From 0913d52203b40b52eca351aed4d3089bf09feacc Mon Sep 17 00:00:00 2001
From: "Harry R. Schwartz" <hello@harryrschwartz.com>
Date: Tue, 16 May 2023 16:05:11 -0700
Subject: [PATCH] Add --no-stemming flag to skip stemming

---
 lib/config.go   |  1 +
 lib/document.go |  7 ++++++-
 lib/main.go     |  2 ++
 man/docsim.1    | 10 ++++++++++
 4 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/lib/config.go b/lib/config.go
index 257bf64..bc47347 100644
--- a/lib/config.go
+++ b/lib/config.go
@@ -3,6 +3,7 @@ package main
 type Config struct {
 	BestFirst  bool
 	Limit      int
+	NoStemming bool
 	NoStoplist bool
 	OmitQuery  bool
 	ShowScores bool
diff --git a/lib/document.go b/lib/document.go
index b66fbc0..0f4a75f 100644
--- a/lib/document.go
+++ b/lib/document.go
@@ -64,7 +64,12 @@ func NewDocument(path string, config *Config) (*Document, error) {
 
 			if word != "" {
 				if config.NoStoplist || !inStoplist(word) {
-					termCount[stem(word)]++
+					if config.NoStemming {
+						termCount[word]++
+					} else {
+						termCount[stem(word)]++
+					}
+
 					totalWordCount++
 				}
 			}
diff --git a/lib/main.go b/lib/main.go
index 687c7fe..71a8f28 100644
--- a/lib/main.go
+++ b/lib/main.go
@@ -57,6 +57,7 @@ func sameFile(a, b string) bool {
 func main() {
 	bestFirstFlag := flag.Bool("best-first", false, "print best matches first")
 	limitFlag := flag.Int("limit", 0, "return at most `limit` results")
+	noStemmingFlag := flag.Bool("no-stemming", false, "don't perform stemming on words")
 	noStoplistFlag := flag.Bool("no-stoplist", false, "don't omit common words by using a stoplist")
 	omitQueryFlag := flag.Bool("omit-query", false, "don't include the query file itself in search results")
 	queryFlag := flag.String("query", "", "path to the file that results should match")
@@ -67,6 +68,7 @@ func main() {
 	config := Config{
 		BestFirst:  *bestFirstFlag,
 		Limit:      *limitFlag,
+		NoStemming: *noStemmingFlag,
 		NoStoplist: *noStoplistFlag,
 		OmitQuery:  *omitQueryFlag,
 		ShowScores: *showScoresFlag,
diff --git a/man/docsim.1 b/man/docsim.1
index ddc7855..9567255 100644
--- a/man/docsim.1
+++ b/man/docsim.1
@@ -26,6 +26,16 @@ default (worst to best).
 .BR \-\-limit " " \fINUM\fR
 Show no more that the best NUM results.
 .TP
+.BR \-\-no\-stemming
+Don't stem words. Stemming reduces inflected words to their word stem before
+comparing documents, which maps morphologically related words to a common term
+(so "spinning", "spinner", and "spins" might all map to "spin").
+.PP
+.RS
+Stemming usually improves scoring accuracy in English, but should probably be
+disabled when searching across code or documents in other languages.
+.RE
+.TP
 .BR \-\-no\-stoplist
 Don't filter out common words, like "the" and "because".
 .TP