updated to pack tesseract 5

README.md

@@ -225,12 +225,13 @@ Normal instructions apply once the application starts.
 ## Windows
 - Requirements
   - **Google Tesseract** installed and in path
-    - A copy of the version 4 is bundled with the release. Just run it, no extra packages needed
-      - You can download the installer here version here: [Installer here](https://digi.bib.uni-mannheim.de/tesseract/tesseract-ocr-w64-setup-v4.1.0.20190314.exe)
+    - A copy of the version 5.0.1 is bundled with the release. Just run it, no extra packages needed
+      - You can download the installer here version here: [Installer here](https://digi.bib.uni-mannheim.de/tesseract/tesseract-ocr-w64-setup-v5.0.1.20220118.exe)
     - Make sure it's in path
   - Python3 installed and in path
 - Running:
   - `source env/bin/activate` (if you use a virtual env)
+  - `python main.py`
 
 # Extra command-line options
 If you run from source, or call the executable from the terminal you can make use of the following flags/arguments to achieve different functionality.

data/versions.json

@@ -1,5 +1,5 @@
 {
-    "app": "1.5.2",
+    "app": "1.5.3",
     "skills": "3.0",
     "languages": {
         "eng": "1.0.1",

src/tesseract/tesseract_utils.py

@@ -25,7 +25,7 @@ def _is_pyinstaller():
 
 def _get_pyinstaller_tesseract_path():
     base_path = sys._MEIPASS
-    bundled_path = os.path.join(base_path, "Tesseract-OCR", "libtesseract-4.dll")
+    bundled_path = os.path.join(base_path, "Tesseract-OCR", "libtesseract-5.dll")
     return bundled_path
 
 
@@ -37,13 +37,26 @@ def find_tesseract():
     # TODO: Make this resilient to "change" (tesseract version), probably not necessary
     locations = [
         ctypes.util.find_library("libtesseract-4"),  # win32
+        ctypes.util.find_library("libtesseract-5"),  # win32
         ctypes.util.find_library("libtesseract302"),  # win32 version 3.2
         ctypes.util.find_library("libtesseract"),  # others
         ctypes.util.find_library("tesseract"),  # others
     ]
 
     if WINDOWS:
         locations += [
+            os.path.join(
+                os.getenv("ProgramW6432"), "Tesseract-OCR", "libtesseract-5.dll"
+            ),
+            os.path.join(
+                os.getenv("LOCALAPPDATA"), "Tesseract-OCR", "libtesseract-5.dll"
+            ),
+            os.path.join(
+                os.getenv("ProgramFiles"), "Tesseract-OCR", "libtesseract-5.dll"
+            ),
+            os.path.join(
+                os.getenv("programfiles(x86)"), "Tesseract-OCR", "libtesseract-5.dll"
+            ),
             os.path.join(
                 os.getenv("ProgramW6432"), "Tesseract-OCR", "libtesseract-4.dll"
             ),