sec-edgar-api 0.3.4 → 0.4.0

package/build/services/DocumentParser/XMLParser.js

@@ -1,304 +1,111 @@
 "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;
-        var xml = params.xml, onCharacter = params.onCharacter, onCloseTag = params.onCloseTag, onOpenTag = params.onOpenTag;
-        var selfEnclosingTags = new Set([
-            'filename',
-            'description',
+        this.selfEnclosingTags = new Set([
+            '!doctype',
+            '?xml',
+            'xml',
+            'hr',
             'br',
-            'meta',
-            'link',
             'img',
             'input',
-            'hr',
-            'area',
-            'base',
-            'col',
-            'command',
-            'embed',
-            'keygen',
-            'param',
-            'source',
-            'track',
-            'wbr',
+            'meta',
+            'filename',
+            'description',
         ]);
-        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];
-            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++;
-                while (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 > 1000000) {
-                        throw new Error('too many iterations');
-                    }
-                }
-                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 (isCloseTag) {
-                while (xml[i] !== '>') {
-                    i++;
-                }
-                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(onCharacterData);
-            }
-        }
-        return pathsArr;
+    }
+    XMLParser.prototype.mapAttributes = function (attributes) {
+        var attributesMap = new Map();
+        attributes.forEach(function (attr) {
+            var _a = attr.split('='), key = _a[0], value = _a[1];
+            if (!value)
+                return;
+            attributesMap.set(key, value.replace(/"/g, '').trim());
+        });
+        return attributesMap;
     };
-    /**
-     * Returns text in each table cell mapped by `${table}.${row}.${col}`
-     */
-    XMLParser.prototype.getTableTextMap = function (params) {
-        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"),
-            "".concat(parentPath, ".table.tfoot.tr"),
-            "".concat(parentPath, ".table.tr"),
-        ]);
-        var colPaths = new Set([
-            "".concat(parentPath, ".table.tbody.tr.td"),
-            "".concat(parentPath, ".table.thead.tr.td"),
-            "".concat(parentPath, ".table.tfoot.tr.td"),
-            "".concat(parentPath, ".table.tr.td"),
-            "".concat(parentPath, ".table.tbody.tr.th"),
-            "".concat(parentPath, ".table.thead.tr.th"),
-            "".concat(parentPath, ".table.tfoot.tr.th"),
-            "".concat(parentPath, ".table.tr.th"),
-        ]);
-        var table = 0;
-        var row = 0;
-        var col = 0;
-        var textByColKey = new Map();
-        var spaceChars = new Set(['\n', '\r', '\t']);
+    XMLParser.prototype.parse = function (xml) {
+        var _this = this;
+        var currentObj = {};
+        var objPath = [currentObj];
         this.iterateXML({
             xml: xml,
-            onOpenTag: function (data) {
-                var _a;
-                var path = data.path;
-                var colKey = "".concat(table, ".").concat(row, ".").concat(col);
-                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);
-                }
-                var isTable = parentPath ? pathLower === "".concat(parentPath, ".table") : pathLower.endsWith('table');
-                var isRow = parentPath ? rowPaths.has(pathLower) : pathLower.endsWith('tr');
-                var isCol = parentPath
-                    ? colPaths.has(pathLower)
-                    : pathLower.endsWith('td') || pathLower.endsWith('th');
-                if (isTable) {
-                    table++;
-                    col = 0;
-                    row = 0;
+            onOpenTag: function (tagName, attributes, isSelfEnclosing) {
+                var newObj = {};
+                var obj = currentObj;
+                var isComment = tagName.startsWith('!--');
+                if (isComment || tagName === '?xml')
+                    return;
+                if (obj[tagName] === undefined) {
+                    obj[tagName] = newObj;
                 }
-                else if (isRow) {
-                    row++;
-                    col = 0;
+                else if (Array.isArray(obj[tagName])) {
+                    ;
+                    obj[tagName].push(newObj);
                 }
-                else if (isCol) {
-                    col++;
+                else if (typeof obj[tagName] === 'object') {
+                    obj[tagName] = [obj[tagName], newObj];
                 }
-                onOpenTag === null || onOpenTag === void 0 ? void 0 : onOpenTag(__assign(__assign({}, data), { textMap: textByColKey }));
-            },
-            onCharacter: function (data) {
-                var _a;
-                var char = spaceChars.has(data.char) ? ' ' : data.char;
-                var colKey = "".concat(table, ".").concat(row, ".").concat(col);
-                var textCur = (_a = textByColKey.get(colKey)) !== null && _a !== void 0 ? _a : '';
-                if (!(trimSpaces && char === ' ' && textCur.endsWith(' '))) {
-                    textByColKey.set(colKey, "".concat(textCur).concat(char));
+                _this.mapAttributes(attributes).forEach(function (value, att) {
+                    newObj["@_".concat(att)] = value;
+                });
+                if (!isSelfEnclosing) {
+                    objPath.push(newObj);
+                    currentObj = newObj;
                 }
-                onCharacter === null || onCharacter === void 0 ? void 0 : onCharacter(__assign(__assign({}, data), { textMap: textByColKey }));
             },
-            onCloseTag: function (data) {
-                var _a;
-                var colKey = "".concat(table, ".").concat(row, ".").concat(col);
-                var textCur = (_a = textByColKey.get(colKey)) !== null && _a !== void 0 ? _a : '';
-                if (textCur.trim().length === 0 && col === 0) {
-                    textByColKey.delete(colKey);
-                }
-                else if (!textCur.endsWith(' ')) {
-                    textByColKey.set(colKey, "".concat(textCur, " "));
-                }
-                onCloseTag === null || onCloseTag === void 0 ? void 0 : onCloseTag(__assign(__assign({}, data), { textMap: textByColKey }));
+            onInnerText: function (text) {
+                var textTrimmed = text.trim();
+                if (!textTrimmed)
+                    return;
+                var obj = currentObj;
+                obj['#text'] = textTrimmed;
             },
-        });
-        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 () {
-                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)
+                if (objPath.length === 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;
-                }
+                objPath.pop();
+                currentObj = objPath[objPath.length - 1];
             },
         });
-        documentNode.setText(xml);
-        return documentNode;
+        return objPath[0];
+    };
+    XMLParser.prototype.iterateXML = function (params) {
+        var onCloseTag = params.onCloseTag, onInnerText = params.onInnerText, onOpenTag = params.onOpenTag, xml = params.xml;
+        for (var i = 0; i < xml.length; i++) {
+            if (xml[i] === '<' && xml[i + 1] !== '/') {
+                i++;
+                var tagEndIndex = xml.indexOf('>', i);
+                var currentTagStr = xml.substring(i, tagEndIndex);
+                var tagName = currentTagStr.split(' ', 1)[0].trim();
+                var attributes = currentTagStr.split(' ').slice(1);
+                var lastAttribute = attributes[attributes.length - 1];
+                if (lastAttribute === null || lastAttribute === void 0 ? void 0 : lastAttribute.endsWith('/')) {
+                    attributes[attributes.length - 1] = lastAttribute.substring(0, lastAttribute.length - 1);
+                }
+                if (!lastAttribute || !lastAttribute.includes('=')) {
+                    attributes.pop();
+                }
+                i = tagEndIndex;
+                var isSelfEnclosing = xml[tagEndIndex - 1] === '/' || this.selfEnclosingTags.has(tagName.toLowerCase());
+                onOpenTag === null || onOpenTag === void 0 ? void 0 : onOpenTag(tagName, attributes, isSelfEnclosing);
+            }
+            else if (xml[i] === '<' && xml[i + 1] === '/') {
+                i += 2;
+                var tagEndIndex = xml.indexOf('>', i);
+                var currentTagStr = xml.substring(i, tagEndIndex);
+                i = tagEndIndex;
+                onCloseTag === null || onCloseTag === void 0 ? void 0 : onCloseTag(currentTagStr);
+            }
+            else {
+                var nextOpenTagIndex = xml.indexOf('<', i);
+                var nextIndex = nextOpenTagIndex === -1 ? xml.length : nextOpenTagIndex;
+                var text = xml.substring(i, nextIndex);
+                onInnerText === null || onInnerText === void 0 ? void 0 : onInnerText(text);
+                i = nextIndex - 1;
+            }
+        }
     };
     return XMLParser;
 }());

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

@@ -0,0 +1,43 @@
+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;
+    onCharacter?: (data: OnCharacterData) => void;
+    onOpenTag?: (data: OnCharacterData) => void;
+    onCloseTag?: (data: OnCharacterData) => void;
+}
+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;
+}
+/**
+ * @deprecated use XMLParser
+ */
+export default class XMLParserLegacy {
+    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 {};