def showDiffRandom(fileName):
# at this point shows the diff of the first article
article = getRandomArticle(fileName)
title = article.getTitle()
convertedArticle = None
if wikipediasql.fConvertedCacheExists(fileName):
convertedArticle = findConvertedArticle(fileName,title)
if not convertedArticle:
print "didn't find article '%s' in the converted cache" % title
origTxt = article.getText()
origTxt = arsutils.normalizeNewlines(origTxt)
if convertedArticle:
converted = arsutils.normalizeNewlines(convertedArticle.getText())
arsutils.showTxtDiff(origTxt, converted)
else:
converted = articleconvert.convertArticle(article.getTitle(), article.getText())
converted = arsutils.normalizeNewlines(converted)
arsutils.showTxtDiff(origTxt,converted)
def showDiffRandom(fileName):
# at this point shows the diff of the first article
article = getRandomArticle(fileName)
title = article.getTitle()
convertedArticle = None
if wikipediasql.fConvertedCacheExists(fileName):
convertedArticle = findConvertedArticle(fileName, title)
if not convertedArticle:
print "didn't find article '%s' in the converted cache" % title
origTxt = article.getText()
origTxt = arsutils.normalizeNewlines(origTxt)
if convertedArticle:
converted = arsutils.normalizeNewlines(convertedArticle.getText())
arsutils.showTxtDiff(origTxt, converted)
else:
converted = articleconvert.convertArticle(article.getTitle(),
article.getText())
converted = arsutils.normalizeNewlines(converted)
arsutils.showTxtDiff(origTxt, converted)
def showDiffTitle(fileName, title, fSave=False, fForceConvert=False):
article = findOrigArticle(fileName, title)
if not article:
print "couldn't find article with the title %s" % title
return
origTxt = article.getText()
origTxt = arsutils.normalizeNewlines(origTxt)
if fForceConvert:
convertedTxt = articleconvert.convertArticle(article.getTitle(),
article.getText())
else:
title = article.getTitle(
) # re-get the title in case this was a redirect
convertedArticle = None
if wikipediasql.fConvertedCacheExists(fileName):
convertedArticle = findConvertedArticle(fileName, title)
else:
print "Converted cache for '%s' doesn't exist" % fileName
sys.exit(0)
if None == convertedArticle:
print "didn't find converted article, generating it myself"
convertedTxt = articleconvert.convertArticle(
article.getTitle(), article.getText())
else:
convertedTxt = convertedArticle.getText()
convertedTxt = arsutils.normalizeNewlines(convertedTxt)
if fSave:
title = article.getTitle()
title = title.replace(" ", "_")
fo = open("%s_orig.txt" % title, "wb")
fo.write(origTxt)
fo.close()
fo = open("%s_conv.txt" % title, "wb")
fo.write(convertedTxt)
fo.close()
arsutils.showTxtDiff(origTxt, convertedTxt)
def showDiffTitle(fileName,title,fSave=False,fForceConvert=False):
article = findOrigArticle(fileName,title)
if not article:
print "couldn't find article with the title %s" % title
return
origTxt = article.getText()
origTxt = arsutils.normalizeNewlines(origTxt)
if fForceConvert:
convertedTxt = articleconvert.convertArticle(article.getTitle(), article.getText())
else:
title = article.getTitle() # re-get the title in case this was a redirect
convertedArticle = None
if wikipediasql.fConvertedCacheExists(fileName):
convertedArticle = findConvertedArticle(fileName,title)
else:
print "Converted cache for '%s' doesn't exist" % fileName
sys.exit(0)
if None == convertedArticle:
print "didn't find converted article, generating it myself"
convertedTxt = articleconvert.convertArticle(article.getTitle(), article.getText())
else:
convertedTxt = convertedArticle.getText()
convertedTxt = arsutils.normalizeNewlines(convertedTxt)
if fSave:
title = article.getTitle()
title = title.replace(" ", "_")
fo = open("%s_orig.txt" % title, "wb")
fo.write(origTxt)
fo.close()
fo = open("%s_conv.txt" % title, "wb")
fo.write(convertedTxt)
fo.close()
arsutils.showTxtDiff(origTxt, convertedTxt)
