Merge pull request #1984 from Sefaria/bug/sc-28448/new-editor-inconsi…

…stency-verse-numbers

Bug/sc 28448/new editor inconsistency verse numbers

sefaria/sheetsUtils.js

@@ -0,0 +1,69 @@
+import Sefaria from "../static/js/sefaria/sefaria";
+
+export async function getSegmentObjs(refs) {
+    /*
+    Given an array of ref-strings (could also be ranged refs),
+    turn each ref to segment object and return an array of all segments
+    */
+    const segments = [];
+
+    for (const ref of refs) {
+      const text = await Sefaria.getText(ref, { stripItags: 1 });
+      const newSegments = Sefaria.makeSegments(text, false);
+      segments.push(...newSegments);
+    }
+    return segments;
+}
+export async function getNormalRef(ref) {
+    /*
+    Given a ref-string, get he and en normal ref-string
+    */
+    const refObj = await Sefaria.getRef(ref);
+    return {en: refObj.ref, he: refObj.heRef}
+}
+function placedSegmentMapper(lang, segmented, includeNumbers, s) {
+    /*
+    Map each segment object to a formatted text string
+    */
+    if (!s[lang]) {return ""}
+
+    let numStr = "";
+    if (includeNumbers) {
+        const num = (lang=="he") ? Sefaria.hebrew.encodeHebrewNumeral(s.number) : s.number;
+        numStr = "<small>(" + num + ")</small> ";
+    }
+    let str = "<span class='segment'>" + numStr + s[lang] + "</span> ";
+    if (segmented) {
+        str = "<p>" + str + "</p>";
+    }
+    str = str.replace(/(<br\/>)+/g, ' ')
+    return str;
+}
+export const segmentsToSourceText = (segments, lan) => {
+    /*
+    Turn array of segment objects into one chunk of formatted text
+    */
+    const segmented = shouldBeSegmented(segments[0].ref);
+    const includeNumbers = shouldIncludeSegmentNums(segments[0].ref);
+    return(segments.map(placedSegmentMapper.bind(this, lan, segmented, includeNumbers))
+    .filter(Boolean)
+    .join(""));
+}
+function shouldIncludeSegmentNums(ref){
+    /*
+    Decide if segment of this ref should have segment numbers when turned into text chunk
+    */
+    const indexTitle = Sefaria.refIndexTitle(ref);
+    const categories =  Sefaria.refCategories(ref);
+    if (categories.includes("Talmud")) {return false}
+    if (indexTitle === "Pesach Haggadah") {return false}
+    if (categories === 1) {return false}
+    return true;
+}
+function shouldBeSegmented(ref){
+    /*
+    Decide if segment of this ref should be followed by new line when turned into text chunk
+    */
+    const categories =  Sefaria.refCategories(ref);
+    return !(categories[0] in {"Tanakh": 1, "Talmud": 1});
+}

static/js/AddToSourceSheet.jsx

@@ -12,6 +12,7 @@ import Component from 'react-class';
 import sanitizeHtml  from 'sanitize-html';
 import { SignUpModalKind } from './sefaria/signupModalContent';
 import { GDocAdvertBox } from './Promotions';
+import * as sheetsUtils from '../../sefaria/sheetsUtils'
 
 
 
@@ -89,7 +90,34 @@ class AddToSourceSheetBox extends Component {
       }, this.confirmAdd);
     }
   }
-  addToSourceSheet() {
+
+  //return the initial index of the suffix of string1 which also constitutes a prefix for string2
+  longestSuffixPrefixIndex(string1, string2) {
+    let longestSuffixIndex = 0;
+    for (let i = 0; i < string1.length; i++){
+      let suffix = string1.slice(i);
+      if (string2.startsWith(suffix)) {
+        longestSuffixIndex = i;
+      }
+    }
+    return longestSuffixIndex;
+  }
+  //return the final index of the prefix of string1 which also constitutes a suffix for string2
+  longestPrefixSuffixIndex(string1, string2) {
+    let longestPrefixIndex = 0;
+    for (let i = 0; i < string1.length; i++) {
+      let prefix = string1.slice(0, i + 1);
+      if (string2.endsWith(prefix)) {
+        longestPrefixIndex = i + 1;
+      }
+    }
+    return longestPrefixIndex;
+  }
+
+  normalize(text){
+    return(text.replaceAll(/(<br\/>)+/g, ' ').replace(/\u2009/g, ' ').replace(/<[^>]*>/g, ''));
+  }
+  async addToSourceSheet() {
     if (!Sefaria._uid) {
       this.props.toggleSignUpModal(SignUpModalKind.AddToSheet);
     }
@@ -111,15 +139,40 @@ class AddToSourceSheetBox extends Component {
         }
       } else if (this.props.srefs) { //regular use - this is currently the case when the component is loaded in the sidepanel or in the modal component via profiles and notes pages
         source.refs = this.props.srefs;
+
+
+
         const { en, he } = this.props.currVersions ? this.props.currVersions : {"en": null, "he": null}; //the text we are adding may be non-default version
         if (he) { source["version-he"] = he; }
         if (en) { source["version-en"] = en; }
 
         // If something is highlighted and main panel language is not bilingual:
         // Use passed in language to determine which version this highlight covers.
-        var selectedWords = this.props.selectedWords; //if there was highlighted single panel
+        let selectedWords = this.props.selectedWords; //if there was highlighted single panel
         if (selectedWords && language != "bilingual") {
-          source[language.slice(0,2)] = selectedWords;
+          let lan = language.slice(0,2);
+          let segments = await sheetsUtils.getSegmentObjs(source.refs);
+          selectedWords = this.normalize(selectedWords);
+          segments = segments.map(segment => ({
+            ...segment,
+            [lan]: this.normalize(segment[lan])
+          }));
+          for (let iSegment = 0; iSegment < segments.length; iSegment++) {
+              const segment = segments[iSegment];
+              if (iSegment == 0){
+                let criticalIndex = this.longestSuffixPrefixIndex(segment[lan], selectedWords);
+                const ellipse = criticalIndex == 0 ? "" : "...";
+                segment[lan] = ellipse + segment[lan].slice(criticalIndex);
+              }
+              else if (iSegment == segments.length-1){
+                let criticalIndex = this.longestPrefixSuffixIndex(segment[lan], selectedWords);
+                const ellipse = criticalIndex == segment[lan].length-1 ? "" : "...";
+                const chunk = segment[lan].slice(0, criticalIndex)
+                segment[lan] = chunk + ellipse;
+              }
+          }
+
+          source[lan] = sheetsUtils.segmentsToSourceText(segments, lan);
         }
       }
       if (this.checkContentForImages(source.refs)) {

static/js/Editor.jsx

@@ -5,6 +5,8 @@ import {Editor, createEditor, Range, Node, Transforms, Path, Text, Point, Elemen
 import {Slate, Editable, ReactEditor, withReact, useSlate, useSelected, useFocused} from 'slate-react'
 import isHotkey from 'is-hotkey'
 import Sefaria from './sefaria/sefaria';
+import * as sheetsUtils from '../../sefaria/sheetsUtils'
+
 
 import {
     SheetMetaDataBox,
@@ -1866,73 +1868,47 @@ const insertMedia = (editor, mediaUrl) => {
   Transforms.move(editor);
 }
 
-
-function placed_segment_mapper(lang, segmented, includeNumbers, s) {
-    if (!s[lang]) {return ""}
-
-    let numStr = "";
-    if (includeNumbers) {
-        const num = (lang=="he") ? Sefaria.hebrew.encodeHebrewNumeral(s.number) : s.number;
-        numStr = "<small>(" + num + ")</small> ";
-    }
-    let str = "<span class='segment'>" + numStr + s[lang] + "</span> ";
-    if (segmented) {
-        str = "<p>" + str + "</p>";
-    }
-    str = str.replace(/(<br\/>)+/g, ' ')
-    return str;
-}
-
-
-const insertSource = (editor, ref) => {
+const insertSource = async (editor, ref) => {
     const path = editor.selection.anchor.path;
 
     Transforms.setNodes(editor, { loading: true }, {at: path});
 
     const nodeAbove = getNodeAbove(path, editor)
     const nodeBelow = getNodeBelow(path, editor)
 
-    Sefaria.getText(ref, {stripItags: 1}).then(text => {
-        let segments = Sefaria.makeSegments(text, false);
-        segments = Sefaria.stripImagesFromSegments(segments);
-
-        let includeNumbers = $.inArray("Talmud", text.categories) == -1;
-        includeNumbers = text.indexTitle === "Pesach Haggadah" ? false : includeNumbers;
-        const segmented = !(text.categories[0] in {"Tanakh": 1, "Talmud": 1});
-
-        const enText = segments.map(placed_segment_mapper.bind(this, "en", segmented, includeNumbers))
-            .filter(Boolean)
-            .join("");
-        const heText = segments.map(placed_segment_mapper.bind(this, "he", segmented, includeNumbers))
-            .filter(Boolean)
-            .join("");
-
-        let fragment = [{
-                type: "SheetSource",
-                node: editor.children[0].nextNode,
-                ref: text.ref,
-                heRef: text.heRef,
-                heText: parseSheetItemHTML(heText),
-                enText: parseSheetItemHTML(enText),
-                title: null,
-                children: [
-                    {text: ""},
-                ]
-        }];
-
-        if (!(nodeBelow.node && (nodeBelow.node.type == "SheetOutsideText" || nodeBelow.node.type == "paragraph" ) )) {
-          fragment.push({type: 'spacer', children: [{text: ""}]})
-        }
-        Transforms.setNodes(editor, { loading: false }, { at: path });
-        addItemToSheet(editor, fragment);
-        checkAndFixDuplicateSheetNodeNumbers(editor)
-        if (nodeAbove.node && (nodeAbove.node.type == "SheetOutsideText" || nodeAbove.node.type == "paragraph" ) ) {
-          Transforms.delete(editor, {at: path})
-        }
-
-
-        Transforms.move(editor, { unit: 'block', distance: 1 })
-    });
+    const {en: normalEnRef, he: normalHeRef} = await sheetsUtils.getNormalRef(ref);
+
+    let segments = await sheetsUtils.getSegmentObjs([ref])
+
+    const enText = sheetsUtils.segmentsToSourceText(segments, 'en');
+
+    const heText = sheetsUtils.segmentsToSourceText(segments, 'he');
+
+    let fragment = [{
+            type: "SheetSource",
+            node: editor.children[0].nextNode,
+            ref: normalEnRef,
+            heRef: normalHeRef,
+            heText: parseSheetItemHTML(heText),
+            enText: parseSheetItemHTML(enText),
+            title: null,
+            children: [
+                {text: ""},
+            ]
+    }];
+
+    if (!(nodeBelow.node && (nodeBelow.node.type == "SheetOutsideText" || nodeBelow.node.type == "paragraph" ) )) {
+      fragment.push({type: 'spacer', children: [{text: ""}]})
+    }
+    Transforms.setNodes(editor, { loading: false }, { at: path });
+    addItemToSheet(editor, fragment);
+    checkAndFixDuplicateSheetNodeNumbers(editor)
+    if (nodeAbove.node && (nodeAbove.node.type == "SheetOutsideText" || nodeAbove.node.type == "paragraph" ) ) {
+      Transforms.delete(editor, {at: path})
+    }
+
+
+    Transforms.move(editor, { unit: 'block', distance: 1 })
 };
 
 

static/js/sefaria/sefaria.js

@@ -274,6 +274,12 @@ Sefaria = extend(Sefaria, {
     let index = Sefaria.index(pRef.index);
     return index && index.categories ? index.categories : [];
   },
+  refIndexTitle: function(ref) {
+    let pRef = Sefaria.parseRef(ref);
+    if ("error" in pRef) { return null; }
+    let index = Sefaria.index(pRef.index);
+    return index?.title
+  },
   sectionRef: function(ref, deriveIfNotFound=false) {
     // Returns the section level ref for `ref` or null if no data is available
     const oref = this.getRefFromCache(ref);