def processCollectedText(collectedText, textElementList, parent_map,
superscriptNode, unicodeFont):
clearedTextElements = []
global debug_output
# First, change the text
if debug_output:
print('** CONVERTING %s to Unicode. ' % collectedText.encode('utf-8'))
convertedText = osageConversion.oldOsageToUnicode(collectedText)
convertedCount = 0
if convertedText != collectedText:
convertedCount = 1
else:
print('---- Not converted: %s' % collectedText.encode('utf-8'))
# 1. Reset text in first element
if not textElementList:
return 0
# ∂textElementList[0].text = convertedText
parent = parent_map[textElementList[0]]
# Fix font and superscripting
fixElementAndParent(textElementList[0], parent, convertedText,
unicodeFont) # Update the font in this item.
if superscriptNode:
superscriptNode.val = 'baseline'
# 2. Clear text in other elements
for element in textElementList[1:]:
element.text = ''
clearedTextElements.append(element)
return convertedCount, clearedTextElements
def replFunc(matchObj):
if matchObj.group(0):
# Avoid converting strings with [aeo] after capital, but in English
if notOsageLatinLower.search(matchObj.group(0)):
return matchObj.group(0)
else:
return osageConversion.oldOsageToUnicode(matchObj.group(0))
def checkAndConvertText(textIn):
if textIn[0] == '=':
# Ignore function calls
return textIn
# Handle text in font-labeled regions
result = osageConversion.oldOsageToUnicode(textIn)
return result
def checkAndConvertText(textIn):
if textIn[0] == '=':
# Ignore function calls
return textIn
#if notOsageLatinLower.search(textIn):
# return textIn
# Handle Latin and TraditionalOsage private use characters.
tryResult = osageConversion.oldOsageToUnicode(textIn)
#tryResult = re.subn(osageConvertPattern, replFunc, textIn)
#if tryResult[1] >= 1:
return tryResult #[0]
def convertToOsageUnicode(oldText):
newText = osageConversion.oldOsageToUnicode(
oldText) # FOR TESTING oldText[::-1]
return newText
def replFunc(matchObj):
if matchObj.group(0):
if notOsageLatinLower.search(matchObj.group(0)):
return matchObj.group(0)
else:
return osageConversion.oldOsageToUnicode(matchObj.group(0))