[Pycodeocr-main] SF.net SVN: pycodeocr:[50] trunk

[<drt...@us...>]

Revision: 50
          http://pycodeocr.svn.sourceforge.net/pycodeocr/?rev=50&view=rev
Author:   drtrigon
Date:     2011-04-27 19:43:41 +0000 (Wed, 27 Apr 2011)

Log Message:
-----------
group the various modules into one single package 'utils'

Modified Paths:
--------------
    trunk/PyCodeOCR.py

Added Paths:
-----------
    trunk/utils/
    trunk/utils/__init__.py
    trunk/utils/checksum.py
    trunk/utils/installbundle.py
    trunk/utils/ocr.py
    trunk/utils/requirements.py

Removed Paths:
-------------
    trunk/checksum.py
    trunk/installbundle.py
    trunk/requirements.py

Modified: trunk/PyCodeOCR.py
===================================================================
--- trunk/PyCodeOCR.py	2011-04-27 19:01:54 UTC (rev 49)
+++ trunk/PyCodeOCR.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -41,12 +41,14 @@
 # python standard modules
 import os, re, sys, subprocess, time, gobject, string, struct
 
+# PyCodeOCR python standard utils (modules)
+import utils
 
+
 # check and install requirements
 #print "To check and install requirements run: 'python %s --install-deps'" % sys.argv[0]
-import requirements
 if "--install-deps" in sys.argv:
-    requirements.check_all_pycodeocr_requirements()
+    utils.requirements.check_all_pycodeocr_requirements()
     sys.exit()
 
 
@@ -56,7 +58,7 @@
 import gtk, gtk.glade
 import gobject
 
-from checksum import modulo10
+from utils.checksum import modulo10
 
 # PIL image library and it's SANE bindings
 import Image, ImageDraw
@@ -76,9 +78,9 @@
 print "To check and install requirements run: 'python %s --install-deps'" % sys.argv[0]
 pakages = []
 # gocr
-paths = { "gocr":             requirements.which("gocr"),
-          "dmtxread":         requirements.which("dmtxread"),
-          "pdf417decode.exe": requirements.which("pdf417decode.exe"), }
+paths = { "gocr":             utils.requirements.which("gocr"),
+          "dmtxread":         utils.requirements.which("dmtxread"),
+          "pdf417decode.exe": utils.requirements.which("pdf417decode.exe"), }
 if os.path.exists(paths["gocr"]):
 	pakages.append( "gocr" )
 	print "gocr found."
@@ -612,7 +614,7 @@
 		if	(op_mode == self.MDE['invoices']):	# 0: invoices
 			self.progress(4./max_steps, "Character correction...")
 			print "*** " * 10
-			data = self.char_correction(data)
+			data = utils.ocr.char_correction(data)
 		self.refresh()
 		if self.__stop: return
 
@@ -775,68 +777,6 @@
 		self.progressbar1.set_text(text)
 		self.refresh()
 
-	## Character correction after recogition (on basis that there should be numbers and few special chars).
-	#
-	# @test Add unittest to check correct operation of this important module/function.
-	#
-	# @todo May be this important function should be placed into a separate module,
-	#       similar to @ref checksum.
-	#
-	def char_correction(self, data):
-		data = re.sub("\n", "", data)
-		print data
-		corrections = [
-		# Zero
-		("O", "0"), ("D", "0"), ("U", "0"), ("Q", "0"), ("\(\)", "0"), ("G", "0"), ("o", "0"),
-		# Two
-		("Z", "2"),
-		# Three
-		("\?>", "3"),
-		# Four
-		("l\.", "4"), ("A", "4"), ("l»", "4"), ("h", "4"), ("i»", "4"), ("l\xe2\x80\xa2", "4"), ("&", "4"),
-		# Five
-		("S", "5"),
-		# Six
-		("c>", "6"),
-		# Seven
-		#("\?", "7"),
-		# One (critical, therefore done as last; has also Overlapps e.g. with Four)
-		(chr(226)+chr(128)+chr(153)+"I", "1"), 
-		(chr(226)+chr(128)+chr(153)+"\|", "1"),
-		(chr(226)+chr(128)+chr(152)+"\|", "1"),
-		(chr(226)+chr(128)+chr(152)+"I", "1"),
-		(chr(39)+"I", "1"),
-		("I", "1"),
-		(chr(153)+"l", "1"),
-		(chr(39)+"l", "1"),
-		("l", "1"),
-		(chr(39)+"!", "1"),
-		(chr(39)+"\|", "1"),
-		("\|", "1"),
-		("'\]", "1"),
-		("'\}", "1"),
-		(r'\\', "1"),
-		# non-numbers
-		(" ", ""),
-		("\xe2", ""),
-		("\x80", ""),
-		("\xbb", ""),
-		("\x99", ""),
-		("\x98", ""),
-		("\x5f", ""),
-		("\x60", ""),
-		("'", ""),
-		#('\xbb', 187)('\x99', 153)('\x98', 152)
-		("\+", "+ "), 
-		("\x92", "1"),	# post correction
-		]
-		for item in corrections:
-			olddata = data
-			data = re.sub(item[0], item[1], data)
-			if not (data == olddata): print 'subst: "%s" => "%s"' % (item[0], item[1])
-
-		return data
-
 ## Run external shell command/application.
 class RunExternal:
 

Deleted: trunk/checksum.py
===================================================================
--- trunk/checksum.py	2011-04-27 19:01:54 UTC (rev 49)
+++ trunk/checksum.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -1,47 +0,0 @@
-##
-# @file    checksum.py
-# @package checksum
-# @authors laserb, drtrigon
-# @version 1.1
-# @brief   Modulo 10 checksum for e-banking codes.
-#
-# @test Add unittest to check correct operation of this important module.
-#
-# $Id$
-#
-# @section Description
-#
-# Modulo 10 checksum for e-banking codes
-#
-
-class modulo10:
-	# thanks to http://www.hosang.ch/modulo10.aspx
-	# much more details http://www.sic.ch/de/dl_tkicch_dta.pdf ( page 51 )
-	# @see http://www.bundesbank.de/download/zahlungsverkehr/zv_pz201012.pdf
-	# @see http://www.bundesbank.de/zahlungsverkehr/zahlungsverkehr_pruefziffernberechnung.php
-	def __init__(self, amount, account, reference):
-		self.tabelle = [0,9,4,6,8,2,7,1,3,5]
-
-		self.account    = account
-		self.reference  = reference
-	
-		self.checksum_b = self._checksum(amount)
-		self.checksum_k = self._checksum(account)
-		self.checksum_r = self._checksum(reference)
-		self.checksum   = self.checksum_b and self.checksum_k and self.checksum_r
-
-		try:
-			self.amount = str(int(amount[2:-1])/100.)
-		except ValueError:
-			self.amount = amount + " (error)"
-
-	def _run(self, nr):
-		self.uebertrag = 0
-		# iterate over each character
-		for i in range(len(nr)):
-			self.uebertrag = self.tabelle[(self.uebertrag + int(nr[i]) )%10]
-		
-		return (10-self.uebertrag)%10
-	
-	def _checksum(self, value):
-		return (int(value[-1]) == self._run(value[:-1]) )

Deleted: trunk/installbundle.py
===================================================================
--- trunk/installbundle.py	2011-04-27 19:01:54 UTC (rev 49)
+++ trunk/installbundle.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -1,196 +0,0 @@
-##
-# @file    installbundle.py
-# @package installbundle
-# @authors University of Utah (VisTrails), drtrigon
-# @version 1.1
-# @brief   Module with utilities to try and install a bundle if possible.
-#
-# $Id$
-#
-# @section Description
-#
-# Module with utilities to try and install a bundle if possible.
-#
-############################################################################
-##
-## Copyright (C) 2006-2010 University of Utah. All rights reserved.
-##
-## This file is part of VisTrails.
-##
-## This file may be used under the terms of the GNU General Public
-## License version 2.0 as published by the Free Software Foundation
-## and appearing in the file LICENSE.GPL included in the packaging of
-## this file.  Please review the following to ensure GNU General Public
-## Licensing requirements will be met:
-## http://www.opensource.org/licenses/gpl-license.php
-##
-## If you are unsure which license is appropriate for your use (for
-## instance, you are interested in developing a commercial derivative
-## of VisTrails), please contact us at vis...@sc....
-##
-## This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
-## WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
-##
-## Changed 2011 by drtrigon. Source from 'core.bundles.installbundle'
-## http://downloads.sourceforge.net/project/vistrails/vistrails/nightly/vistrails-src-nightly.tar.gz
-##
-## This file was modified to be used as part of PyCodeOCR.
-##
-############################################################################
-#
-
-"""Module with utilities to try and install a bundle if possible."""
-
-import os
-
-##############################################################################
-
-def linux_ubuntu_install(package_name):
-    qt = has_qt()
-    hide_splash_if_necessary()
-        
-    if qt:
-        cmd = core.system.vistrails_root_directory()
-        cmd += '/core/bundles/linux_ubuntu_install.py'
-    else:
-        cmd = 'apt-get install -y'
-
-    if type(package_name) == str:
-        cmd += ' ' + package_name
-    elif type(package_name) == list:
-        for package in package_name:
-            if type(package) != str:
-                raise TypeError("Expected string or list of strings")
-            cmd += ' ' + package
-
-    if qt:
-        sucmd = guess_graphical_sudo() + " '" + cmd + "'"
-    else:
-        debug.warning("PyCodeOCR wants to install package(s) '%s'" %
-                      package_name)
-        sucmd = "sudo " + cmd
-
-    result = os.system(sucmd)
-
-    return (result == 0) # 0 indicates success
-
-def linux_fedora_install(package_name):
-    qt = has_qt()
-    hide_splash_if_necessary()
-    if qt:
-        cmd = core.system.vistrails_root_directory()
-        cmd += '/core/bundles/linux_fedora_install.py'
-    else:
-        cmd = 'yum -y install'
-
-    if type(package_name) == str:
-        cmd += ' ' + package_name
-    elif type(package_name) == list:
-        for package in package_name:
-            if type(package) != str:
-                raise TypeError("Expected string or list of strings")
-            cmd += ' ' + package
-
-    if qt:
-        sucmd = guess_graphical_sudo() + " " + cmd
-    else:
-        debug.warning(("PyCodeOCR wants to install package(s) '%s' through "
-                       "_sudo_. Make sure you are a sudoer.") % package_name)
-        sucmd = "sudo " + cmd
-
-    debug.warning("EXECUTING: sucmd")
-    result = os.system(sucmd)
-    debug.warning("RETURN VALUE: %s" % result)
-    return (result == 0)
-
-def show_question(which_files):
-    qt = has_qt()
-    if qt:
-        import gui.utils
-        if type(which_files) == str:
-            which_files = [which_files]
-        v = gui.utils.show_question("Required packages missing",
-                                    "One or more required packages are missing: " +
-                                    " ".join(which_files) +
-                                    ". PyCodeOCR can " +
-                                    "automaticallly install them. " +
-                                    "If you click OK, PyCodeOCR will need "+
-                                    "administrator privileges, and you " +
-                                    "might be asked for the administrator password.",
-                                    buttons=[gui.utils.OK_BUTTON,
-                                             gui.utils.CANCEL_BUTTON],
-                                    default=gui.utils.OK_BUTTON)
-        return v == gui.utils.OK_BUTTON
-    else:
-        print "Required package missing"
-        print ("A required package is missing, but PyCodeOCR can " +
-               "automatically install it. " +
-               "If you say Yes, PyCodeOCR will need "+
-               "administrator privileges, and you" +
-               "might be asked for the administrator password.")
-        print "Give PyCodeOCR permission to try to install package? (y/N)"
-        v = raw_input().upper()
-        return v == 'Y' or v == 'YES'
-
-
-def install(dependency_dictionary):
-    """Tries to import a python module. If unsuccessful, tries to install
-the appropriate bundle and then reimport. py_import tries to be smart
-about which system it runs on."""
-
-    # Ugly fix to avoid circular import
-    distro = guess_system()
-    if not dependency_dictionary.has_key(distro):
-        return False
-    else:
-        files = dependency_dictionary[distro]
-        if show_question(files):
-            callable_ = getattr(installbundle,
-                                distro.replace('-', '_') + '_install')
-            return callable_(files)
-        else:
-            return False
-
-##############################################################################
-
-## Guesses the operation system.
-# 
-# Replaces original VisTrails 'core.bundles.utils.guess_system'.
-#
-# @since (added for PyCodeOCR)
-#
-def guess_system():
-    import platform
-    #print os.name, platform.platform()
-    return ("%s-%s" % (platform.system(), platform.dist()[0])).lower()
-
-## Dummy function, returns False.
-# 
-# Replaces original VisTrails 'has_qt'.
-#
-# @since (added for PyCodeOCR)
-#
-def has_qt():
-    return False
-
-import installbundle
-
-## Dummy function, just passes.
-# 
-# Replaces original VisTrails 'hide_splash_if_necessary'.
-#
-# @since (added for PyCodeOCR)
-#
-def hide_splash_if_necessary():
-    pass
-
-## Debug warning output handler.
-# 
-# Replaces original VisTrails 'core.debug.warning'.
-#
-# @since (added for PyCodeOCR)
-#
-class dbg():
-    def warning(self, *args):
-        print "".join(args)
-debug = dbg()

Deleted: trunk/requirements.py
===================================================================
--- trunk/requirements.py	2011-04-27 19:01:54 UTC (rev 49)
+++ trunk/requirements.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -1,241 +0,0 @@
-##
-# @file    requirements.py
-# @package requirements
-# @authors University of Utah (VisTrails), drtrigon
-# @version 1.1
-# @brief   Online test presence of runtime components.
-#
-# $Id$
-#
-# @section Description
-#
-# module that allows online inspection of environment to test presence of
-# runtime components such as binaries, libraries, other python modules, etc.
-#
-############################################################################
-##
-## Copyright (C) 2006-2010 University of Utah. All rights reserved.
-##
-## This file is part of VisTrails.
-##
-## This file may be used under the terms of the GNU General Public
-## License version 2.0 as published by the Free Software Foundation
-## and appearing in the file LICENSE.GPL included in the packaging of
-## this file.  Please review the following to ensure GNU General Public
-## Licensing requirements will be met:
-## http://www.opensource.org/licenses/gpl-license.php
-##
-## If you are unsure which license is appropriate for your use (for
-## instance, you are interested in developing a commercial derivative
-## of VisTrails), please contact us at vis...@sc....
-##
-## This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
-## WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
-##
-## Changed 2011 by drtrigon. Source from 'core.requirements'
-## http://downloads.sourceforge.net/project/vistrails/vistrails/nightly/vistrails-src-nightly.tar.gz
-##
-## This file was modified to be used as part of PyCodeOCR.
-##
-############################################################################
-#
-
-"""module that allows online inspection of environment to test presence of
-runtime components such as binaries, libraries, other python modules, etc."""
-
-import sys
-
-##############################################################################
-
-def python_module_exists(module_name):
-    """python_module_exists(module_name): Boolean.
-Returns if python module of given name can be safely imported."""
-    
-    try:
-        sys.modules[module_name]
-        return True
-    except KeyError:
-        pass
-    try:
-        __import__(module_name)
-        return True
-    except ImportError:
-        return False
-    
-
-def executable_file_exists(filename):
-    """executable_file_exists(filename): Boolean.
-Returns if certain file is in current path and is executable."""
-    result = executable_is_in_path(filename)
-    if result == "":
-        result = executable_is_in_pythonpath(filename)
-    return result != ""
-
-# FIXME: Add documentation.
-
-def require_python_module(module_name):
-    if not python_module_exists(module_name):
-        raise MissingRequirement(module_name)
-
-def require_executable(filename):
-    if not executable_file_exists(filename):
-        raise MissingRequirement(filename)
-
-def check_pyqt4():
-    # checks for the presence of pyqt4, which is more important than the rest,
-    # since using pyqt requires a qapplication.
-    try:
-        require_python_module('PyQt4.QtGui')
-        require_python_module('PyQt4.QtOpenGL')
-    except MissingRequirement:
-        r = core.bundles.installbundle.install({'linux-ubuntu': ['python-qt4',
-                                                                 'python-qt4-gl',
-                                                                 'python-qt4-sql']})
-        if not r:
-            raise
-
-
-def check_all_vistrails_requirements():
-    pass
-
-    # check scipy
-#     try:
-#         require_python_module('scipy')
-#     except MissingRequirement:
-#         r = core.bundles.installbundle.install({'linux-ubuntu': 'python-scipy'})
-#         if not r:
-#             raise
-        
-
-##############################################################################
-
-class MissingRequirement(Exception):
-    """Raise this exception in packages when necessary items are missing."""
-    def __init__(self, req):
-        self.requirement = req
-    def __str__(self):
-        return "Missing Requirement: %s" % self.requirement
-
-##############################################################################
-
-import installbundle
-import os
-
-
-## Shows the full path of (shell) commands.
-# 
-# According to shell command 'which', look also at:
-# @see http://unixhelp.ed.ac.uk/CGI/man-cgi?which
-# @see http://stackoverflow.com/questions/377017/test-if-executable-exists-in-python
-#
-# @since (added for PyCodeOCR)
-#
-def which(program, search_path=None):
-    is_exe = lambda fpath: os.path.exists(fpath) and os.access(fpath, os.X_OK)
-    #is_exe = lambda fpath: os.path.exists(fpath)
-
-    fpath, fname = os.path.split(program)
-    if fpath:
-        if is_exe(program):
-            return program
-    else:
-        if not search_path:
-            search_path  = os.environ["PATH"].split(os.pathsep)
-            search_path += [ '.' ]    # add local path
-        for path in search_path:
-            exe_file = os.path.join(path, program)
-            if is_exe(exe_file):
-                return exe_file
-    return ""
-
-## Checks if executable is in system PATH.
-# 
-# Replaces original VisTrails 'core.system.executable_is_in_path'.
-#
-# @since (added for PyCodeOCR)
-#
-def executable_is_in_path(filename):
-    #print os.system(filename)
-    return which(filename)
-
-## Checks if executable is in system PYTHONPATH.
-# 
-# Replaces original VisTrails 'core.system.executable_is_in_pythonpath'.
-#
-# @since (added for PyCodeOCR)
-#
-def executable_is_in_pythonpath(filename):
-    return which(filename, search_path=os.environ.get("PYTHONPATH", None))
-
-
-## Checks if pygtk libraries are installed.
-# 
-# @since (added for PyCodeOCR)
-#
-# @todo Check if further dependencies/packages are needed...?
-#
-def check_pygtk():
-    # checks for the presence of pygtk, which is more important than the rest.
-    try:
-        require_python_module('pygtk')
-        #pygtk.require('2.0')
-        require_python_module('gtk')
-        require_python_module('gtk.glade')
-        require_python_module('gobject')
-    except MissingRequirement:
-        r = core.bundles.installbundle.install({'linux-fedora': ['pygtk2']})
-        # other deps?
-        if not r:
-            raise
-
-## Checks if sane libraries are installed.
-# 
-# @since (added for PyCodeOCR)
-#
-# @todo Check if further dependencies/packages are needed...?
-#
-def check_sane():
-    try:
-        require_python_module('sane')
-    except MissingRequirement:
-        r = core.bundles.installbundle.install({'linux-fedora': ['python-imaging-sane']})
-        # other deps?
-        if not r:
-            raise
-
-## Checks if optional libraries are installed.
-# 
-# @since (added for PyCodeOCR)
-#
-def check_optionals():
-    try:
-        require_executable("gocr")
-    except MissingRequirement:
-        r = installbundle.install({'linux-fedora': ['gocr'],
-                                   'linux-ubuntu': ['gocr']})
-
-    try:
-        #require_python_module('pydmtx')
-        require_executable("dmtxread")
-    except MissingRequirement:
-        #r = installbundle.install({'linux-fedora': ['python-libdmtx']})
-        r = installbundle.install({'linux-fedora': ['libdmtx-utils'],
-                                   'linux-ubuntu': ['libdmtx-utils']})
-
-
-## Checks if all libraries whether needed or optional are installed.
-# 
-# @since (added for PyCodeOCR)
-#
-def check_all_pycodeocr_requirements():
-    print "Checking all PyCodeOCR requirements:"
-
-    # needed
-    print "  * NEEDED 'pygtk2'"
-    check_pygtk()
-    print "  * NEEDED 'sane' (scanner)"
-    check_sane()
-
-    # optional
-    print "  * OPTIONAL 'gocr', 'dmtxread'"
-    check_optionals()

Added: trunk/utils/__init__.py
===================================================================
--- trunk/utils/__init__.py	                        (rev 0)
+++ trunk/utils/__init__.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -0,0 +1,18 @@
+# -*- coding: utf-8 -*-
+##
+# @file      __init__.py
+# @namespace utils
+# @authors   drtrigon
+# @version   1.1
+# @brief     This is the utils package with important functions.
+#
+# $Id: checksum.py -1   $
+#
+# @section Description
+#
+# This is the utils package with important functions.
+#
+
+import utils.checksum
+import utils.requirements
+import utils.ocr


Property changes on: trunk/utils/__init__.py
___________________________________________________________________
Added: svn:eol-style
   + native

Copied: trunk/utils/checksum.py (from rev 49, trunk/checksum.py)
===================================================================
--- trunk/utils/checksum.py	                        (rev 0)
+++ trunk/utils/checksum.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -0,0 +1,47 @@
+##
+# @file      utils/checksum.py
+# @namespace utils::checksum
+# @authors   laserb, drtrigon
+# @version   1.1
+# @brief     Modulo 10 checksum for e-banking codes.
+#
+# @test Add unittest to check correct operation of this important module.
+#
+# $Id$
+#
+# @section Description
+#
+# Modulo 10 checksum for e-banking codes
+#
+
+class modulo10:
+	# thanks to http://www.hosang.ch/modulo10.aspx
+	# much more details http://www.sic.ch/de/dl_tkicch_dta.pdf ( page 51 )
+	# @see http://www.bundesbank.de/download/zahlungsverkehr/zv_pz201012.pdf
+	# @see http://www.bundesbank.de/zahlungsverkehr/zahlungsverkehr_pruefziffernberechnung.php
+	def __init__(self, amount, account, reference):
+		self.tabelle = [0,9,4,6,8,2,7,1,3,5]
+
+		self.account    = account
+		self.reference  = reference
+	
+		self.checksum_b = self._checksum(amount)
+		self.checksum_k = self._checksum(account)
+		self.checksum_r = self._checksum(reference)
+		self.checksum   = self.checksum_b and self.checksum_k and self.checksum_r
+
+		try:
+			self.amount = str(int(amount[2:-1])/100.)
+		except ValueError:
+			self.amount = amount + " (error)"
+
+	def _run(self, nr):
+		self.uebertrag = 0
+		# iterate over each character
+		for i in range(len(nr)):
+			self.uebertrag = self.tabelle[(self.uebertrag + int(nr[i]) )%10]
+		
+		return (10-self.uebertrag)%10
+	
+	def _checksum(self, value):
+		return (int(value[-1]) == self._run(value[:-1]) )

Copied: trunk/utils/installbundle.py (from rev 46, trunk/installbundle.py)
===================================================================
--- trunk/utils/installbundle.py	                        (rev 0)
+++ trunk/utils/installbundle.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -0,0 +1,196 @@
+##
+# @file      utils/installbundle.py
+# @namespace utils::installbundle
+# @authors   University of Utah (VisTrails), drtrigon
+# @version   1.1
+# @brief     Module with utilities to try and install a bundle if possible.
+#
+# $Id$
+#
+# @section Description
+#
+# Module with utilities to try and install a bundle if possible.
+#
+############################################################################
+##
+## Copyright (C) 2006-2010 University of Utah. All rights reserved.
+##
+## This file is part of VisTrails.
+##
+## This file may be used under the terms of the GNU General Public
+## License version 2.0 as published by the Free Software Foundation
+## and appearing in the file LICENSE.GPL included in the packaging of
+## this file.  Please review the following to ensure GNU General Public
+## Licensing requirements will be met:
+## http://www.opensource.org/licenses/gpl-license.php
+##
+## If you are unsure which license is appropriate for your use (for
+## instance, you are interested in developing a commercial derivative
+## of VisTrails), please contact us at vis...@sc....
+##
+## This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
+## WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
+##
+## Changed 2011 by drtrigon. Source from 'core.bundles.installbundle'
+## http://downloads.sourceforge.net/project/vistrails/vistrails/nightly/vistrails-src-nightly.tar.gz
+##
+## This file was modified to be used as part of PyCodeOCR.
+##
+############################################################################
+#
+
+"""Module with utilities to try and install a bundle if possible."""
+
+import os
+
+##############################################################################
+
+def linux_ubuntu_install(package_name):
+    qt = has_qt()
+    hide_splash_if_necessary()
+        
+    if qt:
+        cmd = core.system.vistrails_root_directory()
+        cmd += '/core/bundles/linux_ubuntu_install.py'
+    else:
+        cmd = 'apt-get install -y'
+
+    if type(package_name) == str:
+        cmd += ' ' + package_name
+    elif type(package_name) == list:
+        for package in package_name:
+            if type(package) != str:
+                raise TypeError("Expected string or list of strings")
+            cmd += ' ' + package
+
+    if qt:
+        sucmd = guess_graphical_sudo() + " '" + cmd + "'"
+    else:
+        debug.warning("PyCodeOCR wants to install package(s) '%s'" %
+                      package_name)
+        sucmd = "sudo " + cmd
+
+    result = os.system(sucmd)
+
+    return (result == 0) # 0 indicates success
+
+def linux_fedora_install(package_name):
+    qt = has_qt()
+    hide_splash_if_necessary()
+    if qt:
+        cmd = core.system.vistrails_root_directory()
+        cmd += '/core/bundles/linux_fedora_install.py'
+    else:
+        cmd = 'yum -y install'
+
+    if type(package_name) == str:
+        cmd += ' ' + package_name
+    elif type(package_name) == list:
+        for package in package_name:
+            if type(package) != str:
+                raise TypeError("Expected string or list of strings")
+            cmd += ' ' + package
+
+    if qt:
+        sucmd = guess_graphical_sudo() + " " + cmd
+    else:
+        debug.warning(("PyCodeOCR wants to install package(s) '%s' through "
+                       "_sudo_. Make sure you are a sudoer.") % package_name)
+        sucmd = "sudo " + cmd
+
+    debug.warning("EXECUTING: sucmd")
+    result = os.system(sucmd)
+    debug.warning("RETURN VALUE: %s" % result)
+    return (result == 0)
+
+def show_question(which_files):
+    qt = has_qt()
+    if qt:
+        import gui.utils
+        if type(which_files) == str:
+            which_files = [which_files]
+        v = gui.utils.show_question("Required packages missing",
+                                    "One or more required packages are missing: " +
+                                    " ".join(which_files) +
+                                    ". PyCodeOCR can " +
+                                    "automaticallly install them. " +
+                                    "If you click OK, PyCodeOCR will need "+
+                                    "administrator privileges, and you " +
+                                    "might be asked for the administrator password.",
+                                    buttons=[gui.utils.OK_BUTTON,
+                                             gui.utils.CANCEL_BUTTON],
+                                    default=gui.utils.OK_BUTTON)
+        return v == gui.utils.OK_BUTTON
+    else:
+        print "Required package missing"
+        print ("A required package is missing, but PyCodeOCR can " +
+               "automatically install it. " +
+               "If you say Yes, PyCodeOCR will need "+
+               "administrator privileges, and you" +
+               "might be asked for the administrator password.")
+        print "Give PyCodeOCR permission to try to install package? (y/N)"
+        v = raw_input().upper()
+        return v == 'Y' or v == 'YES'
+
+
+def install(dependency_dictionary):
+    """Tries to import a python module. If unsuccessful, tries to install
+the appropriate bundle and then reimport. py_import tries to be smart
+about which system it runs on."""
+
+    # Ugly fix to avoid circular import
+    distro = guess_system()
+    if not dependency_dictionary.has_key(distro):
+        return False
+    else:
+        files = dependency_dictionary[distro]
+        if show_question(files):
+            callable_ = getattr(installbundle,
+                                distro.replace('-', '_') + '_install')
+            return callable_(files)
+        else:
+            return False
+
+##############################################################################
+
+## Guesses the operation system.
+# 
+# Replaces original VisTrails 'core.bundles.utils.guess_system'.
+#
+# @since (added for PyCodeOCR)
+#
+def guess_system():
+    import platform
+    #print os.name, platform.platform()
+    return ("%s-%s" % (platform.system(), platform.dist()[0])).lower()
+
+## Dummy function, returns False.
+# 
+# Replaces original VisTrails 'has_qt'.
+#
+# @since (added for PyCodeOCR)
+#
+def has_qt():
+    return False
+
+import installbundle
+
+## Dummy function, just passes.
+# 
+# Replaces original VisTrails 'hide_splash_if_necessary'.
+#
+# @since (added for PyCodeOCR)
+#
+def hide_splash_if_necessary():
+    pass
+
+## Debug warning output handler.
+# 
+# Replaces original VisTrails 'core.debug.warning'.
+#
+# @since (added for PyCodeOCR)
+#
+class dbg():
+    def warning(self, *args):
+        print "".join(args)
+debug = dbg()

Added: trunk/utils/ocr.py
===================================================================
--- trunk/utils/ocr.py	                        (rev 0)
+++ trunk/utils/ocr.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -0,0 +1,76 @@
+# -*- coding: utf-8 -*-
+##
+# @file      utils/ocr.py
+# @namespace utils::ocr
+# @authors   drtrigon
+# @version   1.1
+# @brief     Character recogition (OCR) functions.
+#
+# $Id$
+#
+# @section Description
+#
+# Character recogition (OCR) functions.
+#
+
+# python standard modules
+import re
+
+## Character correction after recogition (on basis that there should be numbers and few special chars).
+#
+# @test Add unittest to check correct operation of this important module/function.
+#
+def char_correction(data):
+	data = re.sub("\n", "", data)
+	print data
+	corrections = [
+	# Zero
+	("O", "0"), ("D", "0"), ("U", "0"), ("Q", "0"), ("\(\)", "0"), ("G", "0"), ("o", "0"),
+	# Two
+	("Z", "2"),
+	# Three
+	("\?>", "3"),
+	# Four
+	("l\.", "4"), ("A", "4"), ("l»", "4"), ("h", "4"), ("i»", "4"), ("l\xe2\x80\xa2", "4"), ("&", "4"),
+	# Five
+	("S", "5"),
+	# Six
+	("c>", "6"),
+	# Seven
+	#("\?", "7"),
+	# One (critical, therefore done as last; has also Overlapps e.g. with Four)
+	(chr(226)+chr(128)+chr(153)+"I", "1"), 
+	(chr(226)+chr(128)+chr(153)+"\|", "1"),
+	(chr(226)+chr(128)+chr(152)+"\|", "1"),
+	(chr(226)+chr(128)+chr(152)+"I", "1"),
+	(chr(39)+"I", "1"),
+	("I", "1"),
+	(chr(153)+"l", "1"),
+	(chr(39)+"l", "1"),
+	("l", "1"),
+	(chr(39)+"!", "1"),
+	(chr(39)+"\|", "1"),
+	("\|", "1"),
+	("'\]", "1"),
+	("'\}", "1"),
+	(r'\\', "1"),
+	# non-numbers
+	(" ", ""),
+	("\xe2", ""),
+	("\x80", ""),
+	("\xbb", ""),
+	("\x99", ""),
+	("\x98", ""),
+	("\x5f", ""),
+	("\x60", ""),
+	("'", ""),
+	#('\xbb', 187)('\x99', 153)('\x98', 152)
+	("\+", "+ "), 
+	("\x92", "1"),	# post correction
+	]
+	for item in corrections:
+		olddata = data
+		data = re.sub(item[0], item[1], data)
+		if not (data == olddata): print 'subst: "%s" => "%s"' % (item[0], item[1])
+
+	return data


Property changes on: trunk/utils/ocr.py
___________________________________________________________________
Added: svn:keywords
   + Id
Added: svn:eol-style
   + native

Copied: trunk/utils/requirements.py (from rev 49, trunk/requirements.py)
===================================================================
--- trunk/utils/requirements.py	                        (rev 0)
+++ trunk/utils/requirements.py	2011-04-27 19:43:41 UTC (rev 50)
@@ -0,0 +1,241 @@
+##
+# @file      utils/requirements.py
+# @namespace utils::requirements
+# @authors   University of Utah (VisTrails), drtrigon
+# @version   1.1
+# @brief     Online test presence of runtime components.
+#
+# $Id$
+#
+# @section Description
+#
+# module that allows online inspection of environment to test presence of
+# runtime components such as binaries, libraries, other python modules, etc.
+#
+############################################################################
+##
+## Copyright (C) 2006-2010 University of Utah. All rights reserved.
+##
+## This file is part of VisTrails.
+##
+## This file may be used under the terms of the GNU General Public
+## License version 2.0 as published by the Free Software Foundation
+## and appearing in the file LICENSE.GPL included in the packaging of
+## this file.  Please review the following to ensure GNU General Public
+## Licensing requirements will be met:
+## http://www.opensource.org/licenses/gpl-license.php
+##
+## If you are unsure which license is appropriate for your use (for
+## instance, you are interested in developing a commercial derivative
+## of VisTrails), please contact us at vis...@sc....
+##
+## This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
+## WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
+##
+## Changed 2011 by drtrigon. Source from 'core.requirements'
+## http://downloads.sourceforge.net/project/vistrails/vistrails/nightly/vistrails-src-nightly.tar.gz
+##
+## This file was modified to be used as part of PyCodeOCR.
+##
+############################################################################
+#
+
+"""module that allows online inspection of environment to test presence of
+runtime components such as binaries, libraries, other python modules, etc."""
+
+import sys
+
+##############################################################################
+
+def python_module_exists(module_name):
+    """python_module_exists(module_name): Boolean.
+Returns if python module of given name can be safely imported."""
+    
+    try:
+        sys.modules[module_name]
+        return True
+    except KeyError:
+        pass
+    try:
+        __import__(module_name)
+        return True
+    except ImportError:
+        return False
+    
+
+def executable_file_exists(filename):
+    """executable_file_exists(filename): Boolean.
+Returns if certain file is in current path and is executable."""
+    result = executable_is_in_path(filename)
+    if result == "":
+        result = executable_is_in_pythonpath(filename)
+    return result != ""
+
+# FIXME: Add documentation.
+
+def require_python_module(module_name):
+    if not python_module_exists(module_name):
+        raise MissingRequirement(module_name)
+
+def require_executable(filename):
+    if not executable_file_exists(filename):
+        raise MissingRequirement(filename)
+
+def check_pyqt4():
+    # checks for the presence of pyqt4, which is more important than the rest,
+    # since using pyqt requires a qapplication.
+    try:
+        require_python_module('PyQt4.QtGui')
+        require_python_module('PyQt4.QtOpenGL')
+    except MissingRequirement:
+        r = core.bundles.installbundle.install({'linux-ubuntu': ['python-qt4',
+                                                                 'python-qt4-gl',
+                                                                 'python-qt4-sql']})
+        if not r:
+            raise
+
+
+def check_all_vistrails_requirements():
+    pass
+
+    # check scipy
+#     try:
+#         require_python_module('scipy')
+#     except MissingRequirement:
+#         r = core.bundles.installbundle.install({'linux-ubuntu': 'python-scipy'})
+#         if not r:
+#             raise
+        
+
+##############################################################################
+
+class MissingRequirement(Exception):
+    """Raise this exception in packages when necessary items are missing."""
+    def __init__(self, req):
+        self.requirement = req
+    def __str__(self):
+        return "Missing Requirement: %s" % self.requirement
+
+##############################################################################
+
+import installbundle
+import os
+
+
+## Shows the full path of (shell) commands.
+# 
+# According to shell command 'which', look also at:
+# @see http://unixhelp.ed.ac.uk/CGI/man-cgi?which
+# @see http://stackoverflow.com/questions/377017/test-if-executable-exists-in-python
+#
+# @since (added for PyCodeOCR)
+#
+def which(program, search_path=None):
+    is_exe = lambda fpath: os.path.exists(fpath) and os.access(fpath, os.X_OK)
+    #is_exe = lambda fpath: os.path.exists(fpath)
+
+    fpath, fname = os.path.split(program)
+    if fpath:
+        if is_exe(program):
+            return program
+    else:
+        if not search_path:
+            search_path  = os.environ["PATH"].split(os.pathsep)
+            search_path += [ '.' ]    # add local path
+        for path in search_path:
+            exe_file = os.path.join(path, program)
+            if is_exe(exe_file):
+                return exe_file
+    return ""
+
+## Checks if executable is in system PATH.
+# 
+# Replaces original VisTrails 'core.system.executable_is_in_path'.
+#
+# @since (added for PyCodeOCR)
+#
+def executable_is_in_path(filename):
+    #print os.system(filename)
+    return which(filename)
+
+## Checks if executable is in system PYTHONPATH.
+# 
+# Replaces original VisTrails 'core.system.executable_is_in_pythonpath'.
+#
+# @since (added for PyCodeOCR)
+#
+def executable_is_in_pythonpath(filename):
+    return which(filename, search_path=os.environ.get("PYTHONPATH", None))
+
+
+## Checks if pygtk libraries are installed.
+# 
+# @since (added for PyCodeOCR)
+#
+# @todo Check if further dependencies/packages are needed...?
+#
+def check_pygtk():
+    # checks for the presence of pygtk, which is more important than the rest.
+    try:
+        require_python_module('pygtk')
+        #pygtk.require('2.0')
+        require_python_module('gtk')
+        require_python_module('gtk.glade')
+        require_python_module('gobject')
+    except MissingRequirement:
+        r = core.bundles.installbundle.install({'linux-fedora': ['pygtk2']})
+        # other deps?
+        if not r:
+            raise
+
+## Checks if sane libraries are installed.
+# 
+# @since (added for PyCodeOCR)
+#
+# @todo Check if further dependencies/packages are needed...?
+#
+def check_sane():
+    try:
+        require_python_module('sane')
+    except MissingRequirement:
+        r = core.bundles.installbundle.install({'linux-fedora': ['python-imaging-sane']})
+        # other deps?
+        if not r:
+            raise
+
+## Checks if optional libraries are installed.
+# 
+# @since (added for PyCodeOCR)
+#
+def check_optionals():
+    try:
+        require_executable("gocr")
+    except MissingRequirement:
+        r = installbundle.install({'linux-fedora': ['gocr'],
+                                   'linux-ubuntu': ['gocr']})
+
+    try:
+        #require_python_module('pydmtx')
+        require_executable("dmtxread")
+    except MissingRequirement:
+        #r = installbundle.install({'linux-fedora': ['python-libdmtx']})
+        r = installbundle.install({'linux-fedora': ['libdmtx-utils'],
+                                   'linux-ubuntu': ['libdmtx-utils']})
+
+
+## Checks if all libraries whether needed or optional are installed.
+# 
+# @since (added for PyCodeOCR)
+#
+def check_all_pycodeocr_requirements():
+    print "Checking all PyCodeOCR requirements:"
+
+    # needed
+    print "  * NEEDED 'pygtk2'"
+    check_pygtk()
+    print "  * NEEDED 'sane' (scanner)"
+    check_sane()
+
+    # optional
+    print "  * OPTIONAL 'gocr', 'dmtxread'"
+    check_optionals()


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.