@jocmp/mercury-parser 3.0.6 → 3.0.8
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/generate-custom-parser.js +189 -62
- package/dist/generate-custom-parser.js.map +1 -1
- package/dist/mercury.js +151 -37
- package/dist/mercury.js.map +1 -1
- package/dist/mercury.web.js +1 -1
- package/dist/mercury.web.js.map +1 -1
- package/package.json +6 -5
|
@@ -42,7 +42,6 @@ var require$$31 = require('dayjs/plugin/timezone');
|
|
|
42
42
|
var require$$32 = require('dayjs/plugin/customParseFormat');
|
|
43
43
|
var require$$33 = require('wuzzy');
|
|
44
44
|
var require$$34 = require('difflib');
|
|
45
|
-
var require$$35 = require('ellipsize');
|
|
46
45
|
var _taggedTemplateLiteral = require('@babel/runtime-corejs2/helpers/taggedTemplateLiteral');
|
|
47
46
|
|
|
48
47
|
function _interopDefaultLegacy (e) { return e && typeof e === 'object' && 'default' in e ? e : { 'default': e }; }
|
|
@@ -88,7 +87,6 @@ var require$$31__default = /*#__PURE__*/_interopDefaultLegacy(require$$31);
|
|
|
88
87
|
var require$$32__default = /*#__PURE__*/_interopDefaultLegacy(require$$32);
|
|
89
88
|
var require$$33__default = /*#__PURE__*/_interopDefaultLegacy(require$$33);
|
|
90
89
|
var require$$34__default = /*#__PURE__*/_interopDefaultLegacy(require$$34);
|
|
91
|
-
var require$$35__default = /*#__PURE__*/_interopDefaultLegacy(require$$35);
|
|
92
90
|
var _taggedTemplateLiteral__default = /*#__PURE__*/_interopDefaultLegacy(_taggedTemplateLiteral);
|
|
93
91
|
|
|
94
92
|
// Spacer images to be removed
|
|
@@ -222,7 +220,6 @@ function requireMercury() {
|
|
|
222
220
|
var customParseFormat = require$$32__default["default"];
|
|
223
221
|
var wuzzy = require$$33__default["default"];
|
|
224
222
|
var difflib = require$$34__default["default"];
|
|
225
|
-
var ellipsize = require$$35__default["default"];
|
|
226
223
|
function _interopDefaultLegacy(e) {
|
|
227
224
|
return e && _typeof__default["default"](e) === 'object' && 'default' in e ? e : {
|
|
228
225
|
'default': e
|
|
@@ -282,7 +279,6 @@ function requireMercury() {
|
|
|
282
279
|
var customParseFormat__default = /*#__PURE__*/_interopDefaultLegacy(customParseFormat);
|
|
283
280
|
var wuzzy__default = /*#__PURE__*/_interopDefaultLegacy(wuzzy);
|
|
284
281
|
var difflib__default = /*#__PURE__*/_interopDefaultLegacy(difflib);
|
|
285
|
-
var ellipsize__default = /*#__PURE__*/_interopDefaultLegacy(ellipsize);
|
|
286
282
|
var NORMALIZE_RE = /\s{2,}(?![^<>]*<\/(pre|code|textarea)>)/g;
|
|
287
283
|
function normalizeSpaces(text) {
|
|
288
284
|
return text.replace(NORMALIZE_RE, ' ').trim();
|
|
@@ -452,7 +448,7 @@ function requireMercury() {
|
|
|
452
448
|
// Use this setting as the maximum size an article can be
|
|
453
449
|
// for us to attempt parsing. Defaults to 5 MB.
|
|
454
450
|
var MAX_CONTENT_LENGTH = 5242880;
|
|
455
|
-
function ownKeys$
|
|
451
|
+
function ownKeys$h(e, r) {
|
|
456
452
|
var t = _Object$keys__default$1["default"](e);
|
|
457
453
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
458
454
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -462,12 +458,12 @@ function requireMercury() {
|
|
|
462
458
|
}
|
|
463
459
|
return t;
|
|
464
460
|
}
|
|
465
|
-
function _objectSpread$
|
|
461
|
+
function _objectSpread$h(e) {
|
|
466
462
|
for (var r = 1; r < arguments.length; r++) {
|
|
467
463
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
468
|
-
r % 2 ? ownKeys$
|
|
464
|
+
r % 2 ? ownKeys$h(Object(t), !0).forEach(function (r) {
|
|
469
465
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
470
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
466
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$h(Object(t)).forEach(function (r) {
|
|
471
467
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
472
468
|
});
|
|
473
469
|
}
|
|
@@ -547,9 +543,9 @@ function requireMercury() {
|
|
|
547
543
|
case 0:
|
|
548
544
|
headers = _args.length > 2 && _args[2] !== undefined ? _args[2] : {};
|
|
549
545
|
parsedUrl = parsedUrl || URL__default$1["default"].parse(encodeURI(url));
|
|
550
|
-
options = _objectSpread$
|
|
546
|
+
options = _objectSpread$h({
|
|
551
547
|
url: parsedUrl.href,
|
|
552
|
-
headers: _objectSpread$
|
|
548
|
+
headers: _objectSpread$h(_objectSpread$h({}, REQUEST_HEADERS), headers),
|
|
553
549
|
timeout: FETCH_TIMEOUT,
|
|
554
550
|
// Accept cookies
|
|
555
551
|
jar: true,
|
|
@@ -969,7 +965,7 @@ function requireMercury() {
|
|
|
969
965
|
}
|
|
970
966
|
return node;
|
|
971
967
|
}
|
|
972
|
-
function ownKeys$
|
|
968
|
+
function ownKeys$g(e, r) {
|
|
973
969
|
var t = _Object$keys__default$1["default"](e);
|
|
974
970
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
975
971
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -979,12 +975,12 @@ function requireMercury() {
|
|
|
979
975
|
}
|
|
980
976
|
return t;
|
|
981
977
|
}
|
|
982
|
-
function _objectSpread$
|
|
978
|
+
function _objectSpread$g(e) {
|
|
983
979
|
for (var r = 1; r < arguments.length; r++) {
|
|
984
980
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
985
|
-
r % 2 ? ownKeys$
|
|
981
|
+
r % 2 ? ownKeys$g(Object(t), !0).forEach(function (r) {
|
|
986
982
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
987
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
983
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$g(Object(t)).forEach(function (r) {
|
|
988
984
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
989
985
|
});
|
|
990
986
|
}
|
|
@@ -995,7 +991,7 @@ function requireMercury() {
|
|
|
995
991
|
var attrs = getAttrs(node);
|
|
996
992
|
setAttrs(node, _Reflect$ownKeys__default$1["default"](attrs).reduce(function (acc, attr) {
|
|
997
993
|
if (WHITELIST_ATTRS_RE.test(attr)) {
|
|
998
|
-
return _objectSpread$
|
|
994
|
+
return _objectSpread$g(_objectSpread$g({}, acc), {}, _defineProperty__default$1["default"]({}, attr, attrs[attr]));
|
|
999
995
|
}
|
|
1000
996
|
return acc;
|
|
1001
997
|
}, {}));
|
|
@@ -6469,7 +6465,7 @@ function requireMercury() {
|
|
|
6469
6465
|
clean: ['#timezone', '.author', '.article__author__croppimg', '.article_authors_with_thumbnail', '.related_articles__elements', '.gl_plugin.socials', '.gl_plugin.player', '.gl_plugin.video_player', '.gl_plugin + video']
|
|
6470
6466
|
}
|
|
6471
6467
|
};
|
|
6472
|
-
function ownKeys$
|
|
6468
|
+
function ownKeys$f(e, r) {
|
|
6473
6469
|
var t = _Object$keys__default$1["default"](e);
|
|
6474
6470
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6475
6471
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6479,21 +6475,21 @@ function requireMercury() {
|
|
|
6479
6475
|
}
|
|
6480
6476
|
return t;
|
|
6481
6477
|
}
|
|
6482
|
-
function _objectSpread$
|
|
6478
|
+
function _objectSpread$f(e) {
|
|
6483
6479
|
for (var r = 1; r < arguments.length; r++) {
|
|
6484
6480
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6485
|
-
r % 2 ? ownKeys$
|
|
6481
|
+
r % 2 ? ownKeys$f(Object(t), !0).forEach(function (r) {
|
|
6486
6482
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6487
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6483
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$f(Object(t)).forEach(function (r) {
|
|
6488
6484
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6489
6485
|
});
|
|
6490
6486
|
}
|
|
6491
6487
|
return e;
|
|
6492
6488
|
}
|
|
6493
|
-
var SportSePlExtractor = _objectSpread$
|
|
6489
|
+
var SportSePlExtractor = _objectSpread$f(_objectSpread$f({}, WwwSePlExtractor), {}, {
|
|
6494
6490
|
domain: 'sport.se.pl'
|
|
6495
6491
|
});
|
|
6496
|
-
function ownKeys$
|
|
6492
|
+
function ownKeys$e(e, r) {
|
|
6497
6493
|
var t = _Object$keys__default$1["default"](e);
|
|
6498
6494
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6499
6495
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6503,18 +6499,18 @@ function requireMercury() {
|
|
|
6503
6499
|
}
|
|
6504
6500
|
return t;
|
|
6505
6501
|
}
|
|
6506
|
-
function _objectSpread$
|
|
6502
|
+
function _objectSpread$e(e) {
|
|
6507
6503
|
for (var r = 1; r < arguments.length; r++) {
|
|
6508
6504
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6509
|
-
r % 2 ? ownKeys$
|
|
6505
|
+
r % 2 ? ownKeys$e(Object(t), !0).forEach(function (r) {
|
|
6510
6506
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6511
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6507
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$e(Object(t)).forEach(function (r) {
|
|
6512
6508
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6513
6509
|
});
|
|
6514
6510
|
}
|
|
6515
6511
|
return e;
|
|
6516
6512
|
}
|
|
6517
|
-
var PolitykaSePlExtractor = _objectSpread$
|
|
6513
|
+
var PolitykaSePlExtractor = _objectSpread$e(_objectSpread$e({}, WwwSePlExtractor), {}, {
|
|
6518
6514
|
domain: 'polityka.se.pl'
|
|
6519
6515
|
});
|
|
6520
6516
|
var SuperserialeSePlExtractor = {
|
|
@@ -6544,7 +6540,7 @@ function requireMercury() {
|
|
|
6544
6540
|
'.related_articles__elements', '.gl_plugin.socials', '.gl_plugin.player', '.gl_plugin.video_player', '.gl_plugin + video']
|
|
6545
6541
|
}
|
|
6546
6542
|
};
|
|
6547
|
-
function ownKeys$
|
|
6543
|
+
function ownKeys$d(e, r) {
|
|
6548
6544
|
var t = _Object$keys__default$1["default"](e);
|
|
6549
6545
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6550
6546
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6554,21 +6550,21 @@ function requireMercury() {
|
|
|
6554
6550
|
}
|
|
6555
6551
|
return t;
|
|
6556
6552
|
}
|
|
6557
|
-
function _objectSpread$
|
|
6553
|
+
function _objectSpread$d(e) {
|
|
6558
6554
|
for (var r = 1; r < arguments.length; r++) {
|
|
6559
6555
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6560
|
-
r % 2 ? ownKeys$
|
|
6556
|
+
r % 2 ? ownKeys$d(Object(t), !0).forEach(function (r) {
|
|
6561
6557
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6562
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6558
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$d(Object(t)).forEach(function (r) {
|
|
6563
6559
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6564
6560
|
});
|
|
6565
6561
|
}
|
|
6566
6562
|
return e;
|
|
6567
6563
|
}
|
|
6568
|
-
var SzczecinSePlExtractor = _objectSpread$
|
|
6564
|
+
var SzczecinSePlExtractor = _objectSpread$d(_objectSpread$d({}, WwwSePlExtractor), {}, {
|
|
6569
6565
|
domain: 'szczecin.se.pl'
|
|
6570
6566
|
});
|
|
6571
|
-
function ownKeys$
|
|
6567
|
+
function ownKeys$c(e, r) {
|
|
6572
6568
|
var t = _Object$keys__default$1["default"](e);
|
|
6573
6569
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6574
6570
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6578,21 +6574,21 @@ function requireMercury() {
|
|
|
6578
6574
|
}
|
|
6579
6575
|
return t;
|
|
6580
6576
|
}
|
|
6581
|
-
function _objectSpread$
|
|
6577
|
+
function _objectSpread$c(e) {
|
|
6582
6578
|
for (var r = 1; r < arguments.length; r++) {
|
|
6583
6579
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6584
|
-
r % 2 ? ownKeys$
|
|
6580
|
+
r % 2 ? ownKeys$c(Object(t), !0).forEach(function (r) {
|
|
6585
6581
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6586
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6582
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$c(Object(t)).forEach(function (r) {
|
|
6587
6583
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6588
6584
|
});
|
|
6589
6585
|
}
|
|
6590
6586
|
return e;
|
|
6591
6587
|
}
|
|
6592
|
-
var SuperbizSePlExtractor = _objectSpread$
|
|
6588
|
+
var SuperbizSePlExtractor = _objectSpread$c(_objectSpread$c({}, WwwSePlExtractor), {}, {
|
|
6593
6589
|
domain: 'superbiz.se.pl'
|
|
6594
6590
|
});
|
|
6595
|
-
function ownKeys$
|
|
6591
|
+
function ownKeys$b(e, r) {
|
|
6596
6592
|
var t = _Object$keys__default$1["default"](e);
|
|
6597
6593
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6598
6594
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6602,18 +6598,18 @@ function requireMercury() {
|
|
|
6602
6598
|
}
|
|
6603
6599
|
return t;
|
|
6604
6600
|
}
|
|
6605
|
-
function _objectSpread$
|
|
6601
|
+
function _objectSpread$b(e) {
|
|
6606
6602
|
for (var r = 1; r < arguments.length; r++) {
|
|
6607
6603
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6608
|
-
r % 2 ? ownKeys$
|
|
6604
|
+
r % 2 ? ownKeys$b(Object(t), !0).forEach(function (r) {
|
|
6609
6605
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6610
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6606
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$b(Object(t)).forEach(function (r) {
|
|
6611
6607
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6612
6608
|
});
|
|
6613
6609
|
}
|
|
6614
6610
|
return e;
|
|
6615
6611
|
}
|
|
6616
|
-
var PortalobronnySePlExtractor = _objectSpread$
|
|
6612
|
+
var PortalobronnySePlExtractor = _objectSpread$b(_objectSpread$b({}, WwwSePlExtractor), {}, {
|
|
6617
6613
|
domain: 'portalobronny.se.pl'
|
|
6618
6614
|
});
|
|
6619
6615
|
var PolskisamorzadSePlExtractor = {
|
|
@@ -6637,7 +6633,7 @@ function requireMercury() {
|
|
|
6637
6633
|
clean: ['#timezone', '.author', '.article__author__croppimg', '.article_authors_with_thumbnail', '.related_articles__elements', '.gl_plugin.socials', '.gl_plugin.player', '.gl_plugin.video_player', '.gl_plugin + video']
|
|
6638
6634
|
}
|
|
6639
6635
|
};
|
|
6640
|
-
function ownKeys$
|
|
6636
|
+
function ownKeys$a(e, r) {
|
|
6641
6637
|
var t = _Object$keys__default$1["default"](e);
|
|
6642
6638
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6643
6639
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6647,21 +6643,21 @@ function requireMercury() {
|
|
|
6647
6643
|
}
|
|
6648
6644
|
return t;
|
|
6649
6645
|
}
|
|
6650
|
-
function _objectSpread$
|
|
6646
|
+
function _objectSpread$a(e) {
|
|
6651
6647
|
for (var r = 1; r < arguments.length; r++) {
|
|
6652
6648
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6653
|
-
r % 2 ? ownKeys$
|
|
6649
|
+
r % 2 ? ownKeys$a(Object(t), !0).forEach(function (r) {
|
|
6654
6650
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6655
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6651
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$a(Object(t)).forEach(function (r) {
|
|
6656
6652
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6657
6653
|
});
|
|
6658
6654
|
}
|
|
6659
6655
|
return e;
|
|
6660
6656
|
}
|
|
6661
|
-
var LodzSePlExtractor = _objectSpread$
|
|
6657
|
+
var LodzSePlExtractor = _objectSpread$a(_objectSpread$a({}, WwwSePlExtractor), {}, {
|
|
6662
6658
|
domain: 'lodz.se.pl'
|
|
6663
6659
|
});
|
|
6664
|
-
function ownKeys$
|
|
6660
|
+
function ownKeys$9(e, r) {
|
|
6665
6661
|
var t = _Object$keys__default$1["default"](e);
|
|
6666
6662
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6667
6663
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6671,21 +6667,21 @@ function requireMercury() {
|
|
|
6671
6667
|
}
|
|
6672
6668
|
return t;
|
|
6673
6669
|
}
|
|
6674
|
-
function _objectSpread$
|
|
6670
|
+
function _objectSpread$9(e) {
|
|
6675
6671
|
for (var r = 1; r < arguments.length; r++) {
|
|
6676
6672
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6677
|
-
r % 2 ? ownKeys$
|
|
6673
|
+
r % 2 ? ownKeys$9(Object(t), !0).forEach(function (r) {
|
|
6678
6674
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6679
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6675
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$9(Object(t)).forEach(function (r) {
|
|
6680
6676
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6681
6677
|
});
|
|
6682
6678
|
}
|
|
6683
6679
|
return e;
|
|
6684
6680
|
}
|
|
6685
|
-
var WroclawSePlExtractor = _objectSpread$
|
|
6681
|
+
var WroclawSePlExtractor = _objectSpread$9(_objectSpread$9({}, WwwSePlExtractor), {}, {
|
|
6686
6682
|
domain: 'wroclaw.se.pl'
|
|
6687
6683
|
});
|
|
6688
|
-
function ownKeys$
|
|
6684
|
+
function ownKeys$8(e, r) {
|
|
6689
6685
|
var t = _Object$keys__default$1["default"](e);
|
|
6690
6686
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6691
6687
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6695,21 +6691,21 @@ function requireMercury() {
|
|
|
6695
6691
|
}
|
|
6696
6692
|
return t;
|
|
6697
6693
|
}
|
|
6698
|
-
function _objectSpread$
|
|
6694
|
+
function _objectSpread$8(e) {
|
|
6699
6695
|
for (var r = 1; r < arguments.length; r++) {
|
|
6700
6696
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6701
|
-
r % 2 ? ownKeys$
|
|
6697
|
+
r % 2 ? ownKeys$8(Object(t), !0).forEach(function (r) {
|
|
6702
6698
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6703
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6699
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$8(Object(t)).forEach(function (r) {
|
|
6704
6700
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6705
6701
|
});
|
|
6706
6702
|
}
|
|
6707
6703
|
return e;
|
|
6708
6704
|
}
|
|
6709
|
-
var LublinSePlExtractor = _objectSpread$
|
|
6705
|
+
var LublinSePlExtractor = _objectSpread$8(_objectSpread$8({}, WwwSePlExtractor), {}, {
|
|
6710
6706
|
domain: 'lublin.se.pl'
|
|
6711
6707
|
});
|
|
6712
|
-
function ownKeys$
|
|
6708
|
+
function ownKeys$7(e, r) {
|
|
6713
6709
|
var t = _Object$keys__default$1["default"](e);
|
|
6714
6710
|
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
6715
6711
|
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
@@ -6719,18 +6715,18 @@ function requireMercury() {
|
|
|
6719
6715
|
}
|
|
6720
6716
|
return t;
|
|
6721
6717
|
}
|
|
6722
|
-
function _objectSpread$
|
|
6718
|
+
function _objectSpread$7(e) {
|
|
6723
6719
|
for (var r = 1; r < arguments.length; r++) {
|
|
6724
6720
|
var t = null != arguments[r] ? arguments[r] : {};
|
|
6725
|
-
r % 2 ? ownKeys$
|
|
6721
|
+
r % 2 ? ownKeys$7(Object(t), !0).forEach(function (r) {
|
|
6726
6722
|
_defineProperty__default$1["default"](e, r, t[r]);
|
|
6727
|
-
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$
|
|
6723
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$7(Object(t)).forEach(function (r) {
|
|
6728
6724
|
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
6729
6725
|
});
|
|
6730
6726
|
}
|
|
6731
6727
|
return e;
|
|
6732
6728
|
}
|
|
6733
|
-
var BialystokSePlExtractor = _objectSpread$
|
|
6729
|
+
var BialystokSePlExtractor = _objectSpread$7(_objectSpread$7({}, WwwSePlExtractor), {}, {
|
|
6734
6730
|
domain: 'bialystok.se.pl'
|
|
6735
6731
|
});
|
|
6736
6732
|
var WwwLebensmittelwarnungDeExtractor = {
|
|
@@ -7498,6 +7494,84 @@ function requireMercury() {
|
|
|
7498
7494
|
clean: ['.section-header', '.article__footer', '.social-button-container', '.gallery__button', '.gallery__position-label', '.detail-img__caption-toggle', '.nativendo-mid-article', '.taboola-mid-article', 'article > p']
|
|
7499
7495
|
}
|
|
7500
7496
|
};
|
|
7497
|
+
var WwwEuronewsComExtractor = {
|
|
7498
|
+
domain: 'www.euronews.com',
|
|
7499
|
+
title: {
|
|
7500
|
+
selectors: [['meta[name="og:title"]', 'value']]
|
|
7501
|
+
},
|
|
7502
|
+
author: {
|
|
7503
|
+
selectors: [['meta[name="article:author"]', 'value'], '.c-article-contributors']
|
|
7504
|
+
},
|
|
7505
|
+
date_published: {
|
|
7506
|
+
selectors: [['meta[name="article:published_time"]', 'value']]
|
|
7507
|
+
},
|
|
7508
|
+
dek: {
|
|
7509
|
+
selectors: ['h2.c-article-summary']
|
|
7510
|
+
},
|
|
7511
|
+
lead_image_url: {
|
|
7512
|
+
selectors: [['meta[name="og:image"]', 'value']]
|
|
7513
|
+
},
|
|
7514
|
+
content: {
|
|
7515
|
+
selectors: ['.c-article-content', 'article'],
|
|
7516
|
+
transforms: {
|
|
7517
|
+
h2: function h2(node) {
|
|
7518
|
+
return node.attr('class', 'mercury-parser-keep');
|
|
7519
|
+
},
|
|
7520
|
+
'.widget__figure': function widget__figure(node) {
|
|
7521
|
+
return node.addClass('mercury-parser-keep');
|
|
7522
|
+
}
|
|
7523
|
+
},
|
|
7524
|
+
clean: ['.c-ad', '.c-widget-related', '.connatix-container']
|
|
7525
|
+
}
|
|
7526
|
+
};
|
|
7527
|
+
function ownKeys$6(e, r) {
|
|
7528
|
+
var t = _Object$keys__default$1["default"](e);
|
|
7529
|
+
if (_Object$getOwnPropertySymbols__default$1["default"]) {
|
|
7530
|
+
var o = _Object$getOwnPropertySymbols__default$1["default"](e);
|
|
7531
|
+
r && (o = o.filter(function (r) {
|
|
7532
|
+
return _Object$getOwnPropertyDescriptor__default$1["default"](e, r).enumerable;
|
|
7533
|
+
})), t.push.apply(t, o);
|
|
7534
|
+
}
|
|
7535
|
+
return t;
|
|
7536
|
+
}
|
|
7537
|
+
function _objectSpread$6(e) {
|
|
7538
|
+
for (var r = 1; r < arguments.length; r++) {
|
|
7539
|
+
var t = null != arguments[r] ? arguments[r] : {};
|
|
7540
|
+
r % 2 ? ownKeys$6(Object(t), !0).forEach(function (r) {
|
|
7541
|
+
_defineProperty__default$1["default"](e, r, t[r]);
|
|
7542
|
+
}) : _Object$getOwnPropertyDescriptors__default$1["default"] ? _Object$defineProperties__default$1["default"](e, _Object$getOwnPropertyDescriptors__default$1["default"](t)) : ownKeys$6(Object(t)).forEach(function (r) {
|
|
7543
|
+
_Object$defineProperty__default$1["default"](e, r, _Object$getOwnPropertyDescriptor__default$1["default"](t, r));
|
|
7544
|
+
});
|
|
7545
|
+
}
|
|
7546
|
+
return e;
|
|
7547
|
+
}
|
|
7548
|
+
var GrEuronewsComExtractor = _objectSpread$6(_objectSpread$6({}, WwwEuronewsComExtractor), {}, {
|
|
7549
|
+
domain: 'gr.euronews.com'
|
|
7550
|
+
});
|
|
7551
|
+
var WwwIlfattoquotidianoItExtractor = {
|
|
7552
|
+
domain: 'www.ilfattoquotidiano.it',
|
|
7553
|
+
title: {
|
|
7554
|
+
selectors: [['meta[name="og:title"]', 'value']]
|
|
7555
|
+
},
|
|
7556
|
+
author: {
|
|
7557
|
+
selectors: ['.ifq-post__author .ifq-news-meta__author-name'],
|
|
7558
|
+
clean: ['span']
|
|
7559
|
+
},
|
|
7560
|
+
date_published: {
|
|
7561
|
+
selectors: [['meta[name="article:published_time"]', 'value']]
|
|
7562
|
+
},
|
|
7563
|
+
dek: {
|
|
7564
|
+
selectors: [['meta[name="og:description"]', 'value']]
|
|
7565
|
+
},
|
|
7566
|
+
lead_image_url: {
|
|
7567
|
+
selectors: [['meta[name="og:image"]', 'value']]
|
|
7568
|
+
},
|
|
7569
|
+
content: {
|
|
7570
|
+
selectors: ['.ifq-post__content', 'article'],
|
|
7571
|
+
transforms: {},
|
|
7572
|
+
clean: []
|
|
7573
|
+
}
|
|
7574
|
+
};
|
|
7501
7575
|
var CustomExtractors = /*#__PURE__*/_Object$freeze__default["default"]({
|
|
7502
7576
|
__proto__: null,
|
|
7503
7577
|
BalloonJuiceComExtractor: BalloonJuiceComExtractor,
|
|
@@ -7694,7 +7768,10 @@ function requireMercury() {
|
|
|
7694
7768
|
WwwJalopnikComExtractor: WwwJalopnikComExtractor,
|
|
7695
7769
|
Nineto5linuxComExtractor: Nineto5linuxComExtractor,
|
|
7696
7770
|
WwwTransfermarktDeExtractor: WwwTransfermarktDeExtractor,
|
|
7697
|
-
WwwBlickDeExtractor: WwwBlickDeExtractor
|
|
7771
|
+
WwwBlickDeExtractor: WwwBlickDeExtractor,
|
|
7772
|
+
WwwEuronewsComExtractor: WwwEuronewsComExtractor,
|
|
7773
|
+
GrEuronewsComExtractor: GrEuronewsComExtractor,
|
|
7774
|
+
WwwIlfattoquotidianoItExtractor: WwwIlfattoquotidianoItExtractor
|
|
7698
7775
|
});
|
|
7699
7776
|
function ownKeys$5(e, r) {
|
|
7700
7777
|
var t = _Object$keys__default$1["default"](e);
|
|
@@ -9221,11 +9298,61 @@ function requireMercury() {
|
|
|
9221
9298
|
return result(url);
|
|
9222
9299
|
}
|
|
9223
9300
|
};
|
|
9301
|
+
var defaults = {
|
|
9302
|
+
ellipse: '…',
|
|
9303
|
+
chars: [' ', '-'],
|
|
9304
|
+
max: 140,
|
|
9305
|
+
truncate: true
|
|
9306
|
+
};
|
|
9307
|
+
function ellipsizeMiddle(str, max, ellipse, chars) {
|
|
9308
|
+
if (str.length <= max) return str;
|
|
9309
|
+
if (max < 2) return str.slice(0, max - ellipse.length) + ellipse;
|
|
9310
|
+
var maxLen = max - ellipse.length;
|
|
9311
|
+
var middle = Math.floor(maxLen / 2);
|
|
9312
|
+
var left = middle;
|
|
9313
|
+
var right = str.length - middle;
|
|
9314
|
+
for (var i = 0; i < middle; i += 1) {
|
|
9315
|
+
var charLeft = str.charAt(i);
|
|
9316
|
+
var posRight = str.length - i;
|
|
9317
|
+
var charRight = str.charAt(posRight);
|
|
9318
|
+
if (chars.indexOf(charLeft) !== -1) left = i;
|
|
9319
|
+
if (chars.indexOf(charRight) !== -1) right = posRight;
|
|
9320
|
+
}
|
|
9321
|
+
return str.slice(0, left) + ellipse + str.slice(right);
|
|
9322
|
+
}
|
|
9323
|
+
function ellipsize(str, max, ellipse, chars, truncate) {
|
|
9324
|
+
if (str.length <= max) return str;
|
|
9325
|
+
var maxLen = max - ellipse.length;
|
|
9326
|
+
var end = maxLen;
|
|
9327
|
+
var breakpointFound = false;
|
|
9328
|
+
for (var i = 0; i <= maxLen; i += 1) {
|
|
9329
|
+
var _char = str.charAt(i);
|
|
9330
|
+
if (chars.indexOf(_char) !== -1) {
|
|
9331
|
+
end = i;
|
|
9332
|
+
breakpointFound = true;
|
|
9333
|
+
}
|
|
9334
|
+
}
|
|
9335
|
+
if (!truncate && !breakpointFound) return '';
|
|
9336
|
+
return str.slice(0, end) + ellipse;
|
|
9337
|
+
}
|
|
9338
|
+
var ellipsize$1 = function ellipsize$1(str, max, opts) {
|
|
9339
|
+
if (typeof str !== 'string' || str.length === 0) return '';
|
|
9340
|
+
if (max === 0) return '';
|
|
9341
|
+
opts = opts || {};
|
|
9342
|
+
_Object$keys__default$1["default"](defaults).forEach(function (key) {
|
|
9343
|
+
if (opts[key] === null || typeof opts[key] === 'undefined') {
|
|
9344
|
+
opts[key] = defaults[key];
|
|
9345
|
+
}
|
|
9346
|
+
});
|
|
9347
|
+
opts.max = max || opts.max;
|
|
9348
|
+
if (opts.truncate === 'middle') return ellipsizeMiddle(str, opts.max, opts.ellipse, opts.chars);
|
|
9349
|
+
return ellipsize(str, opts.max, opts.ellipse, opts.chars, opts.truncate);
|
|
9350
|
+
};
|
|
9224
9351
|
var EXCERPT_META_SELECTORS = ['og:description', 'twitter:description'];
|
|
9225
9352
|
function clean(content, $) {
|
|
9226
9353
|
var maxLength = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : 200;
|
|
9227
9354
|
content = content.replace(/[\s\n]+/g, ' ').trim();
|
|
9228
|
-
return
|
|
9355
|
+
return ellipsize$1(content, maxLength, {
|
|
9229
9356
|
ellipse: '…'
|
|
9230
9357
|
});
|
|
9231
9358
|
}
|