def runTests(fileName):
redirects = {}
articleTitles = {}
testCount = 0
failedCount = 0
for test in iterTests(fileName):
orig = test.orig
expected = test.expected
converted = articleconvert.convertArticle(test.name,orig)
expected = arsutils.normalizeNewlines(expected)
converted = arsutils.normalizeNewlines(converted)
if converted != expected:
failedCount += 1
test.setConverted(converted)
failedList.append(test)
sys.stdout.write("-")
else:
sys.stdout.write(".")
noLinks = articleconvert.removeInvalidLinks(converted,redirects,articleTitles)
testCount += 1
print
print "Total tests: %d" % testCount
print "Failed tests: %d" % failedCount
dumpFailed()
diffFirstFailed()
def runTests(fileName):
redirects = {}
articleTitles = {}
testCount = 0
failedCount = 0
for test in iterTests(fileName):
orig = test.orig
expected = test.expected
converted = articleconvert.convertArticle(test.name, orig)
expected = arsutils.normalizeNewlines(expected)
converted = arsutils.normalizeNewlines(converted)
if converted != expected:
failedCount += 1
test.setConverted(converted)
failedList.append(test)
sys.stdout.write("-")
else:
sys.stdout.write(".")
noLinks = articleconvert.removeInvalidLinks(converted, redirects,
articleTitles)
testCount += 1
print
print "Total tests: %d" % testCount
print "Failed tests: %d" % failedCount
dumpFailed()
diffFirstFailed()
def dumpArticle(fileName,title):
for article in iterArticlesExactTitle(fileName,title):
if not article:
print "couldn't find the body of article %s" % title
return
title = article.getTitle().strip() + "\n"
txt = article.getTxt()
converted = articleconvert.convertArticle(title,txt)
print "TITLE: %s" % title
print "ORIGINAL: %s" % txt
print "CONVERTED: %s" % converted
return
def dumpArticle(fileName, title):
for article in iterArticlesExactTitle(fileName, title):
if not article:
print "couldn't find the body of article %s" % title
return
title = article.getTitle().strip() + "\n"
txt = article.getTxt()
converted = articleconvert.convertArticle(title, txt)
print "TITLE: %s" % title
print "ORIGINAL: %s" % txt
print "CONVERTED: %s" % converted
return
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)
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 convertArticles(sqlDump, articleLimit):
count = 0
redirects = {}
articleTitles = {}
fTesting = False
if fTesting:
fUseCache = False
fRecreateCache = True
else:
fUseCache = True
fRecreateCache = False
for article in wikipediasql.iterWikipediaArticles(sqlDump, articleLimit, fUseCache, fRecreateCache):
# we only convert article from the main namespace
assert article.getNamespace() == wikipediasql.NS_MAIN
title = article.getTitle()
if article.fRedirect():
redirects[title] = article.getRedirect()
else:
txt = article.getText()
#links = articleconvert.articleExtractLinks(txt)
#articleTitles[title] = links
articleTitles[title] = 1
count += 1
if 0 == count % 1000:
sys.stderr.write("processed %d rows, last title=%s\n" % (count,title.strip()))
if articleLimit and count >= articleLimit:
break
# verify redirects
print "Number of real articles: %d" % len(articleTitles)
print "Number of all redirects: %d (%d in total)" % (len(redirects), len(articleTitles)+len(redirects))
unresolvedCount = 0
setUnresolvedRedirectWriter(sqlDump)
redirectsExisting = {}
for (title,redirect) in redirects.items():
redirectResolved = resolveRedirect(title,redirect,redirects,articleTitles)
if None == redirectResolved:
unresolvedCount +=1
#print "redirect '%s' (to '%s') not resolved" % (title,redirect)
else:
redirectsExisting[title] = redirectResolved
closeUnresolvedRedirectWriter()
print "Number of unresolved redirects: %d" % unresolvedCount
dbName = getDbNameFromFileName(sqlDump)
ipedia_write_cur = getNamedCursor(getIpediaConnection(dbName), "ipedia_write_cur")
# go over articles again (hopefully now using the cache),
# convert them to a destination format (including removing invalid links)
# and insert into a database
sizeStats = {}
count = 0
convWriter = wikipediasql.ConvertedArticleCacheWriter(sqlDump)
convWriter.open()
for article in wikipediasql.iterWikipediaArticles(sqlDump, articleLimit, True, False):
title = article.getTitle()
articleSize = 0 # 0 is for redirects, which we don't log
if article.fRedirect():
convertedArticle = ConvertedArticleRedirect(article.getNamespace(), title, article.getRedirect())
else:
txt = article.getText()
converted = articleconvert.convertArticle(title, txt)
try:
noLinks = articleconvert.removeInvalidLinks(converted,redirects,articleTitles)
except:
print "exception in articleconvert.removeInvalidLinks"
print "title: _%s_" % title
print "txt:\n_%s_" % txt
print "converted:\n_%s_" % converted
raise
if noLinks:
converted = noLinks
convertedArticle = ConvertedArticle(article.getNamespace(), article.getTitle(), converted)
articleSize = len(converted)
if article.fRedirect():
if redirectsExisting.has_key(title):
redirect = redirectsExisting[title]
try:
title = title.replace("_", " ")
redirect = redirect.replace("_", " ")
ipedia_write_cur.execute("""INSERT INTO redirects (title, redirect) VALUES ('%s', '%s')""" % (dbEscape(title), dbEscape(redirect)))
except:
print "DUP REDERICT '%s' => '%s'" % (title, redirect)
else:
title = title.replace("_", " ")
if g_fVerbose:
log_txt = "title: %s " % title
try:
ipedia_write_cur.execute("""INSERT INTO articles (title, body) VALUES ('%s', '%s')""" % (dbEscape(title), dbEscape(converted)))
if g_fVerbose:
log_txt += "*New record"
except:
# assuming that the exception happend because of trying to insert
# item with a duplicate title (duplication due to lower-case
# conversion might convert 2 differnt titles into the same,
# lower-cased title)
if g_fShowDups:
print "dup: " + title
if g_fVerbose:
log_txt += "Update existing record"
print "DUP ARTICLE: '%s'" % title
ipedia_write_cur.execute("""UPDATE articles SET body='%s' WHERE title='%s'""" % (dbEscape(converted), dbEscape(title)))
if g_fVerbose:
print log_txt
convWriter.write(convertedArticle)
if articleSize != 0:
if not sizeStats.has_key(articleSize):
sizeStats[articleSize] = 1
else:
sizeStats[articleSize] = sizeStats[articleSize]+1
count += 1
if count % 1000 == 0:
sys.stderr.write("phase 2 processed %d, last title=%s\n" % (count,article.getTitle()))
convWriter.close()
# dump size stats to a file
statsFileName = wikipediasql.getSizeStatsFileName(sqlDump)
statsFo = open(statsFileName, "wb")
sizes = sizeStats.keys()
sizes.sort()
for size in sizes:
count = sizeStats[size]
statsFo.write("%d\t\t%d\n" % (size,count))
statsFo.close()
def convertArticles(sqlDump, articleLimit):
count = 0
redirects = {}
articleTitles = {}
fTesting = False
if fTesting:
fUseCache = False
fRecreateCache = True
else:
fUseCache = True
fRecreateCache = False
for article in wikipediasql.iterWikipediaArticles(sqlDump, articleLimit,
fUseCache,
fRecreateCache):
# we only convert article from the main namespace
assert article.getNamespace() == wikipediasql.NS_MAIN
title = article.getTitle()
if article.fRedirect():
redirects[title] = article.getRedirect()
else:
txt = article.getText()
#links = articleconvert.articleExtractLinks(txt)
#articleTitles[title] = links
articleTitles[title] = 1
count += 1
if 0 == count % 1000:
sys.stderr.write("processed %d rows, last title=%s\n" %
(count, title.strip()))
if articleLimit and count >= articleLimit:
break
# verify redirects
print "Number of real articles: %d" % len(articleTitles)
print "Number of all redirects: %d (%d in total)" % (
len(redirects), len(articleTitles) + len(redirects))
unresolvedCount = 0
setUnresolvedRedirectWriter(sqlDump)
redirectsExisting = {}
for (title, redirect) in redirects.items():
redirectResolved = resolveRedirect(title, redirect, redirects,
articleTitles)
if None == redirectResolved:
unresolvedCount += 1
#print "redirect '%s' (to '%s') not resolved" % (title,redirect)
else:
redirectsExisting[title] = redirectResolved
closeUnresolvedRedirectWriter()
print "Number of unresolved redirects: %d" % unresolvedCount
dbName = getDbNameFromFileName(sqlDump)
ipedia_write_cur = getNamedCursor(getIpediaConnection(dbName),
"ipedia_write_cur")
# go over articles again (hopefully now using the cache),
# convert them to a destination format (including removing invalid links)
# and insert into a database
sizeStats = {}
count = 0
convWriter = wikipediasql.ConvertedArticleCacheWriter(sqlDump)
convWriter.open()
for article in wikipediasql.iterWikipediaArticles(sqlDump, articleLimit,
True, False):
title = article.getTitle()
articleSize = 0 # 0 is for redirects, which we don't log
if article.fRedirect():
convertedArticle = ConvertedArticleRedirect(
article.getNamespace(), title, article.getRedirect())
else:
txt = article.getText()
converted = articleconvert.convertArticle(title, txt)
try:
noLinks = articleconvert.removeInvalidLinks(
converted, redirects, articleTitles)
except:
print "exception in articleconvert.removeInvalidLinks"
print "title: _%s_" % title
print "txt:\n_%s_" % txt
print "converted:\n_%s_" % converted
raise
if noLinks:
converted = noLinks
convertedArticle = ConvertedArticle(article.getNamespace(),
article.getTitle(), converted)
articleSize = len(converted)
if article.fRedirect():
if redirectsExisting.has_key(title):
redirect = redirectsExisting[title]
try:
title = title.replace("_", " ")
redirect = redirect.replace("_", " ")
ipedia_write_cur.execute(
"""INSERT INTO redirects (title, redirect) VALUES ('%s', '%s')"""
% (dbEscape(title), dbEscape(redirect)))
except:
print "DUP REDERICT '%s' => '%s'" % (title, redirect)
else:
title = title.replace("_", " ")
if g_fVerbose:
log_txt = "title: %s " % title
try:
ipedia_write_cur.execute(
"""INSERT INTO articles (title, body) VALUES ('%s', '%s')"""
% (dbEscape(title), dbEscape(converted)))
if g_fVerbose:
log_txt += "*New record"
except:
# assuming that the exception happend because of trying to insert
# item with a duplicate title (duplication due to lower-case
# conversion might convert 2 differnt titles into the same,
# lower-cased title)
if g_fShowDups:
print "dup: " + title
if g_fVerbose:
log_txt += "Update existing record"
print "DUP ARTICLE: '%s'" % title
ipedia_write_cur.execute(
"""UPDATE articles SET body='%s' WHERE title='%s'""" %
(dbEscape(converted), dbEscape(title)))
if g_fVerbose:
print log_txt
convWriter.write(convertedArticle)
if articleSize != 0:
if not sizeStats.has_key(articleSize):
sizeStats[articleSize] = 1
else:
sizeStats[articleSize] = sizeStats[articleSize] + 1
count += 1
if count % 1000 == 0:
sys.stderr.write("phase 2 processed %d, last title=%s\n" %
(count, article.getTitle()))
convWriter.close()
# dump size stats to a file
statsFileName = wikipediasql.getSizeStatsFileName(sqlDump)
statsFo = open(statsFileName, "wb")
sizes = sizeStats.keys()
sizes.sort()
for size in sizes:
count = sizeStats[size]
statsFo.write("%d\t\t%d\n" % (size, count))
statsFo.close()
