+  for (const _val of data) {
+    const val = safeParse(ZApiChat, _val.chat);
+    if (!val.success) continue;
+    const value = val.output;
+    if (value.deleted !== 1) {
+      const tmpParam: FormattedComment = {
+        id: value.no,
+        vpos: value.vpos,
+        content: value.content || "",
+        date: value.date,
+        date_usec: value.date_usec || 0,
+        owner: !value.user_id,
+        premium: value.premium === 1,
+        mail: [],
+        user_id: -1,
+        layer: -1,
+        is_my_post: false,
+      };
+      if (value.mail) {
+        tmpParam.mail = value.mail.split(/\s+/g);
+      }
+      if (value.content.startsWith("/") && !value.user_id) {
+        tmpParam.mail.push("invisible");
+      }
+      const isUserExist = userList.indexOf(value.user_id);
+      if (isUserExist === -1) {
+        tmpParam.user_id = userList.length;
+        userList.push(value.user_id);
+      } else {
+        tmpParam.user_id = isUserExist;
+      }
+      data_.push(tmpParam);
+    }
+  }
+  return data_;
+};

src/input/legacyOwner.ts

@@ -0,0 +1,55 @@
+import type { InputParser } from "@/@types";
+import { type FormattedComment } from "@/@types";
+import { InvalidFormatError } from "@/errors";
+import typeGuard from "@/typeGuard";
+
+export const LegacyOwnerParser: InputParser = {
+  key: ["legacyOwner"],
+  parse: (input) => {
+    if (!typeGuard.legacyOwner.comments(input)) throw new InvalidFormatError();
+    return fromLegacyOwner(input);
+  },
+};
+
+/**
+ * 旧プレイヤーの投稿者コメントのエディターのデータを処理する
+ * @param data 旧投米のテキストデータ
+ * @returns 変換後のデータ
+ */
+const fromLegacyOwner = (data: string): FormattedComment[] => {
+  const data_: FormattedComment[] = [],
+    comments = data.split("\n");
+  for (let i = 0, n = comments.length; i < n; i++) {
+    const value = comments[i];
+    if (!value) continue;
+    const commentData = value.split(":");
+    if (commentData.length < 3) {
+      continue;
+    } else if (commentData.length > 3) {
+      for (let j = 3, n = commentData.length; j < n; j++) {
+        commentData[2] += `:${commentData[j]}`;
+      }
+    }
+    const tmpParam: FormattedComment = {
+      id: i,
+      vpos: Number(commentData[0]) * 100,
+      content: commentData[2] ?? "",
+      date: i,
+      date_usec: 0,
+      owner: true,
+      premium: true,
+      mail: [],
+      user_id: -1,
+      layer: -1,
+      is_my_post: false,
+    };
+    if (commentData[1]) {
+      tmpParam.mail = commentData[1].split(/[\s+]/g);
+    }
+    if (tmpParam.content.startsWith("/")) {
+      tmpParam.mail.push("invisible");
+    }
+    data_.push(tmpParam);
+  }
+  return data_;
+};

src/input/owner.ts

@@ -0,0 +1,82 @@
+import { array, parse } from "valibot";
+
+import type { InputParser } from "@/@types";
+import {
+  type FormattedComment,
+  type OwnerComment,
+  ZOwnerComment,
+} from "@/@types";
+
+export const OwnerParser: InputParser = {
+  key: ["owner"],
+  parse: (input) => {
+    return fromOwner(parse(array(ZOwnerComment), input));
+  },
+};
+
+/**
+ * 投稿者コメントのエディターのデータを処理する
+ * @param data 投米のデータ
+ * @returns 変換後のデータ
+ */
+const fromOwner = (data: OwnerComment[]): FormattedComment[] => {
+  const data_: FormattedComment[] = [];
+  for (let i = 0, n = data.length; i < n; i++) {
+    const value = data[i];
+    if (!value) continue;
+    const tmpParam: FormattedComment = {
+      id: i,
+      vpos: time2vpos(value.time),
+      content: value.comment,
+      date: i,
+      date_usec: 0,
+      owner: true,
+      premium: true,
+      mail: [],
+      user_id: -1,
+      layer: -1,
+      is_my_post: false,
+    };
+    if (value.command) {
+      tmpParam.mail = value.command.split(/\s+/g);
+    }
+    if (tmpParam.content.startsWith("/")) {
+      tmpParam.mail.push("invisible");
+    }
+    data_.push(tmpParam);
+  }
+  return data_;
+};
+
+/**
+ * 投稿者コメントのエディターは秒数の入力フォーマットに割りと色々対応しているのでvposに変換
+ * @param input 分:秒.秒・分:秒・秒.秒・秒
+ * @returns vpos
+ */
+const time2vpos = (input: string): number => {
+  const time = RegExp(
+    /^(?:(\d+):(\d+)\.(\d+)|(\d+):(\d+)|(\d+)\.(\d+)|(\d+))$/,
+  ).exec(input);
+  if (time) {
+    if (
+      time[1] !== undefined &&
+      time[2] !== undefined &&
+      time[3] !== undefined
+    ) {
+      return (
+        (Number(time[1]) * 60 + Number(time[2])) * 100 +
+        Number(time[3]) / Math.pow(10, time[3].length - 2)
+      );
+    } else if (time[4] !== undefined && time[5] !== undefined) {
+      return (Number(time[4]) * 60 + Number(time[5])) * 100;
+    } else if (time[6] !== undefined && time[7] !== undefined) {
+      return (
+        Number(time[6]) * 100 +
+        Number(time[7]) / Math.pow(10, time[7].length - 2)
+      );
+    } else if (time[8] !== undefined) {
+      return Number(time[8]) * 100;
+    }
+  }
+  return 0;
+};

src/input/v1.ts

@@ -0,0 +1,60 @@
+import { array, parse } from "valibot";
+
+import type { InputParser } from "@/@types";
+import { type FormattedComment, type V1Thread, ZV1Thread } from "@/@types";
+
+export const V1Parser: InputParser = {
+  key: ["v1"],
+  parse: (input: unknown) => {
+    return fromV1(parse(array(ZV1Thread), input));
+  },
+};
+
+/**
+ * ニコニコ公式のv1 apiから帰ってきたデータ処理する
+ * data内threadsのデータを渡されることを想定
+ * @param data v1 apiから帰ってきたデータ
+ * @returns 変換後のデータ
+ */
+const fromV1 = (data: V1Thread[]): FormattedComment[] => {
+  const data_: FormattedComment[] = [],
+    userList: string[] = [];
+  for (const item of data) {
+    const val = item.comments,
+      forkName = item.fork;
+    for (const value of val) {
+      const tmpParam: FormattedComment = {
+        id: value.no,
+        vpos: Math.floor(value.vposMs / 10),
+        content: value.body,
+        date: date2time(value.postedAt),
+        date_usec: 0,
+        owner: forkName === "owner",
+        premium: value.isPremium,
+        mail: value.commands,
+        user_id: -1,
+        layer: -1,
+        is_my_post: value.isMyPost,
+      };
+      if (tmpParam.content.startsWith("/") && tmpParam.owner) {
+        tmpParam.mail.push("invisible");
+      }
+      const isUserExist = userList.indexOf(value.userId);
+      if (isUserExist === -1) {
+        tmpParam.user_id = userList.length;
+        userList.push(value.userId);
+      } else {
+        tmpParam.user_id = isUserExist;
+      }
+      data_.push(tmpParam);
+    }
+  }
+  return data_;
+};
+
+/**
+ * v1 apiのpostedAtはISO 8601のtimestampなのでDate関数を使ってunix timestampに変換
+ * @param date ISO 8601 timestamp
+ * @returns unix timestamp
+ */
+const date2time = (date: string): number => Math.floor(Date.parse(date) / 1000);

src/input/xmlDocument.ts

@@ -0,0 +1,54 @@
+import type { FormattedComment, InputParser } from "@/@types";
+import { InvalidFormatError } from "@/errors";
+import typeGuard from "@/typeGuard";
+
+export const XmlDocumentParser: InputParser = {
+  key: ["formatted", "niconicome"],
+  parse: (input: unknown): FormattedComment[] => {
+    if (!typeGuard.xmlDocument(input)) throw new InvalidFormatError();
+    return parseXMLDocument(input);
+  },
+};
+
+/**
+ * niconicome等が吐き出すxml形式のコメントデータを処理する
+ * @param data 吐き出されたxmlをDOMParserでparseFromStringしたもの
+ * @returns 変換後のデータ
+ */
+const parseXMLDocument = (data: XMLDocument): FormattedComment[] => {
+  const data_: FormattedComment[] = [],
+    userList: string[] = [];
+  let index = Array.from(data.documentElement.children).length;
+  for (const item of Array.from(data.documentElement.children)) {
+    if (item.nodeName !== "chat") continue;
+    const tmpParam: FormattedComment = {
+      id: Number(item.getAttribute("no")) || index++,
+      vpos: Number(item.getAttribute("vpos")),
+      content: item.textContent ?? "",
+      date: Number(item.getAttribute("date")) || 0,
+      date_usec: Number(item.getAttribute("date_usec")) || 0,
+      owner: !item.getAttribute("user_id"),
+      premium: item.getAttribute("premium") === "1",
+      mail: [],
+      user_id: -1,
+      layer: -1,
+      is_my_post: false,
+    };
+    if (item.getAttribute("mail")) {
+      tmpParam.mail = item.getAttribute("mail")?.split(/\s+/g) ?? [];
+    }
+    if (tmpParam.content.startsWith("/") && tmpParam.owner) {
+      tmpParam.mail.push("invisible");
+    }
+    const userId = item.getAttribute("user_id") ?? "";
+    const isUserExist = userList.indexOf(userId);
+    if (isUserExist === -1) {
+      tmpParam.user_id = userList.length;
+      userList.push(userId);
+    } else {
+      tmpParam.user_id = isUserExist;
+    }
+    data_.push(tmpParam);
+  }
+  return data_;
+};