Drop requirement on using tessdata environment variable

As we now require additional tessdata than just the trained languages
and guibot's misc data path is not included in all forms of packaging
it is better to drop any tessdata distributed within guibot and thus
any requirement on explicitly providing a (tess)data path in the form
of CV parameter or an environment variable. These would still take
increasing order of precedence if set though to support any previous
customization.

guibot/finder.py

@@ -2002,19 +2002,22 @@ def __synchronize_backend(self, backend=None, category="text", reset=False):
 
         import cv2
         datapath = self.params["text"]["datapath"].value
+        tessdata_path = os.path.join(datapath, "tessdata")
+        if not os.path.exists(tessdata_path):
+            tessdata_path = os.environ.get("TESSDATA_PREFIX", "./tessdata")
+            if not os.path.exists(tessdata_path):
+                tessdata_path = None
+
         if category == "text" or category in ["contour", "threshold", "threshold2"]:
             # nothing to sync
             return
 
         elif category == "tdetect" and backend == "pytesseract":
-            tessdata_path = os.path.join(datapath, "tessdata")
-            if not os.path.exists(tessdata_path):
-                tessdata_path = os.environ.get("TESSDATA_PREFIX", ".")
-
             import pytesseract
             self.tbox = pytesseract
-            self.tbox_config = r"--tessdata-dir %s --oem %s --psm %s "
-            self.tbox_config %= (tessdata_path,
+            tessdata_dir = "--tessdata-dir '" + tessdata_path + "'" if tessdata_path else ""
+            self.tbox_config = r"%s --oem %s --psm %s "
+            self.tbox_config %= (tessdata_dir,
                                  self.params["tdetect"]["oem"].value,
                                  self.params["tdetect"]["psmode"].value)
             self.tbox_config += r"-c tessedit_char_whitelist='%s' %s batch.nochop wordstrbox"
@@ -2038,33 +2041,36 @@ def __synchronize_backend(self, backend=None, category="text", reset=False):
             return
 
         elif category == "ocr":
-            tessdata_path = os.path.join(datapath, "tessdata")
-            if not os.path.exists(tessdata_path):
-                tessdata_path = os.environ.get("TESSDATA_PREFIX", ".")
-
             if backend == "pytesseract":
                 import pytesseract
                 self.ocr = pytesseract
-                self.ocr_config = r"--tessdata-dir '%s' --oem %s --psm %s "
-                self.ocr_config %= (tessdata_path,
+                tessdata_dir = "--tessdata-dir '" + tessdata_path + "'" if tessdata_path else ""
+                self.ocr_config = r"%s --oem %s --psm %s "
+                self.ocr_config %= (tessdata_dir,
                                     self.params["ocr"]["oem"].value,
                                     self.params["ocr"]["psmode"].value)
                 self.ocr_config += r"-c tessedit_char_whitelist='%s' %s"
                 self.ocr_config %= (self.params["ocr"]["char_whitelist"].value,
                                     self.params["ocr"]["extra_configs"].value)
             elif backend == "tesserocr":
                 from tesserocr import PyTessBaseAPI
-                self.ocr = PyTessBaseAPI(path=tessdata_path,
-                                         lang=self.params["ocr"]["language"].value,
-                                         oem=self.params["ocr"]["oem"].value,
-                                         psm=self.params["ocr"]["psmode"].value)
+                kwargs = {"lang": self.params["ocr"]["language"].value,
+                          "oem": self.params["ocr"]["oem"].value,
+                          "psm": self.params["ocr"]["psmode"].value}
+                if tessdata_path:
+                    self.ocr = PyTessBaseAPI(path=tessdata_path, **kwargs)
+                else:
+                    self.ocr = PyTessBaseAPI(**kwargs)
                 self.ocr.SetVariable("tessedit_char_whitelist", self.params["ocr"]["char_whitelist"].value)
             elif backend == "tesseract":
-                self.ocr = cv2.text.OCRTesseract_create(tessdata_path,
-                                                        language=self.params["ocr"]["language"].value,
-                                                        char_whitelist=self.params["ocr"]["char_whitelist"].value,
-                                                        oem=self.params["ocr"]["oem"].value,
-                                                        psmode=self.params["ocr"]["psmode"].value)
+                kwargs = {"language": self.params["ocr"]["language"].value,
+                          "char_whitelist": self.params["ocr"]["char_whitelist"].value,
+                          "oem": self.params["ocr"]["oem"].value,
+                          "psmode": self.params["ocr"]["psmode"].value}
+                if tessdata_path:
+                    self.ocr = cv2.text.OCRTesseract_create(datapath, **kwargs)
+                else:
+                    self.ocr = cv2.text.OCRTesseract_create(**kwargs)
             elif backend in ["hmm", "beamSearch"]:
 
                 import numpy

packaging/guibot.spec

@@ -61,7 +61,7 @@ pushd packaging
 %{__python3} setup.py install --root %{buildroot}
 popd
 %{__install} -d %{buildroot}%{python3_sitelib}/guibot/tests/images
-%{__install} -d %{buildroot}%{python3_sitelib}/guibot/misc/tessdata
+%{__install} -d %{buildroot}%{python3_sitelib}/guibot/misc
 %{__cp} -a tests/* %{buildroot}%{python3_sitelib}/guibot/tests
 %{__cp} -a misc/* %{buildroot}%{python3_sitelib}/guibot/misc