firecrawl 4.28.2 → 4.28.4
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/{chunk-XZBMU524.js → chunk-VRKVHB52.js} +1 -1
- package/dist/index.cjs +2 -1
- package/dist/index.d.cts +39 -4
- package/dist/index.d.ts +39 -4
- package/dist/index.js +3 -2
- package/dist/{package-7QIHFAN4.js → package-EYT327NL.js} +1 -1
- package/package.json +1 -1
- package/src/__tests__/unit/v2/extract.unit.test.ts +47 -0
- package/src/__tests__/unit/v2/monitor-search-target.test.ts +40 -0
- package/src/v2/methods/extract.ts +3 -1
- package/src/v2/types.ts +49 -3
|
@@ -12,7 +12,7 @@ var require_package = __commonJS({
|
|
|
12
12
|
"package.json"(exports, module) {
|
|
13
13
|
module.exports = {
|
|
14
14
|
name: "@mendable/firecrawl-js",
|
|
15
|
-
version: "4.28.
|
|
15
|
+
version: "4.28.4",
|
|
16
16
|
description: "JavaScript SDK for Firecrawl API",
|
|
17
17
|
main: "dist/index.js",
|
|
18
18
|
types: "dist/index.d.ts",
|
package/dist/index.cjs
CHANGED
|
@@ -39,7 +39,7 @@ var require_package = __commonJS({
|
|
|
39
39
|
"package.json"(exports2, module2) {
|
|
40
40
|
module2.exports = {
|
|
41
41
|
name: "@mendable/firecrawl-js",
|
|
42
|
-
version: "4.28.
|
|
42
|
+
version: "4.28.4",
|
|
43
43
|
description: "JavaScript SDK for Firecrawl API",
|
|
44
44
|
main: "dist/index.js",
|
|
45
45
|
types: "dist/index.d.ts",
|
|
@@ -1226,6 +1226,7 @@ function prepareExtractPayload(args) {
|
|
|
1226
1226
|
if (args.integration && args.integration.trim()) body.integration = args.integration.trim();
|
|
1227
1227
|
if (args.origin) body.origin = args.origin;
|
|
1228
1228
|
if (args.agent) body.agent = args.agent;
|
|
1229
|
+
if (args.webhook != null) body.webhook = args.webhook;
|
|
1229
1230
|
if (args.scrapeOptions) {
|
|
1230
1231
|
ensureValidScrapeOptions(args.scrapeOptions);
|
|
1231
1232
|
body.scrapeOptions = args.scrapeOptions;
|
package/dist/index.d.cts
CHANGED
|
@@ -431,6 +431,7 @@ interface DocumentMetadata {
|
|
|
431
431
|
statusCode?: number;
|
|
432
432
|
scrapeId?: string;
|
|
433
433
|
numPages?: number;
|
|
434
|
+
totalPages?: number;
|
|
434
435
|
contentType?: string;
|
|
435
436
|
timezone?: string;
|
|
436
437
|
proxyUsed?: "basic" | "stealth";
|
|
@@ -705,7 +706,16 @@ interface MonitorCrawlTarget {
|
|
|
705
706
|
crawlOptions?: CrawlOptions;
|
|
706
707
|
scrapeOptions?: ScrapeOptions;
|
|
707
708
|
}
|
|
708
|
-
|
|
709
|
+
interface MonitorSearchTarget {
|
|
710
|
+
id?: string;
|
|
711
|
+
type: "search";
|
|
712
|
+
queries: string[];
|
|
713
|
+
searchWindow?: "5m" | "15m" | "1h" | "6h" | "24h" | "7d";
|
|
714
|
+
includeDomains?: string[];
|
|
715
|
+
excludeDomains?: string[];
|
|
716
|
+
maxResults?: number;
|
|
717
|
+
}
|
|
718
|
+
type MonitorTarget = MonitorScrapeTarget | MonitorCrawlTarget | MonitorSearchTarget;
|
|
709
719
|
interface CreateMonitorRequest {
|
|
710
720
|
name: string;
|
|
711
721
|
schedule: MonitorSchedule;
|
|
@@ -772,10 +782,34 @@ interface MonitorPageJudgment {
|
|
|
772
782
|
reason: string;
|
|
773
783
|
}>;
|
|
774
784
|
}
|
|
785
|
+
interface MonitorScrapeTargetResult {
|
|
786
|
+
targetId: string;
|
|
787
|
+
type: "scrape";
|
|
788
|
+
expectedJobs?: string[];
|
|
789
|
+
}
|
|
790
|
+
interface MonitorCrawlTargetResult {
|
|
791
|
+
targetId: string;
|
|
792
|
+
type: "crawl";
|
|
793
|
+
crawlId?: string;
|
|
794
|
+
}
|
|
795
|
+
interface MonitorSearchTargetResult {
|
|
796
|
+
targetId: string;
|
|
797
|
+
type: "search";
|
|
798
|
+
searchCompleted?: boolean;
|
|
799
|
+
resultCount?: number;
|
|
800
|
+
matches?: number;
|
|
801
|
+
summary?: string;
|
|
802
|
+
judgeDegraded?: boolean;
|
|
803
|
+
degradedReason?: string | null;
|
|
804
|
+
searchCredits?: number;
|
|
805
|
+
judgeCredits?: number;
|
|
806
|
+
resultsJudged?: number;
|
|
807
|
+
}
|
|
808
|
+
type MonitorTargetResult = MonitorScrapeTargetResult | MonitorCrawlTargetResult | MonitorSearchTargetResult;
|
|
775
809
|
interface MonitorCheck {
|
|
776
810
|
id: string;
|
|
777
811
|
monitorId: string;
|
|
778
|
-
status: "queued" | "running" | "completed" | "failed" | "partial" | "skipped_overlap";
|
|
812
|
+
status: "queued" | "running" | "completed" | "failed" | "partial" | "skipped_overlap" | "skipped_no_credits";
|
|
779
813
|
trigger: "scheduled" | "manual";
|
|
780
814
|
scheduledFor?: string | null;
|
|
781
815
|
startedAt?: string | null;
|
|
@@ -785,7 +819,7 @@ interface MonitorCheck {
|
|
|
785
819
|
actualCredits?: number | null;
|
|
786
820
|
billingStatus: "not_applicable" | "reserved" | "confirmed" | "released" | "failed";
|
|
787
821
|
summary: MonitorSummary;
|
|
788
|
-
targetResults?:
|
|
822
|
+
targetResults?: MonitorTargetResult[];
|
|
789
823
|
notificationStatus?: unknown;
|
|
790
824
|
error?: string | null;
|
|
791
825
|
createdAt: string;
|
|
@@ -1209,6 +1243,7 @@ declare function prepareExtractPayload(args: {
|
|
|
1209
1243
|
integration?: string;
|
|
1210
1244
|
origin?: string;
|
|
1211
1245
|
agent?: AgentOptions$1;
|
|
1246
|
+
webhook?: string | WebhookConfig | null;
|
|
1212
1247
|
}): Record<string, unknown>;
|
|
1213
1248
|
/**
|
|
1214
1249
|
* @deprecated The extract endpoint is in maintenance mode and its use is discouraged.
|
|
@@ -2588,4 +2623,4 @@ declare class Firecrawl extends FirecrawlClient {
|
|
|
2588
2623
|
get v1(): FirecrawlApp;
|
|
2589
2624
|
}
|
|
2590
2625
|
|
|
2591
|
-
export { type ActionOption, type ActiveCrawl, type ActiveCrawlsResponse, type AgentOptions$1 as AgentOptions, type AgentResponse, type AgentStatusResponse, type AgentWebhookConfig, type AgentWebhookEvent, type AttributesFormat, type BatchScrapeJob, type BatchScrapeOptions, type BatchScrapeResponse$1 as BatchScrapeResponse, type BrandingProfile, type BrowserCreateResponse, type BrowserDeleteResponse, type BrowserExecuteResponse, type BrowserListResponse, type BrowserSession, type CategoryOption, type ChangeTrackingFormat, type ClickAction, type ConcurrencyCheck, type CrawlErrorsResponse$1 as CrawlErrorsResponse, type CrawlJob, type CrawlOptions, type CrawlResponse$1 as CrawlResponse, type CreateMonitorRequest, type CreditUsage, type CreditUsageHistoricalPeriod, type CreditUsageHistoricalResponse, type Document, type DocumentMetadata, type EndpointFeedbackEndpoint, type EndpointFeedbackRequest, type ErrorDetails, type ExecuteJavascriptAction, type ExtractResponse$1 as ExtractResponse, type FeedbackMissingContent, type FeedbackRating, type FeedbackResponse, type FeedbackValuableSource, Firecrawl, FirecrawlApp as FirecrawlAppV1, FirecrawlClient, type FirecrawlClientInput, type FirecrawlClientOptions, type Format, type FormatOption, type FormatString, type GetMonitorCheckOptions, type GetPaperOptions, type GitHubScoreBreakdown, type GitHubSearchItem, type GitHubSearchResponse, type HighlightsFormat, type IdMap, JobTimeoutError, type JsonFormat, type ListMonitorChecksOptions, type ListMonitorsOptions, type LocationConfig$1 as LocationConfig, type MapData, type MapOptions, type MenuAvailability, type MenuImage, type MenuItem, type MenuItemIdentifiers, type MenuMerchant, type MenuPrice, type MenuProfile, type MenuSection, type Monitor, type MonitorCheck, type MonitorCheckDetail, type MonitorCheckPage, type MonitorCrawlTarget, type MonitorEmailNotification, type MonitorEmailRecipientSubscription, type MonitorJsonFieldDiff, type MonitorNotification, type MonitorPageDiff, type MonitorPageJudgment, type MonitorPageSnapshot, type MonitorSchedule, type MonitorScrapeTarget, type MonitorSummary, type MonitorTarget, type MonitorWebhookConfig, type PDFAction, type PaginationConfig, type PaperMetadata, type PaperMetadataResponse, type PaperResult, type PaperSignals, type ParseFile, type ParseFileData, type ParseFormat, type ParseFormatOption, type ParseFormatString, type ParseOptions, type Passage, type PressAction, type ProductAvailability, type ProductImage, type ProductPrice, type ProductProfile, type ProductSale, type ProductVariant, type QueryFormat, type QuestionFormat, type QueueStatusResponse$1 as QueueStatusResponse, type ReadPaperResponse, type RedactPIIEntity, type RedactPIIOptions, ResearchClient, type ScrapeAction, type ScrapeBrowserDeleteResponse, type ScrapeExecuteRequest, type ScrapeExecuteResponse, type ScrapeOptions, type ScreenshotAction, type ScreenshotFormat, type ScrollAction, SdkError, type SearchData, type SearchFeedbackRequest, type SearchGithubOptions, type SearchPapersOptions, type SearchPapersResponse, type SearchRequest, type SearchResultImages, type SearchResultNews, type SearchResultWeb, type SimilarPapersOptions, type SimilarPapersResponse, type TokenUsage, type TokenUsageHistoricalPeriod, type TokenUsageHistoricalResponse, type UpdateMonitorRequest, type Viewport, type WaitAction, Watcher, type WatcherOptions, type WebhookConfig, type WriteAction, Firecrawl as default };
|
|
2626
|
+
export { type ActionOption, type ActiveCrawl, type ActiveCrawlsResponse, type AgentOptions$1 as AgentOptions, type AgentResponse, type AgentStatusResponse, type AgentWebhookConfig, type AgentWebhookEvent, type AttributesFormat, type BatchScrapeJob, type BatchScrapeOptions, type BatchScrapeResponse$1 as BatchScrapeResponse, type BrandingProfile, type BrowserCreateResponse, type BrowserDeleteResponse, type BrowserExecuteResponse, type BrowserListResponse, type BrowserSession, type CategoryOption, type ChangeTrackingFormat, type ClickAction, type ConcurrencyCheck, type CrawlErrorsResponse$1 as CrawlErrorsResponse, type CrawlJob, type CrawlOptions, type CrawlResponse$1 as CrawlResponse, type CreateMonitorRequest, type CreditUsage, type CreditUsageHistoricalPeriod, type CreditUsageHistoricalResponse, type Document, type DocumentMetadata, type EndpointFeedbackEndpoint, type EndpointFeedbackRequest, type ErrorDetails, type ExecuteJavascriptAction, type ExtractResponse$1 as ExtractResponse, type FeedbackMissingContent, type FeedbackRating, type FeedbackResponse, type FeedbackValuableSource, Firecrawl, FirecrawlApp as FirecrawlAppV1, FirecrawlClient, type FirecrawlClientInput, type FirecrawlClientOptions, type Format, type FormatOption, type FormatString, type GetMonitorCheckOptions, type GetPaperOptions, type GitHubScoreBreakdown, type GitHubSearchItem, type GitHubSearchResponse, type HighlightsFormat, type IdMap, JobTimeoutError, type JsonFormat, type ListMonitorChecksOptions, type ListMonitorsOptions, type LocationConfig$1 as LocationConfig, type MapData, type MapOptions, type MenuAvailability, type MenuImage, type MenuItem, type MenuItemIdentifiers, type MenuMerchant, type MenuPrice, type MenuProfile, type MenuSection, type Monitor, type MonitorCheck, type MonitorCheckDetail, type MonitorCheckPage, type MonitorCrawlTarget, type MonitorCrawlTargetResult, type MonitorEmailNotification, type MonitorEmailRecipientSubscription, type MonitorJsonFieldDiff, type MonitorNotification, type MonitorPageDiff, type MonitorPageJudgment, type MonitorPageSnapshot, type MonitorSchedule, type MonitorScrapeTarget, type MonitorScrapeTargetResult, type MonitorSearchTarget, type MonitorSearchTargetResult, type MonitorSummary, type MonitorTarget, type MonitorTargetResult, type MonitorWebhookConfig, type PDFAction, type PaginationConfig, type PaperMetadata, type PaperMetadataResponse, type PaperResult, type PaperSignals, type ParseFile, type ParseFileData, type ParseFormat, type ParseFormatOption, type ParseFormatString, type ParseOptions, type Passage, type PressAction, type ProductAvailability, type ProductImage, type ProductPrice, type ProductProfile, type ProductSale, type ProductVariant, type QueryFormat, type QuestionFormat, type QueueStatusResponse$1 as QueueStatusResponse, type ReadPaperResponse, type RedactPIIEntity, type RedactPIIOptions, ResearchClient, type ScrapeAction, type ScrapeBrowserDeleteResponse, type ScrapeExecuteRequest, type ScrapeExecuteResponse, type ScrapeOptions, type ScreenshotAction, type ScreenshotFormat, type ScrollAction, SdkError, type SearchData, type SearchFeedbackRequest, type SearchGithubOptions, type SearchPapersOptions, type SearchPapersResponse, type SearchRequest, type SearchResultImages, type SearchResultNews, type SearchResultWeb, type SimilarPapersOptions, type SimilarPapersResponse, type TokenUsage, type TokenUsageHistoricalPeriod, type TokenUsageHistoricalResponse, type UpdateMonitorRequest, type Viewport, type WaitAction, Watcher, type WatcherOptions, type WebhookConfig, type WriteAction, Firecrawl as default };
|
package/dist/index.d.ts
CHANGED
|
@@ -431,6 +431,7 @@ interface DocumentMetadata {
|
|
|
431
431
|
statusCode?: number;
|
|
432
432
|
scrapeId?: string;
|
|
433
433
|
numPages?: number;
|
|
434
|
+
totalPages?: number;
|
|
434
435
|
contentType?: string;
|
|
435
436
|
timezone?: string;
|
|
436
437
|
proxyUsed?: "basic" | "stealth";
|
|
@@ -705,7 +706,16 @@ interface MonitorCrawlTarget {
|
|
|
705
706
|
crawlOptions?: CrawlOptions;
|
|
706
707
|
scrapeOptions?: ScrapeOptions;
|
|
707
708
|
}
|
|
708
|
-
|
|
709
|
+
interface MonitorSearchTarget {
|
|
710
|
+
id?: string;
|
|
711
|
+
type: "search";
|
|
712
|
+
queries: string[];
|
|
713
|
+
searchWindow?: "5m" | "15m" | "1h" | "6h" | "24h" | "7d";
|
|
714
|
+
includeDomains?: string[];
|
|
715
|
+
excludeDomains?: string[];
|
|
716
|
+
maxResults?: number;
|
|
717
|
+
}
|
|
718
|
+
type MonitorTarget = MonitorScrapeTarget | MonitorCrawlTarget | MonitorSearchTarget;
|
|
709
719
|
interface CreateMonitorRequest {
|
|
710
720
|
name: string;
|
|
711
721
|
schedule: MonitorSchedule;
|
|
@@ -772,10 +782,34 @@ interface MonitorPageJudgment {
|
|
|
772
782
|
reason: string;
|
|
773
783
|
}>;
|
|
774
784
|
}
|
|
785
|
+
interface MonitorScrapeTargetResult {
|
|
786
|
+
targetId: string;
|
|
787
|
+
type: "scrape";
|
|
788
|
+
expectedJobs?: string[];
|
|
789
|
+
}
|
|
790
|
+
interface MonitorCrawlTargetResult {
|
|
791
|
+
targetId: string;
|
|
792
|
+
type: "crawl";
|
|
793
|
+
crawlId?: string;
|
|
794
|
+
}
|
|
795
|
+
interface MonitorSearchTargetResult {
|
|
796
|
+
targetId: string;
|
|
797
|
+
type: "search";
|
|
798
|
+
searchCompleted?: boolean;
|
|
799
|
+
resultCount?: number;
|
|
800
|
+
matches?: number;
|
|
801
|
+
summary?: string;
|
|
802
|
+
judgeDegraded?: boolean;
|
|
803
|
+
degradedReason?: string | null;
|
|
804
|
+
searchCredits?: number;
|
|
805
|
+
judgeCredits?: number;
|
|
806
|
+
resultsJudged?: number;
|
|
807
|
+
}
|
|
808
|
+
type MonitorTargetResult = MonitorScrapeTargetResult | MonitorCrawlTargetResult | MonitorSearchTargetResult;
|
|
775
809
|
interface MonitorCheck {
|
|
776
810
|
id: string;
|
|
777
811
|
monitorId: string;
|
|
778
|
-
status: "queued" | "running" | "completed" | "failed" | "partial" | "skipped_overlap";
|
|
812
|
+
status: "queued" | "running" | "completed" | "failed" | "partial" | "skipped_overlap" | "skipped_no_credits";
|
|
779
813
|
trigger: "scheduled" | "manual";
|
|
780
814
|
scheduledFor?: string | null;
|
|
781
815
|
startedAt?: string | null;
|
|
@@ -785,7 +819,7 @@ interface MonitorCheck {
|
|
|
785
819
|
actualCredits?: number | null;
|
|
786
820
|
billingStatus: "not_applicable" | "reserved" | "confirmed" | "released" | "failed";
|
|
787
821
|
summary: MonitorSummary;
|
|
788
|
-
targetResults?:
|
|
822
|
+
targetResults?: MonitorTargetResult[];
|
|
789
823
|
notificationStatus?: unknown;
|
|
790
824
|
error?: string | null;
|
|
791
825
|
createdAt: string;
|
|
@@ -1209,6 +1243,7 @@ declare function prepareExtractPayload(args: {
|
|
|
1209
1243
|
integration?: string;
|
|
1210
1244
|
origin?: string;
|
|
1211
1245
|
agent?: AgentOptions$1;
|
|
1246
|
+
webhook?: string | WebhookConfig | null;
|
|
1212
1247
|
}): Record<string, unknown>;
|
|
1213
1248
|
/**
|
|
1214
1249
|
* @deprecated The extract endpoint is in maintenance mode and its use is discouraged.
|
|
@@ -2588,4 +2623,4 @@ declare class Firecrawl extends FirecrawlClient {
|
|
|
2588
2623
|
get v1(): FirecrawlApp;
|
|
2589
2624
|
}
|
|
2590
2625
|
|
|
2591
|
-
export { type ActionOption, type ActiveCrawl, type ActiveCrawlsResponse, type AgentOptions$1 as AgentOptions, type AgentResponse, type AgentStatusResponse, type AgentWebhookConfig, type AgentWebhookEvent, type AttributesFormat, type BatchScrapeJob, type BatchScrapeOptions, type BatchScrapeResponse$1 as BatchScrapeResponse, type BrandingProfile, type BrowserCreateResponse, type BrowserDeleteResponse, type BrowserExecuteResponse, type BrowserListResponse, type BrowserSession, type CategoryOption, type ChangeTrackingFormat, type ClickAction, type ConcurrencyCheck, type CrawlErrorsResponse$1 as CrawlErrorsResponse, type CrawlJob, type CrawlOptions, type CrawlResponse$1 as CrawlResponse, type CreateMonitorRequest, type CreditUsage, type CreditUsageHistoricalPeriod, type CreditUsageHistoricalResponse, type Document, type DocumentMetadata, type EndpointFeedbackEndpoint, type EndpointFeedbackRequest, type ErrorDetails, type ExecuteJavascriptAction, type ExtractResponse$1 as ExtractResponse, type FeedbackMissingContent, type FeedbackRating, type FeedbackResponse, type FeedbackValuableSource, Firecrawl, FirecrawlApp as FirecrawlAppV1, FirecrawlClient, type FirecrawlClientInput, type FirecrawlClientOptions, type Format, type FormatOption, type FormatString, type GetMonitorCheckOptions, type GetPaperOptions, type GitHubScoreBreakdown, type GitHubSearchItem, type GitHubSearchResponse, type HighlightsFormat, type IdMap, JobTimeoutError, type JsonFormat, type ListMonitorChecksOptions, type ListMonitorsOptions, type LocationConfig$1 as LocationConfig, type MapData, type MapOptions, type MenuAvailability, type MenuImage, type MenuItem, type MenuItemIdentifiers, type MenuMerchant, type MenuPrice, type MenuProfile, type MenuSection, type Monitor, type MonitorCheck, type MonitorCheckDetail, type MonitorCheckPage, type MonitorCrawlTarget, type MonitorEmailNotification, type MonitorEmailRecipientSubscription, type MonitorJsonFieldDiff, type MonitorNotification, type MonitorPageDiff, type MonitorPageJudgment, type MonitorPageSnapshot, type MonitorSchedule, type MonitorScrapeTarget, type MonitorSummary, type MonitorTarget, type MonitorWebhookConfig, type PDFAction, type PaginationConfig, type PaperMetadata, type PaperMetadataResponse, type PaperResult, type PaperSignals, type ParseFile, type ParseFileData, type ParseFormat, type ParseFormatOption, type ParseFormatString, type ParseOptions, type Passage, type PressAction, type ProductAvailability, type ProductImage, type ProductPrice, type ProductProfile, type ProductSale, type ProductVariant, type QueryFormat, type QuestionFormat, type QueueStatusResponse$1 as QueueStatusResponse, type ReadPaperResponse, type RedactPIIEntity, type RedactPIIOptions, ResearchClient, type ScrapeAction, type ScrapeBrowserDeleteResponse, type ScrapeExecuteRequest, type ScrapeExecuteResponse, type ScrapeOptions, type ScreenshotAction, type ScreenshotFormat, type ScrollAction, SdkError, type SearchData, type SearchFeedbackRequest, type SearchGithubOptions, type SearchPapersOptions, type SearchPapersResponse, type SearchRequest, type SearchResultImages, type SearchResultNews, type SearchResultWeb, type SimilarPapersOptions, type SimilarPapersResponse, type TokenUsage, type TokenUsageHistoricalPeriod, type TokenUsageHistoricalResponse, type UpdateMonitorRequest, type Viewport, type WaitAction, Watcher, type WatcherOptions, type WebhookConfig, type WriteAction, Firecrawl as default };
|
|
2626
|
+
export { type ActionOption, type ActiveCrawl, type ActiveCrawlsResponse, type AgentOptions$1 as AgentOptions, type AgentResponse, type AgentStatusResponse, type AgentWebhookConfig, type AgentWebhookEvent, type AttributesFormat, type BatchScrapeJob, type BatchScrapeOptions, type BatchScrapeResponse$1 as BatchScrapeResponse, type BrandingProfile, type BrowserCreateResponse, type BrowserDeleteResponse, type BrowserExecuteResponse, type BrowserListResponse, type BrowserSession, type CategoryOption, type ChangeTrackingFormat, type ClickAction, type ConcurrencyCheck, type CrawlErrorsResponse$1 as CrawlErrorsResponse, type CrawlJob, type CrawlOptions, type CrawlResponse$1 as CrawlResponse, type CreateMonitorRequest, type CreditUsage, type CreditUsageHistoricalPeriod, type CreditUsageHistoricalResponse, type Document, type DocumentMetadata, type EndpointFeedbackEndpoint, type EndpointFeedbackRequest, type ErrorDetails, type ExecuteJavascriptAction, type ExtractResponse$1 as ExtractResponse, type FeedbackMissingContent, type FeedbackRating, type FeedbackResponse, type FeedbackValuableSource, Firecrawl, FirecrawlApp as FirecrawlAppV1, FirecrawlClient, type FirecrawlClientInput, type FirecrawlClientOptions, type Format, type FormatOption, type FormatString, type GetMonitorCheckOptions, type GetPaperOptions, type GitHubScoreBreakdown, type GitHubSearchItem, type GitHubSearchResponse, type HighlightsFormat, type IdMap, JobTimeoutError, type JsonFormat, type ListMonitorChecksOptions, type ListMonitorsOptions, type LocationConfig$1 as LocationConfig, type MapData, type MapOptions, type MenuAvailability, type MenuImage, type MenuItem, type MenuItemIdentifiers, type MenuMerchant, type MenuPrice, type MenuProfile, type MenuSection, type Monitor, type MonitorCheck, type MonitorCheckDetail, type MonitorCheckPage, type MonitorCrawlTarget, type MonitorCrawlTargetResult, type MonitorEmailNotification, type MonitorEmailRecipientSubscription, type MonitorJsonFieldDiff, type MonitorNotification, type MonitorPageDiff, type MonitorPageJudgment, type MonitorPageSnapshot, type MonitorSchedule, type MonitorScrapeTarget, type MonitorScrapeTargetResult, type MonitorSearchTarget, type MonitorSearchTargetResult, type MonitorSummary, type MonitorTarget, type MonitorTargetResult, type MonitorWebhookConfig, type PDFAction, type PaginationConfig, type PaperMetadata, type PaperMetadataResponse, type PaperResult, type PaperSignals, type ParseFile, type ParseFileData, type ParseFormat, type ParseFormatOption, type ParseFormatString, type ParseOptions, type Passage, type PressAction, type ProductAvailability, type ProductImage, type ProductPrice, type ProductProfile, type ProductSale, type ProductVariant, type QueryFormat, type QuestionFormat, type QueueStatusResponse$1 as QueueStatusResponse, type ReadPaperResponse, type RedactPIIEntity, type RedactPIIOptions, ResearchClient, type ScrapeAction, type ScrapeBrowserDeleteResponse, type ScrapeExecuteRequest, type ScrapeExecuteResponse, type ScrapeOptions, type ScreenshotAction, type ScreenshotFormat, type ScrollAction, SdkError, type SearchData, type SearchFeedbackRequest, type SearchGithubOptions, type SearchPapersOptions, type SearchPapersResponse, type SearchRequest, type SearchResultImages, type SearchResultNews, type SearchResultWeb, type SimilarPapersOptions, type SimilarPapersResponse, type TokenUsage, type TokenUsageHistoricalPeriod, type TokenUsageHistoricalResponse, type UpdateMonitorRequest, type Viewport, type WaitAction, Watcher, type WatcherOptions, type WebhookConfig, type WriteAction, Firecrawl as default };
|
package/dist/index.js
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import {
|
|
2
2
|
require_package
|
|
3
|
-
} from "./chunk-
|
|
3
|
+
} from "./chunk-VRKVHB52.js";
|
|
4
4
|
|
|
5
5
|
// src/v2/utils/httpClient.ts
|
|
6
6
|
import axios from "axios";
|
|
@@ -1094,6 +1094,7 @@ function prepareExtractPayload(args) {
|
|
|
1094
1094
|
if (args.integration && args.integration.trim()) body.integration = args.integration.trim();
|
|
1095
1095
|
if (args.origin) body.origin = args.origin;
|
|
1096
1096
|
if (args.agent) body.agent = args.agent;
|
|
1097
|
+
if (args.webhook != null) body.webhook = args.webhook;
|
|
1097
1098
|
if (args.scrapeOptions) {
|
|
1098
1099
|
ensureValidScrapeOptions(args.scrapeOptions);
|
|
1099
1100
|
body.scrapeOptions = args.scrapeOptions;
|
|
@@ -2300,7 +2301,7 @@ var FirecrawlApp = class {
|
|
|
2300
2301
|
if (typeof process !== "undefined" && process.env && process.env.npm_package_version) {
|
|
2301
2302
|
return process.env.npm_package_version;
|
|
2302
2303
|
}
|
|
2303
|
-
const packageJson = await import("./package-
|
|
2304
|
+
const packageJson = await import("./package-EYT327NL.js");
|
|
2304
2305
|
return packageJson.default.version;
|
|
2305
2306
|
} catch (error) {
|
|
2306
2307
|
const isTest = typeof process !== "undefined" && (process.env.JEST_WORKER_ID != null || false);
|
package/package.json
CHANGED
|
@@ -0,0 +1,47 @@
|
|
|
1
|
+
import { describe, expect, test } from "@jest/globals";
|
|
2
|
+
import { startExtract } from "../../../v2/methods/extract";
|
|
3
|
+
import type { WebhookConfig } from "../../../v2/types";
|
|
4
|
+
|
|
5
|
+
describe("v2.extract unit", () => {
|
|
6
|
+
test("startExtract forwards string webhook in request payload", async () => {
|
|
7
|
+
const post = jest.fn().mockResolvedValue({
|
|
8
|
+
status: 200,
|
|
9
|
+
data: { id: "extract-job" },
|
|
10
|
+
});
|
|
11
|
+
|
|
12
|
+
await startExtract({ post } as any, {
|
|
13
|
+
urls: ["https://example.com"],
|
|
14
|
+
prompt: "Extract title",
|
|
15
|
+
webhook: "https://example.com/webhook",
|
|
16
|
+
});
|
|
17
|
+
|
|
18
|
+
expect(post).toHaveBeenCalledWith("/v2/extract", {
|
|
19
|
+
urls: ["https://example.com"],
|
|
20
|
+
prompt: "Extract title",
|
|
21
|
+
webhook: "https://example.com/webhook",
|
|
22
|
+
});
|
|
23
|
+
});
|
|
24
|
+
|
|
25
|
+
test("startExtract forwards object webhook in request payload", async () => {
|
|
26
|
+
const post = jest.fn().mockResolvedValue({
|
|
27
|
+
status: 200,
|
|
28
|
+
data: { id: "extract-job" },
|
|
29
|
+
});
|
|
30
|
+
const webhook: WebhookConfig = {
|
|
31
|
+
url: "https://example.com/webhook",
|
|
32
|
+
headers: { "x-signature": "secret" },
|
|
33
|
+
};
|
|
34
|
+
|
|
35
|
+
await startExtract({ post } as any, {
|
|
36
|
+
urls: ["https://example.com"],
|
|
37
|
+
prompt: "Extract title",
|
|
38
|
+
webhook,
|
|
39
|
+
});
|
|
40
|
+
|
|
41
|
+
expect(post).toHaveBeenCalledWith("/v2/extract", {
|
|
42
|
+
urls: ["https://example.com"],
|
|
43
|
+
prompt: "Extract title",
|
|
44
|
+
webhook,
|
|
45
|
+
});
|
|
46
|
+
});
|
|
47
|
+
});
|
|
@@ -0,0 +1,40 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Unit test: a search-target monitor request is forwarded intact to the API.
|
|
3
|
+
*/
|
|
4
|
+
import { createMonitor } from "../../../v2/methods/monitor";
|
|
5
|
+
import type { CreateMonitorRequest, MonitorSearchTarget } from "../../../v2/types";
|
|
6
|
+
|
|
7
|
+
describe("v2 monitor search target", () => {
|
|
8
|
+
test("createMonitor forwards a search target with its camelCase fields", async () => {
|
|
9
|
+
let captured: any;
|
|
10
|
+
const http: any = {
|
|
11
|
+
post: async (_path: string, body: any) => {
|
|
12
|
+
captured = body;
|
|
13
|
+
return { status: 200, data: { success: true, data: { id: "mon_1", ...body } } };
|
|
14
|
+
},
|
|
15
|
+
};
|
|
16
|
+
|
|
17
|
+
const searchTarget: MonitorSearchTarget = {
|
|
18
|
+
type: "search",
|
|
19
|
+
queries: ["firecrawl launch"],
|
|
20
|
+
searchWindow: "24h",
|
|
21
|
+
includeDomains: ["firecrawl.dev"],
|
|
22
|
+
excludeDomains: ["spam.com"],
|
|
23
|
+
maxResults: 20,
|
|
24
|
+
};
|
|
25
|
+
|
|
26
|
+
const request: CreateMonitorRequest = {
|
|
27
|
+
name: "Search monitor",
|
|
28
|
+
schedule: { text: "every 30 minutes" },
|
|
29
|
+
goal: "Alert when Firecrawl launches a product",
|
|
30
|
+
targets: [searchTarget],
|
|
31
|
+
};
|
|
32
|
+
|
|
33
|
+
await createMonitor(http, request);
|
|
34
|
+
|
|
35
|
+
expect(captured.targets[0]).toEqual(searchTarget);
|
|
36
|
+
expect(captured.targets[0].type).toBe("search");
|
|
37
|
+
expect(captured.targets[0].searchWindow).toBe("24h");
|
|
38
|
+
expect(captured.targets[0].maxResults).toBe(20);
|
|
39
|
+
});
|
|
40
|
+
});
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { type ExtractResponse, type ScrapeOptions, type AgentOptions } from "../types";
|
|
1
|
+
import { type ExtractResponse, type ScrapeOptions, type AgentOptions, type WebhookConfig } from "../types";
|
|
2
2
|
import { HttpClient } from "../utils/httpClient";
|
|
3
3
|
import { ensureValidScrapeOptions } from "../utils/validation";
|
|
4
4
|
import { normalizeAxiosError, throwForBadResponse } from "../utils/errorHandler";
|
|
@@ -18,6 +18,7 @@ function prepareExtractPayload(args: {
|
|
|
18
18
|
integration?: string;
|
|
19
19
|
origin?: string;
|
|
20
20
|
agent?: AgentOptions;
|
|
21
|
+
webhook?: string | WebhookConfig | null;
|
|
21
22
|
}): Record<string, unknown> {
|
|
22
23
|
const body: Record<string, unknown> = {};
|
|
23
24
|
if (args.urls) body.urls = args.urls;
|
|
@@ -33,6 +34,7 @@ function prepareExtractPayload(args: {
|
|
|
33
34
|
if (args.integration && args.integration.trim()) body.integration = args.integration.trim();
|
|
34
35
|
if (args.origin) body.origin = args.origin;
|
|
35
36
|
if (args.agent) body.agent = args.agent;
|
|
37
|
+
if (args.webhook != null) body.webhook = args.webhook;
|
|
36
38
|
if (args.scrapeOptions) {
|
|
37
39
|
ensureValidScrapeOptions(args.scrapeOptions);
|
|
38
40
|
body.scrapeOptions = args.scrapeOptions;
|
package/src/v2/types.ts
CHANGED
|
@@ -570,6 +570,7 @@ export interface DocumentMetadata {
|
|
|
570
570
|
statusCode?: number;
|
|
571
571
|
scrapeId?: string;
|
|
572
572
|
numPages?: number;
|
|
573
|
+
totalPages?: number;
|
|
573
574
|
contentType?: string;
|
|
574
575
|
timezone?: string;
|
|
575
576
|
proxyUsed?: "basic" | "stealth";
|
|
@@ -878,7 +879,20 @@ export interface MonitorCrawlTarget {
|
|
|
878
879
|
scrapeOptions?: ScrapeOptions;
|
|
879
880
|
}
|
|
880
881
|
|
|
881
|
-
export
|
|
882
|
+
export interface MonitorSearchTarget {
|
|
883
|
+
id?: string;
|
|
884
|
+
type: "search";
|
|
885
|
+
queries: string[];
|
|
886
|
+
searchWindow?: "5m" | "15m" | "1h" | "6h" | "24h" | "7d";
|
|
887
|
+
includeDomains?: string[];
|
|
888
|
+
excludeDomains?: string[];
|
|
889
|
+
maxResults?: number;
|
|
890
|
+
}
|
|
891
|
+
|
|
892
|
+
export type MonitorTarget =
|
|
893
|
+
| MonitorScrapeTarget
|
|
894
|
+
| MonitorCrawlTarget
|
|
895
|
+
| MonitorSearchTarget;
|
|
882
896
|
|
|
883
897
|
export interface CreateMonitorRequest {
|
|
884
898
|
name: string;
|
|
@@ -951,6 +965,37 @@ export interface MonitorPageJudgment {
|
|
|
951
965
|
}>;
|
|
952
966
|
}
|
|
953
967
|
|
|
968
|
+
export interface MonitorScrapeTargetResult {
|
|
969
|
+
targetId: string;
|
|
970
|
+
type: "scrape";
|
|
971
|
+
expectedJobs?: string[];
|
|
972
|
+
}
|
|
973
|
+
|
|
974
|
+
export interface MonitorCrawlTargetResult {
|
|
975
|
+
targetId: string;
|
|
976
|
+
type: "crawl";
|
|
977
|
+
crawlId?: string;
|
|
978
|
+
}
|
|
979
|
+
|
|
980
|
+
export interface MonitorSearchTargetResult {
|
|
981
|
+
targetId: string;
|
|
982
|
+
type: "search";
|
|
983
|
+
searchCompleted?: boolean;
|
|
984
|
+
resultCount?: number;
|
|
985
|
+
matches?: number;
|
|
986
|
+
summary?: string;
|
|
987
|
+
judgeDegraded?: boolean;
|
|
988
|
+
degradedReason?: string | null;
|
|
989
|
+
searchCredits?: number;
|
|
990
|
+
judgeCredits?: number;
|
|
991
|
+
resultsJudged?: number;
|
|
992
|
+
}
|
|
993
|
+
|
|
994
|
+
export type MonitorTargetResult =
|
|
995
|
+
| MonitorScrapeTargetResult
|
|
996
|
+
| MonitorCrawlTargetResult
|
|
997
|
+
| MonitorSearchTargetResult;
|
|
998
|
+
|
|
954
999
|
export interface MonitorCheck {
|
|
955
1000
|
id: string;
|
|
956
1001
|
monitorId: string;
|
|
@@ -960,7 +1005,8 @@ export interface MonitorCheck {
|
|
|
960
1005
|
| "completed"
|
|
961
1006
|
| "failed"
|
|
962
1007
|
| "partial"
|
|
963
|
-
| "skipped_overlap"
|
|
1008
|
+
| "skipped_overlap"
|
|
1009
|
+
| "skipped_no_credits";
|
|
964
1010
|
trigger: "scheduled" | "manual";
|
|
965
1011
|
scheduledFor?: string | null;
|
|
966
1012
|
startedAt?: string | null;
|
|
@@ -975,7 +1021,7 @@ export interface MonitorCheck {
|
|
|
975
1021
|
| "released"
|
|
976
1022
|
| "failed";
|
|
977
1023
|
summary: MonitorSummary;
|
|
978
|
-
targetResults?:
|
|
1024
|
+
targetResults?: MonitorTargetResult[];
|
|
979
1025
|
notificationStatus?: unknown;
|
|
980
1026
|
error?: string | null;
|
|
981
1027
|
createdAt: string;
|