[packages/python-tesserocr] - up to 2.3.1 - remove references to legacy API (not exported by tesseract 4.0)
baggins
baggins at pld-linux.org
Thu Aug 16 23:05:13 CEST 2018
commit b02afae566455819a781e966597ec72b8c678d97
Author: Jan Rękorajski <baggins at pld-linux.org>
Date: Thu Aug 16 23:04:32 2018 +0200
- up to 2.3.1
- remove references to legacy API (not exported by tesseract 4.0)
python-tesserocr.spec | 12 +++++----
tesseract4.patch | 67 +++++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 74 insertions(+), 5 deletions(-)
---
diff --git a/python-tesserocr.spec b/python-tesserocr.spec
index 2d22f4e..8c1dc7c 100644
--- a/python-tesserocr.spec
+++ b/python-tesserocr.spec
@@ -7,17 +7,18 @@
%bcond_without python2 # CPython 2.x module
%bcond_without python3 # CPython 3.x module
-%define module tesserocr
-%define egg_name tesserocr
+%define module tesserocr
+%define egg_name tesserocr
%define pypi_name tesserocr
Summary: A simple, Pillow-friendly, Python wrapper around tesseract-ocr API using Cython
Name: python-%{pypi_name}
-Version: 2.2.2
-Release: 3
+Version: 2.3.1
+Release: 1
License: MIT
Group: Libraries/Python
Source0: https://files.pythonhosted.org/packages/source/t/%{pypi_name}/%{pypi_name}-%{version}.tar.gz
-# Source0-md5: e6c9c8f6f6720e16cd612146e20e7feb
+# Source0-md5: 99e2001affe861ae3a5aa2e9f233e2d7
+Patch0: tesseract4.patch
URL: https://github.com/sirfz/tesserocr
BuildRequires: rpm-pythonprov
BuildRequires: rpmbuild(macros) >= 1.714
@@ -60,6 +61,7 @@ image files instead.
%prep
%setup -q -n %{pypi_name}-%{version}
+%patch0 -p1
# Remove bundled egg-info
%{__rm} -r %{egg_name}.egg-info
diff --git a/tesseract4.patch b/tesseract4.patch
new file mode 100644
index 0000000..5813085
--- /dev/null
+++ b/tesseract4.patch
@@ -0,0 +1,67 @@
+--- tesserocr-2.3.1/tesseract.pxd~ 2018-02-04 20:28:35.000000000 +0100
++++ tesserocr-2.3.1/tesseract.pxd 2018-08-16 22:58:33.358727376 +0200
+@@ -198,19 +198,6 @@
+ cdef cppclass TessOsdRenderer(TessResultRenderer):
+ TessOsdRenderer(cchar_t *) except +
+
+-cdef extern from "tesseract/osdetect.h" nogil:
+- struct OSBestResult:
+- int orientation_id
+- int script_id
+- float sconfidence
+- float oconfidence
+-
+- ctypedef int (*get_best_script)(int)
+-
+- struct OSResults:
+- get_best_script get_best_script
+- OSBestResult best_result
+-
+ cdef extern from "tesseract/baseapi.h" namespace "tesseract" nogil:
+
+ IF TESSERACT_VERSION >= 0x040000:
+@@ -312,7 +312,6 @@
+ int IsValidWord(cchar_t *)
+ bool IsValidCharacter(cchar_t *)
+ bool GetTextDirection(int *, float *)
+- bool DetectOS(OSResults *);
+ cchar_t *GetUnichar(int)
+ const OcrEngineMode oem() const
+ void set_min_orientation_margin(double)
+@@ -386,7 +385,6 @@
+ int IsValidWord(cchar_t *)
+ bool IsValidCharacter(cchar_t *)
+ bool GetTextDirection(int *, float *)
+- bool DetectOS(OSResults *);
+ cchar_t *GetUnichar(int)
+ const OcrEngineMode oem() const
+ void set_min_orientation_margin(double)
+--- tesserocr-2.3.1/tesserocr.pyx~ 2018-08-13 19:35:16.000000000 +0200
++++ tesserocr-2.3.1/tesserocr.pyx 2018-08-16 23:00:17.290796571 +0200
+@@ -2292,26 +2292,6 @@
+ self._baseapi.GetTextDirection(&out_offset, &out_slope)
+ return out_offset, out_slope
+
+- def DetectOS(self):
+- """Estimate the Orientation and Script of the image.
+-
+- Returns:
+- `dict` or `None` if image was not successfully processed. dict contains:
+- - orientation: Orientation ids [0..3] map to [0, 270, 180, 90] degree orientations of the
+- page respectively, where the values refer to the amount of clockwise
+- rotation to be applied to the page for the text to be upright and readable.
+- - oconfidence: Orientation confidence.
+- - script: Index of the script with the highest score for this orientation.
+- - sconfidence: script confidence.
+- """
+- cdef OSResults results
+- if self._baseapi.DetectOS(&results):
+- return {'orientation': results.best_result.orientation_id,
+- 'oconfidence': results.best_result.oconfidence,
+- 'script': results.get_best_script(results.best_result.orientation_id),
+- 'sconfidence': results.best_result.sconfidence}
+- return None
+-
+ def GetUnichar(self, int unichar_id):
+ """Return the string form of the specified unichar.
+
================================================================
---- gitweb:
http://git.pld-linux.org/gitweb.cgi/packages/python-tesserocr.git/commitdiff/b02afae566455819a781e966597ec72b8c678d97
More information about the pld-cvs-commit
mailing list