sec-edgar-api 0.0.6 → 0.0.7

package/build/services/DocumentParser/XMLNode/XMLNode.js

@@ -0,0 +1,154 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.XMLNode = void 0;
+var XMLNode = /** @class */ (function () {
+    function XMLNode(args) {
+        this.parent = null;
+        this.previousSibling = null;
+        this.nextSibling = null;
+        this.children = [];
+        this.text = '';
+        this.attributesStr = '';
+        this.path = '';
+        var _a = args !== null && args !== void 0 ? args : {}, _b = _a.attributesStr, attributesStr = _b === void 0 ? '' : _b, _c = _a.path, path = _c === void 0 ? '' : _c;
+        var pathParts = path.split('.').filter(function (str) { return !str.includes(':'); });
+        this.attributesStr = attributesStr;
+        this.path = pathParts.join('.');
+    }
+    XMLNode.prototype.getSiblings = function (params) {
+        var _a = params.stopAtType, stopAtType = _a === void 0 ? XMLNode : _a, dir = params.dir, _b = params.includeStopAtType, includeStopAtType = _b === void 0 ? false : _b;
+        var isPrevious = dir === 'previous';
+        var siblings = [];
+        var nextSibling = isPrevious ? this.getPreviousSibling() : this.getNextSibling();
+        var i = 0;
+        while (nextSibling && !(nextSibling instanceof stopAtType)) {
+            siblings.push(nextSibling);
+            nextSibling = isPrevious
+                ? nextSibling.getPreviousSibling()
+                : nextSibling.getNextSibling();
+            i++;
+            if (i > 1000)
+                throw new Error('infinite loop');
+        }
+        if (includeStopAtType && nextSibling instanceof stopAtType) {
+            siblings.push(nextSibling);
+        }
+        return siblings;
+    };
+    XMLNode.prototype.extractBold = function (str) {
+        var _a, _b;
+        if (str === void 0) { str = this.text; }
+        var boldText = (_b = (_a = str
+            .replace(/\n/g, '')
+            .replace(/\}\}\{\{/g, ' | ')
+            .replace(/ | /g, ' ')
+            .replace(/–/g, '- ')
+            .match(/(?<=\{\{).*?(?=\}\})/g)) === null || _a === void 0 ? void 0 : _a.join(' || ')) !== null && _b !== void 0 ? _b : '';
+        return boldText.replace(/\s+/g, ' ').trim();
+    };
+    XMLNode.prototype.parseValue = function (str) {
+        if (str === void 0) { str = this.text; }
+        if (str === null)
+            return null;
+        var text = str
+            .replace(/\n|&#160;|&nbsp;/g, ' ')
+            .replace(/&#174;/g, '')
+            .replace(/&#8211;|&#8212;|&#x2014;/g, '-')
+            .replace(/&#8217;|&#8220;|&#8221;/g, "'")
+            .replace(/\}\}\{\{/g, ' ')
+            .replace(/\{\{|\}\}/g, '')
+            .replace(/\s+/, ' ')
+            .trim();
+        if (str.replace(/&#8211;|&#8212;|&#x2014;/g, '-') === '-')
+            return '-';
+        if (text === '')
+            return null;
+        var colNum = text.replace(/,|\(|\)|\%/g, '').trim();
+        if (colNum === '-' || colNum === '$')
+            return null;
+        colNum = colNum.replace(/\-|\$/g, '');
+        if (!isNaN(Number(colNum))) {
+            if (text.includes('%'))
+                return text.replace(/[^a-zA-Z\d\s:]/g, '') === '' ? null : text;
+            return text.includes('(') || text.includes('-') ? Number(colNum) * -1 : Number(colNum);
+        }
+        return text;
+    };
+    XMLNode.prototype.setPreviousSibling = function (node) {
+        var prevPreviousSibling = this.previousSibling;
+        this.previousSibling = node;
+        if ((prevPreviousSibling === null || prevPreviousSibling === void 0 ? void 0 : prevPreviousSibling.getNextSibling()) === this) {
+            prevPreviousSibling.setNextSibling(null);
+        }
+        if ((node === null || node === void 0 ? void 0 : node.getNextSibling()) !== this) {
+            node === null || node === void 0 ? void 0 : node.setNextSibling(this);
+        }
+    };
+    XMLNode.prototype.setNextSibling = function (node) {
+        var prevNextSibling = this.nextSibling;
+        this.nextSibling = node;
+        if ((prevNextSibling === null || prevNextSibling === void 0 ? void 0 : prevNextSibling.getPreviousSibling()) === this) {
+            prevNextSibling.setPreviousSibling(null);
+        }
+        if ((node === null || node === void 0 ? void 0 : node.getPreviousSibling()) !== this) {
+            node === null || node === void 0 ? void 0 : node.setPreviousSibling(this);
+        }
+    };
+    XMLNode.prototype.removeChild = function (node) {
+        this.children.splice(this.children.indexOf(node), 1);
+        if (node.getParent() === this) {
+            node.setParent(null);
+        }
+    };
+    XMLNode.prototype.setParent = function (node) {
+        var prevParent = this.parent;
+        this.parent = node;
+        if (prevParent === null || prevParent === void 0 ? void 0 : prevParent.getChildren().includes(this)) {
+            prevParent.removeChild(this);
+        }
+        if (!(node === null || node === void 0 ? void 0 : node.getChildren().includes(this))) {
+            node === null || node === void 0 ? void 0 : node.addChild(this);
+        }
+    };
+    XMLNode.prototype.getNextSibling = function () {
+        return this.nextSibling;
+    };
+    XMLNode.prototype.getPreviousSibling = function () {
+        return this.previousSibling;
+    };
+    XMLNode.prototype.getParent = function () {
+        return this.parent;
+    };
+    XMLNode.prototype.getAttributes = function () {
+        var _a;
+        var attributesObj = {};
+        (_a = this.attributesStr.match(/(\w+)=("[^"]*")/g)) === null || _a === void 0 ? void 0 : _a.forEach(function (attributeStr) {
+            var _a = attributeStr.split('='), key = _a[0], value = _a[1];
+            attributesObj[key.toLowerCase()] = value.replace(/"/g, '');
+        });
+        return attributesObj;
+    };
+    XMLNode.prototype.getAttributesStr = function () {
+        return this.attributesStr;
+    };
+    XMLNode.prototype.getChildren = function () {
+        return this.children;
+    };
+    XMLNode.prototype.addChild = function (node) {
+        this.children.push(node);
+        if (node.getParent() !== this) {
+            node.setParent(this);
+        }
+    };
+    XMLNode.prototype.getText = function () {
+        return this.text;
+    };
+    XMLNode.prototype.setText = function (text) {
+        this.text = text;
+    };
+    XMLNode.prototype.getPath = function () {
+        return this.path;
+    };
+    return XMLNode;
+}());
+exports.XMLNode = XMLNode;

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

@@ -1,8 +1,13 @@
+import { DocumentNode } from './XMLNode/DocumentNode';
 interface OnCharacterData {
     char: string;
     index: number;
     path: string;
     pathOccurrenceCount: number;
+    attributesStr: string;
+}
+interface ParseTableNodesParams {
+    xml: string;
 }
 interface Parse2Params {
     xml: string;
@@ -14,6 +19,15 @@ interface IterateTablesParams {
     xml: string;
     parentPath?: string;
     trimSpaces?: boolean;
+    onCharacter?: (data: OnCharacterData & {
+        textMap: Map<string, string>;
+    }) => void;
+    onOpenTag?: (data: OnCharacterData & {
+        textMap: Map<string, string>;
+    }) => void;
+    onCloseTag?: (data: OnCharacterData & {
+        textMap: Map<string, string>;
+    }) => void;
 }
 export default class XMLParser {
     iterateXML(params: Parse2Params): string[];
@@ -21,5 +35,6 @@ export default class XMLParser {
      * 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

@@ -1,10 +1,27 @@
 "use strict";
+var __assign = (this && this.__assign) || function () {
+    __assign = Object.assign || function(t) {
+        for (var s, i = 1, n = arguments.length; i < n; i++) {
+            s = arguments[i];
+            for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p))
+                t[p] = s[p];
+        }
+        return t;
+    };
+    return __assign.apply(this, arguments);
+};
 Object.defineProperty(exports, "__esModule", { value: true });
+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() {
     }
     XMLParser.prototype.iterateXML = function (params) {
-        var _a, _b, _c, _d;
+        var _a, _b, _c;
         var xml = params.xml, onCharacter = params.onCharacter, onCloseTag = params.onCloseTag, onOpenTag = params.onOpenTag;
         var selfEnclosingTags = new Set([
             'filename',
@@ -26,70 +43,69 @@ var XMLParser = /** @class */ (function () {
             'track',
             'wbr',
         ]);
+        var spaceChars = new Set(['\n', '\r', '\t', ' ']);
         var pathOccurrenceCountMap = new Map();
         var curPath = '';
         var curTag = '';
+        var curAttributes = '';
         var didStart = false;
         var pathsArr = [];
         for (var i = 0; i < xml.length; i++) {
             var char = xml[i];
-            if (char === '<' && xml[i + 1] !== '/' && xml[i + 1] !== '?' && xml[i + 1] !== '!') {
-                var iOpen = i;
+            var isOpenTag = char === '<' && xml[i + 1] !== '/' && xml[i + 1] !== '?' && xml[i + 1] !== '!';
+            var isCloseTag = char === '<' && xml[i + 1] === '/';
+            var onCharacterData = {
+                char: char,
+                index: i,
+                path: curPath,
+                pathOccurrenceCount: (_a = pathOccurrenceCountMap.get(curPath)) !== null && _a !== void 0 ? _a : 0,
+                attributesStr: curAttributes,
+            };
+            if (isOpenTag) {
+                var didEndTagName = false;
+                var j = 0;
                 didStart = true;
                 i++;
-                var j = 0;
-                var didEndTagName = false;
                 while (xml[i] !== '>') {
-                    didEndTagName =
-                        didEndTagName ||
-                            xml[i] === ' ' ||
-                            xml[i] === '\n' ||
-                            xml[i] === '\r' ||
-                            xml[i] === '\t' ||
-                            xml[i] === '/';
+                    didEndTagName = didEndTagName || spaceChars.has(xml[i]) || xml[i] === '/';
                     if (!didEndTagName) {
                         curTag += xml[i].toLowerCase();
                     }
+                    else if (xml[i] !== '/') {
+                        curAttributes += xml[i];
+                    }
                     i++;
                     j++;
-                    if (j > 1000) {
+                    if (j > 1000000) {
                         throw new Error('too many iterations');
                     }
                 }
-                // map path for non-self-enclosing tags
-                if (!selfEnclosingTags.has(curTag)) {
-                    curPath = "".concat(curPath).concat(curPath.length > 0 ? '.' : '').concat(curTag).toLowerCase();
-                    var pathOccurrenceCount = (_a = pathOccurrenceCountMap.get(curPath)) !== null && _a !== void 0 ? _a : 0;
-                    pathOccurrenceCountMap.set(curPath, pathOccurrenceCount + 1);
-                    pathsArr.push(curPath);
-                    onOpenTag === null || onOpenTag === void 0 ? void 0 : onOpenTag({
-                        char: char,
-                        index: iOpen,
-                        path: curPath,
-                        pathOccurrenceCount: (_b = pathOccurrenceCountMap.get(curPath)) !== null && _b !== void 0 ? _b : 0,
-                    });
+                var pathNew = "".concat(curPath).concat(curPath.length > 0 ? '.' : '').concat(curTag).toLowerCase();
+                var countBefore = (_b = pathOccurrenceCountMap.get(pathNew)) !== null && _b !== void 0 ? _b : 0;
+                var pathOccurrenceCount = (_c = pathOccurrenceCountMap.set(pathNew, countBefore + 1).get(pathNew)) !== null && _c !== void 0 ? _c : 0;
+                onCharacterData.path = pathNew;
+                onCharacterData.pathOccurrenceCount = pathOccurrenceCount;
+                onCharacterData.attributesStr = curAttributes;
+                pathsArr.push(pathNew);
+                onOpenTag === null || onOpenTag === void 0 ? void 0 : onOpenTag(onCharacterData);
+                if (selfEnclosingTags.has(curTag)) {
+                    onCloseTag === null || onCloseTag === void 0 ? void 0 : onCloseTag(onCharacterData);
+                }
+                else {
+                    curPath = pathNew;
                 }
                 curTag = '';
             }
-            else if (char === '<' && xml[i + 1] === '/') {
+            else if (isCloseTag) {
                 while (xml[i] !== '>') {
                     i++;
                 }
-                onCloseTag === null || onCloseTag === void 0 ? void 0 : onCloseTag({
-                    char: char,
-                    index: i,
-                    path: curPath,
-                    pathOccurrenceCount: (_c = pathOccurrenceCountMap.get(curPath)) !== null && _c !== void 0 ? _c : 0,
-                });
+                onCloseTag === null || onCloseTag === void 0 ? void 0 : onCloseTag(onCharacterData);
                 curPath = curPath.slice(0, curPath.lastIndexOf('.'));
+                curAttributes = '';
             }
             else if (didStart) {
-                onCharacter === null || onCharacter === void 0 ? void 0 : onCharacter({
-                    char: char,
-                    index: i,
-                    path: curPath,
-                    pathOccurrenceCount: (_d = pathOccurrenceCountMap.get(curPath)) !== null && _d !== void 0 ? _d : 0,
-                });
+                onCharacter === null || onCharacter === void 0 ? void 0 : onCharacter(onCharacterData);
             }
         }
         return pathsArr;
@@ -98,7 +114,7 @@ var XMLParser = /** @class */ (function () {
      * Returns text in each table cell mapped by `${table}.${row}.${col}`
      */
     XMLParser.prototype.getTableTextMap = function (params) {
-        var xml = params.xml, parentPath = params.parentPath, _a = params.trimSpaces, trimSpaces = _a === void 0 ? true : _a;
+        var xml = params.xml, parentPath = params.parentPath, onCharacter = params.onCharacter, onCloseTag = params.onCloseTag, onOpenTag = params.onOpenTag, _a = params.trimSpaces, trimSpaces = _a === void 0 ? true : _a;
         var rowPaths = new Set([
             "".concat(parentPath, ".table.tbody.tr"),
             "".concat(parentPath, ".table.thead.tr"),
@@ -122,11 +138,11 @@ var XMLParser = /** @class */ (function () {
         var spaceChars = new Set(['\n', '\r', '\t']);
         this.iterateXML({
             xml: xml,
-            onOpenTag: function (_a) {
-                var _b;
-                var path = _a.path;
+            onOpenTag: function (data) {
+                var _a;
+                var path = data.path;
                 var colKey = "".concat(table, ".").concat(row, ".").concat(col);
-                var textCur = (_b = textByColKey.get(colKey)) !== null && _b !== void 0 ? _b : '';
+                var textCur = (_a = textByColKey.get(colKey)) !== null && _a !== void 0 ? _a : '';
                 var pathLower = path.toLowerCase();
                 if (textCur.trim().length === 0 && col === 0) {
                     textByColKey.delete(colKey);
@@ -148,18 +164,19 @@ var XMLParser = /** @class */ (function () {
                 else if (isCol) {
                     col++;
                 }
+                onOpenTag === null || onOpenTag === void 0 ? void 0 : onOpenTag(__assign(__assign({}, data), { textMap: textByColKey }));
             },
-            onCharacter: function (_a) {
-                var _b;
-                var char = _a.char;
-                char = spaceChars.has(char) ? ' ' : char;
+            onCharacter: function (data) {
+                var _a;
+                var char = spaceChars.has(data.char) ? ' ' : data.char;
                 var colKey = "".concat(table, ".").concat(row, ".").concat(col);
-                var textCur = (_b = textByColKey.get(colKey)) !== null && _b !== void 0 ? _b : '';
-                if (trimSpaces && char === ' ' && textCur.endsWith(' '))
-                    return;
-                textByColKey.set(colKey, "".concat(textCur).concat(char));
+                var textCur = (_a = textByColKey.get(colKey)) !== null && _a !== void 0 ? _a : '';
+                if (!(trimSpaces && char === ' ' && textCur.endsWith(' '))) {
+                    textByColKey.set(colKey, "".concat(textCur).concat(char));
+                }
+                onCharacter === null || onCharacter === void 0 ? void 0 : onCharacter(__assign(__assign({}, data), { textMap: textByColKey }));
             },
-            onCloseTag: function () {
+            onCloseTag: function (data) {
                 var _a;
                 var colKey = "".concat(table, ".").concat(row, ".").concat(col);
                 var textCur = (_a = textByColKey.get(colKey)) !== null && _a !== void 0 ? _a : '';
@@ -169,10 +186,121 @@ var XMLParser = /** @class */ (function () {
                 else if (!textCur.endsWith(' ')) {
                     textByColKey.set(colKey, "".concat(textCur, " "));
                 }
+                onCloseTag === null || onCloseTag === void 0 ? void 0 : onCloseTag(__assign(__assign({}, data), { textMap: textByColKey }));
             },
         });
         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 colSpan = col.getColSpan();
+            var colIndex = curRowCols.length;
+            col.setIndex(colIndex);
+            for (var i = 0; i < colSpan; i++) {
+                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 () {
+                var _a;
+                if ((curNode === null || curNode === void 0 ? void 0 : curNode.getPath()) === boldPath) {
+                    curNode === null || curNode === void 0 ? void 0 : curNode.setText("".concat((_a = curNode === null || curNode === void 0 ? void 0 : curNode.getText()) !== null && _a !== void 0 ? _a : '', "}}"));
+                    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;
+                }
+            },
+        });
+        return documentNode;
+    };
     return XMLParser;
 }());
 exports.default = XMLParser;

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

@@ -1,7 +1,9 @@
 import { parseForm4 } from './parse-form-4';
 import { parseForm13g } from './parse-form-13g';
+import { parseForm10k } from './parse-form-10k';
 declare const parsers: {
     parseForm4: typeof parseForm4;
     parseForm13g: typeof parseForm13g;
+    parseForm10k: typeof parseForm10k;
 };
 export default parsers;

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

@@ -2,8 +2,10 @@
 Object.defineProperty(exports, "__esModule", { value: true });
 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 parsers = {
     parseForm4: parse_form_4_1.parseForm4,
     parseForm13g: parse_form_13g_1.parseForm13g,
+    parseForm10k: parse_form_10k_1.parseForm10k,
 };
 exports.default = parsers;

package/build/services/DocumentParser/parsers/parse-form-10k.d.ts

@@ -0,0 +1,3 @@
+import { TableData, XMLParams } from '../../../types';
+import XMLParser from '../XMLParser';
+export declare function parseForm10k(params: XMLParams, xmlParser?: XMLParser): TableData[];

package/build/services/DocumentParser/parsers/parse-form-10k.js

@@ -0,0 +1,78 @@
+"use strict";
+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 });
+exports.parseForm10k = void 0;
+var HRNode_1 = require("../XMLNode/HRNode");
+var NonTableNode_1 = require("../XMLNode/NonTableNode");
+var TableNode_1 = require("../XMLNode/TableNode");
+var XMLParser_1 = require("../XMLParser");
+function parseForm10k(params, xmlParser) {
+    var _a, _b, _c, _d;
+    if (xmlParser === void 0) { xmlParser = new XMLParser_1.default(); }
+    var xml = params.xml;
+    var doc = xmlParser.getDocumentNode({ xml: xml });
+    var nodes = doc.getChildren();
+    var tables = nodes.filter(function (child) { return child instanceof TableNode_1.TableNode; });
+    var curSectionIndex = 0;
+    var sectionIndexByTable = new Map();
+    var tableDataArr = [];
+    // map the section index of each table. new section starts for each <hr> tag.
+    for (var _i = 0, nodes_1 = nodes; _i < nodes_1.length; _i++) {
+        var node = nodes_1[_i];
+        if (node instanceof TableNode_1.TableNode) {
+            sectionIndexByTable.set(node, curSectionIndex);
+        }
+        else if (node instanceof HRNode_1.HRNode) {
+            curSectionIndex++;
+        }
+    }
+    var _loop_1 = function (table) {
+        table.mergeHeader();
+        table.removeEmptyTopRows();
+        var siblingsPrev = table.getSiblings({ dir: 'previous', stopAtType: NonTableNode_1.NonTableNode, includeStopAtType: true });
+        var siblingsNext = table.getSiblings({ dir: 'next', stopAtType: NonTableNode_1.NonTableNode, includeStopAtType: true });
+        // set the title based on the bold text in the previous siblings
+        var title = siblingsPrev.map(function (s) { return s.extractBold(); }).join(' | ');
+        var textBefore = siblingsPrev.map(function (s) { return s.getText(); }).join('\n');
+        var textAfter = siblingsNext.map(function (s) { return s.getText(); }).join('\n');
+        var header = table.getHeaderRow();
+        var headerTable = header === null || header === void 0 ? void 0 : header.toTable();
+        var topRow = table.getChildren()[0];
+        var topRowTable = !header ? topRow === null || topRow === void 0 ? void 0 : topRow.toTable() : null;
+        var prevTable = tableDataArr[tableDataArr.length - 1];
+        var isSameTableSize = prevTable ? ((_a = prevTable === null || prevTable === void 0 ? void 0 : prevTable.rows[0]) === null || _a === void 0 ? void 0 : _a.length) === ((_b = topRowTable === null || topRowTable === void 0 ? void 0 : topRowTable[0]) === null || _b === void 0 ? void 0 : _b.length) : false;
+        var isSameSection = prevTable ? prevTable.sectionIndex === sectionIndexByTable.get(table) : false;
+        var isSharedHeader = !!topRowTable && table.getTitle() === '' && isSameTableSize && isSameSection;
+        // some tables will be in the same section directly below another table and expected to have the same header.
+        if (isSharedHeader) {
+            topRowTable.unshift(__spreadArray([], prevTable.rows[0], true));
+            table.setTitle(prevTable.title);
+        }
+        var rows = (_c = headerTable !== null && headerTable !== void 0 ? headerTable : topRowTable) !== null && _c !== void 0 ? _c : [];
+        if (rows.length === 0)
+            return "continue";
+        tableDataArr.push({
+            title: title.trim(),
+            sectionIndex: (_d = sectionIndexByTable.get(table)) !== null && _d !== void 0 ? _d : -1,
+            hasHeader: Boolean(header) || isSharedHeader,
+            textBefore: textBefore,
+            textAfter: textAfter,
+            rows: rows.filter(function (r) { return r.length === rows[0].length; }),
+        });
+    };
+    // create table arrays
+    for (var _e = 0, tables_1 = tables; _e < tables_1.length; _e++) {
+        var table = tables_1[_e];
+        _loop_1(table);
+    }
+    return tableDataArr;
+}
+exports.parseForm10k = parseForm10k;

package/build/services/ReportParser/ReportParser.d.ts

@@ -8,7 +8,7 @@ interface ReportParserArgs {
     propertyResolver?: PropertyResolver;
     keyTranslator?: Record<string, string[]>;
 }
-declare type TranslateRawReportsCallback<T> = (report: T extends undefined ? ReportTranslated : Record<keyof T, string | number>, reportRaw: ReportRaw, keyTranslator: T) => any;
+type TranslateRawReportsCallback<T> = (report: T extends undefined ? ReportTranslated : Record<keyof T, string | number>, reportRaw: ReportRaw, keyTranslator: T) => any;
 /**
  * Takes company facts data from the SEC and translates them to reports as json objects.
  */
@@ -24,11 +24,11 @@ export default class ReportParser {
      *
      * @param companyFactListData This is the json file contents of CIK[number].json file from the SEC website. You can find these using their API or by downloading the companyfacts.zip file: https://www.sec.gov/edgar/sec-api-documentation
      */
-    parseReports(companyFactListData: Pick<CompanyFactListData, 'facts'>): ReportWrapper[];
+    parseReports(companyFactListData: Pick<CompanyFactListData, 'facts'>, usePropertyResolver?: boolean): ReportWrapper[];
     /**
      * Same as parseReports but accepts ReportRaw[] instead of CompanyFactListData
      */
-    parseReportsFromRaw(reportsRaw: ReportRaw[]): ReportWrapper[];
+    parseReportsFromRaw(reportsRaw: ReportRaw[], usePropertyResolver?: boolean): ReportWrapper[];
     /**
      * Note that this includes all reports by default, not just annual and quarterly. use options.reportsToInclude to filter
      *