[incubator-nlpcraft] branch NLPCRAFT-520 updated: WIP.

[sergeykamov]

This is an automated email from the ASF dual-hosted git repository.

sergeykamov pushed a commit to branch NLPCRAFT-520
in repository https://gitbox.apache.org/repos/asf/incubator-nlpcraft.git


The following commit(s) were added to refs/heads/NLPCRAFT-520 by this push:
     new 77c1d91e WIP.
77c1d91e is described below

commit 77c1d91e4faf9cfa3340d19a2b29b5632d1bec1a
Author: Sergey Kamov <skhem...@gmail.com>
AuthorDate: Fri Dec 16 16:44:39 2022 +0400

    WIP.
---
 .../nlp/enrichers/NCQuotesTokenEnricher.scala      | 29 +++++++++++-----------
 1 file changed, 14 insertions(+), 15 deletions(-)

diff --git 
a/nlpcraft/src/main/scala/org/apache/nlpcraft/nlp/enrichers/NCQuotesTokenEnricher.scala
 
b/nlpcraft/src/main/scala/org/apache/nlpcraft/nlp/enrichers/NCQuotesTokenEnricher.scala
index 7433c7b6..30cf49af 100644
--- 
a/nlpcraft/src/main/scala/org/apache/nlpcraft/nlp/enrichers/NCQuotesTokenEnricher.scala
+++ 
b/nlpcraft/src/main/scala/org/apache/nlpcraft/nlp/enrichers/NCQuotesTokenEnricher.scala
@@ -25,8 +25,6 @@ import scala.collection.*
   * Companion helper.
   */
 object NCQuotesTokenEnricher:
-    private val PROP = "quoted"
-
     private case class Range(from: Int, to: Int):
         def in(idx: Int): Boolean = idx >= from && idx <= to
 
@@ -55,32 +53,33 @@ class NCQuotesTokenEnricher extends NCTokenEnricher with 
LazyLogging:
     //noinspection DuplicatedCode
     /** @inheritdoc */
     override def enrich(req: NCRequest, cfg: NCModelConfig, toks: 
List[NCToken]): Unit =
-        def markAllNot(invalidState: Boolean): Unit =
-            if invalidState then logger.warn(s"Detected invalid quotes in: 
${req.getText}")
-            toks.foreach(_.put(PROP, false))
+        def mark(get: NCToken => Boolean): Unit = for (t <- toks) 
t.put("quoted", get(t))
+        def markFalse(invalid: Boolean): Unit =
+            if invalid then logger.warn(s"Detected invalid quotes in: 
${req.getText}")
+            mark(_ => false)
 
         val quotes = toks.filter(isQuote)
 
         if quotes.isEmpty then
-            markAllNot(false)
+            markFalse(false)
         else if quotes.length % 2 != 0 then
-            markAllNot(true)
+            markFalse(true)
         else
-            val quotedRanges = mutable.HashSet.empty[Range]
+            val ranges = mutable.HashSet.empty[Range]
             val stack = mutable.Stack.empty[NCToken]
 
-            for (quote <- quotes)
+            for (q <- quotes)
                 if stack.nonEmpty then
                     val top = stack.top
-                    if top.getText == QUOTES_REVERSED.getOrElse(quote.getText, 
null) then
-                        quotedRanges += Range(top.getIndex + 1, quote.getIndex 
- 1)
+                    if top.getText == QUOTES_REVERSED.getOrElse(q.getText, 
null) then
+                        ranges += Range(top.getIndex + 1, q.getIndex - 1)
                         stack.pop()
                     else
-                        stack.push(quote)
+                        stack.push(q)
                 else
-                    stack.push(quote)
+                    stack.push(q)
 
             if stack.isEmpty then
-                toks.foreach(t => t.put(PROP, 
quotedRanges.exists(_.in(t.getIndex))))
+                mark(t => ranges.exists(_.in(t.getIndex)))
             else
-                markAllNot(true)
\ No newline at end of file
+                markFalse(true)
\ No newline at end of file