def initialize_tesseract(self):
"""Create tesseract api
"""
self.tesseract = tess.get_tesseract(os.path.dirname(__file__))
if not self.tesseract:
self.show_msg('Tesseract initialization failed...')
return
self.api = self.tesseract.TessBaseAPICreate()
tessdata_prefix = tess.get_tessdata_prefix()
#current_locale = locale.getlocale() # Save current locale
# Switch to C locale to handle
# Error: Illegal min or max specification!
# "Fatal error encountered!" == NULL:Error:Assert failed:in file
# ../../tesseract-ocr/ccutil/globaloc.cpp, line 75
locale.setlocale(locale.LC_ALL, 'C')
retc = self.tesseract.TessBaseAPIInit3(self.api,
tessdata_prefix,
self.lang)
# Restore saved locale
# locale.setlocale(locale.LC_ALL, current_locale)
if (retc):
self.tesseract.TessBaseAPIDelete(self.api)
self.show_msg('<span style="color:red">Could not initialize ' \
'tesseract.</span>')
return
self.show_msg('Tesseract %s initialized with language \'%s\'.' % \
(tess.VERSION, self.lang))
def initialize_tesseract(self):
"""Create tesseract api
"""
self.tesseract = tess.get_tesseract(os.path.dirname(__file__))
if not self.tesseract:
self.show_msg('Tesseract initialization failed...')
return
self.api = self.tesseract.TessBaseAPICreate()
tessdata_prefix = tess.get_tessdata_prefix()
#current_locale = locale.getlocale() # Save current locale
# Switch to C locale to handle
# Error: Illegal min or max specification!
# "Fatal error encountered!" == NULL:Error:Assert failed:in file
# ../../tesseract-ocr/ccutil/globaloc.cpp, line 75
locale.setlocale(locale.LC_ALL, 'C')
retc = self.tesseract.TessBaseAPIInit3(self.api, tessdata_prefix,
self.lang)
# Restore saved locale
# locale.setlocale(locale.LC_ALL, current_locale)
if (retc):
self.tesseract.TessBaseAPIDelete(self.api)
self.show_msg('<span style="color:red">Could not initialize ' \
'tesseract.</span>')
return
self.show_msg('Tesseract %s initialized with language \'%s\'.' % \
(tess.VERSION, self.lang))
def __init__(self,parent = None):
QtCore.QObject.__init__(self)
'''
initialize
'''
self.tesseract = tesstool.get_tesseract(os.path.dirname(__file__))
if not self.tesseract:
print('Tesseract initialization failed...')
return
self.api = self.tesseract.TessBaseAPICreate()
self.tessdata_prefix = tesstool.get_tessdata_prefix()
locale.setlocale(locale.LC_ALL, 'C')
#print('Tesseract %s initialized!' % (tesstool.VERSION))
self.pixImage = None
self.VERSION = tesstool.VERSION
self.parent = parent
def on_pushButtonShow_pressed(self):
"""Display rectangles
"""
tessdata_prefix = tess.get_tessdata_prefix()
locale.setlocale(locale.LC_ALL, 'C')
retc = 0
if self.lang != str(self.comboBoxLang.currentText()):
self.lang = str(self.comboBoxLang.currentText())
retc = self.tesseract.TessBaseAPIInit3(self.api,
tessdata_prefix,
self.lang)
self.show_msg('Using language \'%s\'.' % self.lang)
if (retc):
self.tesseract.TessBaseAPIDelete(self.api)
self.show_msg('<span style="color:red">Could not re-initialize ' \
'tesseract.</span>')
return
# Shut up tesseract - there could be a lot of unwanted messages
#tesseract.TessBaseAPISetVariable(api, "debug_file", "/dev/null")
self.tesseract.TessBaseAPIClear(self.api)
# Set PIX structure to tesseract api
self.tesseract.TessBaseAPISetImage2(self.api, self.pix_image)
self.tesseract.TessBaseAPISetPageSegMode(self.api,
self.comboBoxPSM.currentIndex())
# Get info(BOXA structure) about lines(RIL_TEXTLINE) from image in api
boxa = self.tesseract.TessBaseAPIGetComponentImages(self.api,
self.comboBoxRIL.currentIndex(),
1, None, None)
if not boxa:
self.show_msg('No component found. Try to change PSM or RIL.')
return
# Get info about number of items on image
n_items = self.leptonica.boxaGetCount(boxa)
psm = tess.PSM[self.comboBoxPSM.currentIndex()]
ril = tess.RIL[self.comboBoxRIL.currentIndex()]
self.show_msg('<span style="color:green">' \
'Found %d image components with %s and %s.</span>' % \
(n_items, psm, ril))
ocr_psm = tess.PSM_SINGLE_BLOCK
if ril == 'RIL_PARA':
ocr_psm = tess.PSM_SINGLE_BLOCK
elif ril == 'RIL_TEXTLINE':
ocr_psm = tess.PSM_SINGLE_LINE
elif ril == 'RIL_WORD':
ocr_psm = tess.PSM_SINGLE_WORD
elif ril == 'RIL_SYMBOL':
ocr_psm = tess.PSM_SINGLE_CHAR
self.tesseract.TessBaseAPISetPageSegMode(self.api, ocr_psm)
# Set up result type (BOX structure) for leptonica function boxaGetBox
self.leptonica.boxaGetBox.restype = lept.BOX_PTR_T
self.leptonica.boxaGetBox.argtypes = []
n_boxes = len(self.box_data)
if n_boxes:
for idx in xrange(n_boxes):
self.scene.removeItem(self.box_data[idx])
# Display items and print its info
box_items = []
for item in range(0, n_items):
lept_box = self.leptonica.boxaGetBox(boxa, item, lept.L_CLONE)
box = lept_box.contents
self.tesseract.TessBaseAPISetRectangle(self.api,
box.x, box.y, box.w, box.h)
ocr_result = self.tesseract.TessBaseAPIGetUTF8Text(self.api)
result_text = ctypes.string_at(ocr_result).decode('utf-8').strip()
conf = self.tesseract.TessBaseAPIMeanTextConf(self.api)
self.show_msg('Box[%d]: x=%d, y=%d, w=%d, h=%d, ' \
'confidence: %d, ' \
'text: <span style="color:blue">%s</span>' % \
(item, box.x, box.y, box.w, box.h,
conf, result_text
))
box_items.append(self.scene.addRect(box.x, box.y, box.w, box.h,
QPen(QColor(255, 0, 0, 255)),
QBrush(QColor(255, 0, 0, 100))))
box_items[item].setAcceptHoverEvents(True)
box_items[item].setToolTip("Box[%d]: confidence:%s, text:%s" % \
(item, conf, result_text))
QCoreApplication.processEvents()
self.box_data = box_items
def on_pushButtonShow_pressed(self):
"""Display rectangles
"""
tessdata_prefix = tess.get_tessdata_prefix()
locale.setlocale(locale.LC_ALL, 'C')
retc = 0
if self.lang != str(self.comboBoxLang.currentText()):
self.lang = str(self.comboBoxLang.currentText())
retc = self.tesseract.TessBaseAPIInit3(self.api, tessdata_prefix,
self.lang)
self.show_msg('Using language \'%s\'.' % self.lang)
if (retc):
self.tesseract.TessBaseAPIDelete(self.api)
self.show_msg('<span style="color:red">Could not re-initialize ' \
'tesseract.</span>')
return
# Shut up tesseract - there could be a lot of unwanted messages
#tesseract.TessBaseAPISetVariable(api, "debug_file", "/dev/null")
self.tesseract.TessBaseAPIClear(self.api)
# Set PIX structure to tesseract api
self.tesseract.TessBaseAPISetImage2(self.api, self.pix_image)
self.tesseract.TessBaseAPISetPageSegMode(
self.api, self.comboBoxPSM.currentIndex())
# Get info(BOXA structure) about lines(RIL_TEXTLINE) from image in api
boxa = self.tesseract.TessBaseAPIGetComponentImages(
self.api, self.comboBoxRIL.currentIndex(), 1, None, None)
if not boxa:
self.show_msg('No component found. Try to change PSM or RIL.')
return
# Get info about number of items on image
n_items = self.leptonica.boxaGetCount(boxa)
psm = tess.PSM[self.comboBoxPSM.currentIndex()]
ril = tess.RIL[self.comboBoxRIL.currentIndex()]
self.show_msg('<span style="color:green">' \
'Found %d image components with %s and %s.</span>' % \
(n_items, psm, ril))
ocr_psm = tess.PSM_SINGLE_BLOCK
if ril == 'RIL_PARA':
ocr_psm = tess.PSM_SINGLE_BLOCK
elif ril == 'RIL_TEXTLINE':
ocr_psm = tess.PSM_SINGLE_LINE
elif ril == 'RIL_WORD':
ocr_psm = tess.PSM_SINGLE_WORD
elif ril == 'RIL_SYMBOL':
ocr_psm = tess.PSM_SINGLE_CHAR
self.tesseract.TessBaseAPISetPageSegMode(self.api, ocr_psm)
# Set up result type (BOX structure) for leptonica function boxaGetBox
self.leptonica.boxaGetBox.restype = lept.BOX_PTR_T
self.leptonica.boxaGetBox.argtypes = []
n_boxes = len(self.box_data)
if n_boxes:
for idx in xrange(n_boxes):
self.scene.removeItem(self.box_data[idx])
# Display items and print its info
box_items = []
for item in range(0, n_items):
lept_box = self.leptonica.boxaGetBox(boxa, item, lept.L_CLONE)
box = lept_box.contents
self.tesseract.TessBaseAPISetRectangle(self.api, box.x, box.y,
box.w, box.h)
ocr_result = self.tesseract.TessBaseAPIGetUTF8Text(self.api)
result_text = ctypes.string_at(ocr_result).decode('utf-8').strip()
conf = self.tesseract.TessBaseAPIMeanTextConf(self.api)
self.show_msg('Box[%d]: x=%d, y=%d, w=%d, h=%d, ' \
'confidence: %d, ' \
'text: <span style="color:blue">%s</span>' % \
(item, box.x, box.y, box.w, box.h,
conf, result_text
))
box_items.append(
self.scene.addRect(box.x, box.y, box.w, box.h,
QPen(QColor(255, 0, 0, 255)),
QBrush(QColor(255, 0, 0, 100))))
box_items[item].setAcceptHoverEvents(True)
box_items[item].setToolTip("Box[%d]: confidence:%s, text:%s" % \
(item, conf, result_text))
QCoreApplication.processEvents()
self.box_data = box_items