Spaces:
Sleeping
Sleeping
File size: 19,082 Bytes
097eaf0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 | import { parseMarkdown } from "../../../lib/html-to-markdown";
import { Meta } from "..";
import { Document } from "../../../controllers/v2/types";
import { htmlTransform } from "../lib/removeUnwantedElements";
import { extractLinks } from "../lib/extractLinks";
import { extractImages } from "../lib/extractImages";
import { extractMetadata } from "../lib/extractMetadata";
import {
performLLMExtract,
performSummary,
performCleanContent,
} from "./llmExtract";
import { performQuery } from "./query";
import { uploadScreenshot } from "./uploadScreenshot";
import { removeBase64Images } from "./removeBase64Images";
import { performAgent } from "./agent";
import { performAttributes } from "./performAttributes";
import { deriveDiff } from "./diff";
import { fetchAudio } from "./audio";
import { fetchVideo } from "./video";
import { useIndex, useSearchIndex } from "../../../services/index";
import { sendDocumentToIndex } from "../engines/index/index";
import { sendDocumentToSearchIndex } from "./sendToSearchIndex";
import { hasFormatOfType } from "../../../lib/format-utils";
import { brandingTransformer } from "../../../lib/branding/transformer";
import { indexerQueue } from "../../../services/indexing/indexer-queue";
import { config } from "../../../config";
type Transformer = (
meta: Meta,
document: Document,
) => Document | Promise<Document>;
async function deriveMetadataFromRawHTML(
meta: Meta,
document: Document,
): Promise<Document> {
if (document.rawHtml === undefined) {
throw new Error(
"rawHtml is undefined -- this transformer is being called out of order",
);
}
document.metadata = {
...(await extractMetadata(meta, document.rawHtml)),
...document.metadata,
};
return document;
}
async function deriveHTMLFromRawHTML(
meta: Meta,
document: Document,
): Promise<Document> {
if (document.rawHtml === undefined) {
throw new Error(
"rawHtml is undefined -- this transformer is being called out of order",
);
}
document.html = await htmlTransform(
document.rawHtml,
document.metadata.url ??
document.metadata.sourceURL ??
meta.rewrittenUrl ??
meta.url,
meta.options,
);
return document;
}
async function deriveMarkdownFromHTML(
meta: Meta,
document: Document,
): Promise<Document> {
if (document.html === undefined) {
throw new Error(
"html is undefined -- this transformer is being called out of order",
);
}
// Only derive markdown if markdown format is requested or if formats that require markdown are requested:
// - changeTracking requires markdown
// - json format requires markdown (for LLM extraction)
// - summary format requires markdown (for summarization)
// - question/highlights/query formats require markdown (for page-level answers)
const hasMarkdown = hasFormatOfType(meta.options.formats, "markdown");
const hasChangeTracking = hasFormatOfType(
meta.options.formats,
"changeTracking",
);
const hasJson = hasFormatOfType(meta.options.formats, "json");
const hasSummary = hasFormatOfType(meta.options.formats, "summary");
const hasQuestion = hasFormatOfType(meta.options.formats, "question");
const hasHighlights = hasFormatOfType(meta.options.formats, "highlights");
const hasQuery = hasFormatOfType(meta.options.formats, "query");
if (
!hasMarkdown &&
!hasChangeTracking &&
!hasJson &&
!hasSummary &&
!hasQuestion &&
!hasHighlights &&
!hasQuery &&
!meta.options.onlyCleanContent
) {
return document;
}
// Skip markdown derivation if a postprocessor already set it
if (document.metadata.postprocessorsUsed?.length && document.markdown) {
meta.logger.debug(
"Skipping markdown derivation - postprocessor already set markdown",
{ postprocessorsUsed: document.metadata.postprocessorsUsed },
);
return document;
}
if (document.metadata.contentType?.includes("application/json")) {
if (document.rawHtml === undefined) {
throw new Error(
"rawHtml is undefined -- this transformer is being called out of order",
);
}
document.markdown = "```json\n" + document.rawHtml + "\n```";
return document;
}
// Use scrape ID or crawl ID as request_id for tracing
const requestId = meta.id || meta.internalOptions.crawlId;
document.markdown = await parseMarkdown(document.html, {
logger: meta.logger,
requestId,
zeroDataRetention: meta.internalOptions.zeroDataRetention,
});
if (
meta.options.onlyMainContent === true &&
(!document.markdown || document.markdown.trim().length === 0)
) {
meta.logger.info(
"Main content extraction resulted in empty markdown, falling back to full content extraction",
);
const fallbackMeta = {
...meta,
options: {
...meta.options,
onlyMainContent: false,
},
};
document = await deriveHTMLFromRawHTML(fallbackMeta, document);
document.markdown = await parseMarkdown(document.html, {
logger: meta.logger,
requestId,
zeroDataRetention: meta.internalOptions.zeroDataRetention,
});
meta.logger.info("Fallback to full content extraction completed", {
markdownLength: document.markdown?.length || 0,
});
}
return document;
}
async function deriveLinksFromHTML(
meta: Meta,
document: Document,
): Promise<Document> {
if (document.html === undefined) {
throw new Error(
"html is undefined -- this transformer is being called out of order",
);
}
const rate = config.INDEXER_TRAFFIC_SHARE
? Math.max(0, Math.min(1, Number(config.INDEXER_TRAFFIC_SHARE)))
: 0;
const shouldForwardTraffic =
rate > 0 && Math.random() <= rate && !!config.INDEXER_RABBITMQ_URL;
const forwardToIndexer =
!meta.internalOptions.isParse &&
!!meta.internalOptions.teamId &&
!meta.internalOptions.teamId?.includes("robots-txt") &&
!meta.internalOptions.teamId?.includes("sitemap") &&
shouldForwardTraffic;
const requiresLinks = !!hasFormatOfType(meta.options.formats, "links");
if (!forwardToIndexer && !requiresLinks) {
return document;
}
document.links = await extractLinks(
document.html,
document.metadata.url ??
document.metadata.sourceURL ??
meta.rewrittenUrl ??
meta.url,
);
if (forwardToIndexer) {
try {
let linksDeduped: Set<string> = new Set();
if (!!document.links) {
linksDeduped = new Set([...document.links]);
}
indexerQueue
.sendToWorker({
id: meta.id,
type: "links",
discovery_url:
document.metadata.url ??
document.metadata.sourceURL ??
meta.rewrittenUrl ??
meta.url,
urls: [...linksDeduped],
})
.catch(error => {
meta.logger.error("Failed to queue links for indexing", {
error: (error as Error)?.message,
url: meta.url,
});
});
} catch (error) {
meta.logger.error("Failed to queue links for indexing", {
error: (error as Error)?.message,
url: meta.url,
});
}
}
if (!requiresLinks) {
delete document.links;
}
return document;
}
async function deriveImagesFromHTML(
meta: Meta,
document: Document,
): Promise<Document> {
// Only derive if the formats has images
if (hasFormatOfType(meta.options.formats, "images")) {
if (document.html === undefined) {
throw new Error(
"html is undefined -- this transformer is being called out of order",
);
}
document.images = await extractImages(
document.html,
document.metadata.url ??
document.metadata.sourceURL ??
meta.rewrittenUrl ??
meta.url,
);
}
return document;
}
async function deriveBrandingFromActions(
meta: Meta,
document: Document,
): Promise<Document> {
const hasBranding = hasFormatOfType(meta.options.formats, "branding");
if (!hasBranding) {
return document;
}
if (document.branding !== undefined) {
return document;
}
/**
* Find the branding return in the actions javascript returns
* @see src/scraper/scrapeURL/engines/fire-engine/scripts/branding.js
*/
const brandingReturnIndex = document.actions?.javascriptReturns?.findIndex(
x => x.type === "object" && "branding" in (x.value as any),
);
if (brandingReturnIndex === -1 || brandingReturnIndex === undefined) {
return document;
}
// cast as any since this is js return, we might need to validate this
const javascriptReturn = document.actions!.javascriptReturns![
brandingReturnIndex
].value as any;
const rawBranding = javascriptReturn?.branding;
document.actions!.javascriptReturns!.splice(brandingReturnIndex, 1);
document.branding = await brandingTransformer(meta, document, rawBranding);
return document;
}
function coerceFieldsToFormats(meta: Meta, document: Document): Document {
const hasMarkdown = hasFormatOfType(meta.options.formats, "markdown");
const hasRawHtml = hasFormatOfType(meta.options.formats, "rawHtml");
const hasHtml = hasFormatOfType(meta.options.formats, "html");
const hasLinks = hasFormatOfType(meta.options.formats, "links");
const hasImages = hasFormatOfType(meta.options.formats, "images");
const hasChangeTracking = hasFormatOfType(
meta.options.formats,
"changeTracking",
);
const hasJson = hasFormatOfType(meta.options.formats, "json");
const hasScreenshot = hasFormatOfType(meta.options.formats, "screenshot");
const hasSummary = hasFormatOfType(meta.options.formats, "summary");
const hasBranding = hasFormatOfType(meta.options.formats, "branding");
const hasQuestionFormat = hasFormatOfType(meta.options.formats, "question");
const hasHighlightsFormat = hasFormatOfType(
meta.options.formats,
"highlights",
);
const hasLegacyQueryFormat = hasFormatOfType(meta.options.formats, "query");
const hasAnswerFormat = hasQuestionFormat || hasLegacyQueryFormat;
if (!hasMarkdown && document.markdown !== undefined) {
delete document.markdown;
} else if (hasMarkdown && document.markdown === undefined) {
meta.logger.warn(
"Request had format: markdown, but there was no markdown field in the result.",
);
}
if (!hasRawHtml && document.rawHtml !== undefined) {
delete document.rawHtml;
} else if (hasRawHtml && document.rawHtml === undefined) {
meta.logger.warn(
"Request had format: rawHtml, but there was no rawHtml field in the result.",
);
}
if (!hasHtml && document.html !== undefined) {
delete document.html;
} else if (hasHtml && document.html === undefined) {
meta.logger.warn(
"Request had format: html, but there was no html field in the result.",
);
}
if (!hasScreenshot && document.screenshot !== undefined) {
meta.logger.warn(
"Removed screenshot from Document because it wasn't in formats -- this is very wasteful and indicates a bug.",
);
delete document.screenshot;
} else if (hasScreenshot && document.screenshot === undefined) {
meta.logger.warn(
"Request had format: screenshot / screenshot@fullPage, but there was no screenshot field in the result.",
);
}
if (!hasLinks && document.links !== undefined) {
meta.logger.warn(
"Removed links from Document because it wasn't in formats -- this is wasteful and indicates a bug.",
);
delete document.links;
} else if (hasLinks && document.links === undefined) {
meta.logger.warn(
"Request had format: links, but there was no links field in the result.",
);
}
if (!hasImages && document.images !== undefined) {
meta.logger.warn(
"Removed images from Document because it wasn't in formats -- this is wasteful and indicates a bug.",
{ hasImages, hasImagesField: document.images !== undefined },
);
delete document.images;
} else if (hasImages && document.images === undefined) {
meta.logger.warn(
"Request had format: images, but there was no images field in the result.",
{ hasImages, hasImagesField: document.images !== undefined },
);
}
// Handle v1 backward compatibility - don't delete fields based on v1OriginalFormat
const shouldKeepExtract = meta.internalOptions.v1OriginalFormat === "extract";
const shouldKeepJson = meta.internalOptions.v1OriginalFormat === "json";
// Debug logging for v1 format investigation
if (meta.internalOptions.v1OriginalFormat) {
meta.logger.debug("coerceFieldsToFormats v1 format debug", {
v1OriginalFormat: meta.internalOptions.v1OriginalFormat,
hasJson: !!hasJson,
shouldKeepExtract,
shouldKeepJson,
hasExtractField: document.extract !== undefined,
hasJsonField: document.json !== undefined,
});
}
if (
!hasJson &&
(document.extract !== undefined || document.json !== undefined)
) {
// For v1 API, keep the field specified by v1OriginalFormat
if (!shouldKeepExtract && document.extract !== undefined) {
meta.logger.warn(
"Removed extract from Document because it wasn't in formats -- this is extremely wasteful and indicates a bug.",
);
delete document.extract;
}
if (!shouldKeepJson && document.json !== undefined) {
meta.logger.warn(
"Removed json from Document because it wasn't in formats -- this is extremely wasteful and indicates a bug.",
);
delete document.json;
}
} else if (
hasJson &&
document.extract === undefined &&
document.json === undefined
) {
meta.logger.warn(
"Request had format json, but there was no json field in the result.",
);
}
if (!hasSummary && document.summary !== undefined) {
meta.logger.warn(
"Removed summary from Document because it wasn't in formats -- this is wasteful and indicates a bug.",
);
delete document.summary;
} else if (hasSummary && document.summary === undefined) {
meta.logger.warn(
"Request had format summary, but there was no summary field in the result.",
);
}
if (!hasAnswerFormat && document.answer !== undefined) {
meta.logger.warn(
"Removed answer from Document because question/query wasn't in formats -- this is wasteful and indicates a bug.",
);
delete document.answer;
} else if (hasAnswerFormat && document.answer === undefined) {
meta.logger.warn(
"Request had format question/query, but there was no answer field in the result.",
);
}
if (!hasHighlightsFormat && document.highlights !== undefined) {
meta.logger.warn(
"Removed highlights from Document because highlights wasn't in formats -- this is wasteful and indicates a bug.",
);
delete document.highlights;
} else if (hasHighlightsFormat && document.highlights === undefined) {
meta.logger.warn(
"Request had format highlights, but there was no highlights field in the result.",
);
}
if (!hasBranding && document.branding !== undefined) {
meta.logger.warn(
"Removed branding from Document because it wasn't in formats -- this indicates the engine returned unexpected data.",
);
delete document.branding;
} else if (hasBranding && document.branding === undefined) {
meta.logger.warn(
"Request had format branding, but there was no branding field in the result.",
);
}
const hasAudio = hasFormatOfType(meta.options.formats, "audio");
if (!hasAudio && document.audio !== undefined) {
delete document.audio;
} else if (hasAudio && document.audio === undefined) {
meta.logger.warn(
"Request had format: audio, but there was no audio field in the result.",
);
}
const hasVideo = hasFormatOfType(meta.options.formats, "video");
if (!hasVideo && document.video !== undefined) {
delete document.video;
} else if (hasVideo && document.video === undefined) {
meta.logger.warn(
"Request had format: video, but there was no video field in the result.",
);
}
if (!hasChangeTracking && document.changeTracking !== undefined) {
meta.logger.warn(
"Removed changeTracking from Document because it wasn't in formats -- this is extremely wasteful and indicates a bug.",
);
delete document.changeTracking;
} else if (hasChangeTracking && document.changeTracking === undefined) {
meta.logger.warn(
"Request had format changeTracking, but there was no changeTracking field in the result.",
);
}
if (
document.changeTracking &&
!hasChangeTracking?.modes?.includes("git-diff") &&
document.changeTracking.diff !== undefined
) {
meta.logger.warn(
"Removed diff from changeTracking because git-diff mode wasn't specified in changeTrackingOptions.modes.",
);
delete document.changeTracking.diff;
}
if (
document.changeTracking &&
!hasChangeTracking?.modes?.includes("json") &&
document.changeTracking.json !== undefined
) {
meta.logger.warn(
"Removed structured from changeTracking because structured mode wasn't specified in changeTrackingOptions.modes.",
);
delete document.changeTracking.json;
}
if (meta.options.actions === undefined || meta.options.actions.length === 0) {
delete document.actions;
} else if (document.actions) {
// Check if all action arrays are empty
const hasScreenshots =
document.actions.screenshots && document.actions.screenshots.length > 0;
const hasScrapes =
document.actions.scrapes && document.actions.scrapes.length > 0;
const hasJsReturns =
document.actions.javascriptReturns &&
document.actions.javascriptReturns.length > 0;
const hasPdfs = document.actions.pdfs && document.actions.pdfs.length > 0;
if (!hasScreenshots && !hasScrapes && !hasJsReturns && !hasPdfs) {
delete document.actions;
}
}
return document;
}
// TODO: allow some of these to run in parallel
const transformerStack: Transformer[] = [
deriveHTMLFromRawHTML,
deriveMarkdownFromHTML,
performCleanContent,
deriveLinksFromHTML,
deriveImagesFromHTML,
deriveBrandingFromActions,
deriveMetadataFromRawHTML,
uploadScreenshot,
...(useIndex ? [sendDocumentToIndex] : []),
...(useSearchIndex ? [sendDocumentToSearchIndex] : []), // Add to search index for real-time search
performLLMExtract,
performSummary,
performQuery,
performAttributes,
performAgent,
removeBase64Images,
deriveDiff,
fetchAudio,
fetchVideo,
coerceFieldsToFormats,
];
export async function executeTransformers(
meta: Meta,
document: Document,
): Promise<Document> {
const executions: [string, number][] = [];
for (const transformer of transformerStack) {
const _meta = {
...meta,
logger: meta.logger.child({
method: "executeTransformers/" + transformer.name,
}),
};
const start = Date.now();
document = await transformer(_meta, document);
executions.push([transformer.name, Date.now() - start]);
}
meta.logger.debug("Executed transformers.", { executions });
return document;
}
|