sec-edgar-api 0.3.4 → 0.4.1
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/build/services/DocumentParser/DocumentParser.d.ts +17 -1
- package/build/services/DocumentParser/DocumentParser.js +17 -2
- package/build/services/DocumentParser/DocumentXmlSplitter.d.ts +17 -0
- package/build/services/DocumentParser/DocumentXmlSplitter.js +38 -0
- package/build/services/DocumentParser/ObjectUtil.d.ts +20 -0
- package/build/services/DocumentParser/ObjectUtil.js +60 -0
- package/build/services/DocumentParser/XBRLParser/HeaderParser.d.ts +38 -0
- package/build/services/DocumentParser/XBRLParser/HeaderParser.js +138 -0
- package/build/services/DocumentParser/XBRLParser/InstanceParser.d.ts +6 -0
- package/build/services/DocumentParser/XBRLParser/InstanceParser.js +106 -0
- package/build/services/DocumentParser/XBRLParser/LinkbaseParser.d.ts +13 -0
- package/build/services/DocumentParser/XBRLParser/LinkbaseParser.js +98 -0
- package/build/services/DocumentParser/XBRLParser/SchemaParser.d.ts +11 -0
- package/build/services/DocumentParser/XBRLParser/SchemaParser.js +142 -0
- package/build/services/DocumentParser/XBRLParser/XBRLParser.d.ts +34 -0
- package/build/services/DocumentParser/XBRLParser/XBRLParser.js +68 -0
- package/build/services/DocumentParser/XBRLParser/util-xbrl.d.ts +17 -0
- package/build/services/DocumentParser/XBRLParser/util-xbrl.js +50 -0
- package/build/services/DocumentParser/XMLParser.d.ts +9 -38
- package/build/services/DocumentParser/XMLParser.js +87 -280
- package/build/services/DocumentParser/XMLParserLegacy.d.ts +43 -0
- package/build/services/DocumentParser/XMLParserLegacy.js +308 -0
- package/build/services/DocumentParser/parsers/index.d.ts +10 -0
- package/build/services/DocumentParser/parsers/index.js +10 -0
- package/build/services/DocumentParser/parsers/parse-companies.d.ts +5 -0
- package/build/services/DocumentParser/parsers/parse-companies.js +52 -0
- package/build/services/DocumentParser/parsers/parse-current-filings-xbrl.d.ts +2 -0
- package/build/services/DocumentParser/parsers/parse-current-filings-xbrl.js +104 -0
- package/build/services/DocumentParser/parsers/parse-current-filings.d.ts +2 -0
- package/build/services/DocumentParser/parsers/parse-current-filings.js +76 -0
- package/build/services/DocumentParser/parsers/parse-form-10k.d.ts +1 -1
- package/build/services/DocumentParser/parsers/parse-form-10k.js +2 -2
- package/build/services/DocumentParser/parsers/parse-form-13g.d.ts +1 -1
- package/build/services/DocumentParser/parsers/parse-form-13g.js +2 -2
- package/build/services/DocumentParser/parsers/parse-form-4.d.ts +1 -1
- package/build/services/DocumentParser/parsers/parse-form-4.js +2 -2
- package/build/services/DocumentParser/parsers/parse-form-def14a.d.ts +1 -1
- package/build/services/DocumentParser/parsers/parse-form-def14a.js +2 -2
- package/build/services/DocumentParser/parsers/parse-insider-transaction.d.ts +7 -0
- package/build/services/DocumentParser/parsers/parse-insider-transaction.js +199 -0
- package/build/services/DocumentParser/parsers/parse-xbrl.d.ts +8 -0
- package/build/services/DocumentParser/parsers/parse-xbrl.js +179 -0
- package/build/services/HtmlTableExtractor/HtmlTableExtractor.d.ts +45 -0
- package/build/services/HtmlTableExtractor/HtmlTableExtractor.js +430 -0
- package/build/services/HtmlTableExtractor/index.d.ts +2 -0
- package/build/services/HtmlTableExtractor/index.js +4 -0
- package/build/services/ReportRawBuilder/FactGrouper.js +6 -6
- package/build/services/ReportRawBuilder/FactSplitAdjuster.js +2 -1
- package/build/services/SecEdgarApi/SecEdgarApi.d.ts +82 -15
- package/build/services/SecEdgarApi/SecEdgarApi.js +170 -21
- package/build/types/company-search.type.d.ts +5 -0
- package/build/types/company-search.type.js +2 -0
- package/build/types/current-filings.type.d.ts +57 -0
- package/build/types/current-filings.type.js +2 -0
- package/build/types/index.d.ts +3 -0
- package/build/types/index.js +3 -0
- package/build/types/insider-transaction.type.d.ts +32 -0
- package/build/types/insider-transaction.type.js +2 -0
- package/build/types/parsed-filings.type.d.ts +3 -3
- package/build/types/xbrl.type.d.ts +155 -0
- package/build/types/xbrl.type.js +2 -0
- package/build/util/constants.d.ts +2 -0
- package/build/util/constants.js +5 -0
- package/package.json +1 -1
|
@@ -0,0 +1,430 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
var HtmlTableExtractor = /** @class */ (function () {
|
|
4
|
+
function HtmlTableExtractor() {
|
|
5
|
+
}
|
|
6
|
+
HtmlTableExtractor.prototype.extractTables = function (html, options) {
|
|
7
|
+
var _a, _b;
|
|
8
|
+
var tablesOpen = [];
|
|
9
|
+
var tableIndex = -1;
|
|
10
|
+
var htmlBefore = '';
|
|
11
|
+
var tablesData = [];
|
|
12
|
+
for (var i = 0; i < html.length; i++) {
|
|
13
|
+
var isTableStart = html.substring(i, i + 6).toLowerCase() === '<table';
|
|
14
|
+
var isTableEnd = html.substring(i - 7, i + 1).toLowerCase() === '</table>';
|
|
15
|
+
var parentTable = tablesOpen[tablesOpen.length - 1];
|
|
16
|
+
if (isTableStart) {
|
|
17
|
+
tableIndex++;
|
|
18
|
+
tablesOpen.push({
|
|
19
|
+
tableIndex: tableIndex,
|
|
20
|
+
parentTableIndex: (_a = parentTable === null || parentTable === void 0 ? void 0 : parentTable.tableIndex) !== null && _a !== void 0 ? _a : null,
|
|
21
|
+
childTableIndexes: [],
|
|
22
|
+
positionStart: i,
|
|
23
|
+
positionEnd: -1,
|
|
24
|
+
htmlBefore: htmlBefore,
|
|
25
|
+
html: '',
|
|
26
|
+
rows: [],
|
|
27
|
+
});
|
|
28
|
+
parentTable === null || parentTable === void 0 ? void 0 : parentTable.childTableIndexes.push(tableIndex);
|
|
29
|
+
htmlBefore = '';
|
|
30
|
+
}
|
|
31
|
+
if (tablesOpen.length === 0) {
|
|
32
|
+
htmlBefore += html[i];
|
|
33
|
+
}
|
|
34
|
+
for (var a = tablesOpen.length - 1; a >= 0; a--) {
|
|
35
|
+
tablesOpen[a].html += html[i];
|
|
36
|
+
}
|
|
37
|
+
if (isTableEnd && tablesOpen.length > 0) {
|
|
38
|
+
tablesOpen[tablesOpen.length - 1].positionEnd = i;
|
|
39
|
+
var tableData = tablesOpen.pop();
|
|
40
|
+
// tablesData[tableData.tableIndex] = tableData
|
|
41
|
+
tablesData.push(tableData);
|
|
42
|
+
}
|
|
43
|
+
}
|
|
44
|
+
this.addTableCells(tablesData);
|
|
45
|
+
this.addTableCellValues(tablesData, options);
|
|
46
|
+
this.addMissingNameCol(tablesData);
|
|
47
|
+
if ((_b = options === null || options === void 0 ? void 0 : options.removeEmptyColumns) !== null && _b !== void 0 ? _b : true) {
|
|
48
|
+
this.removeEmptyColumns(tablesData);
|
|
49
|
+
}
|
|
50
|
+
this.mergeHeaderRows(tablesData);
|
|
51
|
+
return tablesData.filter(Boolean);
|
|
52
|
+
};
|
|
53
|
+
HtmlTableExtractor.prototype.mergeHeaderRows = function (tables) {
|
|
54
|
+
var _a;
|
|
55
|
+
for (var _i = 0, tables_1 = tables; _i < tables_1.length; _i++) {
|
|
56
|
+
var table = tables_1[_i];
|
|
57
|
+
var bodyRowIndex = table.rows.findIndex(function (row) { return row.some(function (col) { return !col.isHeaderRowCell; }); });
|
|
58
|
+
var headerRowIndex = bodyRowIndex - 1;
|
|
59
|
+
var bodyRow = table.rows[bodyRowIndex];
|
|
60
|
+
var headerRow = table.rows[headerRowIndex];
|
|
61
|
+
if (!bodyRow || headerRowIndex < 0)
|
|
62
|
+
continue;
|
|
63
|
+
for (var i = 0; i < bodyRow.length; i++) {
|
|
64
|
+
var headerCol = headerRow[i];
|
|
65
|
+
var bodyCol = bodyRow[i];
|
|
66
|
+
if (!headerCol || !bodyCol)
|
|
67
|
+
continue;
|
|
68
|
+
headerCol.valueParsed = (_a = bodyCol.headerCol) !== null && _a !== void 0 ? _a : headerCol === null || headerCol === void 0 ? void 0 : headerCol.valueParsed;
|
|
69
|
+
}
|
|
70
|
+
}
|
|
71
|
+
};
|
|
72
|
+
HtmlTableExtractor.prototype.removeEmptyColumns = function (tables) {
|
|
73
|
+
var _a;
|
|
74
|
+
var _loop_1 = function (table) {
|
|
75
|
+
var emptyColumns = new Set();
|
|
76
|
+
var _loop_2 = function (c) {
|
|
77
|
+
var isAllEmpty = table.rows.every(function (row) { var _a, _b; return ((_a = row[c]) === null || _a === void 0 ? void 0 : _a.valueParsed) === null || ((_b = row[c]) === null || _b === void 0 ? void 0 : _b.isHeaderRowCell); });
|
|
78
|
+
if (isAllEmpty) {
|
|
79
|
+
emptyColumns.add(c);
|
|
80
|
+
}
|
|
81
|
+
};
|
|
82
|
+
for (var c = 0; c < ((_a = table.rows[0]) === null || _a === void 0 ? void 0 : _a.length); c++) {
|
|
83
|
+
_loop_2(c);
|
|
84
|
+
}
|
|
85
|
+
for (var r = 0; r < table.rows.length; r++) {
|
|
86
|
+
var row = table.rows[r];
|
|
87
|
+
table.rows[r] = row.filter(function (_, i) { return !emptyColumns.has(i); });
|
|
88
|
+
}
|
|
89
|
+
table.rows = table.rows.filter(function (row) { return row.some(function (col) { return col.valueParsed !== null; }); });
|
|
90
|
+
};
|
|
91
|
+
for (var _i = 0, tables_2 = tables; _i < tables_2.length; _i++) {
|
|
92
|
+
var table = tables_2[_i];
|
|
93
|
+
_loop_1(table);
|
|
94
|
+
}
|
|
95
|
+
return tables;
|
|
96
|
+
};
|
|
97
|
+
HtmlTableExtractor.prototype.addTableCells = function (tables) {
|
|
98
|
+
if (tables.length === 0)
|
|
99
|
+
return;
|
|
100
|
+
var tablesByIndex = new Map(tables.map(function (t) { return [t.tableIndex, t]; }));
|
|
101
|
+
tablesByIndex.forEach(function (table) {
|
|
102
|
+
var _a, _b;
|
|
103
|
+
var skipIndexMap = new Map(table.childTableIndexes.map(function (childIndex) {
|
|
104
|
+
var _a, _b;
|
|
105
|
+
var child = tablesByIndex.get(childIndex);
|
|
106
|
+
return [
|
|
107
|
+
((_a = child === null || child === void 0 ? void 0 : child.positionStart) !== null && _a !== void 0 ? _a : 0) - table.positionStart,
|
|
108
|
+
((_b = child === null || child === void 0 ? void 0 : child.positionEnd) !== null && _b !== void 0 ? _b : 0) - table.positionStart,
|
|
109
|
+
];
|
|
110
|
+
}));
|
|
111
|
+
var grid = [];
|
|
112
|
+
var isInCell = false;
|
|
113
|
+
var isInCellAtts = false;
|
|
114
|
+
var cellAtts = '';
|
|
115
|
+
var cellHTML = '';
|
|
116
|
+
var rowIndex = -1;
|
|
117
|
+
var tableCellIndex = -1;
|
|
118
|
+
var createCell = function (html, atts) {
|
|
119
|
+
var _a, _b, _c, _d, _e, _f;
|
|
120
|
+
var attributePairs = atts
|
|
121
|
+
.toLowerCase()
|
|
122
|
+
.split(' ')
|
|
123
|
+
.map(function (att) { return att.split('='); });
|
|
124
|
+
var rowSpan = Number((_b = (_a = attributePairs.find(function (_a) {
|
|
125
|
+
var key = _a[0];
|
|
126
|
+
return key === 'rowspan';
|
|
127
|
+
})) === null || _a === void 0 ? void 0 : _a[1]) === null || _b === void 0 ? void 0 : _b.replace(/[^0-9]/g, '')) || 1;
|
|
128
|
+
var colSpan = Number((_d = (_c = attributePairs.find(function (_a) {
|
|
129
|
+
var key = _a[0];
|
|
130
|
+
return key === 'colspan';
|
|
131
|
+
})) === null || _c === void 0 ? void 0 : _c[1]) === null || _d === void 0 ? void 0 : _d.replace(/[^0-9]/g, '')) || 1;
|
|
132
|
+
var cell = {
|
|
133
|
+
attributes: atts.length > 4 ? atts.substring(4, atts.length - 1) : '',
|
|
134
|
+
html: html,
|
|
135
|
+
colSpan: colSpan,
|
|
136
|
+
rowSpan: rowSpan,
|
|
137
|
+
tableCellIndex: tableCellIndex,
|
|
138
|
+
rowIndex: rowIndex,
|
|
139
|
+
colIndex: -1,
|
|
140
|
+
isHeaderRowCell: false,
|
|
141
|
+
isBodyTitleRowCell: false,
|
|
142
|
+
valueParsed: null,
|
|
143
|
+
headerCol: null,
|
|
144
|
+
headerRowIndex: null,
|
|
145
|
+
};
|
|
146
|
+
// const hasCopies = cell.colSpan > 1 || cell.rowSpan > 1
|
|
147
|
+
var curRow = (_e = grid[rowIndex]) !== null && _e !== void 0 ? _e : [];
|
|
148
|
+
var nextEmptyCellIndex = curRow.findIndex(function (cell) { return !cell; });
|
|
149
|
+
var idxStart = nextEmptyCellIndex === -1 ? curRow.length : nextEmptyCellIndex;
|
|
150
|
+
for (var r = rowIndex; r < rowIndex + rowSpan; r++) {
|
|
151
|
+
grid[r] = (_f = grid[r]) !== null && _f !== void 0 ? _f : [];
|
|
152
|
+
for (var c = idxStart; c < idxStart + colSpan; c++) {
|
|
153
|
+
cell.colIndex = cell.colIndex > -1 ? cell.colIndex : c;
|
|
154
|
+
grid[r][c] = cell;
|
|
155
|
+
}
|
|
156
|
+
}
|
|
157
|
+
};
|
|
158
|
+
for (var i = 0; i < table.html.length; i++) {
|
|
159
|
+
var skipIndex = (_a = skipIndexMap.get(i)) !== null && _a !== void 0 ? _a : null;
|
|
160
|
+
if (skipIndex) {
|
|
161
|
+
cellHTML += table.html.substring(i, skipIndex + 1);
|
|
162
|
+
i = skipIndex;
|
|
163
|
+
continue;
|
|
164
|
+
}
|
|
165
|
+
var prev5Chars = table.html.substring(i - 4, i + 1).toLowerCase();
|
|
166
|
+
var next3Chars = table.html.substring(i, i + 3).toLowerCase();
|
|
167
|
+
var isCellAttsStart = ['<td', '<th'].includes(next3Chars);
|
|
168
|
+
var isSelfEnclosed = isInCellAtts && table.html[i - 1] === '/' && table.html[i] === '>';
|
|
169
|
+
var isCellAttsEnd = (isInCell && table.html[i] === '>') || isSelfEnclosed;
|
|
170
|
+
var isCellEnd = ['</td>', '</th>'].includes(prev5Chars);
|
|
171
|
+
var isRowStart = next3Chars === '<tr';
|
|
172
|
+
if (isRowStart) {
|
|
173
|
+
rowIndex++;
|
|
174
|
+
grid[rowIndex] = (_b = grid[rowIndex]) !== null && _b !== void 0 ? _b : [];
|
|
175
|
+
}
|
|
176
|
+
if (isCellAttsStart) {
|
|
177
|
+
tableCellIndex++;
|
|
178
|
+
isInCell = true;
|
|
179
|
+
isInCellAtts = true;
|
|
180
|
+
}
|
|
181
|
+
if (isInCellAtts) {
|
|
182
|
+
cellAtts += table.html[i];
|
|
183
|
+
}
|
|
184
|
+
if (isInCell) {
|
|
185
|
+
cellHTML += table.html[i];
|
|
186
|
+
}
|
|
187
|
+
if (isCellAttsEnd) {
|
|
188
|
+
isInCellAtts = false;
|
|
189
|
+
}
|
|
190
|
+
if (isCellEnd || isSelfEnclosed) {
|
|
191
|
+
isInCell = false;
|
|
192
|
+
isInCellAtts = false;
|
|
193
|
+
createCell(cellHTML, cellAtts);
|
|
194
|
+
cellHTML = '';
|
|
195
|
+
cellAtts = '';
|
|
196
|
+
}
|
|
197
|
+
}
|
|
198
|
+
table.rows = grid;
|
|
199
|
+
});
|
|
200
|
+
};
|
|
201
|
+
HtmlTableExtractor.prototype.addMissingNameCol = function (tables) {
|
|
202
|
+
var _a, _b;
|
|
203
|
+
var _loop_3 = function (table) {
|
|
204
|
+
var bodyIndex = table.rows.findIndex(function (row) { return row.some(function (col) { return !col.isHeaderRowCell; }); });
|
|
205
|
+
// get the first column index that has a value
|
|
206
|
+
var firstPopulatedColIndex = Infinity;
|
|
207
|
+
for (var i = bodyIndex; i < table.rows.length; i++) {
|
|
208
|
+
var row = table.rows[i];
|
|
209
|
+
if (!row)
|
|
210
|
+
continue;
|
|
211
|
+
var populatedIndex = row.findIndex(function (col) { return col === null || col === void 0 ? void 0 : col.valueParsed; });
|
|
212
|
+
var isFirstPopulatedIndex = populatedIndex > -1 && populatedIndex < firstPopulatedColIndex;
|
|
213
|
+
if (isFirstPopulatedIndex)
|
|
214
|
+
firstPopulatedColIndex = populatedIndex;
|
|
215
|
+
if (firstPopulatedColIndex === 0)
|
|
216
|
+
break;
|
|
217
|
+
}
|
|
218
|
+
var shouldAddName = table.rows.some(function (row) {
|
|
219
|
+
var firstCol = row[firstPopulatedColIndex];
|
|
220
|
+
var headerCol = firstCol === null || firstCol === void 0 ? void 0 : firstCol.headerCol;
|
|
221
|
+
// skip if the first column has a header col, or if there is no header row
|
|
222
|
+
if (!firstCol || headerCol || firstCol.headerRowIndex === null) {
|
|
223
|
+
return false;
|
|
224
|
+
}
|
|
225
|
+
// if the first col is a string, assume it's a name
|
|
226
|
+
return typeof (firstCol === null || firstCol === void 0 ? void 0 : firstCol.valueParsed) === 'string';
|
|
227
|
+
});
|
|
228
|
+
if (shouldAddName) {
|
|
229
|
+
for (var _c = 0, _d = table.rows; _c < _d.length; _c++) {
|
|
230
|
+
var row = _d[_c];
|
|
231
|
+
var col = row[firstPopulatedColIndex];
|
|
232
|
+
if (!col)
|
|
233
|
+
continue;
|
|
234
|
+
var isEmptyRow = row.every(function (col) { return (col === null || col === void 0 ? void 0 : col.valueParsed) === null; });
|
|
235
|
+
// for header rows, add to valueParsed, body rows, set headerCol
|
|
236
|
+
if (!isEmptyRow && col.isHeaderRowCell) {
|
|
237
|
+
col.valueParsed = (_a = col.valueParsed) !== null && _a !== void 0 ? _a : '[name]';
|
|
238
|
+
}
|
|
239
|
+
else if (!col.isHeaderRowCell) {
|
|
240
|
+
col.headerCol = (_b = col.headerCol) !== null && _b !== void 0 ? _b : '[name]';
|
|
241
|
+
}
|
|
242
|
+
}
|
|
243
|
+
}
|
|
244
|
+
};
|
|
245
|
+
for (var _i = 0, tables_3 = tables; _i < tables_3.length; _i++) {
|
|
246
|
+
var table = tables_3[_i];
|
|
247
|
+
_loop_3(table);
|
|
248
|
+
}
|
|
249
|
+
};
|
|
250
|
+
HtmlTableExtractor.prototype.addTableCellValues = function (tables, options) {
|
|
251
|
+
var _this = this;
|
|
252
|
+
var _a, _b;
|
|
253
|
+
var getHeaderRowIndexDefault = function (data) {
|
|
254
|
+
var rows = data.rows;
|
|
255
|
+
var bodyIndex = rows.findIndex(function (row, r) {
|
|
256
|
+
var _a;
|
|
257
|
+
var prevRow = (_a = rows[r - 1]) !== null && _a !== void 0 ? _a : [];
|
|
258
|
+
var hadUnderlines = prevRow.some(function (col) { return col.attributes.includes('border') && col.attributes.includes('bottom'); });
|
|
259
|
+
var hasUnderline = row.some(function (col) { return col.attributes.includes('border') && col.attributes.includes('bottom'); });
|
|
260
|
+
if (hadUnderlines && !hasUnderline) {
|
|
261
|
+
return true;
|
|
262
|
+
}
|
|
263
|
+
return row.some(function (col) {
|
|
264
|
+
var valueParsed = _this.parseValue(col.html, options);
|
|
265
|
+
var isNumber = typeof valueParsed === 'number';
|
|
266
|
+
var isYear = isNumber && valueParsed > 1900 && valueParsed < 2100;
|
|
267
|
+
var isCol = isNumber && !isYear;
|
|
268
|
+
return isCol;
|
|
269
|
+
});
|
|
270
|
+
});
|
|
271
|
+
return bodyIndex - 1;
|
|
272
|
+
};
|
|
273
|
+
var getHeaderRowIndexCb = (_a = options === null || options === void 0 ? void 0 : options.getHeaderRowIndex) !== null && _a !== void 0 ? _a : getHeaderRowIndexDefault;
|
|
274
|
+
var getHeaderRowIndex = function (data) {
|
|
275
|
+
return Math.max(getHeaderRowIndexCb(data), -1);
|
|
276
|
+
};
|
|
277
|
+
var getNextCell = function (row, colIndex) {
|
|
278
|
+
var startingCol = row[colIndex];
|
|
279
|
+
for (var i = colIndex; i < row.length; i++) {
|
|
280
|
+
if (!row[i])
|
|
281
|
+
continue;
|
|
282
|
+
if (row[i].tableCellIndex !== (startingCol === null || startingCol === void 0 ? void 0 : startingCol.tableCellIndex)) {
|
|
283
|
+
return row[i];
|
|
284
|
+
}
|
|
285
|
+
}
|
|
286
|
+
return null;
|
|
287
|
+
};
|
|
288
|
+
var completedCells = new Set();
|
|
289
|
+
var _loop_4 = function (table) {
|
|
290
|
+
var headerRowIndex = getHeaderRowIndex({ rows: table.rows, table: table });
|
|
291
|
+
var _loop_5 = function (rowIndex) {
|
|
292
|
+
var row = table.rows[rowIndex];
|
|
293
|
+
if (!row)
|
|
294
|
+
return "continue";
|
|
295
|
+
var countUniqueCells = new Set(row.map(function (c) { return c.tableCellIndex; })).size;
|
|
296
|
+
// skip titles in the middle of the body
|
|
297
|
+
var isBodyTitleRow = rowIndex > headerRowIndex && countUniqueCells === 1 && ((_b = row[0]) === null || _b === void 0 ? void 0 : _b.colSpan) > 0;
|
|
298
|
+
var isHeaderRow = rowIndex <= headerRowIndex;
|
|
299
|
+
var headerByIndex = new Map();
|
|
300
|
+
var getHeaderCol = function (c) {
|
|
301
|
+
var _a, _b, _c, _d, _e;
|
|
302
|
+
if (headerByIndex.has(c)) {
|
|
303
|
+
return (_a = headerByIndex.get(c)) !== null && _a !== void 0 ? _a : null;
|
|
304
|
+
}
|
|
305
|
+
if (isHeaderRow) {
|
|
306
|
+
return null;
|
|
307
|
+
}
|
|
308
|
+
for (var r = 0; r <= headerRowIndex; r++) {
|
|
309
|
+
var row_1 = table.rows[r];
|
|
310
|
+
if (!row_1)
|
|
311
|
+
continue;
|
|
312
|
+
for (var c_1 = 0; c_1 < row_1.length; c_1++) {
|
|
313
|
+
var col = (_c = (_b = row_1[c_1]) !== null && _b !== void 0 ? _b : row_1[c_1 - 1]) !== null && _c !== void 0 ? _c : row_1[c_1 + 1];
|
|
314
|
+
if (!col)
|
|
315
|
+
continue;
|
|
316
|
+
var headerCurrent = (_d = headerByIndex.get(c_1)) !== null && _d !== void 0 ? _d : '';
|
|
317
|
+
var value = headerCurrent.endsWith("".concat(col.valueParsed || ''))
|
|
318
|
+
? headerCurrent
|
|
319
|
+
: "".concat(headerCurrent, " ").concat(col.valueParsed || '').trim();
|
|
320
|
+
headerByIndex.set(c_1, value);
|
|
321
|
+
}
|
|
322
|
+
}
|
|
323
|
+
return (_e = headerByIndex.get(c)) !== null && _e !== void 0 ? _e : null;
|
|
324
|
+
};
|
|
325
|
+
for (var colIndex = 0; colIndex < row.length; colIndex++) {
|
|
326
|
+
var cell = row[colIndex];
|
|
327
|
+
if (completedCells.has(cell) || !cell) {
|
|
328
|
+
continue;
|
|
329
|
+
}
|
|
330
|
+
cell.headerRowIndex = headerRowIndex > -1 ? headerRowIndex : null;
|
|
331
|
+
cell.isBodyTitleRowCell = isBodyTitleRow;
|
|
332
|
+
cell.isHeaderRowCell = isHeaderRow;
|
|
333
|
+
// sometimes there is a rogue percent sign that is not in a column, so we need to check the next column
|
|
334
|
+
var nextCell = getNextCell(row, colIndex);
|
|
335
|
+
// const isMissingPercentSign =
|
|
336
|
+
// nextCell?.html.includes('%') && this.parseValue(nextCell?.html) === null
|
|
337
|
+
var isMissingParenthesis = (nextCell === null || nextCell === void 0 ? void 0 : nextCell.html.includes(')')) && cell.html.includes('(') && !cell.html.includes(')');
|
|
338
|
+
var colValue = isMissingParenthesis ? "".concat(cell.html.trim(), ")") : cell.html.trim();
|
|
339
|
+
// colValue = isMissingPercentSign ? `${colValue}` : colValue
|
|
340
|
+
colValue = this_1.parseValue(colValue, options);
|
|
341
|
+
colValue = typeof colValue === 'string' ? colValue.replace(/\s+/g, ' ') : colValue;
|
|
342
|
+
// add parsed value
|
|
343
|
+
cell.valueParsed = colValue;
|
|
344
|
+
cell.headerCol = getHeaderCol(colIndex);
|
|
345
|
+
completedCells.add(cell);
|
|
346
|
+
}
|
|
347
|
+
};
|
|
348
|
+
for (var rowIndex = 0; rowIndex < table.rows.length; rowIndex++) {
|
|
349
|
+
_loop_5(rowIndex);
|
|
350
|
+
}
|
|
351
|
+
};
|
|
352
|
+
var this_1 = this;
|
|
353
|
+
for (var _i = 0, tables_4 = tables; _i < tables_4.length; _i++) {
|
|
354
|
+
var table = tables_4[_i];
|
|
355
|
+
_loop_4(table);
|
|
356
|
+
}
|
|
357
|
+
};
|
|
358
|
+
HtmlTableExtractor.prototype.stripHtml = function (str, options) {
|
|
359
|
+
var _a = (options !== null && options !== void 0 ? options : {}).tagsToExclude, tagsToExclude = _a === void 0 ? [] : _a;
|
|
360
|
+
var strNew = str;
|
|
361
|
+
if (tagsToExclude.length > 0) {
|
|
362
|
+
strNew = '';
|
|
363
|
+
var _loop_6 = function (i) {
|
|
364
|
+
var char = str[i];
|
|
365
|
+
if (char !== '<') {
|
|
366
|
+
strNew += char;
|
|
367
|
+
return out_i_1 = i, "continue";
|
|
368
|
+
}
|
|
369
|
+
var matchedTag = tagsToExclude.find(function (tag) { return str.substring(i, i + tag.length + 1).toLowerCase() === "<".concat(tag); });
|
|
370
|
+
if (!matchedTag) {
|
|
371
|
+
strNew += char;
|
|
372
|
+
return out_i_1 = i, "continue";
|
|
373
|
+
}
|
|
374
|
+
var endTag = "</".concat(matchedTag, ">");
|
|
375
|
+
var endTagIndex = str.indexOf(endTag, i);
|
|
376
|
+
if (endTagIndex > -1) {
|
|
377
|
+
i = endTagIndex + endTag.length - 1;
|
|
378
|
+
}
|
|
379
|
+
out_i_1 = i;
|
|
380
|
+
};
|
|
381
|
+
var out_i_1;
|
|
382
|
+
for (var i = 0; i < str.length; i++) {
|
|
383
|
+
_loop_6(i);
|
|
384
|
+
i = out_i_1;
|
|
385
|
+
}
|
|
386
|
+
}
|
|
387
|
+
return strNew.replace(/<.*?>/gm, '');
|
|
388
|
+
};
|
|
389
|
+
HtmlTableExtractor.prototype.parseValue = function (str, options) {
|
|
390
|
+
var _a;
|
|
391
|
+
if (str === null)
|
|
392
|
+
return null;
|
|
393
|
+
if (typeof str === 'number')
|
|
394
|
+
return str;
|
|
395
|
+
var _b = options !== null && options !== void 0 ? options : {}, _c = _b.stripHtml, stripHtml = _c === void 0 ? true : _c, _d = _b.tagsToExclude, tagsToExclude = _d === void 0 ? [] : _d, _e = _b.stripParenthesis, stripParenthesis = _e === void 0 ? false : _e;
|
|
396
|
+
var strNew = stripHtml ? this.stripHtml(str, { tagsToExclude: tagsToExclude }) : str;
|
|
397
|
+
var text = strNew
|
|
398
|
+
.replace(/ | |\n/g, ' ')
|
|
399
|
+
.replace(/®|☐/g, '')
|
|
400
|
+
.replace(/–|—|—|—/g, '-')
|
|
401
|
+
.replace(/’|“|”|’/g, "'");
|
|
402
|
+
if (stripParenthesis) {
|
|
403
|
+
text = text.replace(/\(.*?\)/g, '');
|
|
404
|
+
}
|
|
405
|
+
text = text
|
|
406
|
+
.replace(/\s+/, ' ')
|
|
407
|
+
.replace(/&([a-z0-9]+|#[0-9]{1,6}|#x[0-9a-fA-F]{1,6});/g, ' ')
|
|
408
|
+
.trim();
|
|
409
|
+
if (str.replace(/–|—|—/g, '-') === '-')
|
|
410
|
+
return '-';
|
|
411
|
+
if (text === '')
|
|
412
|
+
return null;
|
|
413
|
+
var colNum = text.replace(/,|\(|\)|%/g, '').trim();
|
|
414
|
+
if (colNum === '-' || colNum === '$')
|
|
415
|
+
return null;
|
|
416
|
+
colNum = colNum.replace(/-|\$/g, '');
|
|
417
|
+
var hasNumBeforeParenthesis = Boolean(/\d+\s*(?=\()/.test(text));
|
|
418
|
+
colNum = hasNumBeforeParenthesis ? (_a = colNum.split(' ')[0]) === null || _a === void 0 ? void 0 : _a.trim() : colNum;
|
|
419
|
+
if (!isNaN(Number(colNum))) {
|
|
420
|
+
if (text.includes('%'))
|
|
421
|
+
return text.replace(/[^a-zA-Z\d\s:]/g, '') === '' ? null : text;
|
|
422
|
+
return (text.trim().includes('(') && !hasNumBeforeParenthesis) || text.includes('-')
|
|
423
|
+
? Number(colNum) * -1
|
|
424
|
+
: Number(colNum);
|
|
425
|
+
}
|
|
426
|
+
return text;
|
|
427
|
+
};
|
|
428
|
+
return HtmlTableExtractor;
|
|
429
|
+
}());
|
|
430
|
+
exports.default = HtmlTableExtractor;
|
|
@@ -37,8 +37,8 @@ var FactGrouper = /** @class */ (function () {
|
|
|
37
37
|
*/
|
|
38
38
|
FactGrouper.prototype.buildFactGroupsByReportKey = function (params) {
|
|
39
39
|
var _this = this;
|
|
40
|
-
var _a, _b, _c;
|
|
41
|
-
var facts = params.facts, cik = params.cik, fiscalCalculator = params.fiscalCalculator, resolvePeriodValues = params.resolvePeriodValues,
|
|
40
|
+
var _a, _b, _c, _d;
|
|
41
|
+
var facts = params.facts, cik = params.cik, fiscalCalculator = params.fiscalCalculator, resolvePeriodValues = params.resolvePeriodValues, _e = params.generateMissingGroups, generateMissingGroups = _e === void 0 ? false : _e, splits = params.splits;
|
|
42
42
|
// min and max year will be used to sort the reports
|
|
43
43
|
var minYear = 0;
|
|
44
44
|
var maxYear = 9999;
|
|
@@ -51,8 +51,8 @@ var FactGrouper = /** @class */ (function () {
|
|
|
51
51
|
// Create groups from facts.
|
|
52
52
|
for (var _i = 0, facts_1 = facts; _i < facts_1.length; _i++) {
|
|
53
53
|
var fact = facts_1[_i];
|
|
54
|
-
var
|
|
55
|
-
var
|
|
54
|
+
var _f = fiscalCalculator.getFiscalYearQuarter({ dateStr: fact.end }), quarter = _f.quarter, year = _f.year;
|
|
55
|
+
var _g = (_a = fiscalCalculator.getDatesByYearQuarter({ quarter: quarter, year: year })) !== null && _a !== void 0 ? _a : {}, _h = _g.end, reportEnd = _h === void 0 ? '' : _h, _j = _g.filed, reportFiled = _j === void 0 ? '' : _j;
|
|
56
56
|
var groupKey = this.createGroupKey({ fiscalYear: year, quarter: quarter, name: fact.name });
|
|
57
57
|
if (year < minYear)
|
|
58
58
|
minYear = year;
|
|
@@ -103,8 +103,8 @@ var FactGrouper = /** @class */ (function () {
|
|
|
103
103
|
if (factValue !== null && !group.values.includes(factValue)) {
|
|
104
104
|
group.values.push(factValue);
|
|
105
105
|
}
|
|
106
|
-
if (
|
|
107
|
-
group.accn = (_c = fact.accn) !== null && _c !== void 0 ? _c : '';
|
|
106
|
+
if (periodResolver.isOriginalFiling({ end: fact.end, filed: fact.filed })) {
|
|
107
|
+
group.accn = (_d = (_c = fact.accn) !== null && _c !== void 0 ? _c : group.accn) !== null && _d !== void 0 ? _d : '';
|
|
108
108
|
accnByFiled.set(fact.filed, group.accn);
|
|
109
109
|
}
|
|
110
110
|
if (isPeriodFact) {
|
|
@@ -20,6 +20,7 @@ var __spreadArray = (this && this.__spreadArray) || function (to, from, pack) {
|
|
|
20
20
|
return to.concat(ar || Array.prototype.slice.call(from));
|
|
21
21
|
};
|
|
22
22
|
Object.defineProperty(exports, "__esModule", { value: true });
|
|
23
|
+
var constants_1 = require("../../util/constants");
|
|
23
24
|
var FactPeriodResolver_1 = require("./FactPeriodResolver");
|
|
24
25
|
/**
|
|
25
26
|
* Splits can be filed multiple times throughout different reports. There is no clear
|
|
@@ -29,7 +30,7 @@ var FactPeriodResolver_1 = require("./FactPeriodResolver");
|
|
|
29
30
|
*/
|
|
30
31
|
var FactSplitAdjuster = /** @class */ (function () {
|
|
31
32
|
function FactSplitAdjuster() {
|
|
32
|
-
this.keySplit =
|
|
33
|
+
this.keySplit = constants_1.KEY_SPLIT;
|
|
33
34
|
this.preferFirstValue = true;
|
|
34
35
|
}
|
|
35
36
|
FactSplitAdjuster.prototype.getGroupValue = function (factGroup, isTrailing) {
|
|
@@ -53,11 +53,37 @@ export interface GetFactFrameParams {
|
|
|
53
53
|
unit?: 'pure' | 'USD' | 'shares' | string;
|
|
54
54
|
taxonomy?: 'us-gaap' | 'dei' | 'invest' | string;
|
|
55
55
|
}
|
|
56
|
-
export interface
|
|
57
|
-
|
|
56
|
+
export interface GetInsiderTransactionsParams {
|
|
57
|
+
page: number;
|
|
58
58
|
symbol: string | number;
|
|
59
|
-
|
|
60
|
-
|
|
59
|
+
itemsPerPage?: number;
|
|
60
|
+
isOwnerCik?: boolean;
|
|
61
|
+
}
|
|
62
|
+
export interface SearchCompaniesParams {
|
|
63
|
+
sic?: number;
|
|
64
|
+
/** Abbreviation of state: IL, CA, NY, etc... */
|
|
65
|
+
state?: string;
|
|
66
|
+
page: number;
|
|
67
|
+
itemsPerPage?: number;
|
|
68
|
+
/** Partial name of the company for search. */
|
|
69
|
+
company?: string;
|
|
70
|
+
companyMatch?: 'startsWith' | 'contains';
|
|
71
|
+
}
|
|
72
|
+
interface GetCurrentFilingsParams {
|
|
73
|
+
formType?: '4' | '3' | '10-Q' | '10-K' | 'SC 13G' | '13F-HR' | '8-K';
|
|
74
|
+
page?: number;
|
|
75
|
+
itemsPerPage?: number;
|
|
76
|
+
symbol?: string | number;
|
|
77
|
+
searchType?: 'include' | 'exclude' | 'only';
|
|
78
|
+
}
|
|
79
|
+
export interface GetDocumentXbrlParams {
|
|
80
|
+
url?: string;
|
|
81
|
+
symbol?: string | number;
|
|
82
|
+
accessionNumber?: string;
|
|
83
|
+
includeReport?: boolean;
|
|
84
|
+
includeInstance?: boolean;
|
|
85
|
+
includeLinkbases?: boolean;
|
|
86
|
+
includeSchema?: boolean;
|
|
61
87
|
}
|
|
62
88
|
/**
|
|
63
89
|
* Gets reports from companies filed with the SEC
|
|
@@ -168,7 +194,8 @@ export default class SecEdgarApi {
|
|
|
168
194
|
*/
|
|
169
195
|
getMutualFundList(): Promise<FieldDataResponse<'cik' | 'seriesId' | 'classId' | 'symbol'>>;
|
|
170
196
|
/**
|
|
171
|
-
*
|
|
197
|
+
* If url is provided, all other props are ignored. Otherwise, both accessionNumber
|
|
198
|
+
* and symbol (symbol or cik) are required. provide fileName if different from `${accessionNumber}.txt`
|
|
172
199
|
*
|
|
173
200
|
* Some form types can be parsed using the DocumentParser such as form 4 (insider transactions) and form 13g (institutional holders)
|
|
174
201
|
*
|
|
@@ -176,19 +203,32 @@ export default class SecEdgarApi {
|
|
|
176
203
|
*
|
|
177
204
|
* @see https://www.sec.gov/forms for a list of form types
|
|
178
205
|
*/
|
|
179
|
-
|
|
206
|
+
getDocument(params: {
|
|
207
|
+
url?: string;
|
|
208
|
+
symbol?: string | number;
|
|
209
|
+
accessionNumber?: string;
|
|
210
|
+
fileName?: string;
|
|
211
|
+
}): Promise<string>;
|
|
180
212
|
/**
|
|
181
|
-
*
|
|
182
|
-
*
|
|
183
|
-
* Some form types can be parsed using the DocumentParser such as form 4 (insider transactions) and form 13g (institutional holders)
|
|
213
|
+
* Fetches SEC document and parses XBRL data. If url is provided, symbol and accessionNumber are ignored.
|
|
184
214
|
*
|
|
185
|
-
*
|
|
215
|
+
* Use "include" params to specify what to parse. If not provided, all data is parsed.
|
|
216
|
+
*/
|
|
217
|
+
getDocumentXbrl(params: GetDocumentXbrlParams): Promise<import("../DocumentParser/XBRLParser/XBRLParser").XbrlParseResult & {
|
|
218
|
+
report: ReportRaw | null;
|
|
219
|
+
facts: import("../../types").FactItem[];
|
|
220
|
+
xml: string;
|
|
221
|
+
}>;
|
|
222
|
+
/**
|
|
223
|
+
* Builds a url for a document. If fileName is not provided, it defaults to `${accessionNumber}.txt`
|
|
186
224
|
*
|
|
187
|
-
*
|
|
225
|
+
* format: `https://www.sec.gov/Archives/edgar/data/${cik}/${accessionNumberNoHyphen)}/${fileNameAccessionFile}`
|
|
188
226
|
*/
|
|
189
|
-
|
|
190
|
-
|
|
191
|
-
|
|
227
|
+
buildDocumentUrl(params: {
|
|
228
|
+
symbol: string | number;
|
|
229
|
+
accessionNumber: string;
|
|
230
|
+
fileName?: string;
|
|
231
|
+
}): string;
|
|
192
232
|
/**
|
|
193
233
|
* Used for getting insider transactions. extracts insider transaction urls from submission list response, and parses the xml doc.
|
|
194
234
|
*
|
|
@@ -237,12 +277,13 @@ export default class SecEdgarApi {
|
|
|
237
277
|
*/
|
|
238
278
|
createRequestProxyStatement(params: CreateRequestWrapperParams): SubmissionRequestWrapper<FormDef14aData>;
|
|
239
279
|
/**
|
|
240
|
-
* Gets list of
|
|
280
|
+
* Gets list of filings for a day up to 5 days ago.
|
|
241
281
|
*
|
|
242
282
|
* @see https://www.sec.gov/edgar/searchedgar/currentevents
|
|
243
283
|
*/
|
|
244
284
|
getCurrentFilingsDaily(params?: {
|
|
245
285
|
formType?: DailyFilingFormType;
|
|
286
|
+
/** max 5 */
|
|
246
287
|
lookbackDays?: number;
|
|
247
288
|
}): Promise<{
|
|
248
289
|
date: string;
|
|
@@ -256,5 +297,31 @@ export default class SecEdgarApi {
|
|
|
256
297
|
filedDate: string;
|
|
257
298
|
}[];
|
|
258
299
|
}>;
|
|
300
|
+
getCurrentFilings(params?: GetCurrentFilingsParams): Promise<import("../../types").CurrentFilingsList>;
|
|
301
|
+
/**
|
|
302
|
+
* @see https://www.sec.gov/structureddata/rss-feeds-submitted-filings
|
|
303
|
+
*/
|
|
304
|
+
getCurrentFilingsXBRL(params?: {
|
|
305
|
+
taxonomy?: 'usGaap' | 'mutualFund' | 'inlineXbrl' | 'allXbrl';
|
|
306
|
+
}): Promise<import("../../types").CurrentFilingsXBRL>;
|
|
307
|
+
/**
|
|
308
|
+
* Gets insider transactions for a provided symbol or CIK.
|
|
309
|
+
*
|
|
310
|
+
* To get transactions by a specific owner, set isOwnerCik to true and provide
|
|
311
|
+
* the owner CIK for the symbol parameter.
|
|
312
|
+
*
|
|
313
|
+
* example at https://www.sec.gov/cgi-bin/own-disp?action=getissuer&CIK=0000320193
|
|
314
|
+
*/
|
|
315
|
+
getInsiderTransactions(params: GetInsiderTransactionsParams): Promise<{
|
|
316
|
+
transactions: import("../../types").InsiderTransaction[];
|
|
317
|
+
owners: import("../../types").Owner[];
|
|
318
|
+
issuers: import("../../types").Issuer[];
|
|
319
|
+
}>;
|
|
320
|
+
/**
|
|
321
|
+
* Search for companies from by name, sic code, or state.
|
|
322
|
+
*/
|
|
323
|
+
searchCompanies(params: SearchCompaniesParams): Promise<{
|
|
324
|
+
items: import("../../types").CompanySearchResult[];
|
|
325
|
+
}>;
|
|
259
326
|
}
|
|
260
327
|
export {};
|