rescribe: Set up so only Tesseract needed for the build platform is embedded

5 files changed, 46 insertions, 8 deletions

diff --git a/cmd/rescribe/embed_darwin.go b/cmd/rescribe/embed_darwin.go
new file mode 100644
index 0000000..cad8a29
--- /dev/null
+++ b/cmd/rescribe/embed_darwin.go
@@ -0,0 +1,10 @@
+// Copyright 2021 Nick White.
+// Use of this source code is governed by the GPLv3
+// license that can be found in the LICENSE file.
+
+package main
+
+import _ "embed"
+
+// TODO: add go:embed here
+var tesszip []byte
diff --git a/cmd/rescribe/embed_linux.go b/cmd/rescribe/embed_linux.go
new file mode 100644
index 0000000..cad8a29
--- /dev/null
+++ b/cmd/rescribe/embed_linux.go
@@ -0,0 +1,10 @@
+// Copyright 2021 Nick White.
+// Use of this source code is governed by the GPLv3
+// license that can be found in the LICENSE file.
+
+package main
+
+import _ "embed"
+
+// TODO: add go:embed here
+var tesszip []byte
diff --git a/cmd/rescribe/embed_other.go b/cmd/rescribe/embed_other.go
new file mode 100644
index 0000000..fe51fd0
--- /dev/null
+++ b/cmd/rescribe/embed_other.go
@@ -0,0 +1,13 @@
+// Copyright 2021 Nick White.
+// Use of this source code is governed by the GPLv3
+// license that can be found in the LICENSE file.
+
+// +build !darwin
+// +build !linux
+// +build !windows
+
+package main
+
+// if not one of the above platforms, we won't embed anything, so
+// just create an empty byte slice
+var tesszip []byte
diff --git a/cmd/rescribe/embed_windows.go b/cmd/rescribe/embed_windows.go
new file mode 100644
index 0000000..c447624
--- /dev/null
+++ b/cmd/rescribe/embed_windows.go
@@ -0,0 +1,10 @@
+// Copyright 2021 Nick White.
+// Use of this source code is governed by the GPLv3
+// license that can be found in the LICENSE file.
+
+package main
+
+import _ "embed"
+
+//go:embed tesseract-w32-v5.0.0-alpha.20210506.zip
+var tesszip []byte
diff --git a/cmd/rescribe/main.go b/cmd/rescribe/main.go
index c47de00..51a33b2 100644
--- a/cmd/rescribe/main.go
+++ b/cmd/rescribe/main.go
@@ -1,4 +1,4 @@
-// Copyright 2019 Nick White.
+// Copyright 2021 Nick White.
 // Use of this source code is governed by the GPLv3
 // license that can be found in the LICENSE file.
 
@@ -43,9 +43,6 @@ Note that embedded Tesseract includes these training files:
 - rescribev8_fast.traineddata (Latin historic printing)
 `
 
-//go:embed tesseract-w32-v5.0.0-alpha.20210506.zip
-var w32zip []byte
-
 const QueueTimeoutSecs = 2 * 60
 const PauseBetweenChecks = 1 * time.Second
 const LogSaveTime = 1 * time.Minute
@@ -184,15 +181,13 @@ func main() {
 		}
 		switch runtime.GOOS {
 		case "windows":
-			err = unpackZip(w32zip, tessdir)
+			err = unpackZip(tesszip, tessdir)
 			if err != nil {
-				log.Fatalln("Error unpacking embedded w32 zip:", err)
+				log.Fatalln("Error unpacking embedded Tesseract zip:", err)
 			}
 			tessCommand = filepath.Join(tessdir, "tesseract.exe")
 		// TODO: add linux and osx
 		}
-
-		trainingPath = filepath.Join(tessdir, "tessdata", strings.Replace(trainingPath, "trainings/", "", 1))
 	}
 
 	f, err := os.Open(trainingPath)