sec-edgar-api 0.2.2 → 0.2.3

package/README.md

@@ -24,10 +24,14 @@ Reports are all returned as a uniform interface:
 
 ```TS
 interface ReportTranslated {
-    dateReport: string
+	cik: number
+	url: string | null
+	dateReport: string
 	dateFiled: string
 	fiscalPeriod: FiscalPeriod
 	fiscalYear: number
+	splitDate: string | null
+	splitRatio: number | null
 
 	assetTotal: number | null
 	assetCurrent: number | null
@@ -92,8 +96,6 @@ interface ReportTranslated {
 	cashFlowDeferredTax: number | null
 
 	cashFlowWorkingCapitalNonCash: number | null
-
-	splitRatio: number | null
 }
 ```
 

package/build/services/DocumentParser/XMLParser.d.ts

@@ -1,3 +1,4 @@
+import { DocumentNode } from './XMLNode/DocumentNode';
 interface OnCharacterData {
     char: string;
     index: number;
@@ -5,7 +6,9 @@ interface OnCharacterData {
     pathOccurrenceCount: number;
     attributesStr: string;
 }
-type ObjOrArr = Record<string | symbol | number, any> | any[];
+interface ParseTableNodesParams {
+    xml: string;
+}
 interface Parse2Params {
     xml: string;
     onCharacter?: (data: OnCharacterData) => void;
@@ -26,30 +29,12 @@ interface IterateTablesParams {
         textMap: Map<string, string>;
     }) => void;
 }
-interface Parser {
-    parse(xml: string): Record<string, any>;
-}
-interface XMLParserArgs {
-    parser: Parser;
-}
-interface Section {
-    text: string;
-    tables: string[][][][];
-}
 export default class XMLParser {
-    private readonly parser;
-    constructor(args?: XMLParserArgs);
-    extractJsonKey(json: Record<string, any>, extractKey: string): Record<string, any>;
-    parse(xml: string): any;
-    /**
-     * Return true in the callback if you want the loop to stop
-     */
-    iterateKeysDeep(obj: ObjOrArr, cb: (key: string, currentObj: ObjOrArr, originalObj: ObjOrArr) => void | boolean): void;
-    parseSectionTables(xml: string): Section[];
     iterateXML(params: Parse2Params): string[];
     /**
      * Returns text in each table cell mapped by `${table}.${row}.${col}`
      */
     getTableTextMap(params: IterateTablesParams): Map<string, string>;
+    getDocumentNode(params: ParseTableNodesParams): DocumentNode;
 }
 export {};

package/build/services/DocumentParser/XMLParser.js

@@ -10,131 +10,16 @@ var __assign = (this && this.__assign) || function () {
     };
     return __assign.apply(this, arguments);
 };
-var __spreadArray = (this && this.__spreadArray) || function (to, from, pack) {
-    if (pack || arguments.length === 2) for (var i = 0, l = from.length, ar; i < l; i++) {
-        if (ar || !(i in from)) {
-            if (!ar) ar = Array.prototype.slice.call(from, 0, i);
-            ar[i] = from[i];
-        }
-    }
-    return to.concat(ar || Array.prototype.slice.call(from));
-};
 Object.defineProperty(exports, "__esModule", { value: true });
-var fast_xml_parser_1 = require("fast-xml-parser");
+var ColNode_1 = require("./XMLNode/ColNode");
+var DocumentNode_1 = require("./XMLNode/DocumentNode");
+var HRNode_1 = require("./XMLNode/HRNode");
+var NonTableNode_1 = require("./XMLNode/NonTableNode");
+var RowNode_1 = require("./XMLNode/RowNode");
+var TableNode_1 = require("./XMLNode/TableNode");
 var XMLParser = /** @class */ (function () {
-    function XMLParser(args) {
-        var _a = (args !== null && args !== void 0 ? args : {}).parser, parser = _a === void 0 ? new fast_xml_parser_1.XMLParser() : _a;
-        this.parser = parser;
+    function XMLParser() {
     }
-    XMLParser.prototype.extractJsonKey = function (json, extractKey) {
-        var obj = {};
-        this.iterateKeysDeep(json, function (key, currentObj) {
-            if (key === extractKey) {
-                obj = currentObj[key];
-                return true;
-            }
-        });
-        return obj;
-    };
-    XMLParser.prototype.parse = function (xml) {
-        return this.parser.parse(xml);
-    };
-    /**
-     * Return true in the callback if you want the loop to stop
-     */
-    XMLParser.prototype.iterateKeysDeep = function (obj, cb) {
-        var originalObj = obj;
-        // prevents original object from being modified
-        var _iterateKeys = function (obj, cb) {
-            for (var key in obj) {
-                var shouldBreak = cb(key, obj, originalObj) === true;
-                if (shouldBreak)
-                    return;
-                var value = obj[key];
-                var isObjOrArr = typeof value === 'object' && value !== null;
-                if (isObjOrArr)
-                    _iterateKeys(obj[key], cb);
-            }
-        };
-        _iterateKeys(obj, cb);
-    };
-    XMLParser.prototype.parseSectionTables = function (xml) {
-        var _this = this;
-        var xmlSplit = xml.split(/<HR.*?>/gi);
-        var fastParser = new fast_xml_parser_1.XMLParser({
-            htmlEntities: true,
-            ignoreAttributes: false,
-            alwaysCreateTextNode: true,
-        });
-        var toArray = function (value) { return (Array.isArray(value) ? value : [value]); };
-        var parseTable = function (obj) {
-            var _a, _b, _c, _d;
-            var getRowStrings = function (rows) {
-                var _a, _b, _c;
-                var results = [];
-                for (var i = 0; i < rows.length; i++) {
-                    var obj_1 = rows[i];
-                    var isObject = typeof obj_1 === 'object' && obj_1;
-                    var cols = isObject ? (_c = (_b = (_a = obj_1['TD']) !== null && _a !== void 0 ? _a : obj_1['td']) !== null && _b !== void 0 ? _b : obj_1['TH']) !== null && _c !== void 0 ? _c : obj_1['th'] : [];
-                    var rowCols = [];
-                    var _loop_1 = function (a) {
-                        var col = cols[a];
-                        if (typeof col !== 'object' || !col) {
-                            rowCols.push(String(col || '').trim());
-                        }
-                        var colStr = '';
-                        _this.iterateKeysDeep(col, function (key, curObj) {
-                            var _a;
-                            if (!key.startsWith('@_')) {
-                                colStr = "".concat((_a = curObj[key]) !== null && _a !== void 0 ? _a : '').trim();
-                            }
-                        });
-                        rowCols.push(String(colStr || '').trim());
-                        var colSpan = !Array.isArray(col) ? Number(col['@_COLSPAN']) || 1 : 1;
-                        for (var c = 1; c < colSpan; c++) {
-                            rowCols.push('@_SPAN');
-                        }
-                    };
-                    for (var a = 0; a < (cols !== null && cols !== void 0 ? cols : []).length; a++) {
-                        _loop_1(a);
-                    }
-                    results.push(rowCols);
-                }
-                return results;
-            };
-            var table = toArray(obj);
-            var thead = toArray((_b = (_a = table['THEAD']) !== null && _a !== void 0 ? _a : table['thead']) !== null && _b !== void 0 ? _b : []);
-            var tbody = toArray((_d = (_c = table['TBODY']) !== null && _c !== void 0 ? _c : table['tbody']) !== null && _d !== void 0 ? _d : []);
-            var tableRows = __spreadArray(__spreadArray(__spreadArray([], thead, true), table, true), tbody, true).map(function (obj) { var _a, _b; return toArray((_b = (_a = obj['TR']) !== null && _a !== void 0 ? _a : obj['tr']) !== null && _b !== void 0 ? _b : []); }).flat();
-            var tableData = [];
-            tableData.push(getRowStrings(tableRows));
-            return tableData;
-        };
-        var sections = [];
-        xmlSplit.forEach(function (xml) {
-            var tableIndex = xml.indexOf('<TABLE');
-            tableIndex = tableIndex === -1 ? xml.indexOf('<table') : tableIndex;
-            var text = xml
-                .substring(0, tableIndex)
-                .replace(/<\/?(.|\n)*?>/gi, ' ')
-                .replace(/\s+/g, ' ');
-            var xmlTables = xml
-                .replace(/<(?!(\/?TABLE|\/?TR|\/?TD|\/?TH|\/?HTML))(.|\n)*?>/gi, ' ')
-                .replace(/\s+/g, ' ');
-            var tableObjects = fastParser.parse(xmlTables);
-            var tableObjectsFlat = [];
-            _this.iterateKeysDeep(tableObjects, function (key, curObj) {
-                if (key.toLowerCase() === 'table') {
-                    tableObjectsFlat.push.apply(tableObjectsFlat, toArray(curObj[key]));
-                }
-            });
-            var tables = tableObjectsFlat.map(function (t) { return parseTable(t)[0]; });
-            if (tables.length > 0) {
-                sections.push({ text: text, tables: tables });
-            }
-        });
-        return sections;
-    };
     XMLParser.prototype.iterateXML = function (params) {
         var _a, _b, _c;
         var xml = params.xml, onCharacter = params.onCharacter, onCloseTag = params.onCloseTag, onOpenTag = params.onOpenTag;
@@ -306,6 +191,117 @@ var XMLParser = /** @class */ (function () {
         });
         return textByColKey;
     };
+    XMLParser.prototype.getDocumentNode = function (params) {
+        var xml = params.xml;
+        var rowsArr = [];
+        var colsArr = [];
+        var documentNode = new DocumentNode_1.DocumentNode();
+        var curNode = null;
+        var prevRowCols = [];
+        var curRowCols = [];
+        var isBold = false;
+        var boldPath = null;
+        var pushColToRow = function (col) {
+            var _a;
+            var colIndex = curRowCols.length;
+            col.setIndex(colIndex);
+            var colSpan = col.getColSpan();
+            Array.from({ length: colSpan }).forEach(function () { return curRowCols.push(col); });
+            var topSibling = (_a = prevRowCols[colIndex]) !== null && _a !== void 0 ? _a : null;
+            topSibling === null || topSibling === void 0 ? void 0 : topSibling.addBottomSibling(col);
+        };
+        this.iterateXML({
+            xml: xml,
+            onCloseTag: function (_a) {
+                var _b;
+                var path = _a.path;
+                var tag = path.split('.').pop();
+                if ((curNode === null || curNode === void 0 ? void 0 : curNode.getPath()) === boldPath) {
+                    curNode === null || curNode === void 0 ? void 0 : curNode.setText("".concat((_b = curNode === null || curNode === void 0 ? void 0 : curNode.getText()) !== null && _b !== void 0 ? _b : '', "}}"));
+                    boldPath = null;
+                }
+            },
+            onCharacter: function (_a) {
+                var _b;
+                var char = _a.char;
+                curNode === null || curNode === void 0 ? void 0 : curNode.setText(((_b = curNode === null || curNode === void 0 ? void 0 : curNode.getText()) !== null && _b !== void 0 ? _b : '') + char);
+            },
+            onOpenTag: function (_a) {
+                var _b;
+                var path = _a.path, attributesStr = _a.attributesStr;
+                // skip nested tables
+                if (path.split('.').reduce(function (acc, cur) { return (cur === 'table' ? acc + 1 : acc); }, 0) > 1)
+                    return;
+                var tag = path.split('.').pop();
+                var isInTable = path.includes('table');
+                var topLevelNodes = documentNode.getChildren();
+                var prevTopLevelNode = topLevelNodes[topLevelNodes.length - 1];
+                var wasHorizontalLine = prevTopLevelNode instanceof HRNode_1.HRNode;
+                var wasNonTableNode = prevTopLevelNode instanceof NonTableNode_1.NonTableNode;
+                var wasBold = isBold;
+                var attributesLower = attributesStr.toLowerCase().replace(/\s/g, '');
+                isBold =
+                    tag === 'b' ||
+                        tag === 'strong' ||
+                        attributesLower.includes('font-weight:bold') ||
+                        attributesLower.includes('font-weight:700') ||
+                        attributesLower.includes('font-weight:800') ||
+                        attributesLower.includes('font-weight:900');
+                if (!isInTable) {
+                    prevRowCols = [];
+                    curRowCols = [];
+                }
+                if (tag === 'hr' && !isInTable) {
+                    var hr = new HRNode_1.HRNode({ attributesStr: attributesStr, path: path });
+                    hr.setPreviousSibling(prevTopLevelNode !== null && prevTopLevelNode !== void 0 ? prevTopLevelNode : null);
+                    topLevelNodes.push(hr);
+                    curNode = hr;
+                }
+                else if (tag === 'table') {
+                    var table = new TableNode_1.TableNode({ attributesStr: attributesStr, path: path });
+                    table.setPreviousSibling(prevTopLevelNode !== null && prevTopLevelNode !== void 0 ? prevTopLevelNode : null);
+                    topLevelNodes.push(table);
+                    curNode = table;
+                }
+                else if (tag === 'tr') {
+                    var row = new RowNode_1.RowNode({ attributesStr: attributesStr, path: path });
+                    var prevRow = rowsArr[rowsArr.length - 1];
+                    row.setParent(prevTopLevelNode);
+                    row.setPreviousSibling((prevRow === null || prevRow === void 0 ? void 0 : prevRow.getParent()) === row.getParent() ? prevRow : null);
+                    rowsArr.push(row);
+                    prevRowCols = curRowCols;
+                    curRowCols = [];
+                    curNode = row;
+                }
+                else if (tag === 'td' || tag === 'th') {
+                    var col = new ColNode_1.ColNode({ attributesStr: attributesStr, path: path });
+                    var prevCol = colsArr[colsArr.length - 1];
+                    col.setParent(rowsArr[rowsArr.length - 1]);
+                    col.setPreviousSibling((prevCol === null || prevCol === void 0 ? void 0 : prevCol.getParent()) === col.getParent() ? prevCol : null);
+                    colsArr.push(col);
+                    pushColToRow(col);
+                    curNode = col;
+                }
+                else if ((!isInTable && !wasNonTableNode) || (wasHorizontalLine && tag !== 'hr')) {
+                    var node = new NonTableNode_1.NonTableNode({ attributesStr: attributesStr, path: path });
+                    node.setPreviousSibling(prevTopLevelNode !== null && prevTopLevelNode !== void 0 ? prevTopLevelNode : null);
+                    topLevelNodes.push(node);
+                    curNode = node;
+                }
+                else if (curNode && !curNode.getText().endsWith('\n')) {
+                    curNode.setText("".concat(curNode.getText().trim(), "\n"));
+                }
+                if (isBold && !wasBold && !(curNode === null || curNode === void 0 ? void 0 : curNode.getText().endsWith('{{'))) {
+                    curNode === null || curNode === void 0 ? void 0 : curNode.setText("".concat(curNode === null || curNode === void 0 ? void 0 : curNode.getText().trim(), "{{"));
+                }
+                if (isBold) {
+                    boldPath = (_b = curNode === null || curNode === void 0 ? void 0 : curNode.getPath()) !== null && _b !== void 0 ? _b : null;
+                }
+            },
+        });
+        documentNode.setText(xml);
+        return documentNode;
+    };
     return XMLParser;
 }());
 exports.default = XMLParser;

package/build/services/DocumentParser/parsers/index.d.ts

@@ -1,13 +1,11 @@
-import { parseCurrentFilings } from './parse-current-filings';
-import { parseForm13f } from './parse-form-13f';
-import { parseForm13g } from './parse-form-13g';
 import { parseForm4 } from './parse-form-4';
+import { parseForm13g } from './parse-form-13g';
+import { parseForm10k } from './parse-form-10k';
 import { parseFormDef14a } from './parse-form-def14a';
 declare const parsers: {
     parseForm4: typeof parseForm4;
     parseForm13g: typeof parseForm13g;
-    parseForm13f: typeof parseForm13f;
+    parseForm10k: typeof parseForm10k;
     parseFormDef14a: typeof parseFormDef14a;
-    parseCurrentFilings: typeof parseCurrentFilings;
 };
 export default parsers;

package/build/services/DocumentParser/parsers/index.js

@@ -1,15 +1,13 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-var parse_current_filings_1 = require("./parse-current-filings");
-var parse_form_13f_1 = require("./parse-form-13f");
-var parse_form_13g_1 = require("./parse-form-13g");
 var parse_form_4_1 = require("./parse-form-4");
+var parse_form_13g_1 = require("./parse-form-13g");
+var parse_form_10k_1 = require("./parse-form-10k");
 var parse_form_def14a_1 = require("./parse-form-def14a");
 var parsers = {
     parseForm4: parse_form_4_1.parseForm4,
     parseForm13g: parse_form_13g_1.parseForm13g,
-    parseForm13f: parse_form_13f_1.parseForm13f,
+    parseForm10k: parse_form_10k_1.parseForm10k,
     parseFormDef14a: parse_form_def14a_1.parseFormDef14a,
-    parseCurrentFilings: parse_current_filings_1.parseCurrentFilings,
 };
 exports.default = parsers;

package/build/services/DocumentParser/parsers/parse-form-13g.js

@@ -58,7 +58,7 @@ function parseForm13g(params, xmlParser) {
             return "continue";
         var textParts = text.split(' ').filter(function (t) { return t.trim() !== ''; });
         var colNameIndex = textParts.findIndex(function (t) { return getKey(t) === colName; });
-        var value = ((_c = (_b = textParts[colNameIndex + 1]) === null || _b === void 0 ? void 0 : _b.trim()) !== null && _c !== void 0 ? _c : '').replace(/&.*?;/g, ' ').replace(/\s+/g, ' ').trim();
+        var value = (_c = (_b = textParts[colNameIndex + 1]) === null || _b === void 0 ? void 0 : _b.trim()) !== null && _c !== void 0 ? _c : '';
         switch (colName) {
             case 'shares':
                 holder.shares = Number(value.replace(/[^0-9]/g, '')) || 0;
@@ -83,6 +83,6 @@ function parseForm13g(params, xmlParser) {
     if (((_d = holders[holders.length - 1]) === null || _d === void 0 ? void 0 : _d.name) === '') {
         holders.pop();
     }
-    return { institutionalHolders: holders };
+    return { holders: holders };
 }
 exports.parseForm13g = parseForm13g;

package/build/services/DocumentParser/parsers/parse-form-4.d.ts

@@ -1,12 +1,7 @@
 import { Form4Data, XMLParams } from '../../../types';
 import XMLParser from '../XMLParser';
 /**
- * Form 4 and form 5 - Insider Transactions
- *
- * TODO: sometimes not working. keys not consistently coming back objects vs arrays:
- *
- * here reporting owner is array, so not being extracted correctly
- * https://www.sec.gov/Archives/edgar/data/1067983/000095017024044198/0000950170-24-044198.txt
+ * Form 4 - Insider Transactions
  *
  * example at https://www.sec.gov/Archives/edgar/data/320193/000032019323000079/xslF345X05/wk-form4_1691533817.xml
  */