Optimize span and anchor parsing

evt-project · Dec 28, 2023 · 2280610 · 2280610
1 parent ee0fd42
commit 2280610
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 66 deletions.
diff --git a/src/app/services/xml-parsers/basic-parsers.ts b/src/app/services/xml-parsers/basic-parsers.ts
@@ -4,8 +4,8 @@ import {
     Addition, Analogue, Anchor, Attributes, Damage, Deletion, Gap, GenericElement, Lb, Milestone, Note, NoteLayout,
     Paragraph, PlacementType, Ptr, QuoteEntry, Span, SpanGrp, Supplied, Term, Text, Verse, VersesGroup, Word, XMLElement,
 } from '../../models/evt-models';
-import { isNestedInElem, xpath } from '../../utils/dom-utils';
-import { getContentBetweenElementAndId, getExternalElements, isAnalogue, isSource, replaceMultispaces } from '../../utils/xml-utils';
+import { getElementsBetweenTreeNode, isNestedInElem, xpath } from '../../utils/dom-utils';
+import { getExternalElements, isAnalogue, isSource, replaceMultispaces } from '../../utils/xml-utils';
 import { createParser, getClass, getDefaultN, getID, parseChildren, ParseFn, Parser } from './parser-models';
 import { AppConfig } from 'src/app/app.config';
 import { AnalogueParser } from './analogue-parser';
@@ -429,15 +429,17 @@ export class TermParser extends GenericElemParser implements Parser<XMLElement>
 export class MilestoneParser extends GenericElemParser implements Parser<XMLElement> {
     parse(xml: XMLElement): Milestone {
 
-        const elements = getContentBetweenElementAndId(xml, xml.getAttribute('spanTo'));
-        const parsedElements = elements.elements.map((x) => super.parse(x));
+        const endElement = (xml.getAttribute('spanTo')) ? getExternalElements(xml, ['spanTo'], 'xml:id', 'anchor') : [];
+        const includedElements = (endElement.length !== 0) ? getElementsBetweenTreeNode(xml, endElement[0]) : [];
+        const parsedElements = (includedElements.length !== 0) ?
+            includedElements.map((x: XMLElement) => (x.nodeType !== 3 && x.nodeType !== 8) ? super.parse(x) : x) : [];
 
         return {
             type: Milestone,
             id: xml.getAttribute('xml:id'),
             attributes: this.attributeParser.parse(xml),
             unit: xml.getAttribute('unit'),
-            spanText: elements.text,
+            spanText: '',
             spanElements: parsedElements,
             content: parseChildren(xml, this.genericParse),
         };
@@ -481,21 +483,18 @@ export class SpanParser extends GenericElemParser implements Parser<XMLElement>
             }
 
         } else if (xml.tagName === 'span') {
-            let included = { text: '', elements: [] };
-            let parsedElements = [];
-            const startingElement = getExternalElements(xml, ['from'], 'xml:id', 'anchor');
-            if (startingElement.length > 0) {
-                included = getContentBetweenElementAndId(startingElement[0], xml.getAttribute('to'));
-                parsedElements = included.elements.map((x) => super.parse(x));
-            }
+            const endElement = (xml.getAttribute('spanTo')) ? getExternalElements(xml, ['from'], 'xml:id', 'anchor') : [];
+            const includedElements = (endElement.length !== 0) ? getElementsBetweenTreeNode(xml, endElement[0]) : [];
+            const parsedElements = (includedElements.length !== 0) ?
+                includedElements.map((x: XMLElement) => (x.nodeType !== 3 && x.nodeType !== 8) ? super.parse(x) : x) : [];
 
             return <Span> {
                 type: Span,
                 id: xml.getAttribute('xml:id'),
                 attributes: this.attributeParser.parse(xml),
                 from: xml.getAttribute('from'),
                 to: xml.getAttribute('to'),
-                includedText: included.text,
+                includedText: '',
                 includedElements: parsedElements,
                 content: parseChildren(xml, this.genericParse),
             };

diff --git a/src/app/utils/xml-utils.ts b/src/app/utils/xml-utils.ts
@@ -107,56 +107,3 @@ export function isSource(elem: XMLElement, attrs: string[]): boolean {
 
   return (validAttrs);
 }
-
-
-/**
- * Retrieve textContent and elements between the provided element and the one found with the xml:id provided
- */
-export function getContentBetweenElementAndId(fromElement: XMLElement, toXMLID: string): { text: string, elements: any } {
-
-  if ((fromElement === null) || (fromElement === undefined) || (toXMLID === null)) {
-    return { text: '', elements: [] };
-  }
-
-  const cleanID = toXMLID.replace('#','');
-  let found = false;
-  // text after the milestone but still inside the parent element
-  let foundText = (fromElement.nextSibling !== null) ? fromElement.nextSibling.textContent : '';
-  // the milestone is always inside another element?
-  // otherwise const foundElements = [fromElement.nextSibling];
-  let next = (fromElement.nextElementSibling !== null) ?
-    fromElement.nextElementSibling as XMLElement :
-    fromElement.parentElement.nextElementSibling as XMLElement;
-
-  // creating a fake element for partial text included from milestone on
-  let foundElements = [];
-  if (fromElement.parentElement.nextElementSibling !== null) {
-    const nextEdited = fromElement.parentElement.nextElementSibling.cloneNode(true);
-    nextEdited.textContent = foundText;
-    foundElements.push(nextEdited);
-  }
-
-  let maxExec = 50;
-
-  while(!found && next !== null && maxExec !== 0) {
-    foundElements.push(next);
-    foundText = foundText + next.textContent;
-    if (next.getAttribute('xml:id') === cleanID) {
-      found = true;
-    } else {
-      maxExec--;
-      if (next.nextElementSibling === null) {
-        if (next.parentElement !== null) {
-          next = next.parentElement.nextElementSibling as XMLElement;
-          maxExec--;
-        } else {
-          next = null;
-        }
-      } else {
-        next = next.nextElementSibling as XMLElement;
-      }
-    }
-  }
-
-  return { 'text': foundText, 'elements': foundElements };
-}