Merge pull request #8 from translationCoreApps/feature-mcleanb-5051.2

Feature 5051/Fix edge cases for merge and unmerge

.gitignore

@@ -59,3 +59,6 @@ typings/
 
 # IntelliJ config
 .idea
+
+# compiler output
+lib

__tests__/AlignmentHelpers.test.js

@@ -1,46 +1,11 @@
+/* eslint-disable no-use-before-define */
 /* eslint-env jest */
 import fs from 'fs-extra';
 import path from 'path';
+import usfmjs from 'usfm-js';
 jest.unmock('fs-extra');
 import wordaligner, {VerseObjectUtils} from '../src/';
 const RESOURCES = path.join('__tests__', 'fixtures', 'pivotAlignmentVerseObjects');
-/**
- * Reads a usfm file from the resources dir
- * @param {string} filename relative path to usfm file
- * @return {Object} - The read JSON object
- */
-const readJSON = filename => {
-  const fullPath = path.join(RESOURCES, filename);
-  if (fs.existsSync(fullPath)) {
-    const json = fs.readJsonSync(fullPath);
-    return json;
-  }
-  console.log('File not found.');
-  return false;
-};
-
-/**
- * Generator for testing merging of alignment into verseObjects
- * @param {string} name - the name of the test files to use. e.g. `valid` will test `valid.usfm` to `valid.json`
- */
-const mergeTest = (name = {}) => {
-  const json = readJSON(`${name}.json`);
-  expect(json).toBeTruthy();
-  const {alignment, verseObjects, verseString, wordBank} = json;
-  const output = wordaligner.merge(alignment, wordBank, verseString);
-  expect(output).toEqual(verseObjects);
-};
-/**
- * Generator for testing unmerging of alignment from verseObjects
- * @param {string} name - the name of the test files to use. e.g. `valid` will test `valid.usfm` to `valid.json`
- */
-const unmergeTest = (name = {}) => {
-  const json = readJSON(`${name}.json`);
-  expect(json).toBeTruthy();
-  const {verseObjects, alignment, wordBank, alignedVerseString} = json;
-  const output = wordaligner.unmerge(verseObjects, alignedVerseString);
-  expect(output).toEqual({alignment, wordBank});
-};
 
 describe("Merge Alignment into Verse Objects", () => {
   it('handles one to one', () => {
@@ -102,6 +67,9 @@ describe("Merge Alignment into Verse Objects", () => {
     }
     expect(fail).toBeTruthy();
   });
+  it('handles titus 1-12', () => {
+    mergeTest('tit1-12');
+  });
 });
 
 describe("UnMerge Alignment from Verse Objects", () => {
@@ -153,6 +121,18 @@ describe("UnMerge Alignment from Verse Objects", () => {
   it('handles acts 1-4', () => {
     unmergeTest('acts-1-4');
   });
+  it('handles titus 1-12', () => {
+    unmergeTest('tit1-12');
+  });
+});
+
+describe("export USFM3 from Verse Objects", () => {
+  it('handles acts-1-11', () => {
+    exportTest('acts-1-11');
+  });
+  it('handles acts 1-4', () => {
+    exportTest('acts-1-4');
+  });
 });
 
 describe('wordaligner.generateBlankAlignments', () => {
@@ -242,8 +222,90 @@ describe('wordaligner.generateWordBank', () => {
     // then
     expect(results).toEqual(wordBank);
   });
-
-  //
-  // helpers
-  //
 });
+
+//
+// helpers
+//
+
+/**
+ * Reads a json file from the resources dir
+ * @param {string} filename relative path to usfm file
+ * @return {Object} - The read JSON object
+ */
+const readJSON = filename => {
+  const fullPath = path.join(RESOURCES, filename);
+  if (fs.existsSync(fullPath)) {
+    const json = fs.readJsonSync(fullPath);
+    return json;
+  }
+  console.log('File not found.');
+  return false;
+};
+
+/**
+ * Reads a usfm file from the resources dir
+ * @param {string} filename relative path to usfm file
+ * @return {Object} - The read JSON object
+ */
+const readUSFM = filename => {
+  const fullPath = path.join(RESOURCES, filename);
+  if (fs.existsSync(fullPath)) {
+    const usfm = fs.readFileSync(fullPath, 'UTF-8').toString();
+    return usfm;
+  }
+  console.log('File not found.');
+  return false;
+};
+
+/**
+ * Generator for testing merging of alignment into verseObjects
+ * @param {string} name - the name of the test files to use. e.g. `valid` will test `valid.usfm` to `valid.json`
+ */
+const mergeTest = (name = {}) => {
+  const json = readJSON(`${name}.json`);
+  expect(json).toBeTruthy();
+  const {alignment, verseObjects, verseString, wordBank} = json;
+  const output = wordaligner.merge(alignment, wordBank, verseString);
+  expect(output).toEqual(verseObjects);
+};
+
+/**
+ * Generator for testing unmerging of alignment from verseObjects
+ * @param {string} name - the name of the test files to use. e.g. `valid` will test `valid.usfm` to `valid.json`
+ */
+const unmergeTest = (name = {}) => {
+  const json = readJSON(`${name}.json`);
+  expect(json).toBeTruthy();
+  const {verseObjects, alignment, wordBank, alignedVerseString} = json;
+  const output = wordaligner.unmerge(verseObjects, alignedVerseString);
+  expect(output).toEqual({alignment, wordBank});
+};
+
+/**
+ * Generator for testing merging of alignment into verseObjects
+ * @param {string} name - the name of the test files to use. e.g. `valid` will test `valid.usfm` to `valid.json`
+ */
+const exportTest = (name = {}) => {
+  const json = readJSON(`${name}.json`);
+  expect(json).toBeTruthy();
+  const expectedUsfm = readUSFM(`${name}.usfm`);
+  expect(expectedUsfm).toBeTruthy();
+  const {alignment, verseString, wordBank} = json;
+  const output = wordaligner.merge(alignment, wordBank, verseString);
+  const outputData = {
+    chapters: {},
+    headers: [],
+    verses: {
+      1: output
+    }
+  };
+  let usfm = usfmjs.toUSFM(outputData, {chunk: true});
+  const split = usfm.split("\\v 1");
+  usfm = split.length > 1 ? split[1] : "";
+  if (usfm.substr(0, 1) === ' ') {
+    usfm = usfm.substr(1);
+  }
+  expect(usfm).toEqual(expectedUsfm);
+};
+

__tests__/VerseObjectUtils.js

@@ -67,7 +67,7 @@ describe('VerseObjectUtils.getWordsFromVerseObjects', () => {
       occurrence: 1,
       occurrences: 1
     },
-    {type: 'text', text: ','},
+    {type: 'text', text: ', '},
     {
       tag: 'w',
       type: 'word',
@@ -147,7 +147,7 @@ describe("getOrderedVerseObjectsFromString", () => {
       },
       {
         type: "text",
-        text: ","
+        text: ", "
       },
       {
         tag: "w",
@@ -193,7 +193,7 @@ describe("getOrderedVerseObjectsFromString", () => {
       },
       {
         type: "text",
-        text: ","
+        text: ", "
       },
       {
         tag: "w",
@@ -253,7 +253,7 @@ describe("getOrderedVerseObjectsFromString", () => {
       },
       {
         type: "text",
-        text: ","
+        text: ", "
       },
       {
         tag: "w",
@@ -278,7 +278,7 @@ describe("getOrderedVerseObjectsFromString", () => {
       },
       {
         type: "text",
-        text: "."
+        text: ". "
       },
       {
         tag: "f",

__tests__/fixtures/pivotAlignmentVerseObjects/acts-1-11.json

@@ -1,6 +1,6 @@
 {
   "alignedVerseString": "οἳ καὶ εἶπαν, “ ἄνδρες, Γαλιλαῖοι, τί ἑστήκατε βλέποντες εἰς τὸν οὐρανόν? οὗτος ὁ Ἰησοῦς ὁ ἀναλημφθεὶς ἀφ’ ὑμῶν εἰς τὸν οὐρανὸν, οὕτως ἐλεύσεται ὃν τρόπον ἐθεάσασθε αὐτὸν πορευόμενον εἰς τὸν οὐρανόν.”",
-  "verseString": "And they said,\"Men of Galilee, why do you stand looking into heaven? This Jesus who has been taken up from you into heaven, will likewise return in the same manner as you saw him going into heaven.\"\\p",
+  "verseString": "And they said,\"Men of Galilee, why do you stand looking into heaven? This Jesus who has been taken up from you into heaven, will likewise return in the same manner as you saw him going into heaven.\"\n\\p",
   "alignment": [
     {
       "topWords": [
@@ -709,7 +709,7 @@
       "type": "milestone"
     },
     {
-      "text": ",",
+      "text": ", ",
       "type": "text"
     },
     {
@@ -834,7 +834,7 @@
       "type": "milestone"
     },
     {
-      "text": "?",
+      "text": "? ",
       "type": "text"
     },
     {
@@ -1035,7 +1035,7 @@
       "type": "milestone"
     },
     {
-      "text": ",",
+      "text": ", ",
       "type": "text"
     },
     {
@@ -1273,7 +1273,7 @@
       "type": "text"
     },
     {
-      "text": "\"",
+      "text": "\"\n",
       "type": "text"
     },
     {

__tests__/fixtures/pivotAlignmentVerseObjects/acts-1-11.usfm

@@ -0,0 +1,104 @@
+
+\zaln-s | x-strong="G25320" x-lemma="καί" x-morph="Gr,D,,,,,,,,," x-occurrence="1" x-occurrences="1" x-content="καὶ"
+\w And|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G37390" x-lemma="ὅς" x-morph="Gr,RR,,,,NMP," x-occurrence="1" x-occurrences="1" x-content="οἳ"
+\w they|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G30040" x-lemma="λέγω" x-morph="Gr,V,IAA3,,P," x-occurrence="1" x-occurrences="1" x-content="εἶπαν"
+\w said|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*,"
+\zaln-s | x-strong="G04350" x-lemma="ἀνήρ" x-morph="Gr,N,,,,,VMP," x-occurrence="1" x-occurrences="1" x-content="ἄνδρες"
+\w Men|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G10570" x-lemma="Γαλιλαῖος" x-morph="Gr,NS,,,,VMP," x-occurrence="1" x-occurrences="1" x-content="Γαλιλαῖοι"
+\w of|x-occurrence="1" x-occurrences="1"\w*
+\w Galilee|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*, 
+\zaln-s | x-strong="G51010" x-lemma="τίς" x-morph="Gr,RT,,,,ANS," x-occurrence="1" x-occurrences="1" x-content="τί"
+\w why|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G24760" x-lemma="ἵστημι" x-morph="Gr,V,IEA2,,P," x-occurrence="1" x-occurrences="1" x-content="ἑστήκατε"
+\w do|x-occurrence="1" x-occurrences="1"\w*
+\w you|x-occurrence="1" x-occurrences="3"\w*
+\w stand|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G09910" x-lemma="βλέπω" x-morph="Gr,V,PPA,NMP," x-occurrence="1" x-occurrences="1" x-content="βλέποντες"
+\w looking|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G15190" x-lemma="εἰς" x-morph="Gr,P,,,,,A,,," x-occurrence="1" x-occurrences="3" x-content="εἰς"
+\w into|x-occurrence="1" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-s | x-strong="G35880" x-lemma="ὁ" x-morph="Gr,EA,,,,AMS," x-occurrence="1" x-occurrences="3" x-content="τὸν"
+\zaln-s | x-strong="G37720" x-lemma="οὐρανός" x-morph="Gr,N,,,,,AMS," x-occurrence="1" x-occurrences="2" x-content="οὐρανόν"
+\w heaven|x-occurrence="1" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-e\*? 
+\zaln-s | x-strong="G37780" x-lemma="οὗτος" x-morph="Gr,ED,,,,NMS," x-occurrence="1" x-occurrences="1" x-content="οὗτος"
+\w This|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G35880" x-lemma="ὁ" x-morph="Gr,EA,,,,NMS," x-occurrence="1" x-occurrences="2" x-content="ὁ"
+\zaln-s | x-strong="G24240" x-lemma="Ἰησοῦς" x-morph="Gr,N,,,,,NMS," x-occurrence="1" x-occurrences="1" x-content="Ἰησοῦς"
+\w Jesus|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-e\*
+\zaln-s | x-strong="G35880" x-lemma="ὁ" x-morph="Gr,RD,,,,NMS," x-occurrence="2" x-occurrences="2" x-content="ὁ"
+\zaln-s | x-strong="G03530" x-lemma="ἀναλαμβάνω" x-morph="Gr,V,PAP,NMS," x-occurrence="1" x-occurrences="1" x-content="ἀναλημφθεὶς"
+\w who|x-occurrence="1" x-occurrences="1"\w*
+\w has|x-occurrence="1" x-occurrences="1"\w*
+\w been|x-occurrence="1" x-occurrences="1"\w*
+\w taken|x-occurrence="1" x-occurrences="1"\w*
+\w up|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-e\*
+\zaln-s | x-strong="G05750" x-lemma="ἀπό" x-morph="Gr,P,,,,,G,,," x-occurrence="1" x-occurrences="1" x-content="ἀφ’"
+\w from|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G47710" x-lemma="σύ" x-morph="Gr,RP,,,2G,P," x-occurrence="1" x-occurrences="1" x-content="ὑμῶν"
+\w you|x-occurrence="2" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-s | x-strong="G15190" x-lemma="εἰς" x-morph="Gr,P,,,,,A,,," x-occurrence="2" x-occurrences="3" x-content="εἰς"
+\w into|x-occurrence="2" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-s | x-strong="G35880" x-lemma="ὁ" x-morph="Gr,EA,,,,AMS," x-occurrence="2" x-occurrences="3" x-content="τὸν"
+\zaln-s | x-strong="G37720" x-lemma="οὐρανός" x-morph="Gr,N,,,,,AMS," x-occurrence="1" x-occurrences="1" x-content="οὐρανὸν"
+\w heaven|x-occurrence="2" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-e\*, 
+\zaln-s | x-strong="G20640" x-lemma="ἔρχομαι" x-morph="Gr,V,IFM3,,S," x-occurrence="1" x-occurrences="1" x-content="ἐλεύσεται"
+\w will|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G37790" x-lemma="οὕτως" x-morph="Gr,D,,,,,,,,," x-occurrence="1" x-occurrences="1" x-content="οὕτως"
+\w likewise|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G20640" x-lemma="ἔρχομαι" x-morph="Gr,V,IFM3,,S," x-occurrence="1" x-occurrences="1" x-content="ἐλεύσεται"
+\w return|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G37390" x-lemma="ὅς" x-morph="Gr,ER,,,,AMS," x-occurrence="1" x-occurrences="1" x-content="ὃν"
+\w in|x-occurrence="1" x-occurrences="1"\w*
+\w the|x-occurrence="1" x-occurrences="1"\w*
+\w same|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G51580" x-lemma="τρόπος" x-morph="Gr,N,,,,,AMS," x-occurrence="1" x-occurrences="1" x-content="τρόπον"
+\w manner|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G23000" x-lemma="θεάομαι" x-morph="Gr,V,IAM2,,P," x-occurrence="1" x-occurrences="1" x-content="ἐθεάσασθε"
+\w as|x-occurrence="1" x-occurrences="1"\w*
+\w you|x-occurrence="3" x-occurrences="3"\w*
+\w saw|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G08460" x-lemma="αὐτός" x-morph="Gr,RP,,,3AMS," x-occurrence="1" x-occurrences="1" x-content="αὐτὸν"
+\w him|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G41980" x-lemma="πορεύω" x-morph="Gr,V,PPM,AMS," x-occurrence="1" x-occurrences="1" x-content="πορευόμενον"
+\w going|x-occurrence="1" x-occurrences="1"\w*
+\zaln-e\*
+\zaln-s | x-strong="G15190" x-lemma="εἰς" x-morph="Gr,P,,,,,A,,," x-occurrence="3" x-occurrences="3" x-content="εἰς"
+\w into|x-occurrence="3" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-s | x-strong="G35880" x-lemma="ὁ" x-morph="Gr,EA,,,,AMS," x-occurrence="3" x-occurrences="3" x-content="τὸν"
+\zaln-s | x-strong="G37720" x-lemma="οὐρανός" x-morph="Gr,N,,,,,AMS," x-occurrence="2" x-occurrences="2" x-content="οὐρανόν"
+\w heaven|x-occurrence="3" x-occurrences="3"\w*
+\zaln-e\*
+\zaln-e\*."
+\p