Bugfix #7490/Fix broken ampersand (#70)

* added unit tests to check behavior of original quote parsing.

* fixed issue that flattening tWords tags in original language loses space.
added unit tests to check behavior of original quote parsing.
moved code from tc-source-content-updater
export wordOccurrenceHelpers

* clean up

* add more test cases

__tests__/__snapshots__/tsvToGroupdataSingle.test.js.snap

@@ -0,0 +1,367 @@
+// Jest Snapshot v1, https://goo.gl/fbAQLP
+
+exports[`tsvToGroupData() - testing quotes Test -1 occurrence 1`] = `"καὶ"`;
+
+exports[`tsvToGroupData() - testing quotes Test all occurrences 1`] = `"καὶ"`;
+
+exports[`tsvToGroupData() - testing quotes Test fifth occurrence 1`] = `"καὶ"`;
+
+exports[`tsvToGroupData() - testing quotes Test multiple ellipsis 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "καὶ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test multiple ellipsis 2 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 3,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 3,
+ "word": "καὶ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test multiple ellipsis 3 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 3,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 3,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 4,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 4,
+ "word": "καὶ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test multiple ellipsis and phrases 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "τοῖς",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καθαροῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "ἀπίστοις",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 3,
+ "word": "τοῖς",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "καθαροῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "καὶ",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "ἀπίστοις",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test second occurrence 1`] = `"καὶ"`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "καθαροῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "τοῖς",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand 2 1`] = `
+Array [
+ Object {
+ "occurrence": 2,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand 3 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "ἐν",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand with tWord tag 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "ἐν",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand with verse range 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 3,
+ "word": "ἐν",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand with verse range 2 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "δοῦλοι",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "Ἰησοῦ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand with verse range 3 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "Χριστῷ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "Ἰησοῦ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand with verse range 4 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "ἡμῶν",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "Χριστοῦ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ampersand with verse range and tWord tag 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "ἐν",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "Χριστῷ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test simple ellipsis 1`] = `
+Array [
+ Object {
+ "occurrence": 2,
+ "word": "τοῖς",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 1,
+ "word": "καὶ",
+ },
+]
+`;
+
+exports[`tsvToGroupData() - testing quotes Test three dot with verse range 1`] = `
+Array [
+ Object {
+ "occurrence": 1,
+ "word": "ἡμῶν",
+ },
+ Object {
+ "word": "…",
+ },
+ Object {
+ "occurrence": 2,
+ "word": "Χριστοῦ",
+ },
+]
+`;