firecrawl 1.15.3 → 1.16.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/index.cjs +6 -5
- package/dist/index.d.cts +5 -0
- package/dist/index.d.ts +5 -0
- package/dist/index.js +6 -5
- package/package.json +1 -1
- package/src/index.ts +13 -5
package/dist/index.cjs
CHANGED
|
@@ -276,7 +276,7 @@ var FirecrawlApp = class {
|
|
|
276
276
|
targetURL.searchParams.set("skip", skip.toString());
|
|
277
277
|
}
|
|
278
278
|
if (limit !== void 0) {
|
|
279
|
-
targetURL.searchParams.set("
|
|
279
|
+
targetURL.searchParams.set("limit", limit.toString());
|
|
280
280
|
}
|
|
281
281
|
try {
|
|
282
282
|
const response = await this.getRequest(
|
|
@@ -530,7 +530,7 @@ var FirecrawlApp = class {
|
|
|
530
530
|
targetURL.searchParams.set("skip", skip.toString());
|
|
531
531
|
}
|
|
532
532
|
if (limit !== void 0) {
|
|
533
|
-
targetURL.searchParams.set("
|
|
533
|
+
targetURL.searchParams.set("limit", limit.toString());
|
|
534
534
|
}
|
|
535
535
|
try {
|
|
536
536
|
const response = await this.getRequest(
|
|
@@ -629,7 +629,7 @@ var FirecrawlApp = class {
|
|
|
629
629
|
try {
|
|
630
630
|
const response = await this.postRequest(
|
|
631
631
|
this.apiUrl + `/v1/extract`,
|
|
632
|
-
{ ...jsonData, schema: jsonSchema, origin: "api-sdk" },
|
|
632
|
+
{ ...jsonData, schema: jsonSchema, origin: params?.origin || "api-sdk" },
|
|
633
633
|
headers
|
|
634
634
|
);
|
|
635
635
|
if (response.status === 200) {
|
|
@@ -841,7 +841,7 @@ var FirecrawlApp = class {
|
|
|
841
841
|
throw new FirecrawlError(
|
|
842
842
|
`Failed to ${action}. Status code: ${response.status}. Error: ${errorMessage}${details}`,
|
|
843
843
|
response.status,
|
|
844
|
-
response
|
|
844
|
+
response?.data?.details
|
|
845
845
|
);
|
|
846
846
|
} else {
|
|
847
847
|
throw new FirecrawlError(
|
|
@@ -859,7 +859,8 @@ var CrawlWatcher = class extends import_typescript_event_target.TypedEventTarget
|
|
|
859
859
|
constructor(id, app) {
|
|
860
860
|
super();
|
|
861
861
|
this.id = id;
|
|
862
|
-
|
|
862
|
+
const wsUrl = app.apiUrl.replace(/^http/, "ws");
|
|
863
|
+
this.ws = new import_isows.WebSocket(`${wsUrl}/v1/crawl/${id}`, app.apiKey);
|
|
863
864
|
this.status = "scraping";
|
|
864
865
|
this.data = [];
|
|
865
866
|
const messageHandler = (msg) => {
|
package/dist/index.d.cts
CHANGED
|
@@ -87,6 +87,7 @@ interface CrawlScrapeOptions {
|
|
|
87
87
|
mobile?: boolean;
|
|
88
88
|
skipTlsVerification?: boolean;
|
|
89
89
|
removeBase64Images?: boolean;
|
|
90
|
+
blockAds?: boolean;
|
|
90
91
|
}
|
|
91
92
|
type Action = {
|
|
92
93
|
type: "wait";
|
|
@@ -156,6 +157,7 @@ interface CrawlParams {
|
|
|
156
157
|
url: string;
|
|
157
158
|
headers?: Record<string, string>;
|
|
158
159
|
metadata?: Record<string, string>;
|
|
160
|
+
events?: ["completed", "failed", "page", "started"][number][];
|
|
159
161
|
};
|
|
160
162
|
deduplicateSimilarURLs?: boolean;
|
|
161
163
|
ignoreQueryParameters?: boolean;
|
|
@@ -219,6 +221,7 @@ interface MapParams {
|
|
|
219
221
|
includeSubdomains?: boolean;
|
|
220
222
|
sitemapOnly?: boolean;
|
|
221
223
|
limit?: number;
|
|
224
|
+
timeout?: number;
|
|
222
225
|
}
|
|
223
226
|
/**
|
|
224
227
|
* Response interface for mapping operations.
|
|
@@ -238,7 +241,9 @@ interface ExtractParams<LLMSchema extends zt.ZodSchema = any> {
|
|
|
238
241
|
schema?: LLMSchema | object;
|
|
239
242
|
systemPrompt?: string;
|
|
240
243
|
allowExternalLinks?: boolean;
|
|
244
|
+
enableWebSearch?: boolean;
|
|
241
245
|
includeSubdomains?: boolean;
|
|
246
|
+
origin?: string;
|
|
242
247
|
}
|
|
243
248
|
/**
|
|
244
249
|
* Response interface for extracting information from URLs.
|
package/dist/index.d.ts
CHANGED
|
@@ -87,6 +87,7 @@ interface CrawlScrapeOptions {
|
|
|
87
87
|
mobile?: boolean;
|
|
88
88
|
skipTlsVerification?: boolean;
|
|
89
89
|
removeBase64Images?: boolean;
|
|
90
|
+
blockAds?: boolean;
|
|
90
91
|
}
|
|
91
92
|
type Action = {
|
|
92
93
|
type: "wait";
|
|
@@ -156,6 +157,7 @@ interface CrawlParams {
|
|
|
156
157
|
url: string;
|
|
157
158
|
headers?: Record<string, string>;
|
|
158
159
|
metadata?: Record<string, string>;
|
|
160
|
+
events?: ["completed", "failed", "page", "started"][number][];
|
|
159
161
|
};
|
|
160
162
|
deduplicateSimilarURLs?: boolean;
|
|
161
163
|
ignoreQueryParameters?: boolean;
|
|
@@ -219,6 +221,7 @@ interface MapParams {
|
|
|
219
221
|
includeSubdomains?: boolean;
|
|
220
222
|
sitemapOnly?: boolean;
|
|
221
223
|
limit?: number;
|
|
224
|
+
timeout?: number;
|
|
222
225
|
}
|
|
223
226
|
/**
|
|
224
227
|
* Response interface for mapping operations.
|
|
@@ -238,7 +241,9 @@ interface ExtractParams<LLMSchema extends zt.ZodSchema = any> {
|
|
|
238
241
|
schema?: LLMSchema | object;
|
|
239
242
|
systemPrompt?: string;
|
|
240
243
|
allowExternalLinks?: boolean;
|
|
244
|
+
enableWebSearch?: boolean;
|
|
241
245
|
includeSubdomains?: boolean;
|
|
246
|
+
origin?: string;
|
|
242
247
|
}
|
|
243
248
|
/**
|
|
244
249
|
* Response interface for extracting information from URLs.
|
package/dist/index.js
CHANGED
|
@@ -240,7 +240,7 @@ var FirecrawlApp = class {
|
|
|
240
240
|
targetURL.searchParams.set("skip", skip.toString());
|
|
241
241
|
}
|
|
242
242
|
if (limit !== void 0) {
|
|
243
|
-
targetURL.searchParams.set("
|
|
243
|
+
targetURL.searchParams.set("limit", limit.toString());
|
|
244
244
|
}
|
|
245
245
|
try {
|
|
246
246
|
const response = await this.getRequest(
|
|
@@ -494,7 +494,7 @@ var FirecrawlApp = class {
|
|
|
494
494
|
targetURL.searchParams.set("skip", skip.toString());
|
|
495
495
|
}
|
|
496
496
|
if (limit !== void 0) {
|
|
497
|
-
targetURL.searchParams.set("
|
|
497
|
+
targetURL.searchParams.set("limit", limit.toString());
|
|
498
498
|
}
|
|
499
499
|
try {
|
|
500
500
|
const response = await this.getRequest(
|
|
@@ -593,7 +593,7 @@ var FirecrawlApp = class {
|
|
|
593
593
|
try {
|
|
594
594
|
const response = await this.postRequest(
|
|
595
595
|
this.apiUrl + `/v1/extract`,
|
|
596
|
-
{ ...jsonData, schema: jsonSchema, origin: "api-sdk" },
|
|
596
|
+
{ ...jsonData, schema: jsonSchema, origin: params?.origin || "api-sdk" },
|
|
597
597
|
headers
|
|
598
598
|
);
|
|
599
599
|
if (response.status === 200) {
|
|
@@ -805,7 +805,7 @@ var FirecrawlApp = class {
|
|
|
805
805
|
throw new FirecrawlError(
|
|
806
806
|
`Failed to ${action}. Status code: ${response.status}. Error: ${errorMessage}${details}`,
|
|
807
807
|
response.status,
|
|
808
|
-
response
|
|
808
|
+
response?.data?.details
|
|
809
809
|
);
|
|
810
810
|
} else {
|
|
811
811
|
throw new FirecrawlError(
|
|
@@ -823,7 +823,8 @@ var CrawlWatcher = class extends TypedEventTarget {
|
|
|
823
823
|
constructor(id, app) {
|
|
824
824
|
super();
|
|
825
825
|
this.id = id;
|
|
826
|
-
|
|
826
|
+
const wsUrl = app.apiUrl.replace(/^http/, "ws");
|
|
827
|
+
this.ws = new WebSocket(`${wsUrl}/v1/crawl/${id}`, app.apiKey);
|
|
827
828
|
this.status = "scraping";
|
|
828
829
|
this.data = [];
|
|
829
830
|
const messageHandler = (msg) => {
|
package/package.json
CHANGED
package/src/index.ts
CHANGED
|
@@ -93,6 +93,7 @@ export interface CrawlScrapeOptions {
|
|
|
93
93
|
mobile?: boolean;
|
|
94
94
|
skipTlsVerification?: boolean;
|
|
95
95
|
removeBase64Images?: boolean;
|
|
96
|
+
blockAds?: boolean;
|
|
96
97
|
}
|
|
97
98
|
|
|
98
99
|
export type Action = {
|
|
@@ -167,6 +168,7 @@ export interface CrawlParams {
|
|
|
167
168
|
url: string;
|
|
168
169
|
headers?: Record<string, string>;
|
|
169
170
|
metadata?: Record<string, string>;
|
|
171
|
+
events?: ["completed", "failed", "page", "started"][number][];
|
|
170
172
|
};
|
|
171
173
|
deduplicateSimilarURLs?: boolean;
|
|
172
174
|
ignoreQueryParameters?: boolean;
|
|
@@ -235,6 +237,7 @@ export interface MapParams {
|
|
|
235
237
|
includeSubdomains?: boolean;
|
|
236
238
|
sitemapOnly?: boolean;
|
|
237
239
|
limit?: number;
|
|
240
|
+
timeout?: number;
|
|
238
241
|
}
|
|
239
242
|
|
|
240
243
|
/**
|
|
@@ -256,7 +259,9 @@ export interface ExtractParams<LLMSchema extends zt.ZodSchema = any> {
|
|
|
256
259
|
schema?: LLMSchema | object;
|
|
257
260
|
systemPrompt?: string;
|
|
258
261
|
allowExternalLinks?: boolean;
|
|
262
|
+
enableWebSearch?: boolean;
|
|
259
263
|
includeSubdomains?: boolean;
|
|
264
|
+
origin?: string;
|
|
260
265
|
}
|
|
261
266
|
|
|
262
267
|
/**
|
|
@@ -606,7 +611,7 @@ export default class FirecrawlApp {
|
|
|
606
611
|
targetURL.searchParams.set("skip", skip.toString());
|
|
607
612
|
}
|
|
608
613
|
if (limit !== undefined) {
|
|
609
|
-
targetURL.searchParams.set("
|
|
614
|
+
targetURL.searchParams.set("limit", limit.toString());
|
|
610
615
|
}
|
|
611
616
|
|
|
612
617
|
try {
|
|
@@ -908,7 +913,7 @@ export default class FirecrawlApp {
|
|
|
908
913
|
targetURL.searchParams.set("skip", skip.toString());
|
|
909
914
|
}
|
|
910
915
|
if (limit !== undefined) {
|
|
911
|
-
targetURL.searchParams.set("
|
|
916
|
+
targetURL.searchParams.set("limit", limit.toString());
|
|
912
917
|
}
|
|
913
918
|
|
|
914
919
|
try {
|
|
@@ -1013,10 +1018,11 @@ export default class FirecrawlApp {
|
|
|
1013
1018
|
throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);
|
|
1014
1019
|
}
|
|
1015
1020
|
|
|
1021
|
+
|
|
1016
1022
|
try {
|
|
1017
1023
|
const response: AxiosResponse = await this.postRequest(
|
|
1018
1024
|
this.apiUrl + `/v1/extract`,
|
|
1019
|
-
{ ...jsonData, schema: jsonSchema, origin: "api-sdk" },
|
|
1025
|
+
{ ...jsonData, schema: jsonSchema, origin: params?.origin || "api-sdk" },
|
|
1020
1026
|
headers
|
|
1021
1027
|
);
|
|
1022
1028
|
|
|
@@ -1262,7 +1268,7 @@ export default class FirecrawlApp {
|
|
|
1262
1268
|
throw new FirecrawlError(
|
|
1263
1269
|
`Failed to ${action}. Status code: ${response.status}. Error: ${errorMessage}${details}`,
|
|
1264
1270
|
response.status,
|
|
1265
|
-
response
|
|
1271
|
+
response?.data?.details
|
|
1266
1272
|
);
|
|
1267
1273
|
} else {
|
|
1268
1274
|
throw new FirecrawlError(
|
|
@@ -1295,7 +1301,9 @@ export class CrawlWatcher extends TypedEventTarget<CrawlWatcherEvents> {
|
|
|
1295
1301
|
constructor(id: string, app: FirecrawlApp) {
|
|
1296
1302
|
super();
|
|
1297
1303
|
this.id = id;
|
|
1298
|
-
|
|
1304
|
+
// replace `http` with `ws` (`http://` -> `ws://` and `https://` -> `wss://`)
|
|
1305
|
+
const wsUrl = app.apiUrl.replace(/^http/, "ws");
|
|
1306
|
+
this.ws = new WebSocket(`${wsUrl}/v1/crawl/${id}`, app.apiKey);
|
|
1299
1307
|
this.status = "scraping";
|
|
1300
1308
|
this.data = [];
|
|
1301
1309
|
|