fix: Fix CAPS sentense splitting

__tests__/rules/abbreviations.ts

@@ -123,32 +123,28 @@ describe('pairAbbreviation', () => {
 });
 
 describe('leftPairsTailAbbreviation', () => {
-    it('evaluates to true if left is tail of the pair abbreviation', () => {
-        const go = compose(reduce(and, true), map(leftPairsTailAbbreviation));
-        const generateInput = compose(
-            unnest,
-            juxt([keysUpperWithRightUpper, keysSpacesWithRightLower]),
-            keys,
-        );
-        const input = [
-            ...generateInput(HEAD_PAIR),
-            ...generateInput(TAIL_PAIR),
-            ...generateInput(OTHER_PAIR),
-        ];
-        const expected = true;
-        const actual = go(input);
-        expect(actual).toBe(expected);
-    });
+    const test = (expected: boolean) =>  (pair: string[]) => {
+        it('handles "' + pair.join('') + '"', () => {
+            expect(leftPairsTailAbbreviation(pair)).toBe(expected);
+        });
+    }
 
-    it('evaluates to false otherwise', () => {
-        const go = compose(reduce(or, false), map(leftPairsTailAbbreviation));
-        const input = [
-            ['фоо.бар. ', 'амбар'],
-            ['not.an.abbr.', ' not happening'],
-            ['not. an. abbr.', ' not happening'],
-        ];
-        const expected = false;
-        const actual = go(input);
-        expect(actual).toBe(expected);
-    });
+    const generateInput = compose(
+        unnest,
+        juxt([keysUpperWithRightUpper, keysSpacesWithRightLower]),
+        keys,
+    );
+
+    [
+        ...generateInput(HEAD_PAIR),
+        ...generateInput(TAIL_PAIR),
+        ...generateInput(OTHER_PAIR),
+    ].forEach(test(true));
+
+    [
+        ['фоо.бар. ', 'амбар'],
+        ['not.an.abbr.', ' not happening'],
+        ['not. an. abbr.', ' not happening'],
+        ['и т.п.', ' В очереди'],
+    ].forEach(test(false));
 });

src/rules/abbreviations.ts

@@ -12,6 +12,8 @@ import {
     anyPass,
     toLower,
     Pred,
+    length,
+    lt,
 } from 'ramda';
 
 import {
@@ -98,8 +100,10 @@ export const leftAbbreviation = compose(
     fst,
 );
 
+const isCaps = allPass([isUpper, compose(lt(1), length)]);
+
 // right join condition is to be uppercase or lowercase word
-const rightLowercaseOrCaps = compose(anyPass([startsWithLower, isUpper]), fstWord, snd);
+const rightLowercaseOrCaps = compose(anyPass([startsWithLower, isCaps]), fstWord, snd);
 
 // portion of the source <s> before target <t>
 const before = (s: string) => (t: string) => s.slice(0, Math.max(s.indexOf(t), 0));

src/rules/base.ts

@@ -1,12 +1,27 @@
-import { call, zipWith, compose, map, all, not, always, Pred, identity, allPass } from 'ramda';
+import {
+    call,
+    zipWith,
+    compose,
+    map,
+    all,
+    not,
+    always,
+    Pred,
+    identity,
+    allPass,
+    equals,
+    length,
+    juxt,
+    toUpper
+} from 'ramda';
 
 import {
     startsWithLower,
     startsWithUpper,
     startsWithNewline,
     startsWithHardbreak,
     endsWithHardbreak,
-    lengthNonZero,
+    lengthNonZero, allEqual, hasAlpha,
 } from '../utilities';
 import {
     fstToken,
@@ -17,9 +32,19 @@ import {
     quotationClosePrefix,
     bracketsClosePrefix,
     delimiterPrefix,
-    spaces,
+    spaces, dotSuffix, lstWord,
 } from '../parsers';
 
+// determine if left is part of the initials
+// conditions:
+//     * left delimiter is dot
+//     * left last word is single letter
+//     * left last word is in upper case
+//     * left has alpha characters
+const isLeftDotDelimiter = compose(lengthNonZero, dotSuffix);
+const isLeftSingleLetter = compose(equals(1), length, lstWord);
+const isLeftUpper = compose(allEqual, juxt([toUpper, identity]), lstWord);
+const leftHasAlpha = compose(hasAlpha, lstWord);
 const isSpaceSuffix = compose(lengthNonZero, spaceSuffix);
 const isSpacePrefix = compose(lengthNonZero, spacePrefix);
 
@@ -72,4 +97,6 @@ export const leftEndsWithHardbreak = rule('leftEndsWithHardbreak', [endsWithHard
 
 export const rightStartsWithHardbreak = rule('rightStartsWithHardbreak', [_, startsWithHardbreak]);
 
-export const rightStartsNewlineUppercased = rule('rightStartsNewlineUppercased', [_, allPass([startsWithNewline, startsWithUpper])]);
+export const rightStartsNewlineUppercased = rule('rightStartsNewlineUppercased', [_, allPass([startsWithNewline, startsWithUpper])]);
+
+export const leftInitials = rule('leftInitials', [allPass([isLeftDotDelimiter, isLeftSingleLetter, isLeftUpper, leftHasAlpha]), _]);

src/rules/index.ts

@@ -1,3 +1,2 @@
 export * from './base';
-export * from './initials';
 export * from './abbreviations';