github/v-firecrawl
github/v-firecrawl
0
Fork 0
Code Actions

adding option to replace all relative paths with absolute paths

Browse Source
This commit is contained in:
rafaelsideguide 2024-04-19 11:47:20 -03:00
parent 2c0660653d
commit 72e1dadccd
7 changed files with 257 additions and 250 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
apps/api/src/lib/entities.ts
View File

@ -22,6 +22,7 @@ export type WebScraperOptions = {
maxCrawledLinks?: number;
limit?: number;
generateImgAltText?: boolean;
replaceAllPathsWithAbsolutePaths?: boolean;
};
pageOptions?: PageOptions;
concurrentRequests?: number;

179
apps/api/src/scraper/WebScraper/__tests__/index.test.ts
View File

@ -1,179 +0,0 @@
import { WebScraperDataProvider } from "../index";
describe("WebScraperDataProvider", () => {
describe("replaceImgPathsWithAbsolutePaths", () => {
it("should replace image paths with absolute paths", () => {
const webScraperDataProvider = new WebScraperDataProvider();
const documents = [
{
metadata: { sourceURL: "https://example.com/page" },
content: "![alt text](/image.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content: "![another alt text](./another-image.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content: "![another alt text](./another-image.webp)",
},
{
metadata: { sourceURL: "https://example.com/data-image" },
content: "![data image](data:image/png;base64,...)",
},
];
const expectedDocuments = [
{
metadata: { sourceURL: "https://example.com/page" },
content: "![alt text](https://example.com/image.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content: "![another alt text](https://example.com/another-image.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content: "![another alt text](https://example.com/another-image.webp)",
},
{
metadata: { sourceURL: "https://example.com/data-image" },
content: "![data image](data:image/png;base64,...)",
},
];
const result =
webScraperDataProvider.replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it("should handle absolute URLs without modification", () => {
const webScraperDataProvider = new WebScraperDataProvider();
const documents = [
{
metadata: { sourceURL: "https://example.com/page" },
content: "![alt text](https://example.com/image.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content:
"![another alt text](http://anotherexample.com/another-image.png)",
},
];
const expectedDocuments = [
{
metadata: { sourceURL: "https://example.com/page" },
content: "![alt text](https://example.com/image.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content:
"![another alt text](http://anotherexample.com/another-image.png)",
},
];
const result =
webScraperDataProvider.replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it("should not replace non-image content within the documents", () => {
const webScraperDataProvider = new WebScraperDataProvider();
const documents = [
{
metadata: { sourceURL: "https://example.com/page" },
content:
"This is a test. ![alt text](/image.png) Here is a link: [Example](https://example.com).",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content:
"Another test. ![another alt text](./another-image.png) Here is some **bold text**.",
},
];
const expectedDocuments = [
{
metadata: { sourceURL: "https://example.com/page" },
content:
"This is a test. ![alt text](https://example.com/image.png) Here is a link: [Example](https://example.com).",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content:
"Another test. ![another alt text](https://example.com/another-image.png) Here is some **bold text**.",
},
];
const result =
webScraperDataProvider.replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it("should replace multiple image paths within the documents", () => {
const webScraperDataProvider = new WebScraperDataProvider();
const documents = [
{
metadata: { sourceURL: "https://example.com/page" },
content:
"This is a test. ![alt text](/image1.png) Here is a link: [Example](https://example.com). ![alt text](/image2.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content:
"Another test. ![another alt text](./another-image1.png) Here is some **bold text**. ![another alt text](./another-image2.png)",
},
];
const expectedDocuments = [
{
metadata: { sourceURL: "https://example.com/page" },
content:
"This is a test. ![alt text](https://example.com/image1.png) Here is a link: [Example](https://example.com). ![alt text](https://example.com/image2.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page" },
content:
"Another test. ![another alt text](https://example.com/another-image1.png) Here is some **bold text**. ![another alt text](https://example.com/another-image2.png)",
},
];
const result =
webScraperDataProvider.replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it("should replace image paths within the documents with complex URLs", () => {
const webScraperDataProvider = new WebScraperDataProvider();
const documents = [
{
metadata: { sourceURL: "https://example.com/page/subpage" },
content:
"This is a test. ![alt text](/image1.png) Here is a link: [Example](https://example.com). ![alt text](/sub/image2.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page/subpage" },
content:
"Another test. ![another alt text](/another-page/another-image1.png) Here is some **bold text**. ![another alt text](/another-page/sub/another-image2.png)",
},
];
const expectedDocuments = [
{
metadata: { sourceURL: "https://example.com/page/subpage" },
content:
"This is a test. ![alt text](https://example.com/image1.png) Here is a link: [Example](https://example.com). ![alt text](https://example.com/sub/image2.png)",
},
{
metadata: { sourceURL: "https://example.com/another-page/subpage" },
content:
"Another test. ![another alt text](https://example.com/another-page/another-image1.png) Here is some **bold text**. ![another alt text](https://example.com/another-page/sub/another-image2.png)",
},
];
const result =
webScraperDataProvider.replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
});
});

63
apps/api/src/scraper/WebScraper/index.ts
View File

@ -6,6 +6,7 @@ import { WebCrawler } from "./crawler";
import { getValue, setValue } from "../../services/redis";
import { getImageDescription } from "./utils/gptVision";
import { fetchAndProcessPdf } from "./utils/pdfProcessor";
import { replaceImgPathsWithAbsolutePaths, replacePathsWithAbsolutePaths } from "./utils/replacePaths";
export class WebScraperDataProvider {
@ -19,6 +20,7 @@ export class WebScraperDataProvider {
private concurrentRequests: number = 20;
private generateImgAltText: boolean = false;
private pageOptions?: PageOptions;
private replaceAllPathsWithAbsolutePaths?: boolean = false;
authorize(): void {
throw new Error("Method not implemented.");
@ -100,7 +102,13 @@ export class WebScraperDataProvider {
let documents = await this.convertUrlsToDocuments(links, inProgress);
documents = await this.getSitemapData(this.urls[0], documents);
documents = this.replaceImgPathsWithAbsolutePaths(documents);
if (this.replaceAllPathsWithAbsolutePaths) {
documents = replacePathsWithAbsolutePaths(documents);
} else {
documents = replaceImgPathsWithAbsolutePaths(documents);
}
if (this.generateImgAltText) {
documents = await this.generatesImgAltText(documents);
}
@ -164,7 +172,13 @@ export class WebScraperDataProvider {
this.urls.filter((link) => !link.endsWith(".pdf")),
inProgress
);
documents = this.replaceImgPathsWithAbsolutePaths(documents);
if (this.replaceAllPathsWithAbsolutePaths) {
documents = replacePathsWithAbsolutePaths(documents);
} else {
documents = replaceImgPathsWithAbsolutePaths(documents);
}
if (this.generateImgAltText) {
documents = await this.generatesImgAltText(documents);
}
@ -197,7 +211,13 @@ export class WebScraperDataProvider {
);
documents = await this.getSitemapData(this.urls[0], documents);
documents = this.replaceImgPathsWithAbsolutePaths(documents);
if (this.replaceAllPathsWithAbsolutePaths) {
documents = replacePathsWithAbsolutePaths(documents);
} else {
documents = replaceImgPathsWithAbsolutePaths(documents);
}
if (this.generateImgAltText) {
documents = await this.generatesImgAltText(documents);
}
@ -351,6 +371,7 @@ export class WebScraperDataProvider {
this.generateImgAltText =
options.crawlerOptions?.generateImgAltText ?? false;
this.pageOptions = options.pageOptions ?? {onlyMainContent: false};
this.replaceAllPathsWithAbsolutePaths = options.crawlerOptions?.replaceAllPathsWithAbsolutePaths ?? false;
//! @nicolas, for some reason this was being injected and breakign everything. Don't have time to find source of the issue so adding this check
this.excludes = this.excludes.filter((item) => item !== "");
@ -436,40 +457,4 @@ export class WebScraperDataProvider {
return documents;
};
replaceImgPathsWithAbsolutePaths = (documents: Document[]): Document[] => {
try {
documents.forEach((document) => {
const baseUrl = new URL(document.metadata.sourceURL).origin;
const images =
document.content.match(
/!\[.*?\]\(((?:[^()]+|\((?:[^()]+|\([^()]*\))*\))*)\)/g
) || [];
images.forEach((image: string) => {
let imageUrl = image.match(/\(([^)]+)\)/)[1];
let altText = image.match(/\[(.*?)\]/)[1];
if (!imageUrl.startsWith("data:image")) {
if (!imageUrl.startsWith("http")) {
if (imageUrl.startsWith("/")) {
imageUrl = imageUrl.substring(1);
}
imageUrl = new URL(imageUrl, baseUrl).toString();
}
}
document.content = document.content.replace(
image,
`![${altText}](${imageUrl})`
);
});
});
return documents;
} catch (error) {
console.error("Error replacing img paths with absolute paths", error);
return documents;
}
};
}

69
apps/api/src/scraper/WebScraper/utils/__tests__/pdfProcessor.test.ts
View File

@ -1,40 +1,47 @@
import * as pdfProcessor from '../pdfProcessor';
describe('PDF Processing Module - Integration Test', () => {
it('should download and read a simple PDF file by URL', async () => {
it('should correctly process a simple PDF file without the LLAMAPARSE_API_KEY', async () => {
delete process.env.LLAMAPARSE_API_KEY;
const pdfContent = await pdfProcessor.fetchAndProcessPdf('https://s3.us-east-1.amazonaws.com/storage.mendable.ai/rafa-testing/test%20%281%29.pdf');
expect(pdfContent).toEqual("Dummy PDF file");
expect(pdfContent.trim()).toEqual("Dummy PDF file");
});
it('should download and read a complex PDF file by URL', async () => {
const pdfContent = await pdfProcessor.fetchAndProcessPdf('https://arxiv.org/pdf/2307.06435.pdf');
// We're hitting the LLAMAPARSE rate limit 🫠
// it('should download and read a simple PDF file by URL', async () => {
// const pdfContent = await pdfProcessor.fetchAndProcessPdf('https://s3.us-east-1.amazonaws.com/storage.mendable.ai/rafa-testing/test%20%281%29.pdf');
// expect(pdfContent).toEqual("Dummy PDF file");
// });
const expectedContent = 'A Comprehensive Overview of Large Language Models\n' +
' a a,∗ b,∗ c,d,∗ e,f e,f g,i\n' +
' Humza Naveed , Asad Ullah Khan , Shi Qiu , Muhammad Saqib , Saeed Anwar , Muhammad Usman , Naveed Akhtar ,\n' +
' Nick Barnes h, Ajmal Mian i\n' +
' aUniversity of Engineering and Technology (UET), Lahore, Pakistan\n' +
' bThe Chinese University of Hong Kong (CUHK), HKSAR, China\n' +
' cUniversity of Technology Sydney (UTS), Sydney, Australia\n' +
' dCommonwealth Scientific and Industrial Research Organisation (CSIRO), Sydney, Australia\n' +
' eKing Fahd University of Petroleum and Minerals (KFUPM), Dhahran, Saudi Arabia\n' +
' fSDAIA-KFUPM Joint Research Center for Artificial Intelligence (JRCAI), Dhahran, Saudi Arabia\n' +
' gThe University of Melbourne (UoM), Melbourne, Australia\n' +
' hAustralian National University (ANU), Canberra, Australia\n' +
' iThe University of Western Australia (UWA), Perth, Australia\n' +
' Abstract\n' +
' Large Language Models (LLMs) have recently demonstrated remarkable capabilities in natural language processing tasks and\n' +
' beyond. This success of LLMs has led to a large influx of research contributions in this direction. These works encompass diverse\n' +
' topics such as architectural innovations, better training strategies, context length improvements, fine-tuning, multi-modal LLMs,\n' +
' robotics, datasets, benchmarking, efficiency, and more. With the rapid development of techniques and regular breakthroughs in\n' +
' LLM research, it has become considerably challenging to perceive the bigger picture of the advances in this direction. Considering\n' +
' the rapidly emerging plethora of literature on LLMs, it is imperative that the research community is able to benefit from a concise\n' +
' yet comprehensive overview of the recent developments in this field. This article provides an overview of the existing literature\n' +
' on a broad range of LLM-related concepts. Our self-contained comprehensive overview of LLMs discusses relevant background\n' +
' concepts along with covering the advanced topics at the frontier of research in LLMs. This review article is intended to not only\n' +
' provide a systematic survey but also a quick comprehensive reference for the researchers and practitioners to draw insights from\n' +
' extensive informative summaries of the existing works to advance the LLM research.\n'
expect(pdfContent).toContain(expectedContent);
}, 60000);
// it('should download and read a complex PDF file by URL', async () => {
// const pdfContent = await pdfProcessor.fetchAndProcessPdf('https://arxiv.org/pdf/2307.06435.pdf');
// const expectedContent = 'A Comprehensive Overview of Large Language Models\n' +
// ' a a,∗ b,∗ c,d,∗ e,f e,f g,i\n' +
// ' Humza Naveed , Asad Ullah Khan , Shi Qiu , Muhammad Saqib , Saeed Anwar , Muhammad Usman , Naveed Akhtar ,\n' +
// ' Nick Barnes h, Ajmal Mian i\n' +
// ' aUniversity of Engineering and Technology (UET), Lahore, Pakistan\n' +
// ' bThe Chinese University of Hong Kong (CUHK), HKSAR, China\n' +
// ' cUniversity of Technology Sydney (UTS), Sydney, Australia\n' +
// ' dCommonwealth Scientific and Industrial Research Organisation (CSIRO), Sydney, Australia\n' +
// ' eKing Fahd University of Petroleum and Minerals (KFUPM), Dhahran, Saudi Arabia\n' +
// ' fSDAIA-KFUPM Joint Research Center for Artificial Intelligence (JRCAI), Dhahran, Saudi Arabia\n' +
// ' gThe University of Melbourne (UoM), Melbourne, Australia\n' +
// ' hAustralian National University (ANU), Canberra, Australia\n' +
// ' iThe University of Western Australia (UWA), Perth, Australia\n' +
// ' Abstract\n' +
// ' Large Language Models (LLMs) have recently demonstrated remarkable capabilities in natural language processing tasks and\n' +
// ' beyond. This success of LLMs has led to a large influx of research contributions in this direction. These works encompass diverse\n' +
// ' topics such as architectural innovations, better training strategies, context length improvements, fine-tuning, multi-modal LLMs,\n' +
// ' robotics, datasets, benchmarking, efficiency, and more. With the rapid development of techniques and regular breakthroughs in\n' +
// ' LLM research, it has become considerably challenging to perceive the bigger picture of the advances in this direction. Considering\n' +
// ' the rapidly emerging plethora of literature on LLMs, it is imperative that the research community is able to benefit from a concise\n' +
// ' yet comprehensive overview of the recent developments in this field. This article provides an overview of the existing literature\n' +
// ' on a broad range of LLM-related concepts. Our self-contained comprehensive overview of LLMs discusses relevant background\n' +
// ' concepts along with covering the advanced topics at the frontier of research in LLMs. This review article is intended to not only\n' +
// ' provide a systematic survey but also a quick comprehensive reference for the researchers and practitioners to draw insights from\n' +
// ' extensive informative summaries of the existing works to advance the LLM research.\n'
// expect(pdfContent).toContain(expectedContent);
// }, 60000);
});

114
apps/api/src/scraper/WebScraper/utils/__tests__/replacePaths.test.ts Normal file
View File

@ -0,0 +1,114 @@
import { Document } from "../../../../lib/entities";
import { replacePathsWithAbsolutePaths, replaceImgPathsWithAbsolutePaths } from "../replacePaths";
describe('replacePaths', () => {
describe('replacePathsWithAbsolutePaths', () => {
it('should replace relative paths with absolute paths', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'This is a [link](/path/to/resource) and an image ![alt text](/path/to/image.jpg).'
}];
const expectedDocuments: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'This is a [link](https://example.com/path/to/resource) and an image ![alt text](https://example.com/path/to/image.jpg).'
}];
const result = replacePathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it('should not alter absolute URLs', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'This is an [external link](https://external.com/path) and an image ![alt text](https://example.com/path/to/image.jpg).'
}];
const result = replacePathsWithAbsolutePaths(documents);
expect(result).toEqual(documents); // Expect no change
});
it('should not alter data URLs for images', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'This is an image: ![alt text](data:image/png;base64,ABC123==).'
}];
const result = replacePathsWithAbsolutePaths(documents);
expect(result).toEqual(documents); // Expect no change
});
it('should handle multiple links and images correctly', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Here are two links: [link1](/path1) and [link2](/path2), and two images: ![img1](/img1.jpg) ![img2](/img2.jpg).'
}];
const expectedDocuments: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Here are two links: [link1](https://example.com/path1) and [link2](https://example.com/path2), and two images: ![img1](https://example.com/img1.jpg) ![img2](https://example.com/img2.jpg).'
}];
const result = replacePathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it('should correctly handle a mix of absolute and relative paths', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Mixed paths: [relative](/path), [absolute](https://example.com/path), and [data image](data:image/png;base64,ABC123==).'
}];
const expectedDocuments: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Mixed paths: [relative](https://example.com/path), [absolute](https://example.com/path), and [data image](data:image/png;base64,ABC123==).'
}];
const result = replacePathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
});
describe('replaceImgPathsWithAbsolutePaths', () => {
it('should replace relative image paths with absolute paths', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Here is an image: ![alt text](/path/to/image.jpg).'
}];
const expectedDocuments: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Here is an image: ![alt text](https://example.com/path/to/image.jpg).'
}];
const result = replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
it('should not alter data:image URLs', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'An image with a data URL: ![alt text](data:image/png;base64,ABC123==).'
}];
const result = replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(documents); // Expect no change
});
it('should handle multiple images with a mix of data and relative URLs', () => {
const documents: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Multiple images: ![img1](/img1.jpg) ![img2](data:image/png;base64,ABC123==) ![img3](/img3.jpg).'
}];
const expectedDocuments: Document[] = [{
metadata: { sourceURL: 'https://example.com' },
content: 'Multiple images: ![img1](https://example.com/img1.jpg) ![img2](data:image/png;base64,ABC123==) ![img3](https://example.com/img3.jpg).'
}];
const result = replaceImgPathsWithAbsolutePaths(documents);
expect(result).toEqual(expectedDocuments);
});
});
});

80
apps/api/src/scraper/WebScraper/utils/replacePaths.ts Normal file
View File

@ -0,0 +1,80 @@
import { Document } from "../../../lib/entities";
export const replacePathsWithAbsolutePaths = (documents: Document[]): Document[] => {
try {
documents.forEach((document) => {
const baseUrl = new URL(document.metadata.sourceURL).origin;
const paths =
document.content.match(
/(!?\[.*?\])\(((?:[^()]+|\((?:[^()]+|\([^()]*\))*\))*)\)|href="([^"]+)"/g
) || [];
paths.forEach((path: string) => {
const isImage = path.startsWith("!");
let matchedUrl = path.match(/\(([^)]+)\)/) || path.match(/href="([^"]+)"/);
let url = matchedUrl[1];
if (!url.startsWith("data:") && !url.startsWith("http")) {
if (url.startsWith("/")) {
url = url.substring(1);
}
url = new URL(url, baseUrl).toString();
}
const markdownLinkOrImageText = path.match(/(!?\[.*?\])/)[0];
if (isImage) {
document.content = document.content.replace(
path,
`${markdownLinkOrImageText}(${url})`
);
} else {
document.content = document.content.replace(
path,
`${markdownLinkOrImageText}(${url})`
);
}
});
});
return documents;
} catch (error) {
console.error("Error replacing paths with absolute paths", error);
return documents;
}
};
export const replaceImgPathsWithAbsolutePaths = (documents: Document[]): Document[] => {
try {
documents.forEach((document) => {
const baseUrl = new URL(document.metadata.sourceURL).origin;
const images =
document.content.match(
/!\[.*?\]\(((?:[^()]+|\((?:[^()]+|\([^()]*\))*\))*)\)/g
) || [];
images.forEach((image: string) => {
let imageUrl = image.match(/\(([^)]+)\)/)[1];
let altText = image.match(/\[(.*?)\]/)[1];
if (!imageUrl.startsWith("data:image")) {
if (!imageUrl.startsWith("http")) {
if (imageUrl.startsWith("/")) {
imageUrl = imageUrl.substring(1);
}
imageUrl = new URL(imageUrl, baseUrl).toString();
}
}
document.content = document.content.replace(
image,
`![${altText}](${imageUrl})`
);
});
});
return documents;
} catch (error) {
console.error("Error replacing img paths with absolute paths", error);
return documents;
}
};

1
apps/api/src/services/queue-worker.ts
View File

@ -3,7 +3,6 @@ import { getWebScraperQueue } from "./queue-service";
import "dotenv/config";
import { logtail } from "./logtail";
import { startWebScraperPipeline } from "../main/runWebScraper";
import { WebScraperDataProvider } from "../scraper/WebScraper";
import { callWebhook } from "./webhook";
getWebScraperQueue().process(