filters: accelerate text search

hpi-swa-lab · Jun 19, 2024 · db5869b · db5869b
1 parent 5bc9a27
commit db5869b
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 7 deletions.
diff --git a/packages/SqueakInboxTalk.package/TalkTextSearcher.class/instance/filter..st b/packages/SqueakInboxTalk.package/TalkTextSearcher.class/instance/filter..st
@@ -0,0 +1,12 @@
+filtering
+filter: conversations
+	"Overridden for performance."
+
+	| partSearchers |
+	self isActive ifFalse: [^ conversations].
+
+	partSearchers := self partSearchers.
+
+	^ conversations select: [:conversation |
+		partSearchers allSatisfy: [:searcher |
+			(searcher matchesIn: conversation) talkIsEmpty not]]
diff --git a/packages/SqueakInboxTalk.package/TalkTextSearcher.class/instance/partTerms.st b/packages/SqueakInboxTalk.package/TalkTextSearcher.class/instance/partTerms.st
@@ -2,7 +2,7 @@ private
 partTerms
 	"Split the current term into part terms that form a complex term together."
 
-	| patternFull patternMulti patternSingle patternTerm regexFull regexMulti regexSingle termStream |
+	| patternMulti patternSingle regexFull regexMulti regexSingle termStream |
 	(self usesRegex not and: [self hasComplexTerm]) ifFalse: [^ {self term}].
 
 	self term withBlanksTrimmed ifEmpty: [^ {self term}].
@@ -12,13 +12,16 @@ partTerms
 	self flag: #regexFamily.
 	patternSingle := '[^\s"]+'.
 	patternMulti := '"([^"\\]|\\["\\])+"'.
-	regexSingle := patternSingle asRegex.
-	regexMulti := patternMulti asRegex.
-	patternTerm := '(?%<=^|\s)(<1s>|<2s>)(?=\s|$)'
+	regexSingle := [patternSingle asRegex] once.
+	regexMulti := [patternMulti asRegex] once.
+	regexFull :=
+		[| patternTerm patternFull |
+		patternTerm := '(?%<=^|\s)(<1s>|<2s>)(?=\s|$)'
 		expandMacrosWith: patternSingle
 		with: patternMulti.
-	patternFull := '^(\s*(<1s>)\s*)*$' expandMacrosWith: patternTerm.
-	regexFull := patternFull asRegex.
+		patternFull := '^(\s*(<1s>)\s*)*$' expandMacrosWith: patternTerm.
+		patternFull asRegex]
+			once.
 
 	termStream := self term readStream.
 	(regexFull matchesStream: termStream) ifFalse: [

diff --git a/packages/SqueakInboxTalk.package/TalkTextSearcher.class/methodProperties.json b/packages/SqueakInboxTalk.package/TalkTextSearcher.class/methodProperties.json
@@ -17,6 +17,7 @@
 		"doMatchWildcard" : "ct 12/4/2021 23:45",
 		"doUseRegex" : "ct 6/17/2021 15:26",
 		"errors" : "ct 6/18/2021 00:33",
+		"filter:" : "ct 6/19/2024 18:10",
 		"fullyMatches:" : "ct 7/24/2021 00:12",
 		"fuzzyStrategies" : "ct 7/24/2021 00:57",
 		"hasComplexTerm" : "ct 7/24/2021 00:47",
@@ -51,7 +52,7 @@
 		"numberOfSurroundingLines" : "ct 7/24/2021 00:52",
 		"numberOfSurroundingLines:" : "ct 7/24/2021 00:52",
 		"partSearchers" : "ct 7/24/2021 00:59",
-		"partTerms" : "ct 7/24/2021 01:00",
+		"partTerms" : "ct 6/7/2024 17:02",
 		"postCopy" : "ct 6/11/2021 18:11",
 		"regexIntervalsIn:" : "ct 7/22/2021 19:40",
 		"removeRedundantLabelsFromMatches:" : "ct 1/5/2024 16:25",