Remove fuzzy search from board search command

arduino · silvanocerza · Mar 10, 2021 · Mar 4, 2021 · Mar 4, 2021 · Mar 9, 2021
commit 181252c30b6f54395bd046cee5a65bc072021bd8
diff --git a/arduino/utils/search.go b/arduino/utils/search.go
@@ -0,0 +1,66 @@
+// This file is part of arduino-cli.
+//
+// Copyright 2020 ARDUINO SA (http://www.arduino.cc/)
+//
+// This software is released under the GNU General Public License version 3,
+// which covers the main part of arduino-cli.
+// The terms of this license can be found at:
+// https://www.gnu.org/licenses/gpl-3.0.en.html
+//
+// You can be released from the requirements of the above licenses by purchasing
+// a commercial license. Buying such a license is mandatory if you want to
+// modify or otherwise use the software for commercial activities involving the
+// Arduino software without disclosing the source code of your own applications.
+// To purchase a commercial license, send an email to license@arduino.cc.
+
+package utils
+
+import (
+	"strings"
+	"unicode"
+
+	"golang.org/x/text/runes"
+	"golang.org/x/text/transform"
+	"golang.org/x/text/unicode/norm"
+)
+
+// removeDiatrics removes accents and similar diatrics from unicode characters.
+// An empty string is returned in case of errors.
+// This might not be the best solution but it works well enough for our usecase,
+// in the future we might want to use the golang.org/x/text/secure/precis package
+// when its API will be finalized.
+// From https://stackoverflow.com/a/26722698
+func removeDiatrics(s string) (string, error) {
+	transformer := transform.Chain(
+		norm.NFD,
+		runes.Remove(runes.In(unicode.Mn)),
+		norm.NFC,
+	)
+	s, _, err := transform.String(transformer, s)
+	if err != nil {
+		return "", err
+	}
+	return s, nil
+}
+
+// Match returns true if all substrings are contained in str.
+// Both str and substrings are transforms to lower case and have their
+// accents and other unicode diatrics removed.
+// If strings transformation fails an error is returned.
+func Match(str string, substrings []string) (bool, error) {
+	str, err := removeDiatrics(strings.ToLower(str))
+	if err != nil {
+		return false, err
+	}
+
+	for _, sub := range substrings {
+		cleanSub, err := removeDiatrics(strings.ToLower(sub))
+		if err != nil {
+			return false, err
+		}
+		if !strings.Contains(str, cleanSub) {
+			return false, nil
+		}
+	}
+	return true, nil
+}
diff --git a/commands/board/search.go b/commands/board/search.go
@@ -21,9 +21,9 @@ import (
 	"sort"
 	"strings"
 
+	"github.com/arduino/arduino-cli/arduino/utils"
 	"github.com/arduino/arduino-cli/commands"
 	rpc "github.com/arduino/arduino-cli/rpc/commands"
-	"github.com/lithammer/fuzzysearch/fuzzy"
 )
 
 // Search returns all boards that match the search arg.
@@ -36,20 +36,23 @@ func Search(ctx context.Context, req *rpc.BoardSearchReq) (*rpc.BoardSearchResp,
 		return nil, errors.New("invalid instance")
 	}
 
-	searchArgs := strings.Trim(req.SearchArgs, " ")
+	searchArgs := strings.Split(strings.Trim(req.SearchArgs, " "), " ")
 
-	match := func(toTest []string) bool {
+	match := func(toTest []string) (bool, error) {
 		if len(searchArgs) == 0 {
-			return true
+			return true, nil
 		}
-		const maximumSearchDistance = 20
 
-		for _, rank := range fuzzy.RankFindNormalizedFold(searchArgs, toTest) {
-			if rank.Distance < maximumSearchDistance {
-				return true
+		for _, t := range toTest {
+			matches, err := utils.Match(t, searchArgs)
+			if err != nil {
+				return false, err
+			}
+			if matches {
+				return matches, nil
 			}
 		}
-		return false
+		return false, nil
 	}
 
 	res := &rpc.BoardSearchResp{Boards: []*rpc.BoardListItem{}}
@@ -87,7 +90,9 @@ func Search(ctx context.Context, req *rpc.BoardSearchReq) (*rpc.BoardSearchResp,
 					}
 
 					toTest := append(strings.Split(board.Name(), " "), board.Name(), board.FQBN())
-					if !match(toTest) {
+					if ok, err := match(toTest); err != nil {
+						return nil, err
+					} else if !ok {
 						continue
 					}
 
@@ -101,7 +106,9 @@ func Search(ctx context.Context, req *rpc.BoardSearchReq) (*rpc.BoardSearchResp,
 			} else {
 				for _, board := range latestPlatformRelease.BoardsManifest {
 					toTest := append(strings.Split(board.Name, " "), board.Name)
-					if !match(toTest) {
+					if ok, err := match(toTest); err != nil {
+						return nil, err
+					} else if !ok {
 						continue
 					}
 

diff --git a/test/test_board.py b/test/test_board.py
@@ -606,6 +606,18 @@ def test_board_search(run_command, data_dir):
     assert "Arduino Nano 33 BLE" in names
     assert "Arduino Portenta H7" in names
 
+    # Search in non installed boards
+    res = run_command("board search --format json nano 33")
+    assert res.ok
+    data = json.loads(res.stdout)
+    # Verifies boards are returned
+    assert len(data) > 0
+    # Verifies no board has FQBN set since no platform is installed
+    assert len([board["FQBN"] for board in data if "FQBN" in board]) == 0
+    names = [board["name"] for board in data if "name" in board]
+    assert "Arduino Nano 33 BLE" in names
+    assert "Arduino Nano 33 IoT" in names
+
     # Install a platform from index
     assert run_command("core install arduino:avr@1.8.3")
 
@@ -621,6 +633,14 @@ def test_board_search(run_command, data_dir):
     assert "arduino:avr:yun" in installed_boards
     assert "Arduino Yún" == installed_boards["arduino:avr:yun"]["name"]
 
+    res = run_command("board search --format json arduino yun")
+    assert res.ok
+    data = json.loads(res.stdout)
+    assert len(data) > 0
+    installed_boards = {board["FQBN"]: board for board in data if "FQBN" in board}
+    assert "arduino:avr:yun" in installed_boards
+    assert "Arduino Yún" == installed_boards["arduino:avr:yun"]["name"]
+
     # Manually installs a core in sketchbooks hardware folder
     git_url = "https://github.com/arduino/ArduinoCore-samd.git"
     repo_dir = Path(data_dir, "hardware", "arduino-beta-development", "samd")
@@ -647,38 +667,6 @@ def test_board_search(run_command, data_dir):
     assert "Arduino NANO 33 IoT" == installed_boards["arduino-beta-development:samd:nano_33_iot"]["name"]
     assert "arduino-beta-development:samd:arduino_zero_native" in installed_boards
 
-
-def test_board_search_fuzzy(run_command, data_dir):
-    assert run_command("update")
-
-    # Search in non installed boards
-    res = run_command("board search --format json nano 33")
-    assert res.ok
-    data = json.loads(res.stdout)
-    # Verifies boards are returned
-    assert len(data) > 0
-    # Verifies no board has FQBN set since no platform is installed
-    assert len([board["FQBN"] for board in data if "FQBN" in board]) == 0
-    names = [board["name"] for board in data if "name" in board]
-    assert "Arduino Nano 33 BLE" in names
-    assert "Arduino Nano 33 IoT" in names
-
-    # Install a platform from index
-    assert run_command("core install arduino:avr@1.8.3")
-
-    res = run_command("board search --format json arduino yun")
-    assert res.ok
-    data = json.loads(res.stdout)
-    assert len(data) > 0
-    installed_boards = {board["FQBN"]: board for board in data if "FQBN" in board}
-    assert "arduino:avr:yun" in installed_boards
-    assert "Arduino Yún" == installed_boards["arduino:avr:yun"]["name"]
-
-    # Manually installs a core in sketchbooks hardware folder
-    git_url = "https://github.com/arduino/ArduinoCore-samd.git"
-    repo_dir = Path(data_dir, "hardware", "arduino-beta-development", "samd")
-    assert Repo.clone_from(git_url, repo_dir, multi_options=["-b 1.8.11"])
-
     res = run_command("board search --format json mkr1000")
     assert res.ok
     data = json.loads(res.stdout)