From b95bdbcca270889527159b3b2a5a4e20a7738a57 Mon Sep 17 00:00:00 2001
From: David Martin <david_martin@fastmail.com>
Date: Mon, 11 Jun 2018 16:30:49 +1000
Subject: [PATCH] Determine the psm parameter based on the Tesseract version.

It turns out that for versions before the current 4 beta only '-psm' is
allowed, and the latest build only allows '--psm'.
---
 src/pyocr/builders.py  |  7 ++++---
 src/pyocr/tesseract.py | 11 ++++++++++-
 2 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/src/pyocr/builders.py b/src/pyocr/builders.py
index 5959a7c..f27b829 100644
--- a/src/pyocr/builders.py
+++ b/src/pyocr/builders.py
@@ -14,6 +14,7 @@
 import xml.dom.minidom
 import logging
 
+from .tesseract import psm_parameter
 from .util import to_unicode
 
 logger = logging.getLogger(__name__)
@@ -305,7 +306,7 @@ class TextBuilder(BaseBuilder):
     def __init__(self, tesseract_layout=3, cuneiform_dotmatrix=False,
                  cuneiform_fax=False, cuneiform_singlecolumn=False):
         file_ext = ["txt"]
-        tess_flags = ["--psm", str(tesseract_layout)]
+        tess_flags = [psm_parameter(), str(tesseract_layout)]
         cun_args = ["-f", "text"]
         # Add custom cuneiform parameters if needed
         for par, arg in [(cuneiform_dotmatrix, "--dotmatrix"),
@@ -562,7 +563,7 @@ class WordBoxBuilder(BaseBuilder):
 
     def __init__(self, tesseract_layout=1):
         file_ext = ["html", "hocr"]
-        tess_flags = ["--psm", str(tesseract_layout)]
+        tess_flags = [psm_parameter(), str(tesseract_layout)]
         tess_conf = ["hocr"]
         cun_args = ["-f", "hocr"]
         super(WordBoxBuilder, self).__init__(file_ext, tess_flags, tess_conf,
@@ -638,7 +639,7 @@ class LineBoxBuilder(BaseBuilder):
 
     def __init__(self, tesseract_layout=1):
         file_ext = ["html", "hocr"]
-        tess_flags = ["--psm", str(tesseract_layout)]
+        tess_flags = [psm_parameter(), str(tesseract_layout)]
         tess_conf = ["hocr"]
         cun_args = ["-f", "hocr"]
         super(LineBoxBuilder, self).__init__(file_ext, tess_flags, tess_conf,
diff --git a/src/pyocr/tesseract.py b/src/pyocr/tesseract.py
index 5b8e002..c1329e0 100755
--- a/src/pyocr/tesseract.py
+++ b/src/pyocr/tesseract.py
@@ -161,6 +161,15 @@ def can_detect_orientation():
     )
 
 
+def psm_parameter():
+    """Return the psm option string depending on the Tesseract version."""
+    version = get_version()
+    if version[0] <= 3:
+        return "-psm"
+
+    return "--psm"
+
+
 def detect_orientation(image, lang=None):
     """
     Arguments:
@@ -178,7 +187,7 @@ def detect_orientation(image, lang=None):
     """
     _set_environment()
     with temp_dir() as tmpdir:
-        command = [TESSERACT_CMD, "input.bmp", 'stdout', "--psm", "0"]
+        command = [TESSERACT_CMD, "input.bmp", 'stdout', psm_parameter(), "0"]
         version = get_version()
         if version[0] >= 4:
             # XXX: temporary fix to remove once Tesseract 4 is stable