def getDipValue(element, attributeName) :
if attributeName in element.attrib:
value = element.get(attributeName)
if (not TextUtils.isEmpty(value) and value.endswith(Constants.UNIT_DIP)) :
dValue = float(value[0:len(value) - 3])
return dValue
return 0
def containAllSpacesOrInvalidChars(text):
if TextUtils.isEmpty(text):
return True
asciiContain = all([ord(char) < 33 or ord(char) > 126 for char in text])
invalidChar = len([char for char in text if char not in printable]) != 0
allSpace = all([" " == c or '\n' == c for c in text])
return not (allSpace or invalidChar or asciiContain)
def __init__(self, appName):
super().__init__()
self.mAppName = appName
self.mRoot = Element(Constants.ELEMENT_RESOURCE)
element = SubElement(self.mRoot, Constants.ELEMENT_STRING)
element.set(Constants.ATTRIBUTE_NAME, "app_name")
element.text = TextUtils.formatText(self.mAppName)
self.mDataIndexMap = {}
def getOriginalIdFromRectView(self, baseView):
if (baseView.mType == RectView.VIEW_TYPE_IMAGE):
return baseView.getImageInfo().drawableId
elif baseView.mType == RectView.VIEW_TYPE_TEXT:
if (TextUtils.isEmpty(baseView.getTextInfo().id)):
return self.mWriter.addResource(
baseView.getTextInfo().textWrapper.text)
else:
return baseView.getTextInfo().id
else:
return None
def getLineText(self, rect):
try:
self.mHandle.SetImage(self.mBufferedImageRgbaImage)
self.mHandle.SetRectangle(rect.x, rect.y, rect.width, rect.height)
text = self.mHandle.GetUTF8Text()
if (TextUtils.isEmpty(text)):
self.mHandle = PyTessBaseAPI(psm=PSM.SINGLE_LINE)
self.mHandle.SetImage(self.mBufferedImageRgbaImage)
self.mHandle.SetRectangle(rect.x, rect.y, rect.width,
rect.height)
text = self.mHandle.GetUTF8Text()
if (TextUtils.isEmpty(text)):
self.mHandle.SetImage(self.getImage(rect))
text = self.mHandle.GetUTF8Text()
self.mHandle = PyTessBaseAPI(psm=PSM.AUTO)
return text
except Exception as error:
print('Caught this error: ' + repr(error))
return ""
def addResource(self, value):
formatText = TextUtils.formatText(value)
_id = "string_"
if formatText in self.mDataIndexMap:
index = self.mDataIndexMap[formatText]
_id += str(index)
else:
_id += str(self.mId)
self.mDataIndexMap[value] = self.mId
self.mId = self.mId + 1
element = SubElement(self.mRoot, Constants.ELEMENT_STRING)
element.set(Constants.ATTRIBUTE_NAME, _id)
element.text = formatText
return _id
def hasDipValue(element, attributeName) :
value = element.get(attributeName)
if (not TextUtils.isEmpty(value) and value.endswith(Constants.UNIT_DIP)) :
return True
return False
def isValidTextUsingBoundaryCheck(self, ocrTextWrapper):
# confident between TextProcessor.TEXT_CONFIDENT_THRESHOLD and
# TextProcessor.TEXT_CONFIDENT_THRESHOLD +
# TextProcessor.TEXT_CONFIDENT_THRESHOLD_SECONDARY_RANGE
if (TextUtils.isEmpty(ocrTextWrapper.text)):
# We cannot calculate width of empty text
return True
# return True
# frc = mGraphics.getFontRenderContext()
# font = QFont(ocrTextWrapper.fontName,ocrTextWrapper.fontSize)
# loc = Point(0, 0)
# layout = TextLayout(ocrTextWrapper.text,font, frc)
# layout.draw(mGraphics, loc.getX(), loc.getY())
# bound = layout.getBounds()
width, height = self.getTextDimensions(ocrTextWrapper.text,
ocrTextWrapper.fontName,
ocrTextWrapper.fontSize)
fontRatio = float(height / width)
boundRatio = float(ocrTextWrapper.height / ocrTextWrapper.width)
fontArea = self.mDipCalculator.dipToHeightPx(
height) * self.mDipCalculator.dipToWidthPx(width)
boundArea = float(ocrTextWrapper.width * ocrTextWrapper.height)
#
# the different between dimensions of the text should be smaller than
# 10% of the max dimension.
# System.out.prln(" Ratio: " + fontRatio + ", " + boundRatio + ", "
# + Math.abs(boundRatio - fontRatio)
# / Math.max(boundRatio, fontRatio) + "," + fontArea + ", "
# + boundArea + ", " + Math.min(fontArea, boundArea)
# / Math.max(fontArea, boundArea))
# It the bound is square, it less likely that this text is correct
# TODO: This rule may not need it
# if (float(min(ocrTextWrapper.getWidth(),ocrTextWrapper.getHeight()) / max( ocrTextWrapper.getWidth(),
# ocrTextWrapper.getHeight())) > 0.95) :
# # if drawing text cannot create square, sorry -> invalid
# if (float(min(width, height) / max(width, height)) <= 0.95 and not validWord(ocrTextWrapper.text)) :
# return False
#
#
#
# print(self.mDipCalculator.dipToWidthPx(width), self.mDipCalculator.dipToHeightPx(height))
# print( ocrTextWrapper.width, ocrTextWrapper.height)
dimension = abs(boundRatio - fontRatio) / max(boundRatio, fontRatio)
# print(dimension)
dimensionCheck = abs(boundRatio - fontRatio) / max(
boundRatio, fontRatio
) <= Constants.TEXT_CONFIDENT_ACCEPTANCE_DIMENSION_RATIO_DIFFERENCE_THRESHOLD
areaCheckVal = min(fontArea, boundArea) / max(fontArea, boundArea)
# print(areaCheckVal)
# print(ocrTextWrapper.text)
areaCheck = min(fontArea, boundArea) / max(
fontArea,
boundArea) >= Constants.TEXT_AREA_ACCEPTANCE_DIFFERENCE_THRESHOLD
return dimensionCheck and areaCheck
def pruneBasicInternal(self, parent, view):
# TODO: if this view is too small and it has no children,so we don't
# need them
if self.mDipCalculator.isViewToBeIgnore(view.width, view.height):
if (parent != None):
parent.mChildren.remove(view)
return
#
#
allChildrenAreTooSmall = self.isAllChildrenTooSmall(view)
if not allChildrenAreTooSmall and len(view.mChildren) != 0:
removedChildren = []
for childView in view.mChildren:
if self.mDipCalculator.isViewToBeIgnore(
childView.width, childView.height):
removedChildren.append(childView)
view.mChildren = [
x for x in view.mChildren if x not in removedChildren
]
for childView in view.mChildren:
self.pruneBasicInternal(view, childView)
# # add this drawable if we did not want to show any children
# # here
#
isAImageView = self.isFullImage(view)
if isAImageView:
# if not view.hasText() and allChildrenAreTooSmall and len(view.mChildren) == 0 :
currentMat = ImageUtil.getImageFromRect(self.mImage, view.bound())
iconInfo = IconInfo(currentMat)
drawableId = ""
if iconInfo in self.interestedIcons:
drawableId = self.interestedIcons[iconInfo]
viewsSameDrawable = None
if (TextUtils.isEmpty(drawableId)):
drawableId = self.mDrawableWriter.addResourceDirectly(
currentMat, view)
self.interestedIcons[iconInfo] = drawableId
viewsSameDrawable = []
self.mDrawableMap[drawableId] = viewsSameDrawable
else:
viewsSameDrawable = self.mDrawableMap[drawableId]
view.mType = RectViewTypes.VIEW_TYPE_IMAGE
view.mImageInfo.iconInfo = iconInfo
view.mImageInfo.drawableId = drawableId
view.mChildren = []
viewsSameDrawable.append(view)
elif view.hasTextRecusive():
# process text view
textWithLocations = view.mTextWithLocations
view.mColor = ColorUtil.findDominateColor(view, self.mImage)
for textWrapper in textWithLocations:
newHeight = TesseractOCR.increaseHeight(textWrapper.height)
textView = textWrapper.boundRectView
newY = textView.y - (newHeight - textView.height) / 2.0
textView.y = newY
textView.x = textWrapper.x
textView.width = textWrapper.width
textView.height = newHeight
textView.mType = RectViewTypes.VIEW_TYPE_TEXT
textView.mTextInfo.textWrapper = textWrapper
textView.rect = Rect(textView.x, textView.y, textWrapper.width,
textView.height)
color = ColorUtil.findDominateColorForTextView(
textView, self.mImage)
textView.mColor = color[0]
textView.textColor = color[1]
# textView.mColor = ColorUtil.findDominateColor(textView,self.mImage)
view.addChild(textView)
# currentMat = ImageUtil.getImageFromRect(self.mImage, textView.bound())
# iconInfo = IconInfo(currentMat)
# drawableId = ""
# if iconInfo in self.interestedIcons:
# drawableId = self.interestedIcons[iconInfo]
# viewsSameDrawable = None
# if (TextUtils.isEmpty(drawableId)) :
# drawableId = self.mDrawableWriter.addResourceDirectly(currentMat,view)
# self.interestedIcons[iconInfo] = drawableId
# viewsSameDrawable = []
# self.mDrawableMap[drawableId] = viewsSameDrawable
# else :
# viewsSameDrawable = self.mDrawableMap[drawableId]
#
# textView.mType = RectViewTypes.VIEW_TYPE_IMAGE
# textView.mImageInfo.iconInfo = iconInfo
# textView.mImageInfo.drawableId = drawableId
# textView.mChildren = []
# viewsSameDrawable.append(textView)
# Update Bound of parent text view
if len(view.mChildren) > 0:
allViews = []
allViews.extend(view.mChildren)
allViews.append(view)
view.bound = RectUtil.findBoundRectangle(allViews)
def getLayoutId(self):
if (TextUtils.isEmpty(self.layoutId)):
return self.getVariableName()
return self.layoutId
def findInInvalidTextByList(self, ocrs):
invalidTexts = []
groups = GroupUtil.group(ocrs, self.sameGroup)
for group in groups:
if len(
group
) >= Constants.MIN_ACCEPTABLE_LIST_SIZE_FOR_INVALID_LIST_TEXT:
for ocrTextWrapper in group:
if ocrTextWrapper.confidence <= Constants.MIN_INVALID_LIST_TEXT_THRESHOLD and not TextUtils.isEmpty(
ocrTextWrapper.text) and len(
ocrTextWrapper.text) == 1:
invalidTexts.append(ocrTextWrapper)
return invalidTexts
def __init__(self, name):
self.mName = TextUtils.removeInvalidProjectNameChars(name.lower())
self.mPath = ""
def getBaseNameRemoveInvalidChars(fileName):
return TextUtils.removeInvalidProjectNameChars(basename(fileName))