hydra-crawler 2.8.3 → 2.8.4

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.
@@ -10,7 +10,7 @@ var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, ge
10
10
  import { commonsTypeHasPropertyNumber } from 'tscommons-es-core';
11
11
  import { isIUrl } from 'hydra-crawler-ts-assets';
12
12
  import { EStatus } from 'hydra-crawler-ts-assets';
13
- import { commonsOutputDoing, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
13
+ import { commonsOutputDoing, commonsOutputPercent, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
14
14
  import { CommonsApp } from 'nodecommons-es-app';
15
15
  import { Lists } from '../classes/lists';
16
16
  import { Cleaner } from '../classes/cleaner';
@@ -47,7 +47,8 @@ export class CleanupApp extends CommonsApp {
47
47
  return __awaiter(this, void 0, void 0, function* () {
48
48
  if (!this.databaseService)
49
49
  throw new Error('Database service has not been set');
50
- commonsOutputDoing('Reseting urls for DENY, FAILED, DEAD and DISALLOWED');
50
+ commonsOutputDoing('Enumerating urls for DENY, FAILED, DEAD and DISALLOWED');
51
+ const urls = [];
51
52
  let tally = 0;
52
53
  while (true) {
53
54
  const result = this.databaseService.getUrls()
@@ -67,7 +68,7 @@ export class CleanupApp extends CommonsApp {
67
68
  const row = yield result.next();
68
69
  if (row === null)
69
70
  break;
70
- yield this.databaseService.getLinks().deleteMany({ url: row.url });
71
+ urls.push(row.url);
71
72
  }
72
73
  break;
73
74
  }
@@ -77,6 +78,15 @@ export class CleanupApp extends CommonsApp {
77
78
  }
78
79
  }
79
80
  commonsOutputResult(tally);
81
+ commonsOutputDoing('Resetting links for DENY, FAILED, DEAD and DISALLOWED');
82
+ let i = 0;
83
+ for (const url of urls) {
84
+ i++;
85
+ if ((i % 100) === 0)
86
+ commonsOutputPercent(i, urls.length);
87
+ yield this.databaseService.getLinks().deleteMany({ url: url });
88
+ }
89
+ commonsOutputSuccess();
80
90
  });
81
91
  }
82
92
  links() {
@@ -1 +1 @@
1
- {"version":3,"file":"cleanup.app.js","sourceRoot":"","sources":["../../src/apps/cleanup.app.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,4BAA4B,EAAE,MAAM,mBAAmB,CAAC;AAEjE,OAAO,EAAQ,MAAM,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAElD,OAAO,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC1H,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AACzC,OAAO,EAAE,OAAO,EAAE,MAAM,oBAAoB,CAAC;AAU7C,MAAM,OAAO,UAAW,SAAQ,UAAU;IAKzC,YACU,eAAwB,KAAK,EAC7B,eAAwB,KAAK;QAEtC,KAAK,CAAC,eAAe,CAAC,CAAC;QAHd,iBAAY,GAAZ,YAAY,CAAiB;QAC7B,iBAAY,GAAZ,YAAY,CAAiB;QAItC,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,iBAAiB,CAAC;IAC1B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEa,KAAK;;YAClB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,qDAAqD,CAAC,CAAC;YAE1E,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,OAAO,IAAI,EAAE;gBACZ,MAAM,MAAM,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBACpE,IAAI,CAAC,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE;4BACrB,OAAO,CAAC,IAAI;4BACZ,OAAO,CAAC,MAAM;4BACd,OAAO,CAAC,IAAI;4BACZ,OAAO,CAAC,UAAU;yBACnB,EAAE,EAAE,CAAC;qBACL,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;qBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;gBAEf,IAAI;oBACH,OAAO,IAAI,EAAE;wBACZ,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;wBAEtD,MAAM,GAAG,GAAc,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;wBAC3C,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;qBACnE;oBAED,MAAM;iBACN;gBAAC,OAAO,GAAG,EAAE;oBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;iBAC7E;aACD;YAED,mBAAmB,CAAC,KAAK,CAAC,CAAC;QAC5B,CAAC;KAAA;IAEa,KAAK;;YAClB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,4BAA4B,CAAC,CAAC;YACjD,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,MAAM,MAAM,GAA8B,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;YAE9F,IAAI,KAAK,GAAa,EAAE,CAAC;YACzB,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,OAAO,IAAI,EAAE;gBACZ,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,KAAK,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;gBAE/E,MAAM,IAAI,GAAe,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBAC7C,IAAI,IAAI,KAAK,IAAI;oBAAE,MAAM;gBAEzB,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACpD,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAE9D,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK;oBAAE,SAAS;gBAEnC,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAC3E,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,CACxB,CAAC;gBAEF,MAAM,OAAO,GAAa,CAAC,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;qBACrF,GAAG,CAAC,CAAC,CAAO,EAAU,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;gBAEnC,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE;oBACtB,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC;wBAAE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;iBAC1C;gBAED,KAAK,GAAG,EAAE,CAAC;aACX;YACD,mBAAmB,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;YAEpC,kBAAkB,CAAC,uBAAuB,CAAC,CAAC;YAC5C,KAAK,GAAG,CAAC,CAAC;YACV,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;gBAC7B,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,EAAE,CAAC,CAAC;gBAC3D,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC;gBAClE,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC;aACvE;YACD,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,IAAI,IAAI,CAAC,YAAY;gBAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;YAC1C,IAAI,IAAI,CAAC,YAAY;gBAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;YAE1C,MAAM,SAAS,GAAqB,IAAI,CAAC,OAAO,EAAE,CAAC,oBAAoB,CAAC,YAAY,CAAC,CAAC;YAEtF,MAAM,OAAO,GAAY,IAAI,OAAO,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,eAAe,CAAC,CAAC;YACvE,MAAM,OAAO,CAAC,eAAe,CAAC,SAAS,CAAC,CAAC;YACzC,MAAM,OAAO,CAAC,iBAAiB,EAAE,CAAC;QACnC,CAAC;KAAA;CACD"}
1
+ {"version":3,"file":"cleanup.app.js","sourceRoot":"","sources":["../../src/apps/cleanup.app.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,4BAA4B,EAAE,MAAM,mBAAmB,CAAC;AAEjE,OAAO,EAAQ,MAAM,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAElD,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAChJ,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AACzC,OAAO,EAAE,OAAO,EAAE,MAAM,oBAAoB,CAAC;AAU7C,MAAM,OAAO,UAAW,SAAQ,UAAU;IAKzC,YACU,eAAwB,KAAK,EAC7B,eAAwB,KAAK;QAEtC,KAAK,CAAC,eAAe,CAAC,CAAC;QAHd,iBAAY,GAAZ,YAAY,CAAiB;QAC7B,iBAAY,GAAZ,YAAY,CAAiB;QAItC,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,iBAAiB,CAAC;IAC1B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEa,KAAK;;YAClB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,wDAAwD,CAAC,CAAC;YAE7E,MAAM,IAAI,GAAa,EAAE,CAAC;YAE1B,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,OAAO,IAAI,EAAE;gBACZ,MAAM,MAAM,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBACpE,IAAI,CAAC,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE;4BACrB,OAAO,CAAC,IAAI;4BACZ,OAAO,CAAC,MAAM;4BACd,OAAO,CAAC,IAAI;4BACZ,OAAO,CAAC,UAAU;yBACnB,EAAE,EAAE,CAAC;qBACL,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;qBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;gBAEf,IAAI;oBACH,OAAO,IAAI,EAAE;wBACZ,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;wBAEtD,MAAM,GAAG,GAAc,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;wBAC3C,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;qBACnB;oBAED,MAAM;iBACN;gBAAC,OAAO,GAAG,EAAE;oBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;iBAC7E;aACD;YAED,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,kBAAkB,CAAC,uDAAuD,CAAC,CAAC;YAC5E,IAAI,CAAC,GAAW,CAAC,CAAC;YAClB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE;gBACvB,CAAC,EAAE,CAAC;gBACJ,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,oBAAoB,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAE1D,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC;aAC/D;YACD,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;IAEa,KAAK;;YAClB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,4BAA4B,CAAC,CAAC;YACjD,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,MAAM,MAAM,GAA8B,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;YAE9F,IAAI,KAAK,GAAa,EAAE,CAAC;YACzB,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,OAAO,IAAI,EAAE;gBACZ,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,KAAK,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;gBAE/E,MAAM,IAAI,GAAe,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBAC7C,IAAI,IAAI,KAAK,IAAI;oBAAE,MAAM;gBAEzB,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACpD,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAE9D,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK;oBAAE,SAAS;gBAEnC,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAC3E,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,CACxB,CAAC;gBAEF,MAAM,OAAO,GAAa,CAAC,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;qBACrF,GAAG,CAAC,CAAC,CAAO,EAAU,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;gBAEnC,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE;oBACtB,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC;wBAAE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;iBAC1C;gBAED,KAAK,GAAG,EAAE,CAAC;aACX;YACD,mBAAmB,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;YAEpC,kBAAkB,CAAC,uBAAuB,CAAC,CAAC;YAC5C,KAAK,GAAG,CAAC,CAAC;YACV,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;gBAC7B,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,EAAE,CAAC,CAAC;gBAC3D,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC;gBAClE,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC;aACvE;YACD,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,IAAI,IAAI,CAAC,YAAY;gBAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;YAC1C,IAAI,IAAI,CAAC,YAAY;gBAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;YAE1C,MAAM,SAAS,GAAqB,IAAI,CAAC,OAAO,EAAE,CAAC,oBAAoB,CAAC,YAAY,CAAC,CAAC;YAEtF,MAAM,OAAO,GAAY,IAAI,OAAO,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,eAAe,CAAC,CAAC;YACvE,MAAM,OAAO,CAAC,eAAe,CAAC,SAAS,CAAC,CAAC;YACzC,MAAM,OAAO,CAAC,iBAAiB,EAAE,CAAC;QACnC,CAAC;KAAA;CACD"}
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "hydra-crawler",
3
- "version": "2.8.3",
3
+ "version": "2.8.4",
4
4
  "description": "Node.js Hydra web crawler",
5
5
  "author": "Pete Morris",
6
6
  "license": "ISC",