mintlify 1.0.6 → 1.0.8

package/bin/navigation.js

@@ -0,0 +1,4 @@
+export function isNavigation(navEntry) {
+    return typeof navEntry !== "string";
+}
+//# sourceMappingURL=navigation.js.map

package/bin/navigation.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"navigation.js","sourceRoot":"","sources":["../src/navigation.ts"],"names":[],"mappings":"AAOA,MAAM,UAAU,YAAY,CAC1B,QAAyB;IAEzB,OAAO,OAAO,QAAQ,KAAK,QAAQ,CAAC;AACtC,CAAC"}

package/bin/pageTemplate.js

@@ -0,0 +1,30 @@
+import inquirer from "inquirer";
+import { createPage } from "./util.js";
+const generatePageTemplate = () => {
+    inquirer
+        .prompt([
+        {
+            type: "input",
+            name: "title",
+            message: "What is the title of the new page?",
+        },
+        {
+            type: "input",
+            name: "description",
+            message: "What is the description?",
+            default: "",
+        },
+    ])
+        .then((answers) => {
+        const { title, description } = answers;
+        createPage(title, description);
+        console.log("🌱 Created initial files for Mintlify docs");
+        process.exit(0);
+    })
+        .catch((error) => {
+        console.error(error);
+        process.exit(1);
+    });
+};
+export default generatePageTemplate;
+//# sourceMappingURL=pageTemplate.js.map

package/bin/pageTemplate.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pageTemplate.js","sourceRoot":"","sources":["../src/pageTemplate.ts"],"names":[],"mappings":"AAAA,OAAO,QAAQ,MAAM,UAAU,CAAC;AAChC,OAAO,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AAEvC,MAAM,oBAAoB,GAAG,GAAG,EAAE;IAChC,QAAQ;SACL,MAAM,CAAC;QACN;YACE,IAAI,EAAE,OAAO;YACb,IAAI,EAAE,OAAO;YACb,OAAO,EAAE,oCAAoC;SAC9C;QACD;YACE,IAAI,EAAE,OAAO;YACb,IAAI,EAAE,aAAa;YACnB,OAAO,EAAE,0BAA0B;YACnC,OAAO,EAAE,EAAE;SACZ;KACF,CAAC;SACD,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE;QAChB,MAAM,EAAE,KAAK,EAAE,WAAW,EAAE,GAAG,OAAO,CAAC;QAEvC,UAAU,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;QAC/B,OAAO,CAAC,GAAG,CAAC,4CAA4C,CAAC,CAAC;QAC1D,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAClB,CAAC,CAAC;SACD,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE;QACf,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;QACrB,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAClB,CAAC,CAAC,CAAC;AACP,CAAC,CAAC;AAEF,eAAe,oBAAoB,CAAC"}

package/bin/scraping/combineNavWithEmptyGroupTitles.js

@@ -0,0 +1,20 @@
+export default function combineNavWithEmptyGroupTitles(navArray) {
+    let newNavArray = [];
+    navArray.forEach((nav) => {
+        // The first run through the loop will always have -1 as the index.
+        // JavaScript returns undefined when we look for an index outside the size of the array.
+        const prev = newNavArray[newNavArray.length - 1];
+        if (prev == null) {
+            newNavArray.push(nav);
+        }
+        else if (!nav.group && !prev.group) {
+            // Joins multiple groups without a title together IF they occur side by side
+            prev.pages = prev.pages.concat(nav.pages);
+        }
+        else {
+            newNavArray.push(nav);
+        }
+    });
+    return newNavArray;
+}
+//# sourceMappingURL=combineNavWithEmptyGroupTitles.js.map

package/bin/scraping/combineNavWithEmptyGroupTitles.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"combineNavWithEmptyGroupTitles.js","sourceRoot":"","sources":["../../src/scraping/combineNavWithEmptyGroupTitles.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,OAAO,UAAU,8BAA8B,CAAC,QAAsB;IAC3E,IAAI,WAAW,GAAG,EAAE,CAAC;IAErB,QAAQ,CAAC,OAAO,CAAC,CAAC,GAAe,EAAE,EAAE;QACnC,mEAAmE;QACnE,wFAAwF;QACxF,MAAM,IAAI,GAAG,WAAW,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACjD,IAAI,IAAI,IAAI,IAAI,EAAE;YAChB,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;SACvB;aAAM,IAAI,CAAC,GAAG,CAAC,KAAK,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE;YACpC,4EAA4E;YAC5E,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;SAC3C;aAAM;YACL,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;SACvB;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,WAAW,CAAC;AACrB,CAAC"}

package/bin/scraping/detectFramework.js

@@ -0,0 +1,25 @@
+import cheerio from "cheerio";
+export var Frameworks;
+(function (Frameworks) {
+    Frameworks["DOCUSAURUS"] = "DOCUSAURUS";
+    Frameworks["GITBOOK"] = "GITBOOK";
+    Frameworks["README"] = "README";
+})(Frameworks = Frameworks || (Frameworks = {}));
+export function detectFramework(html) {
+    const $ = cheerio.load(html);
+    const docusaurusMeta = $('meta[name="generator"]');
+    if (docusaurusMeta.length > 0 &&
+        docusaurusMeta.attr("content").includes("Docusaurus")) {
+        return Frameworks.DOCUSAURUS;
+    }
+    const isGitBook = $(".gitbook-root").length > 0;
+    if (isGitBook) {
+        return Frameworks.GITBOOK;
+    }
+    const isReadMe = $('meta[name="readme-deploy"]').length > 0;
+    if (isReadMe) {
+        return Frameworks.README;
+    }
+    return undefined;
+}
+//# sourceMappingURL=detectFramework.js.map

package/bin/scraping/detectFramework.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"detectFramework.js","sourceRoot":"","sources":["../../src/scraping/detectFramework.ts"],"names":[],"mappings":"AAAA,OAAO,OAAO,MAAM,SAAS,CAAC;AAE9B,MAAM,CAAN,IAAY,UAIX;AAJD,WAAY,UAAU;IACpB,uCAAyB,CAAA;IACzB,iCAAmB,CAAA;IACnB,+BAAiB,CAAA;AACnB,CAAC,EAJW,UAAU,GAAV,UAAU,KAAV,UAAU,QAIrB;AAED,MAAM,UAAU,eAAe,CAAC,IAAI;IAClC,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC7B,MAAM,cAAc,GAAG,CAAC,CAAC,wBAAwB,CAAC,CAAC;IAEnD,IACE,cAAc,CAAC,MAAM,GAAG,CAAC;QACzB,cAAc,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,QAAQ,CAAC,YAAY,CAAC,EACrD;QACA,OAAO,UAAU,CAAC,UAAU,CAAC;KAC9B;IAED,MAAM,SAAS,GAAG,CAAC,CAAC,eAAe,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;IAChD,IAAI,SAAS,EAAE;QACb,OAAO,UAAU,CAAC,OAAO,CAAC;KAC3B;IAED,MAAM,QAAQ,GAAG,CAAC,CAAC,4BAA4B,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;IAC5D,IAAI,QAAQ,EAAE;QACZ,OAAO,UAAU,CAAC,MAAM,CAAC;KAC1B;IAED,OAAO,SAAS,CAAC;AACnB,CAAC"}

package/bin/scraping/downloadAllImages.js

@@ -0,0 +1,57 @@
+import path from "path";
+import downloadImage from "../downloadImage.js";
+// To Do: Use CheerioElement instead of any when we bump the cheerio version
+export default async function downloadAllImages($, content, origin, baseDir, modifyFileName) {
+    if (!baseDir) {
+        console.debug("Skipping image downloading");
+        return;
+    }
+    // We remove duplicates because some frameworks duplicate img tags
+    // to show the image larger when clicked on.
+    const imageSrcs = [
+        ...new Set(content
+            .find("img[src]")
+            .map((i, image) => $(image).attr("src"))
+            .toArray()),
+    ];
+    // Wait to all images to download before continuing
+    const origToNewArray = await Promise.all(imageSrcs.map(async (origImageSrc) => {
+        // Add origin if the image tags are using relative sources
+        const imageHref = origImageSrc.startsWith("http")
+            ? origImageSrc
+            : new URL(origImageSrc, origin).href;
+        let fileName = removeMetadataFromExtension(path.basename(imageHref));
+        if (modifyFileName) {
+            fileName = modifyFileName(fileName);
+        }
+        if (!fileName) {
+            console.error("Invalid image path " + imageHref);
+            return;
+        }
+        const writePath = path.join(baseDir, fileName);
+        await downloadImage(imageHref, writePath)
+            .then(() => {
+            console.log("🖼️ - " + writePath);
+        })
+            .catch((e) => {
+            if (e.code === "EEXIST") {
+                console.log(`❌ Skipping existing image ${writePath}`);
+            }
+            else {
+                console.error(e);
+            }
+        });
+        return { [origImageSrc]: writePath };
+    }));
+    return origToNewArray.reduce((result, current) => Object.assign(result, current), {});
+}
+function removeMetadataFromExtension(src) {
+    // Part of the URL standard
+    const metadataSymbols = ["?", "#"];
+    metadataSymbols.forEach((dividerSymbol) => {
+        // Some frameworks add metadata after the file extension, we need to remove that.
+        src = src.split(dividerSymbol)[0];
+    });
+    return src;
+}
+//# sourceMappingURL=downloadAllImages.js.map

package/bin/scraping/downloadAllImages.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"downloadAllImages.js","sourceRoot":"","sources":["../../src/scraping/downloadAllImages.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,MAAM,CAAC;AACxB,OAAO,aAAa,MAAM,qBAAqB,CAAC;AAEhD,4EAA4E;AAC5E,MAAM,CAAC,OAAO,CAAC,KAAK,UAAU,iBAAiB,CAC7C,CAAM,EACN,OAAY,EACZ,MAAc,EACd,OAAe,EACf,cAAoB;IAEpB,IAAI,CAAC,OAAO,EAAE;QACZ,OAAO,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;QAC5C,OAAO;KACR;IAED,kEAAkE;IAClE,4CAA4C;IAC5C,MAAM,SAAS,GAAG;QAChB,GAAG,IAAI,GAAG,CACR,OAAO;aACJ,IAAI,CAAC,UAAU,CAAC;aAChB,GAAG,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;aACvC,OAAO,EAAE,CACb;KACF,CAAC;IAEF,mDAAmD;IACnD,MAAM,cAAc,GAAG,MAAM,OAAO,CAAC,GAAG,CACtC,SAAS,CAAC,GAAG,CAAC,KAAK,EAAE,YAAoB,EAAE,EAAE;QAC3C,0DAA0D;QAC1D,MAAM,SAAS,GAAG,YAAY,CAAC,UAAU,CAAC,MAAM,CAAC;YAC/C,CAAC,CAAC,YAAY;YACd,CAAC,CAAC,IAAI,GAAG,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC;QAEvC,IAAI,QAAQ,GAAG,2BAA2B,CAAC,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC;QACrE,IAAI,cAAc,EAAE;YAClB,QAAQ,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC;SACrC;QAED,IAAI,CAAC,QAAQ,EAAE;YACb,OAAO,CAAC,KAAK,CAAC,qBAAqB,GAAG,SAAS,CAAC,CAAC;YACjD,OAAO;SACR;QAED,MAAM,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;QAE/C,MAAM,aAAa,CAAC,SAAS,EAAE,SAAS,CAAC;aACtC,IAAI,CAAC,GAAG,EAAE;YACT,OAAO,CAAC,GAAG,CAAC,QAAQ,GAAG,SAAS,CAAC,CAAC;QACpC,CAAC,CAAC;aACD,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE;YACX,IAAI,CAAC,CAAC,IAAI,KAAK,QAAQ,EAAE;gBACvB,OAAO,CAAC,GAAG,CAAC,6BAA6B,SAAS,EAAE,CAAC,CAAC;aACvD;iBAAM;gBACL,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;aAClB;QACH,CAAC,CAAC,CAAC;QAEL,OAAO,EAAE,CAAC,YAAY,CAAC,EAAE,SAAS,EAAE,CAAC;IACvC,CAAC,CAAC,CACH,CAAC;IAEF,OAAO,cAAc,CAAC,MAAM,CAC1B,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACnD,EAAE,CACH,CAAC;AACJ,CAAC;AAED,SAAS,2BAA2B,CAAC,GAAW;IAC9C,2BAA2B;IAC3B,MAAM,eAAe,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;IAEnC,eAAe,CAAC,OAAO,CAAC,CAAC,aAAa,EAAE,EAAE;QACxC,iFAAiF;QACjF,GAAG,GAAG,GAAG,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC;IACpC,CAAC,CAAC,CAAC;IACH,OAAO,GAAG,CAAC;AACb,CAAC"}

package/bin/scraping/getSitemapLinks.js

@@ -0,0 +1,18 @@
+import axios from "axios";
+// Not in use.
+// Gets all links in a sitemap.
+export const getSitemapLinks = async (url) => {
+    const hostname = url.hostname.replace(".", "\\.");
+    const regex = new RegExp(`https?:\/\/${hostname}.+?(?=<\/loc>)`, "gmi");
+    try {
+        const indexData = (await axios.default.get(url.href)).data;
+        const array = indexData.match(regex);
+        return array || [];
+    }
+    catch (err) {
+        console.error(err);
+        console.log("Skipping sitemap links because we encountered an error.");
+        return [];
+    }
+};
+//# sourceMappingURL=getSitemapLinks.js.map

package/bin/scraping/getSitemapLinks.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"getSitemapLinks.js","sourceRoot":"","sources":["../../src/scraping/getSitemapLinks.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAE1B,cAAc;AACd,+BAA+B;AAC/B,MAAM,CAAC,MAAM,eAAe,GAAG,KAAK,EAAE,GAAQ,EAAE,EAAE;IAChD,MAAM,QAAQ,GAAG,GAAG,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;IAClD,MAAM,KAAK,GAAG,IAAI,MAAM,CAAC,cAAc,QAAQ,gBAAgB,EAAE,KAAK,CAAC,CAAC;IAExE,IAAI;QACF,MAAM,SAAS,GAAG,CAAC,MAAM,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,IAAc,CAAC;QACrE,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,KAAK,CAAoB,CAAC;QACxD,OAAO,KAAK,IAAI,EAAE,CAAC;KACpB;IAAC,OAAO,GAAG,EAAE;QACZ,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QACnB,OAAO,CAAC,GAAG,CAAC,yDAAyD,CAAC,CAAC;QACvE,OAAO,EAAE,CAAC;KACX;AACH,CAAC,CAAC"}

package/bin/scraping/replaceImagePaths.js

@@ -0,0 +1,17 @@
+export default function replaceImagePaths(origToWritePath, cliDir, markdown) {
+    if (origToWritePath == null) {
+        return markdown;
+    }
+    // Change image paths to use the downloaded locations
+    for (const [origHref, writePath] of Object.entries(origToWritePath)) {
+        // Use relative paths within the folder we are in
+        if (writePath.startsWith(cliDir)) {
+            markdown = markdown.replaceAll(origHref, writePath.slice(cliDir.length));
+        }
+        else {
+            markdown = markdown.replaceAll(origHref, writePath);
+        }
+    }
+    return markdown;
+}
+//# sourceMappingURL=replaceImagePaths.js.map

package/bin/scraping/replaceImagePaths.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"replaceImagePaths.js","sourceRoot":"","sources":["../../src/scraping/replaceImagePaths.ts"],"names":[],"mappings":"AAAA,MAAM,CAAC,OAAO,UAAU,iBAAiB,CACvC,eAAuB,EACvB,MAAc,EACd,QAAgB;IAEhB,IAAI,eAAe,IAAI,IAAI,EAAE;QAC3B,OAAO,QAAQ,CAAC;KACjB;IAED,qDAAqD;IACrD,KAAK,MAAM,CAAC,QAAQ,EAAE,SAAS,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,eAAe,CAAC,EAAE;QACnE,iDAAiD;QACjD,IAAI,SAAS,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE;YAChC,QAAQ,GAAG,QAAQ,CAAC,UAAU,CAAC,QAAQ,EAAE,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC;SAC1E;aAAM;YACL,QAAQ,GAAG,QAAQ,CAAC,UAAU,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;SACrD;KACF;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC"}

package/bin/scraping/scrapeFileGettingFileNameFromUrl.js

@@ -0,0 +1,43 @@
+import path from "path";
+import axios from "axios";
+import { getHtmlWithPuppeteer } from "../browser.js";
+import { createPage } from "../util.js";
+export async function scrapeFileGettingFileNameFromUrl(pathname, cliDir, origin, overwrite, scrapePageFunc, puppeteer = false, baseToRemove) {
+    // Skip scraping external links
+    if (pathname.startsWith("https://") || pathname.startsWith("http://")) {
+        return pathname;
+    }
+    // Removes file name from the end
+    const splitSubpath = pathname.split("/");
+    let folders = splitSubpath.slice(0, splitSubpath.length - 1).join("/");
+    // Remove base dir if passed in
+    if (baseToRemove && folders.startsWith(baseToRemove)) {
+        folders = folders.replace(baseToRemove, "");
+    }
+    // TO DO: Improve this by putting each page's images in a separate
+    // folder named after the title of the page.
+    const imageBaseDir = path.join(cliDir, "images", folders);
+    // Scrape each page separately
+    const href = new URL(pathname, origin).href;
+    let html;
+    if (puppeteer) {
+        html = await getHtmlWithPuppeteer(href);
+    }
+    else {
+        const res = await axios.default.get(href);
+        html = res.data;
+    }
+    const { title, description, markdown } = await scrapePageFunc(html, origin, cliDir, imageBaseDir);
+    // Check if page didn't have content
+    if (!title && !markdown) {
+        return undefined;
+    }
+    const newFileLocation = folders ? path.join(cliDir, folders) : cliDir;
+    // Default to introduction.mdx if we encountered index.html
+    const fileName = splitSubpath[splitSubpath.length - 1] || "introduction";
+    // Will create subfolders as needed
+    createPage(title, description, markdown, overwrite, newFileLocation, fileName);
+    // Removes first slash if we are in a folder, Mintlify doesn't need it
+    return folders ? path.join(folders, fileName).substring(1) : fileName;
+}
+//# sourceMappingURL=scrapeFileGettingFileNameFromUrl.js.map

package/bin/scraping/scrapeFileGettingFileNameFromUrl.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapeFileGettingFileNameFromUrl.js","sourceRoot":"","sources":["../../src/scraping/scrapeFileGettingFileNameFromUrl.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,MAAM,CAAC;AACxB,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAExC,MAAM,CAAC,KAAK,UAAU,gCAAgC,CACpD,QAAgB,EAChB,MAAc,EACd,MAAc,EACd,SAAkB,EAClB,cASE,EACF,SAAS,GAAG,KAAK,EACjB,YAAqB;IAErB,+BAA+B;IAC/B,IAAI,QAAQ,CAAC,UAAU,CAAC,UAAU,CAAC,IAAI,QAAQ,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE;QACrE,OAAO,QAAQ,CAAC;KACjB;IAED,iCAAiC;IACjC,MAAM,YAAY,GAAG,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IACzC,IAAI,OAAO,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAEvE,+BAA+B;IAC/B,IAAI,YAAY,IAAI,OAAO,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE;QACpD,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC,CAAC;KAC7C;IAED,kEAAkE;IAClE,4CAA4C;IAC5C,MAAM,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAC;IAE1D,8BAA8B;IAC9B,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC;IAC5C,IAAI,IAAY,CAAC;IACjB,IAAI,SAAS,EAAE;QACb,IAAI,GAAG,MAAM,oBAAoB,CAAC,IAAI,CAAC,CAAC;KACzC;SAAM;QACL,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC1C,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC;KACjB;IAED,MAAM,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,GAAG,MAAM,cAAc,CAC3D,IAAI,EACJ,MAAM,EACN,MAAM,EACN,YAAY,CACb,CAAC;IAEF,oCAAoC;IACpC,IAAI,CAAC,KAAK,IAAI,CAAC,QAAQ,EAAE;QACvB,OAAO,SAAS,CAAC;KAClB;IAED,MAAM,eAAe,GAAG,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;IAEtE,2DAA2D;IAC3D,MAAM,QAAQ,GAAG,YAAY,CAAC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,IAAI,cAAc,CAAC;IAEzE,mCAAmC;IACnC,UAAU,CACR,KAAK,EACL,WAAW,EACX,QAAQ,EACR,SAAS,EACT,eAAe,EACf,QAAQ,CACT,CAAC;IAEF,sEAAsE;IACtE,OAAO,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC;AACxE,CAAC"}

package/bin/scraping/scrapeGettingFileNameFromUrl.js

@@ -0,0 +1,14 @@
+import { isNavigation } from "../navigation.js";
+import { scrapeFileGettingFileNameFromUrl } from "./scrapeFileGettingFileNameFromUrl.js";
+export async function scrapeGettingFileNameFromUrl(navEntry, cliDir, origin, overwrite, scrapePageFunc, puppeteer = false, baseToRemove) {
+    if (isNavigation(navEntry)) {
+        const newPages = [];
+        for (const nestedNavEntry of navEntry.pages) {
+            newPages.push(await scrapeGettingFileNameFromUrl(nestedNavEntry, cliDir, origin, overwrite, scrapePageFunc, puppeteer, baseToRemove));
+        }
+        navEntry.pages = newPages;
+        return navEntry;
+    }
+    return await scrapeFileGettingFileNameFromUrl(navEntry, cliDir, origin, overwrite, scrapePageFunc, puppeteer, baseToRemove);
+}
+//# sourceMappingURL=scrapeGettingFileNameFromUrl.js.map

package/bin/scraping/scrapeGettingFileNameFromUrl.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapeGettingFileNameFromUrl.js","sourceRoot":"","sources":["../../src/scraping/scrapeGettingFileNameFromUrl.ts"],"names":[],"mappings":"AAAA,OAAO,EAAmB,YAAY,EAAE,MAAM,kBAAkB,CAAC;AACjE,OAAO,EAAE,gCAAgC,EAAE,MAAM,uCAAuC,CAAC;AAEzF,MAAM,CAAC,KAAK,UAAU,4BAA4B,CAChD,QAAyB,EACzB,MAAc,EACd,MAAc,EACd,SAAkB,EAClB,cASE,EACF,SAAS,GAAG,KAAK,EACjB,YAAqB;IAErB,IAAI,YAAY,CAAC,QAAQ,CAAC,EAAE;QAC1B,MAAM,QAAQ,GAAG,EAAE,CAAC;QACpB,KAAK,MAAM,cAAc,IAAI,QAAQ,CAAC,KAAK,EAAE;YAC3C,QAAQ,CAAC,IAAI,CACX,MAAM,4BAA4B,CAChC,cAAc,EACd,MAAM,EACN,MAAM,EACN,SAAS,EACT,cAAc,EACd,SAAS,EACT,YAAY,CACb,CACF,CAAC;SACH;QACD,QAAQ,CAAC,KAAK,GAAG,QAAQ,CAAC;QAC1B,OAAO,QAAQ,CAAC;KACjB;IAED,OAAO,MAAM,gCAAgC,CAC3C,QAAQ,EACR,MAAM,EACN,MAAM,EACN,SAAS,EACT,cAAc,EACd,SAAS,EACT,YAAY,CACb,CAAC;AACJ,CAAC"}

package/bin/scraping/scrapePage.js

@@ -0,0 +1,9 @@
+import path from "path";
+import { createPage, getOrigin } from "../util.js";
+export async function scrapePage(scrapeFunc, href, html, overwrite) {
+    const origin = getOrigin(href);
+    const imageBaseDir = path.join(process.cwd(), "images");
+    const { title, description, markdown } = await scrapeFunc(html, origin, process.cwd(), imageBaseDir);
+    createPage(title, description, markdown, overwrite, process.cwd());
+}
+//# sourceMappingURL=scrapePage.js.map

package/bin/scraping/scrapePage.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapePage.js","sourceRoot":"","sources":["../../src/scraping/scrapePage.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,MAAM,CAAC;AACxB,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,MAAM,YAAY,CAAC;AAEnD,MAAM,CAAC,KAAK,UAAU,UAAU,CAC9B,UAKiB,EACjB,IAAY,EACZ,IAAY,EACZ,SAAkB;IAElB,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,QAAQ,CAAC,CAAC;IACxD,MAAM,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,GAAG,MAAM,UAAU,CACvD,IAAI,EACJ,MAAM,EACN,OAAO,CAAC,GAAG,EAAE,EACb,YAAY,CACb,CAAC;IACF,UAAU,CAAC,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC,CAAC;AACrE,CAAC"}

package/bin/scraping/scrapePageCommands.js

@@ -0,0 +1,48 @@
+import axios from "axios";
+import { scrapePage } from "./scrapePage.js";
+import { scrapeDocusaurusPage } from "./site-scrapers/scrapeDocusaurusPage.js";
+import { scrapeGitBookPage } from "./site-scrapers/scrapeGitBookPage.js";
+import { scrapeReadMePage } from "./site-scrapers/scrapeReadMePage.js";
+import { detectFramework, Frameworks } from "./detectFramework.js";
+import { getHrefFromArgs } from "../util.js";
+import { getHtmlWithPuppeteer } from "../browser.js";
+function validateFramework(framework) {
+    if (!framework) {
+        console.log("Could not detect the framework automatically. Please use one of:");
+        console.log("scrape-page-docusaurus");
+        console.log("scrape-page-gitbook");
+        console.log("scrape-page-readme");
+        return process.exit(1);
+    }
+}
+export async function scrapePageWrapper(argv, scrapeFunc, puppeteer = false) {
+    const href = getHrefFromArgs(argv);
+    let html;
+    if (puppeteer) {
+        html = await getHtmlWithPuppeteer(href);
+    }
+    else {
+        const res = await axios.default.get(href);
+        html = res.data;
+    }
+    await scrapePage(scrapeFunc, href, html, argv.overwrite);
+    process.exit(0);
+}
+export async function scrapePageAutomatically(argv) {
+    const href = getHrefFromArgs(argv);
+    const res = await axios.default.get(href);
+    const html = res.data;
+    const framework = detectFramework(html);
+    validateFramework(framework);
+    console.log("Detected framework: " + framework);
+    if (framework === Frameworks.DOCUSAURUS) {
+        await scrapePageWrapper(argv, scrapeDocusaurusPage);
+    }
+    else if (framework === Frameworks.GITBOOK) {
+        await scrapePageWrapper(argv, scrapeGitBookPage, true);
+    }
+    else if (framework === Frameworks.README) {
+        await scrapePageWrapper(argv, scrapeReadMePage);
+    }
+}
+//# sourceMappingURL=scrapePageCommands.js.map

package/bin/scraping/scrapePageCommands.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapePageCommands.js","sourceRoot":"","sources":["../../src/scraping/scrapePageCommands.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAC7C,OAAO,EAAE,oBAAoB,EAAE,MAAM,yCAAyC,CAAC;AAC/E,OAAO,EAAE,iBAAiB,EAAE,MAAM,sCAAsC,CAAC;AACzE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qCAAqC,CAAC;AACvE,OAAO,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AACnE,OAAO,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAC7C,OAAO,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAC;AAErD,SAAS,iBAAiB,CAAC,SAAS;IAClC,IAAI,CAAC,SAAS,EAAE;QACd,OAAO,CAAC,GAAG,CACT,kEAAkE,CACnE,CAAC;QACF,OAAO,CAAC,GAAG,CAAC,wBAAwB,CAAC,CAAC;QACtC,OAAO,CAAC,GAAG,CAAC,qBAAqB,CAAC,CAAC;QACnC,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,CAAC;QAClC,OAAO,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;KACxB;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,IAAI,EAAE,UAAU,EAAE,SAAS,GAAG,KAAK;IACzE,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IACnC,IAAI,IAAY,CAAC;IACjB,IAAI,SAAS,EAAE;QACb,IAAI,GAAG,MAAM,oBAAoB,CAAC,IAAI,CAAC,CAAC;KACzC;SAAM;QACL,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC1C,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC;KACjB;IACD,MAAM,UAAU,CAAC,UAAU,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;IACzD,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAAC,IAAS;IACrD,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IACnC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC;IACtB,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IAExC,iBAAiB,CAAC,SAAS,CAAC,CAAC;IAE7B,OAAO,CAAC,GAAG,CAAC,sBAAsB,GAAG,SAAS,CAAC,CAAC;IAEhD,IAAI,SAAS,KAAK,UAAU,CAAC,UAAU,EAAE;QACvC,MAAM,iBAAiB,CAAC,IAAI,EAAE,oBAAoB,CAAC,CAAC;KACrD;SAAM,IAAI,SAAS,KAAK,UAAU,CAAC,OAAO,EAAE;QAC3C,MAAM,iBAAiB,CAAC,IAAI,EAAE,iBAAiB,EAAE,IAAI,CAAC,CAAC;KACxD;SAAM,IAAI,SAAS,KAAK,UAAU,CAAC,MAAM,EAAE;QAC1C,MAAM,iBAAiB,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAC;KACjD;AACH,CAAC"}

package/bin/scraping/scrapeSection.js

@@ -0,0 +1,9 @@
+import { objToReadableString } from "../util.js";
+export async function scrapeSection(scrapeFunc, html, origin, overwrite) {
+    console.log(`Started scraping${overwrite ? ", overwrite mode is on" : ""}...`);
+    const groupsConfig = await scrapeFunc(html, origin, process.cwd(), overwrite);
+    console.log("Finished scraping.");
+    console.log("Add the following to your navigation in mint.json:");
+    console.log(objToReadableString(groupsConfig));
+}
+//# sourceMappingURL=scrapeSection.js.map

package/bin/scraping/scrapeSection.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapeSection.js","sourceRoot":"","sources":["../../src/scraping/scrapeSection.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAAE,MAAM,YAAY,CAAC;AAEjD,MAAM,CAAC,KAAK,UAAU,aAAa,CACjC,UAAe,EACf,IAAY,EACZ,MAAc,EACd,SAAkB;IAElB,OAAO,CAAC,GAAG,CACT,mBAAmB,SAAS,CAAC,CAAC,CAAC,wBAAwB,CAAC,CAAC,CAAC,EAAE,KAAK,CAClE,CAAC;IACF,MAAM,YAAY,GAAG,MAAM,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,CAAC,GAAG,EAAE,EAAE,SAAS,CAAC,CAAC;IAC9E,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,CAAC;IAClC,OAAO,CAAC,GAAG,CAAC,oDAAoD,CAAC,CAAC;IAClE,OAAO,CAAC,GAAG,CAAC,mBAAmB,CAAC,YAAY,CAAC,CAAC,CAAC;AACjD,CAAC"}

package/bin/scraping/scrapeSectionCommands.js

@@ -0,0 +1,90 @@
+import axios from "axios";
+import { detectFramework, Frameworks } from "./detectFramework.js";
+import { getHrefFromArgs, getOrigin } from "../util.js";
+import { scrapeSection } from "./scrapeSection.js";
+import { scrapeDocusaurusSection } from "./site-scrapers/scrapeDocusaurusSection.js";
+import { scrapeGitBookSection } from "./site-scrapers/scrapeGitBookSection.js";
+import { scrapeReadMeSection } from "./site-scrapers/scrapeReadMeSection.js";
+import { startBrowser } from "../browser.js";
+function validateFramework(framework) {
+    if (!framework) {
+        console.log("Could not detect the framework automatically. Please use one of:");
+        console.log("scrape-page-docusaurus");
+        console.log("scrape-page-gitbook");
+        console.log("scrape-page-readme");
+        return process.exit(1);
+    }
+}
+export async function scrapeSectionAxiosWrapper(argv, scrapeFunc) {
+    const href = getHrefFromArgs(argv);
+    const res = await axios.default.get(href);
+    const html = res.data;
+    await scrapeSection(scrapeFunc, html, getOrigin(href), argv.overwrite);
+    process.exit(0);
+}
+export async function scrapeGitbookSectionCommand(argv) {
+    await scrapeSectionGitBookWrapper(argv, scrapeGitBookSection);
+}
+async function scrapeSectionGitBookWrapper(argv, scrapeFunc) {
+    const href = getHrefFromArgs(argv);
+    const browser = await startBrowser();
+    const page = await browser.newPage();
+    await page.goto(href, {
+        waitUntil: "networkidle2",
+    });
+    let prevEncountered = [];
+    let encounteredHref = ["fake"];
+    // Loop until we've encountered every link
+    while (!encounteredHref.every((href) => prevEncountered.includes(href))) {
+        prevEncountered = encounteredHref;
+        encounteredHref = await page.evaluate((encounteredHref) => {
+            const icons = Array.from(document.querySelectorAll('path[d="M9 18l6-6-6-6"]'));
+            const linksFound = [];
+            icons.forEach(async (icon) => {
+                var _a, _b;
+                const toClick = (_a = icon === null || icon === void 0 ? void 0 : icon.parentElement) === null || _a === void 0 ? void 0 : _a.parentElement;
+                const link = (_b = toClick === null || toClick === void 0 ? void 0 : toClick.parentElement) === null || _b === void 0 ? void 0 : _b.parentElement;
+                // Skip icons not in the side navigation
+                if (!(link === null || link === void 0 ? void 0 : link.hasAttribute("href"))) {
+                    return;
+                }
+                const href = link.getAttribute("href");
+                // Should never occur but we keep it as a fail-safe
+                if ((href === null || href === void 0 ? void 0 : href.startsWith("https://")) || (href === null || href === void 0 ? void 0 : href.startsWith("http://"))) {
+                    return;
+                }
+                // Click any links we haven't seen before
+                if (href && !encounteredHref.includes(href)) {
+                    toClick === null || toClick === void 0 ? void 0 : toClick.click();
+                }
+                if (href) {
+                    linksFound.push(href);
+                }
+            });
+            return linksFound;
+        }, encounteredHref // Need to pass array into the browser
+        );
+    }
+    const html = await page.content();
+    browser.close();
+    await scrapeSection(scrapeFunc, html, getOrigin(href), argv.overwrite);
+    process.exit(0);
+}
+export async function scrapeSectionAutomatically(argv) {
+    const href = getHrefFromArgs(argv);
+    const res = await axios.default.get(href);
+    const html = res.data;
+    const framework = detectFramework(html);
+    validateFramework(framework);
+    console.log("Detected framework: " + framework);
+    if (framework === Frameworks.DOCUSAURUS) {
+        await scrapeSectionAxiosWrapper(argv, scrapeDocusaurusSection);
+    }
+    else if (framework === Frameworks.GITBOOK) {
+        await scrapeGitbookSectionCommand(argv);
+    }
+    else if (framework === Frameworks.README) {
+        await scrapeSectionAxiosWrapper(argv, scrapeReadMeSection);
+    }
+}
+//# sourceMappingURL=scrapeSectionCommands.js.map

package/bin/scraping/scrapeSectionCommands.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapeSectionCommands.js","sourceRoot":"","sources":["../../src/scraping/scrapeSectionCommands.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AACnE,OAAO,EAAE,eAAe,EAAE,SAAS,EAAE,MAAM,YAAY,CAAC;AACxD,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAE,uBAAuB,EAAE,MAAM,4CAA4C,CAAC;AACrF,OAAO,EAAE,oBAAoB,EAAE,MAAM,yCAAyC,CAAC;AAC/E,OAAO,EAAE,mBAAmB,EAAE,MAAM,wCAAwC,CAAC;AAC7E,OAAO,EAAE,YAAY,EAAE,MAAM,eAAe,CAAC;AAE7C,SAAS,iBAAiB,CAAC,SAAiC;IAC1D,IAAI,CAAC,SAAS,EAAE;QACd,OAAO,CAAC,GAAG,CACT,kEAAkE,CACnE,CAAC;QACF,OAAO,CAAC,GAAG,CAAC,wBAAwB,CAAC,CAAC;QACtC,OAAO,CAAC,GAAG,CAAC,qBAAqB,CAAC,CAAC;QACnC,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,CAAC;QAClC,OAAO,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;KACxB;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAAC,IAAS,EAAE,UAAe;IACxE,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IACnC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC;IACtB,MAAM,aAAa,CAAC,UAAU,EAAE,IAAI,EAAE,SAAS,CAAC,IAAI,CAAC,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;IACvE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAAC,IAAS;IACzD,MAAM,2BAA2B,CAAC,IAAI,EAAE,oBAAoB,CAAC,CAAC;AAChE,CAAC;AAED,KAAK,UAAU,2BAA2B,CAAC,IAAS,EAAE,UAAe;IACnE,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IAEnC,MAAM,OAAO,GAAG,MAAM,YAAY,EAAE,CAAC;IACrC,MAAM,IAAI,GAAG,MAAM,OAAO,CAAC,OAAO,EAAE,CAAC;IACrC,MAAM,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE;QACpB,SAAS,EAAE,cAAc;KAC1B,CAAC,CAAC;IAEH,IAAI,eAAe,GAAa,EAAE,CAAC;IACnC,IAAI,eAAe,GAAG,CAAC,MAAM,CAAC,CAAC;IAE/B,0CAA0C;IAC1C,OAAO,CAAC,eAAe,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,EAAE;QACvE,eAAe,GAAG,eAAe,CAAC;QAClC,eAAe,GAAG,MAAM,IAAI,CAAC,QAAQ,CACnC,CAAC,eAAe,EAAE,EAAE;YAClB,MAAM,KAAK,GAAkB,KAAK,CAAC,IAAI,CACrC,QAAQ,CAAC,gBAAgB,CAAC,yBAAyB,CAAC,CACrD,CAAC;YAEF,MAAM,UAAU,GAAa,EAAE,CAAC;YAChC,KAAK,CAAC,OAAO,CAAC,KAAK,EAAE,IAAiB,EAAE,EAAE;;gBACxC,MAAM,OAAO,GAAG,MAAA,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,aAAa,0CAAE,aAAa,CAAC;gBACnD,MAAM,IAAI,GAAG,MAAA,OAAO,aAAP,OAAO,uBAAP,OAAO,CAAE,aAAa,0CAAE,aAAa,CAAC;gBAEnD,wCAAwC;gBACxC,IAAI,CAAC,CAAA,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,YAAY,CAAC,MAAM,CAAC,CAAA,EAAE;oBAC/B,OAAO;iBACR;gBAED,MAAM,IAAI,GAAG,IAAI,CAAC,YAAY,CAAC,MAAM,CAAC,CAAC;gBAEvC,mDAAmD;gBACnD,IAAI,CAAA,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,UAAU,CAAC,UAAU,CAAC,MAAI,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,UAAU,CAAC,SAAS,CAAC,CAAA,EAAE;oBAC/D,OAAO;iBACR;gBAED,yCAAyC;gBACzC,IAAI,IAAI,IAAI,CAAC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE;oBAC3C,OAAO,aAAP,OAAO,uBAAP,OAAO,CAAE,KAAK,EAAE,CAAC;iBAClB;gBACD,IAAI,IAAI,EAAE;oBACR,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;iBACvB;YACH,CAAC,CAAC,CAAC;YAEH,OAAO,UAAU,CAAC;QACpB,CAAC,EACD,eAAe,CAAC,sCAAsC;SACvD,CAAC;KACH;IAED,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC;IAClC,OAAO,CAAC,KAAK,EAAE,CAAC;IAChB,MAAM,aAAa,CAAC,UAAU,EAAE,IAAI,EAAE,SAAS,CAAC,IAAI,CAAC,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;IACvE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,IAAS;IACxD,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IACnC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC;IACtB,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;IAExC,iBAAiB,CAAC,SAAS,CAAC,CAAC;IAE7B,OAAO,CAAC,GAAG,CAAC,sBAAsB,GAAG,SAAS,CAAC,CAAC;IAEhD,IAAI,SAAS,KAAK,UAAU,CAAC,UAAU,EAAE;QACvC,MAAM,yBAAyB,CAAC,IAAI,EAAE,uBAAuB,CAAC,CAAC;KAChE;SAAM,IAAI,SAAS,KAAK,UAAU,CAAC,OAAO,EAAE;QAC3C,MAAM,2BAA2B,CAAC,IAAI,CAAC,CAAC;KACzC;SAAM,IAAI,SAAS,KAAK,UAAU,CAAC,MAAM,EAAE;QAC1C,MAAM,yBAAyB,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAC;KAC5D;AACH,CAAC"}

package/bin/scraping/site-scrapers/getLinksRecursively.js

@@ -0,0 +1,33 @@
+// Used by GitBook and ReadMe section scrapers
+export default function getLinksRecursively(linkSections, $) {
+    if (linkSections == null || linkSections.length === 0) {
+        return [];
+    }
+    return linkSections
+        .map((i, s) => {
+        const subsection = $(s);
+        const link = subsection.children().first();
+        const linkHref = link.attr("href");
+        // Skip missing links. For example, GitBook uses
+        // empty divs are used for styling a line beside the nav.
+        // Skip external links until Mintlify supports them
+        if (!linkHref ||
+            linkHref.startsWith("https://") ||
+            linkHref.startsWith("http://")) {
+            return undefined;
+        }
+        const childLinks = subsection.children().eq(1).children();
+        if (childLinks.length > 0) {
+            // Put the section link in the list of pages.
+            // When we support the section itself being a link we should update this
+            return {
+                group: link.text(),
+                pages: [linkHref, ...getLinksRecursively(childLinks, $)],
+            };
+        }
+        return linkHref;
+    })
+        .toArray()
+        .filter(Boolean);
+}
+//# sourceMappingURL=getLinksRecursively.js.map

package/bin/scraping/site-scrapers/getLinksRecursively.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"getLinksRecursively.js","sourceRoot":"","sources":["../../../src/scraping/site-scrapers/getLinksRecursively.ts"],"names":[],"mappings":"AAAA,8CAA8C;AAC9C,MAAM,CAAC,OAAO,UAAU,mBAAmB,CAAC,YAAiB,EAAE,CAAM;IACnE,IAAI,YAAY,IAAI,IAAI,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC,EAAE;QACrD,OAAO,EAAE,CAAC;KACX;IAED,OAAO,YAAY;SAChB,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;QACZ,MAAM,UAAU,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QACxB,MAAM,IAAI,GAAG,UAAU,CAAC,QAAQ,EAAE,CAAC,KAAK,EAAE,CAAC;QAE3C,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAEnC,gDAAgD;QAChD,yDAAyD;QACzD,mDAAmD;QACnD,IACE,CAAC,QAAQ;YACT,QAAQ,CAAC,UAAU,CAAC,UAAU,CAAC;YAC/B,QAAQ,CAAC,UAAU,CAAC,SAAS,CAAC,EAC9B;YACA,OAAO,SAAS,CAAC;SAClB;QAED,MAAM,UAAU,GAAG,UAAU,CAAC,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC;QAE1D,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE;YACzB,6CAA6C;YAC7C,wEAAwE;YACxE,OAAO;gBACL,KAAK,EAAE,IAAI,CAAC,IAAI,EAAE;gBAClB,KAAK,EAAE,CAAC,QAAQ,EAAE,GAAG,mBAAmB,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC;aACzD,CAAC;SACH;QAED,OAAO,QAAQ,CAAC;IAClB,CAAC,CAAC;SACD,OAAO,EAAE;SACT,MAAM,CAAC,OAAO,CAAC,CAAC;AACrB,CAAC"}

package/bin/scraping/site-scrapers/scrapeDocusaurusPage.js

@@ -0,0 +1,43 @@
+import cheerio from "cheerio";
+import { NodeHtmlMarkdown } from "node-html-markdown";
+import downloadAllImages from "../downloadAllImages.js";
+import replaceImagePaths from "../replaceImagePaths.js";
+export async function scrapeDocusaurusPage(html, origin, cliDir, imageBaseDir) {
+    const $ = cheerio.load(html);
+    const content = $(".theme-doc-markdown").first();
+    // Index pages with no additional text don't have the markdown class
+    if (content.length === 0) {
+        return {};
+    }
+    const titleComponent = content.find("h1");
+    const title = titleComponent.text().trim();
+    // Do not include title in the content when we insert it in our metadata
+    titleComponent.remove();
+    const origToWritePath = await downloadAllImages($, content, origin, imageBaseDir);
+    const contentHtml = content.html();
+    const nhm = new NodeHtmlMarkdown();
+    let markdown = nhm.translate(contentHtml);
+    // Description only exists in meta tags. The code is commented out because its prone to incorrectly
+    // including a description if the first line of text had markdown annotations like `.
+    // The commented out alternative is to ignore description if it's the first line of text,
+    // this means it was not set in the metadata and Docusaurus defaulted to the text.
+    const description = null;
+    // let description = $('meta[property="og:description"]').attr("content");
+    // if (markdown.startsWith(description)) {
+    //   description = null;
+    // }
+    // Remove Docusaurus links from headers
+    // When we parse their HTML the parser adds things like:
+    // [](#setup "Direct link to heading")
+    // to the end of each header.
+    markdown = markdown.replace(/\[\]\(#.+ ".+"\)\n/g, "\n");
+    // Remove unnecessary nonwidth blank space characters
+    markdown = markdown.replace(/\u200b/g, "");
+    // Reduce unnecessary blank lines
+    markdown = markdown.replace(/\n\n\n/g, "\n\n");
+    // Mintlify doesn't support bolded headers, remove the asterisks
+    markdown = markdown.replace(/(\n#+) \*\*(.*)\*\*\n/g, "$1 $2\n");
+    markdown = replaceImagePaths(origToWritePath, cliDir, markdown);
+    return { title, description, markdown };
+}
+//# sourceMappingURL=scrapeDocusaurusPage.js.map

package/bin/scraping/site-scrapers/scrapeDocusaurusPage.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapeDocusaurusPage.js","sourceRoot":"","sources":["../../../src/scraping/site-scrapers/scrapeDocusaurusPage.ts"],"names":[],"mappings":"AAAA,OAAO,OAAO,MAAM,SAAS,CAAC;AAC9B,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACtD,OAAO,iBAAiB,MAAM,yBAAyB,CAAC;AACxD,OAAO,iBAAiB,MAAM,yBAAyB,CAAC;AAExD,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACxC,IAAY,EACZ,MAAc,EACd,MAAc,EACd,YAAoB;IAEpB,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAE7B,MAAM,OAAO,GAAG,CAAC,CAAC,qBAAqB,CAAC,CAAC,KAAK,EAAE,CAAC;IAEjD,oEAAoE;IACpE,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE;QACxB,OAAO,EAAE,CAAC;KACX;IAED,MAAM,cAAc,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,KAAK,GAAG,cAAc,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;IAE3C,wEAAwE;IACxE,cAAc,CAAC,MAAM,EAAE,CAAC;IAExB,MAAM,eAAe,GAAG,MAAM,iBAAiB,CAC7C,CAAC,EACD,OAAO,EACP,MAAM,EACN,YAAY,CACb,CAAC;IAEF,MAAM,WAAW,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC;IAEnC,MAAM,GAAG,GAAG,IAAI,gBAAgB,EAAE,CAAC;IACnC,IAAI,QAAQ,GAAG,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,CAAC;IAE1C,mGAAmG;IACnG,qFAAqF;IACrF,yFAAyF;IACzF,kFAAkF;IAClF,MAAM,WAAW,GAAG,IAAI,CAAC;IACzB,0EAA0E;IAC1E,0CAA0C;IAC1C,wBAAwB;IACxB,IAAI;IAEJ,uCAAuC;IACvC,wDAAwD;IACxD,sCAAsC;IACtC,6BAA6B;IAC7B,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,qBAAqB,EAAE,IAAI,CAAC,CAAC;IAEzD,qDAAqD;IACrD,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;IAE3C,iCAAiC;IACjC,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC;IAE/C,gEAAgE;IAChE,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,wBAAwB,EAAE,SAAS,CAAC,CAAC;IAEjE,QAAQ,GAAG,iBAAiB,CAAC,eAAe,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAC;IAEhE,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC;AAC1C,CAAC"}

package/bin/scraping/site-scrapers/scrapeDocusaurusSection.js

@@ -0,0 +1,52 @@
+import cheerio from "cheerio";
+import { scrapeGettingFileNameFromUrl } from "../scrapeGettingFileNameFromUrl.js";
+import { scrapeDocusaurusPage } from "./scrapeDocusaurusPage.js";
+export async function scrapeDocusaurusSection(html, origin, cliDir, overwrite) {
+    const $ = cheerio.load(html);
+    // Get all the navigation sections
+    const navigationSections = $(".theme-doc-sidebar-menu").first().children();
+    // Get all links per group
+    const groupsConfig = navigationSections
+        .map((i, section) => {
+        const sectionComponent = $(section);
+        // Links without a group
+        if (sectionComponent.hasClass("theme-doc-sidebar-item-link")) {
+            const linkHref = sectionComponent.find("a[href]").first().attr("href");
+            return {
+                group: "",
+                pages: [linkHref],
+            };
+        }
+        const sectionTitle = sectionComponent
+            .find(".menu__list-item-collapsible")
+            .first()
+            .text();
+        // The category title can be a page too so we find from the
+        // section component instead of the more specific menu__list child
+        const linkPaths = sectionComponent
+            .find("a[href]")
+            .map((i, link) => {
+            return $(link).attr("href");
+        })
+            .filter((i, link) => link !== "#")
+            .toArray();
+        // Follows the same structure as mint.json
+        return {
+            group: sectionTitle,
+            pages: linkPaths,
+        };
+    })
+        .toArray();
+    // Scrape each link in the navigation.
+    const groupsConfigCleanPaths = await Promise.all(groupsConfig.map(async (groupConfig) => {
+        groupConfig.pages = (await Promise.all(groupConfig.pages.map(async (pathname) => 
+        // Docusaurus requires a directory on all sections wheras we use root.
+        // /docs is their default directory so we remove it
+        scrapeGettingFileNameFromUrl(pathname, cliDir, origin, overwrite, scrapeDocusaurusPage, false, "/docs"))))
+            // Remove skipped index pages (they return undefined from the above function)
+            .filter(Boolean);
+        return groupConfig;
+    }));
+    return groupsConfigCleanPaths;
+}
+//# sourceMappingURL=scrapeDocusaurusSection.js.map

package/bin/scraping/site-scrapers/scrapeDocusaurusSection.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scrapeDocusaurusSection.js","sourceRoot":"","sources":["../../../src/scraping/site-scrapers/scrapeDocusaurusSection.ts"],"names":[],"mappings":"AAAA,OAAO,OAAO,MAAM,SAAS,CAAC;AAC9B,OAAO,EAAE,4BAA4B,EAAE,MAAM,oCAAoC,CAAC;AAClF,OAAO,EAAE,oBAAoB,EAAE,MAAM,2BAA2B,CAAC;AAEjE,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAC3C,IAAY,EACZ,MAAc,EACd,MAAc,EACd,SAAkB;IAElB,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAE7B,kCAAkC;IAClC,MAAM,kBAAkB,GAAG,CAAC,CAAC,yBAAyB,CAAC,CAAC,KAAK,EAAE,CAAC,QAAQ,EAAE,CAAC;IAE3E,0BAA0B;IAC1B,MAAM,YAAY,GAAG,kBAAkB;SACpC,GAAG,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,EAAE;QAClB,MAAM,gBAAgB,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC;QAEpC,wBAAwB;QACxB,IAAI,gBAAgB,CAAC,QAAQ,CAAC,6BAA6B,CAAC,EAAE;YAC5D,MAAM,QAAQ,GAAG,gBAAgB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YACvE,OAAO;gBACL,KAAK,EAAE,EAAE;gBACT,KAAK,EAAE,CAAC,QAAQ,CAAC;aAClB,CAAC;SACH;QAED,MAAM,YAAY,GAAG,gBAAgB;aAClC,IAAI,CAAC,8BAA8B,CAAC;aACpC,KAAK,EAAE;aACP,IAAI,EAAE,CAAC;QAEV,2DAA2D;QAC3D,kEAAkE;QAClE,MAAM,SAAS,GAAG,gBAAgB;aAC/B,IAAI,CAAC,SAAS,CAAC;aACf,GAAG,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,EAAE;YACf,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAC9B,CAAC,CAAC;aACD,MAAM,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,GAAG,CAAC;aACjC,OAAO,EAAE,CAAC;QAEb,0CAA0C;QAC1C,OAAO;YACL,KAAK,EAAE,YAAY;YACnB,KAAK,EAAE,SAAS;SACjB,CAAC;IACJ,CAAC,CAAC;SACD,OAAO,EAAE,CAAC;IAEb,sCAAsC;IACtC,MAAM,sBAAsB,GAAG,MAAM,OAAO,CAAC,GAAG,CAC9C,YAAY,CAAC,GAAG,CAAC,KAAK,EAAE,WAAW,EAAE,EAAE;QACrC,WAAW,CAAC,KAAK,GAAG,CAClB,MAAM,OAAO,CAAC,GAAG,CACf,WAAW,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,QAAgB,EAAE,EAAE;QAC/C,sEAAsE;QACtE,mDAAmD;QACnD,4BAA4B,CAC1B,QAAQ,EACR,MAAM,EACN,MAAM,EACN,SAAS,EACT,oBAAoB,EACpB,KAAK,EACL,OAAO,CACR,CACF,CACF,CACF;YACC,6EAA6E;aAC5E,MAAM,CAAC,OAAO,CAAC,CAAC;QACnB,OAAO,WAAW,CAAC;IACrB,CAAC,CAAC,CACH,CAAC;IAEF,OAAO,sBAAsB,CAAC;AAChC,CAAC"}