@@ -38,7 +38,7 @@ import {
38
38
CrawlHTMLDetailTargetConfig ,
39
39
CrawlHTMLAdvancedConfig
40
40
} from './types/api'
41
- import { LogConfig , XCrawlInstanceConfig } from './types'
41
+ import { LogOptions , CrawlBaseConfig } from './types'
42
42
import { fingerprints } from './default'
43
43
44
44
/* Types */
@@ -49,7 +49,7 @@ export interface InfoCommonConfig {
49
49
serialNumber : string
50
50
mode : 'async' | 'sync'
51
51
type : 'page' | 'html' | 'data' | 'file'
52
- logConfig : LogConfig
52
+ logConfig : LogOptions
53
53
intervalTime : IntervalTime | undefined
54
54
}
55
55
@@ -403,7 +403,7 @@ function loaderPageFingerprintToDetailTarget(
403
403
}
404
404
405
405
function loaderCommonConfigToCrawlConfig (
406
- xCrawlInstanceConfig : XCrawlInstanceConfig ,
406
+ crawlBaseConfig : CrawlBaseConfig ,
407
407
advancedDetailTargetsConfig :
408
408
| CrawlPageAdvancedDetailTargetsConfig
409
409
| CrawlHTMLAdvancedDetailTargetsConfig
@@ -429,16 +429,16 @@ function loaderCommonConfigToCrawlConfig(
429
429
detail
430
430
431
431
// 1.1.baseUrl
432
- if ( xCrawlInstanceConfig . baseUrl ) {
433
- detail . url = xCrawlInstanceConfig . baseUrl + url
432
+ if ( crawlBaseConfig . baseUrl ) {
433
+ detail . url = crawlBaseConfig . baseUrl + url
434
434
}
435
435
436
436
// 1.2.timeout
437
437
if ( isUndefined ( timeout ) ) {
438
438
if ( ! isUndefined ( advancedDetailTargetsConfig . timeout ) ) {
439
439
detail . timeout = advancedDetailTargetsConfig . timeout ?? undefined
440
440
} else {
441
- detail . timeout = xCrawlInstanceConfig . timeout
441
+ detail . timeout = crawlBaseConfig . timeout
442
442
}
443
443
}
444
444
@@ -447,16 +447,16 @@ function loaderCommonConfigToCrawlConfig(
447
447
if ( ! isUndefined ( advancedDetailTargetsConfig . maxRetry ) ) {
448
448
detail . maxRetry = advancedDetailTargetsConfig . maxRetry ?? 0
449
449
} else {
450
- detail . maxRetry = xCrawlInstanceConfig . maxRetry
450
+ detail . maxRetry = crawlBaseConfig . maxRetry
451
451
}
452
452
}
453
453
454
454
// 1.4.proxy
455
455
if ( isUndefined ( proxy ) ) {
456
456
if ( ! isUndefined ( advancedDetailTargetsConfig . proxy ) ) {
457
457
detail . proxy = advancedDetailTargetsConfig . proxy
458
- } else if ( ! isUndefined ( xCrawlInstanceConfig . proxy ) ) {
459
- detail . proxy = xCrawlInstanceConfig . proxy
458
+ } else if ( ! isUndefined ( crawlBaseConfig . proxy ) ) {
459
+ detail . proxy = crawlBaseConfig . proxy
460
460
}
461
461
}
462
462
@@ -503,9 +503,9 @@ function loaderCommonConfigToCrawlConfig(
503
503
} else if (
504
504
isUndefined ( fingerprint ) &&
505
505
! isArray ( advancedDetailTargetsConfig . fingerprints ) &&
506
- xCrawlInstanceConfig . enableRandomFingerprint
506
+ crawlBaseConfig . enableRandomFingerprint
507
507
) {
508
- // xCrawlInstanceConfig
508
+ // crawlBaseConfig
509
509
const fingerprint = fingerprints [ random ( fingerprints . length ) ]
510
510
511
511
loaderCommonFingerprintToDetailTarget ( detail , fingerprint )
@@ -519,9 +519,9 @@ function loaderCommonConfigToCrawlConfig(
519
519
crawlConfig . intervalTime = advancedDetailTargetsConfig . intervalTime
520
520
if (
521
521
isUndefined ( advancedDetailTargetsConfig . intervalTime ) &&
522
- ! isUndefined ( xCrawlInstanceConfig . intervalTime )
522
+ ! isUndefined ( crawlBaseConfig . intervalTime )
523
523
) {
524
- crawlConfig . intervalTime = xCrawlInstanceConfig . intervalTime
524
+ crawlConfig . intervalTime = crawlBaseConfig . intervalTime
525
525
}
526
526
527
527
// 3.onCrawlItemComplete
@@ -540,7 +540,7 @@ function loaderCommonConfigToCrawlConfig(
540
540
*/
541
541
542
542
function createCrawlPageConfig (
543
- xCrawlInstanceConfig : XCrawlInstanceConfig ,
543
+ crawlBaseConfig : CrawlBaseConfig ,
544
544
originalConfig : UniteCrawlPageConfig
545
545
) : CrawlPageConfig {
546
546
const crawlPageConfig : CrawlPageConfig = {
@@ -577,7 +577,7 @@ function createCrawlPageConfig(
577
577
578
578
// 装载公共配置
579
579
loaderCommonConfigToCrawlConfig (
580
- xCrawlInstanceConfig ,
580
+ crawlBaseConfig ,
581
581
advancedDetailTargetsConfig ,
582
582
crawlPageConfig
583
583
)
@@ -618,7 +618,7 @@ function createCrawlPageConfig(
618
618
}
619
619
620
620
function createCrawlHTMLConfig (
621
- xCrawlInstanceConfig : XCrawlInstanceConfig ,
621
+ crawlBaseConfig : CrawlBaseConfig ,
622
622
originalConfig : UniteCrawlHTMLConfig
623
623
) : CrawlHTMLConfig {
624
624
const crawlHTMLConfig : CrawlHTMLConfig = {
@@ -658,7 +658,7 @@ function createCrawlHTMLConfig(
658
658
}
659
659
660
660
loaderCommonConfigToCrawlConfig (
661
- xCrawlInstanceConfig ,
661
+ crawlBaseConfig ,
662
662
advancedDetailTargetsConfig ,
663
663
crawlHTMLConfig
664
664
)
@@ -667,7 +667,7 @@ function createCrawlHTMLConfig(
667
667
}
668
668
669
669
function createCrawlDataConfig < T > (
670
- xCrawlInstanceConfig : XCrawlInstanceConfig ,
670
+ crawlBaseConfig : CrawlBaseConfig ,
671
671
originalConfig : UniteCrawlDataConfig < T >
672
672
) : CrawlDataConfig {
673
673
const crawlDataConfig : CrawlDataConfig = {
@@ -704,7 +704,7 @@ function createCrawlDataConfig<T>(
704
704
}
705
705
706
706
loaderCommonConfigToCrawlConfig (
707
- xCrawlInstanceConfig ,
707
+ crawlBaseConfig ,
708
708
advancedDetailTargetsConfig ,
709
709
crawlDataConfig
710
710
)
@@ -713,7 +713,7 @@ function createCrawlDataConfig<T>(
713
713
}
714
714
715
715
function createCrawlFileConfig (
716
- xCrawlInstanceConfig : XCrawlInstanceConfig ,
716
+ crawlBaseConfig : CrawlBaseConfig ,
717
717
originalConfig : UniteCrawlFileConfig
718
718
) : CrawlFileConfig {
719
719
const crawlFileConfig : CrawlFileConfig = {
@@ -748,7 +748,7 @@ function createCrawlFileConfig(
748
748
}
749
749
750
750
loaderCommonConfigToCrawlConfig (
751
- xCrawlInstanceConfig ,
751
+ crawlBaseConfig ,
752
752
advancedDetailTargetsConfig ,
753
753
crawlFileConfig
754
754
)
@@ -1074,13 +1074,13 @@ function fileSingleResultHandle(
1074
1074
1075
1075
/* Create crawl API */
1076
1076
1077
- export function createCrawlPage ( xCrawlInstanceConfig : XCrawlInstanceConfig ) {
1077
+ export function createCrawlPage ( crawlBaseConfig : CrawlBaseConfig ) {
1078
1078
const {
1079
1079
id : xId ,
1080
1080
mode,
1081
- logConfig,
1082
- crawlPage : crawlPageConfig
1083
- } = xCrawlInstanceConfig
1081
+ logOptions : logConfig ,
1082
+ crawlPage : crawlPageOptions
1083
+ } = crawlBaseConfig
1084
1084
1085
1085
let id = 0
1086
1086
let browser : Browser | null = null
@@ -1109,7 +1109,7 @@ export function createCrawlPage(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1109
1109
if ( ! haveCreateBrowser ) {
1110
1110
haveCreateBrowser = true
1111
1111
createBrowserPending = puppeteer
1112
- . launch ( crawlPageConfig ?. puppeteerLaunch )
1112
+ . launch ( crawlPageOptions ?. puppeteerLaunchOptions )
1113
1113
. then ( ( result ) => {
1114
1114
browser = result
1115
1115
} )
@@ -1124,7 +1124,7 @@ export function createCrawlPage(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1124
1124
1125
1125
// 创建新配置
1126
1126
const { detailTargets, intervalTime, onCrawlItemComplete } =
1127
- createCrawlPageConfig ( xCrawlInstanceConfig , config )
1127
+ createCrawlPageConfig ( crawlBaseConfig , config )
1128
1128
1129
1129
const infoConfig : InfoPageConfig = {
1130
1130
serialNumber : `${ xId } -${ type } -${ ++ id } ` ,
@@ -1154,8 +1154,8 @@ export function createCrawlPage(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1154
1154
return crawlPage
1155
1155
}
1156
1156
1157
- export function createCrawlHTML ( xCrawlInstanceConfig : XCrawlInstanceConfig ) {
1158
- const { id : xId , mode, logConfig } = xCrawlInstanceConfig
1157
+ export function createCrawlHTML ( crawlBaseConfig : CrawlBaseConfig ) {
1158
+ const { id : xId , mode, logOptions : logConfig } = crawlBaseConfig
1159
1159
let id = 0
1160
1160
const type = 'html'
1161
1161
@@ -1177,7 +1177,7 @@ export function createCrawlHTML(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1177
1177
config : UniteCrawlHTMLConfig
1178
1178
) : Promise < CrawlHTMLSingleResult | CrawlHTMLSingleResult [ ] > {
1179
1179
const { detailTargets, intervalTime, onCrawlItemComplete } =
1180
- createCrawlHTMLConfig ( xCrawlInstanceConfig , config )
1180
+ createCrawlHTMLConfig ( crawlBaseConfig , config )
1181
1181
1182
1182
const infoConfig : InfoHTMLConfig = {
1183
1183
serialNumber : `${ xId } -${ type } -${ ++ id } ` ,
@@ -1206,8 +1206,8 @@ export function createCrawlHTML(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1206
1206
return crawlHTML
1207
1207
}
1208
1208
1209
- export function createCrawlData ( xCrawlInstanceConfig : XCrawlInstanceConfig ) {
1210
- const { id : xId , mode, logConfig } = xCrawlInstanceConfig
1209
+ export function createCrawlData ( crawlBaseConfig : CrawlBaseConfig ) {
1210
+ const { id : xId , mode, logOptions : logConfig } = crawlBaseConfig
1211
1211
let id = 0
1212
1212
const type = 'data'
1213
1213
@@ -1229,7 +1229,7 @@ export function createCrawlData(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1229
1229
config : UniteCrawlDataConfig < T >
1230
1230
) : Promise < CrawlDataSingleResult < T > | CrawlDataSingleResult < T > [ ] > {
1231
1231
const { detailTargets, intervalTime, onCrawlItemComplete } =
1232
- createCrawlDataConfig ( xCrawlInstanceConfig , config )
1232
+ createCrawlDataConfig ( crawlBaseConfig , config )
1233
1233
1234
1234
const infoConfig : InfoDataConfig < T > = {
1235
1235
serialNumber : `${ xId } -${ type } -${ ++ id } ` ,
@@ -1258,8 +1258,8 @@ export function createCrawlData(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1258
1258
return crawlData
1259
1259
}
1260
1260
1261
- export function createCrawlFile ( xCrawlInstanceConfig : XCrawlInstanceConfig ) {
1262
- const { id : xId , mode, logConfig } = xCrawlInstanceConfig
1261
+ export function createCrawlFile ( crawlBaseConfig : CrawlBaseConfig ) {
1262
+ const { id : xId , mode, logOptions : logConfig } = crawlBaseConfig
1263
1263
let id = 0
1264
1264
const type = 'file'
1265
1265
@@ -1283,7 +1283,7 @@ export function createCrawlFile(xCrawlInstanceConfig: XCrawlInstanceConfig) {
1283
1283
intervalTime,
1284
1284
onBeforeSaveItemFile,
1285
1285
onCrawlItemComplete
1286
- } = createCrawlFileConfig ( xCrawlInstanceConfig , config )
1286
+ } = createCrawlFileConfig ( crawlBaseConfig , config )
1287
1287
1288
1288
const infoConfig : InfoFileConfig = {
1289
1289
serialNumber : `${ xId } -${ type } -${ ++ id } ` ,
0 commit comments