document-ir 0.0.14 → 0.1.0

package/script/WhitespaceTransformer.js

@@ -1,79 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.WhitespaceTransformer = void 0;
-const IdentityTransformer_js_1 = require("./IdentityTransformer.js");
-class RemoveEmptyTextTransformer extends IdentityTransformer_js_1.IdentityTransformer {
-    // deno-lint-ignore require-await
-    async text(node) {
-        if (node.text == "") {
-            return null;
-        }
-        return node;
-    }
-}
-class WhitespaceTransformer extends IdentityTransformer_js_1.IdentityTransformer {
-    constructor() {
-        super();
-        Object.defineProperty(this, "stripWhitespace", {
-            enumerable: true,
-            configurable: true,
-            writable: true,
-            value: void 0
-        });
-        Object.defineProperty(this, "lastText", {
-            enumerable: true,
-            configurable: true,
-            writable: true,
-            value: void 0
-        });
-        this.stripWhitespace = true;
-        this.lastText = null;
-    }
-    // deno-lint-ignore require-await
-    async text(node) {
-        let result = "";
-        for (const c of node.text) {
-            if (c == " " || c == "\n" || c == "\t" || c == "\r") {
-                if (!this.stripWhitespace) {
-                    result += " ";
-                    this.stripWhitespace = true;
-                }
-            }
-            else {
-                result += c;
-                this.stripWhitespace = false;
-            }
-        }
-        if (result.length == 0) {
-            return null;
-        }
-        const text = {
-            type: "text",
-            text: result,
-        };
-        this.lastText = text;
-        return text;
-    }
-    stripLastText() {
-        if (this.lastText) {
-            if (this.lastText.text.endsWith(" ")) {
-                this.lastText.text = this.lastText.text.slice(0, -1);
-            }
-            this.lastText = null;
-        }
-        this.stripWhitespace = true;
-    }
-    // deno-lint-ignore require-await
-    async beforeBlock() {
-        this.stripLastText();
-    }
-    // deno-lint-ignore require-await
-    async afterBlock() {
-        this.stripLastText();
-    }
-    async transform(node) {
-        const result = await super.transform(node);
-        return await new RemoveEmptyTextTransformer().transform(result);
-    }
-}
-exports.WhitespaceTransformer = WhitespaceTransformer;

package/script/WordCountTransformer.js

@@ -1,87 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.WordCounterTransformer = void 0;
-const DocumentThinningTransformer_js_1 = require("./DocumentThinningTransformer.js");
-const WordCounterVisitor_js_1 = require("./WordCounterVisitor.js");
-const index_js_1 = require("./index.js");
-function convertHierarchy(parent) {
-    const docHierarchy = {
-        headerText: parent.header,
-        headerId: parent.headerId,
-        words: 0,
-        totalWords: 0,
-        children: [],
-    };
-    const visitor = new WordCounterVisitor_js_1.WordCounterVisitor();
-    for (const node of parent.nodes) {
-        visitor.visit(node);
-    }
-    docHierarchy.words = visitor.getCount();
-    docHierarchy.totalWords = docHierarchy.words;
-    for (const child of parent.children) {
-        const childHierarchy = convertHierarchy(child);
-        docHierarchy.children.push(childHierarchy);
-        docHierarchy.totalWords += childHierarchy.totalWords;
-    }
-    return docHierarchy;
-}
-class WordCounterTransformer extends index_js_1.IdentityTransformer {
-    constructor() {
-        super();
-    }
-    async transform(node) {
-        // Isolate it
-        const jsonNode = JSON.parse(JSON.stringify(node));
-        const thinned = await new DocumentThinningTransformer_js_1.DocumentThinningTransformer().transform(jsonNode);
-        const stack = [];
-        const root = {
-            header: node.title,
-            headerId: "title",
-            nodes: [],
-            children: [],
-            depth: 1,
-        };
-        stack.push(root);
-        let depth = 1;
-        for (const node of thinned.content) {
-            if (node.type == "header") {
-                if (node.level == 1) {
-                    // never pop the root
-                    continue;
-                }
-                else if (node.level <= depth) {
-                    for (let i = stack.length - 1; i > 0; i--) {
-                        if (stack[i].depth >= node.level) {
-                            stack.pop();
-                        }
-                    }
-                }
-                const visitor = new index_js_1.TextVisitor();
-                visitor.visit(node);
-                const h = {
-                    header: visitor.getText(),
-                    depth: node.level,
-                    children: [],
-                    nodes: [],
-                };
-                if (node.htmlId) {
-                    h.headerId = node.htmlId;
-                }
-                stack[stack.length - 1].children.push(h);
-                stack.push(h);
-                depth = node.level;
-            }
-            else {
-                stack[stack.length - 1].nodes.push(node);
-            }
-        }
-        // The transformer does not actually walk through the document tree
-        // We just append a newly calculated hierarchy object
-        const doc = {
-            ...node,
-        };
-        doc.hierarchy = convertHierarchy(root);
-        return doc;
-    }
-}
-exports.WordCounterTransformer = WordCounterTransformer;

package/script/WordCounterVisitor.js

@@ -1,47 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.WordCounterVisitor = void 0;
-const index_js_1 = require("./index.js");
-class WordCounterVisitor extends index_js_1.NodeVisitor {
-    constructor() {
-        super();
-        Object.defineProperty(this, "count", {
-            enumerable: true,
-            configurable: true,
-            writable: true,
-            value: void 0
-        });
-        Object.defineProperty(this, "texts", {
-            enumerable: true,
-            configurable: true,
-            writable: true,
-            value: void 0
-        });
-        this.count = 0;
-        this.texts = [];
-    }
-    countText() {
-        if (this.texts.length > 0) {
-            this.count += this.texts.join("").split(" ").length;
-            this.texts = [];
-        }
-    }
-    beforeBlock() {
-        this.countText();
-    }
-    afterBlock() {
-        this.countText();
-    }
-    text(node) {
-        this.texts.push(node.text);
-    }
-    document(node) {
-        super.document(node);
-        this.countText();
-    }
-    getCount() {
-        this.countText();
-        return this.count;
-    }
-}
-exports.WordCounterVisitor = WordCounterVisitor;

package/script/index.d.ts

@@ -1,11 +0,0 @@
-export * from "./types.js";
-export { ArrayCollapseTransformer } from "./ArrayCollapseTransformer.js";
-export { IdentityTransformer } from "./IdentityTransformer.js";
-export { TextCollapseTransformer } from "./TextCollapseTransformer.js";
-export { WhitespaceTransformer } from "./WhitespaceTransformer.js";
-export { WhitespaceStretchingTransformer } from "./WhitespaceStretchingTransformer.js";
-export { NodeVisitor } from "./NodeVisitor.js";
-export { TextVisitor } from "./TextVisitor.js";
-export { WordCounterTransformer } from "./WordCountTransformer.js";
-export { WordCounterVisitor } from "./WordCounterVisitor.js";
-export { DocumentThinningTransformer } from "./DocumentThinningTransformer.js";

package/script/index.js

@@ -1,38 +0,0 @@
-"use strict";
-var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
-    if (k2 === undefined) k2 = k;
-    var desc = Object.getOwnPropertyDescriptor(m, k);
-    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
-      desc = { enumerable: true, get: function() { return m[k]; } };
-    }
-    Object.defineProperty(o, k2, desc);
-}) : (function(o, m, k, k2) {
-    if (k2 === undefined) k2 = k;
-    o[k2] = m[k];
-}));
-var __exportStar = (this && this.__exportStar) || function(m, exports) {
-    for (var p in m) if (p !== "default" && !Object.prototype.hasOwnProperty.call(exports, p)) __createBinding(exports, m, p);
-};
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.DocumentThinningTransformer = exports.WordCounterVisitor = exports.WordCounterTransformer = exports.TextVisitor = exports.NodeVisitor = exports.WhitespaceStretchingTransformer = exports.WhitespaceTransformer = exports.TextCollapseTransformer = exports.IdentityTransformer = exports.ArrayCollapseTransformer = void 0;
-__exportStar(require("./types.js"), exports);
-var ArrayCollapseTransformer_js_1 = require("./ArrayCollapseTransformer.js");
-Object.defineProperty(exports, "ArrayCollapseTransformer", { enumerable: true, get: function () { return ArrayCollapseTransformer_js_1.ArrayCollapseTransformer; } });
-var IdentityTransformer_js_1 = require("./IdentityTransformer.js");
-Object.defineProperty(exports, "IdentityTransformer", { enumerable: true, get: function () { return IdentityTransformer_js_1.IdentityTransformer; } });
-var TextCollapseTransformer_js_1 = require("./TextCollapseTransformer.js");
-Object.defineProperty(exports, "TextCollapseTransformer", { enumerable: true, get: function () { return TextCollapseTransformer_js_1.TextCollapseTransformer; } });
-var WhitespaceTransformer_js_1 = require("./WhitespaceTransformer.js");
-Object.defineProperty(exports, "WhitespaceTransformer", { enumerable: true, get: function () { return WhitespaceTransformer_js_1.WhitespaceTransformer; } });
-var WhitespaceStretchingTransformer_js_1 = require("./WhitespaceStretchingTransformer.js");
-Object.defineProperty(exports, "WhitespaceStretchingTransformer", { enumerable: true, get: function () { return WhitespaceStretchingTransformer_js_1.WhitespaceStretchingTransformer; } });
-var NodeVisitor_js_1 = require("./NodeVisitor.js");
-Object.defineProperty(exports, "NodeVisitor", { enumerable: true, get: function () { return NodeVisitor_js_1.NodeVisitor; } });
-var TextVisitor_js_1 = require("./TextVisitor.js");
-Object.defineProperty(exports, "TextVisitor", { enumerable: true, get: function () { return TextVisitor_js_1.TextVisitor; } });
-var WordCountTransformer_js_1 = require("./WordCountTransformer.js");
-Object.defineProperty(exports, "WordCounterTransformer", { enumerable: true, get: function () { return WordCountTransformer_js_1.WordCounterTransformer; } });
-var WordCounterVisitor_js_1 = require("./WordCounterVisitor.js");
-Object.defineProperty(exports, "WordCounterVisitor", { enumerable: true, get: function () { return WordCounterVisitor_js_1.WordCounterVisitor; } });
-var DocumentThinningTransformer_js_1 = require("./DocumentThinningTransformer.js");
-Object.defineProperty(exports, "DocumentThinningTransformer", { enumerable: true, get: function () { return DocumentThinningTransformer_js_1.DocumentThinningTransformer; } });

package/script/package.json

@@ -1,3 +0,0 @@
-{
-  "type": "commonjs"
-}

package/script/types.d.ts

@@ -1,343 +0,0 @@
-export interface ArrayNode {
-    type: "array";
-    content: Node[];
-}
-export interface BlockNode {
-    type: "block";
-    content: Node[];
-}
-export interface BlockQuoteNode {
-    type: "block-quote";
-    content: Node[];
-}
-export interface BoldNode {
-    type: "bold";
-    content: Node[];
-}
-export interface BreakNode {
-    type: "break";
-}
-export interface BubbleNode {
-    type: "bubble";
-    orientation: "left" | "right";
-    content: Node[];
-}
-export interface CenterNode {
-    type: "center";
-    content: Node[];
-}
-export interface CodeNode {
-    type: "code";
-    content: Node[];
-}
-export interface ColumnsNode {
-    type: "columns";
-    "column-count": 1 | 2 | 3 | 4;
-    columns: Node[][];
-}
-export interface DefinitionNode {
-    type: "definition";
-    content: Node[];
-    title: Node[];
-    abbreviation: Node[];
-    key: string;
-}
-export interface DefinitionListNode {
-    type: "definition-list";
-    content: DefinitionNode[];
-}
-export interface DefinitionReferenceNode {
-    type: "definition-reference";
-    definition: {
-        abbreviation: Node[];
-        key: string;
-    };
-    content: Node[];
-}
-export interface EmbedImagePreview {
-    url: string;
-    blurhash: string;
-    width: number;
-    height: number;
-}
-export interface YoutubeEmbed {
-    type: "youtube";
-    id: string;
-    imagePreview?: EmbedImagePreview;
-}
-export interface EmbedNode {
-    type: "embed";
-    content: YoutubeEmbed;
-}
-export interface EmojiNode {
-    type: "emoji";
-    url: string;
-    alt: string;
-}
-export interface FigureNode {
-    type: "figure";
-    content: Node[];
-}
-export interface FigureCaptionNode {
-    type: "figure-caption";
-    content: Node[];
-}
-export interface FigureImageNode {
-    type: "figure-image";
-    content: Node[];
-    width?: number;
-    height?: number;
-    blurhash?: string;
-    url: string;
-    image?: string;
-    alt: string;
-    hero?: true;
-}
-export interface FormattedTextNode {
-    type: "formatted-text";
-    language?: string;
-    text: string;
-}
-export interface HeaderNode {
-    type: "header";
-    level: 1 | 2 | 3 | 4 | 5 | 6;
-    content: Node[];
-    htmlId?: string;
-}
-export interface HighTechAlertNode {
-    type: "high-tech-alert";
-    content: Node[];
-    warning: Node[];
-}
-export interface HorizontalRuleNode {
-    type: "horizontal-rule";
-}
-export interface ImageNode {
-    type: "image";
-    url: string;
-    alt: string;
-    width?: number;
-    height?: number;
-    blurhash?: string;
-    image?: string;
-    hero?: true;
-}
-export interface ItalicNode {
-    type: "italic";
-    content: Node[];
-}
-export interface LinkNode {
-    type: "link";
-    content: Node[];
-    url: string;
-    title?: string;
-    target?: "_blank" | "_self" | "_top";
-    userGeneratedContent?: true;
-    noReferrer?: true;
-    noFollow?: true;
-    noOpener?: true;
-}
-export interface ListNode {
-    type: "list";
-    style: "ordered" | "unordered";
-    content: ListItem[];
-}
-export interface ListItem {
-    type: "list-item";
-    content: Node[];
-}
-export interface NoteNode {
-    type: "note";
-    content: Node[];
-}
-export interface ParagraphNode {
-    type: "paragraph";
-    content: Node[];
-}
-export interface QuoteNode {
-    type: "quote";
-    name: string;
-    icon: string;
-    url?: string;
-    orientation?: "left" | "right";
-    content: Node[];
-}
-export interface RedactedNode {
-    type: "redacted";
-    style: "inline" | "block";
-    content: Node[];
-}
-export interface RegionNode {
-    type: "region";
-    mode: "deny" | "allow";
-    content: Node[];
-    regions: string;
-}
-export interface ScriptNode {
-    type: "script";
-    "mime-type": string;
-    source: string;
-}
-export interface SecretNode {
-    type: "secret";
-    content: Node[];
-}
-export interface SmallerNode {
-    type: "smaller";
-    content: Node[];
-}
-export interface StrikeThroughNode {
-    type: "strike-through";
-    content: Node[];
-}
-export interface StickerNode {
-    type: "sticker";
-    size?: number;
-    orientation: "center" | "left" | "right";
-    character: string;
-    content: Node[];
-    name: string;
-    width?: number;
-    height?: number;
-}
-export interface TableNode {
-    type: "table";
-    content: TableCellNode[][];
-}
-export interface TableCellNode {
-    type: "table-cell";
-    header?: true;
-    span: [number, number];
-    content: Node[];
-}
-export interface TextNode {
-    type: "text";
-    text: string;
-}
-export interface UnderlineNode {
-    type: "underline";
-    content: Node[];
-}
-export interface VideoNode {
-    type: "video";
-    poster: string;
-    alt: string;
-    mp4: string;
-    width?: number;
-    webm?: string;
-    blurhash?: string;
-    height?: number;
-    muted?: true;
-    autoplay?: true;
-    loop?: true;
-    content?: Node[];
-    controls?: true;
-}
-export interface WarningNode {
-    type: "warning";
-    content: Node[];
-}
-export interface TweetNode {
-    type: "tweet";
-    id: string;
-}
-export interface TootNode {
-    type: "toot";
-    id: string;
-}
-export interface VimeoNode {
-    type: "vimeo";
-    id: string;
-}
-export interface YoutubeNode {
-    type: "youtube";
-    id: string;
-}
-export type SocialNode = TweetNode | TootNode | VimeoNode | YoutubeNode;
-export interface CardHeader {
-    type: "card-header";
-    imageUrl?: string;
-    imageBlurhash?: string;
-    backgroundImage?: string;
-    backgroundColor?: string;
-    backgroundBlurhash?: string;
-    title: Node[];
-    url?: string;
-    username?: string;
-    usernameDomain?: string;
-}
-export interface CardMedia {
-    type: "card-media";
-    content: (ImageNode | VideoNode | EmbedNode)[];
-}
-export interface CardAttribution {
-    type: "card-attribution";
-    title?: Node[];
-    url?: string;
-    date?: string;
-    archiveUrl?: string;
-}
-export interface CardContent {
-    type: "card-content";
-    content: Node[];
-}
-export type CardContentNode = CardHeader | CardContent | CardMedia | CardAttribution;
-export interface CardNode {
-    type: "card";
-    header?: CardHeader;
-    content?: CardContent;
-    media?: CardMedia;
-    attribution?: CardAttribution;
-    original?: Node;
-}
-export interface SuperTextNode {
-    type: "super";
-    content: Node[];
-}
-export interface SubTextNode {
-    type: "sub";
-    content: Node[];
-}
-export interface DateNode {
-    type: "date";
-    isoDate: string;
-    content: Node[];
-}
-export interface TimeNode {
-    type: "time";
-    isoTime: string;
-    content: Node[];
-}
-export interface DateTimeNode {
-    type: "datetime";
-    iso8601: string;
-    content: Node[];
-}
-export type Node = ArrayNode | BlockNode | BlockQuoteNode | BoldNode | BreakNode | BubbleNode | CardNode | CenterNode | CodeNode | ColumnsNode | DefinitionNode | DefinitionListNode | DefinitionReferenceNode | EmbedNode | EmojiNode | FigureNode | FigureCaptionNode | FigureImageNode | FormattedTextNode | HeaderNode | HighTechAlertNode | HorizontalRuleNode | ImageNode | ItalicNode | LinkNode | ListNode | NoteNode | ParagraphNode | QuoteNode | RedactedNode | RegionNode | ScriptNode | SecretNode | SmallerNode | StickerNode | StrikeThroughNode | TextNode | TableNode | SocialNode | UnderlineNode | VideoNode | DateNode | TimeNode | DateTimeNode | SuperTextNode | SubTextNode | WarningNode;
-export interface DocumentMeta {
-    hidden?: boolean;
-    noindex?: boolean;
-    title: string;
-    author?: string;
-    description?: string;
-    image?: string;
-    guid?: string;
-    "pub-date"?: number;
-    date?: string;
-    url: string;
-    contentDigest?: string;
-    readingDifficultyMultiplier?: number;
-}
-export interface DocumentHierarchy {
-    headerText: string;
-    headerId?: string;
-    words: number;
-    totalWords: number;
-    children: DocumentHierarchy[];
-}
-export interface DocumentNode extends DocumentMeta {
-    type: "document";
-    content: Node[];
-    definitions?: DefinitionNode[];
-    hierarchy?: DocumentHierarchy;
-}

package/script/types.js

@@ -1,2 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });