From 90a388afa9e3314a59ae6f20c1eb47b03faa5ade Mon Sep 17 00:00:00 2001
From: spencer kelly <spencermountain@gmail.com>
Date: Tue, 23 Jan 2024 16:26:03 -0500
Subject: [PATCH] fix clauses issues

---
 scratch.js                                      |  8 +++-----
 src/1-one/match/methods/termMethods.js          |  4 ++--
 src/3-three/chunker/api/clauses.js              | 17 ++++++++++-------
 .../clause.ignore.js => three/clause.test.js}   |  2 +-
 4 files changed, 16 insertions(+), 15 deletions(-)
 rename tests/{_ignore/clause.ignore.js => three/clause.test.js} (98%)

diff --git a/scratch.js b/scratch.js
index 118aafe21..2c8c7c875 100644
--- a/scratch.js
+++ b/scratch.js
@@ -6,11 +6,9 @@ import nlp from './src/three.js'
 
 // let doc = nlp('one two three four five. one three four')
 // doc.before('three four').debug()
-
-let m = nlp("i said, 'did you have to do that' and then left, like nothing happened (which it didn't).")
-  .clauses()
-  .debug()
-console.log(m.eq(1).text())
+let doc = nlp(`...and my butt smells, and i like to kiss my own butt`)
+doc.match('@hasEllipses').debug()
+// let m = doc.clauses() //.debug()
 
 let arr = [
   // 'I left the window open for fresh air.',
diff --git a/src/1-one/match/methods/termMethods.js b/src/1-one/match/methods/termMethods.js
index 71df13afd..545f63fea 100644
--- a/src/1-one/match/methods/termMethods.js
+++ b/src/1-one/match/methods/termMethods.js
@@ -11,7 +11,7 @@ const hasDash = / [-–—]{1,3} /
 /** search the term's 'post' punctuation  */
 const hasPost = (term, punct) => term.post.indexOf(punct) !== -1
 /** search the term's 'pre' punctuation  */
-const hasPre = (term, punct) => term.pre.indexOf(punct) !== -1
+// const hasPre = (term, punct) => term.pre.indexOf(punct) !== -1
 
 const methods = {
   /** does it have a quotation symbol?  */
@@ -25,7 +25,7 @@ const methods = {
   /** does it end with a question mark? */
   hasQuestionMark: term => hasPost(term, '?') || hasPost(term, '¿'),
   /** is there a ... at the end? */
-  hasEllipses: term => hasPost(term, '..') || hasPost(term, '…') || hasPre(term, '..') || hasPre(term, '…'),
+  hasEllipses: term => hasPost(term, '..') || hasPost(term, '…'),
   /** is there a semicolon after term word? */
   hasSemicolon: term => hasPost(term, ';'),
   /** is there a colon after term word? */
diff --git a/src/3-three/chunker/api/clauses.js b/src/3-three/chunker/api/clauses.js
index 982d7392f..1a0d2f57a 100644
--- a/src/3-three/chunker/api/clauses.js
+++ b/src/3-three/chunker/api/clauses.js
@@ -10,7 +10,7 @@ const byComma = function (doc) {
     if (m.growRight('. .').wordCount() === 1) {
       return false
     }
-    let more = m.grow(".") // grow by 1 word in either direction
+    let more = m.grow('.') // grow by 1 word in either direction
     more = more.ifNo('@hasComma @hasComma') //fun, cool...
     more = more.ifNo('@hasComma (and|or) .') //cool, and fun
     more = more.ifNo('(#City && @hasComma) #Country') //'toronto, canada'
@@ -70,13 +70,18 @@ const clauses = function (n) {
   found = found.splitBefore('as (though|if)')
   found = found.splitBefore('(til|until)')
 
+  // it is cool but it is ..
+  let m = found.match('#Verb .* [but] .* #Verb', 0)
+  if (m.found) {
+    found = found.splitBefore(m)
+  }
   // it is cool and it is ..
   // let conjunctions = found.if('#Copula #Adjective #Conjunction (#Pronoun|#Determiner) #Verb').match('#Conjunction')
   // found = found.splitBefore(conjunctions)
 
-  // // if it is this then that
-  // let condition = found.if('if .{2,9} then .').match('then')
-  // found = found.splitBefore(condition)
+  // if it is this then that
+  let condition = found.if('if .{2,9} then .').match('then')
+  found = found.splitBefore(condition)
 
   // // misc clause partitions
   // found = found.splitBefore('as well as .')
@@ -87,14 +92,12 @@ const clauses = function (n) {
   // found = found.splitAfter('@hasSemicolon')
   // found = found.splitAfter('@hasDash')
 
-  // // 
+  // //
   // found = found.splitBefore('which (were|are|will)')
 
   // // he said [...]
   // found = found.splitAfter('#Noun (said|say|says)')
 
-
-
   // passive voice verb - '.. which was robbed is empty'
   // let passive = found.match('#Noun (which|that) (was|is) #Adverb? #PastTense #Adverb?')
   // if (passive.found) {
diff --git a/tests/_ignore/clause.ignore.js b/tests/three/clause.test.js
similarity index 98%
rename from tests/_ignore/clause.ignore.js
rename to tests/three/clause.test.js
index a34988eb6..f6edb5101 100644
--- a/tests/_ignore/clause.ignore.js
+++ b/tests/three/clause.test.js
@@ -1,5 +1,5 @@
 import test from 'tape'
-import nlp from '../three/_lib.js'
+import nlp from './_lib.js'
 const here = '[three/clause] '
 
 test('clauses-parentheses:', function (t) {