def findOrigArticlesUnderThreshold(fileName, thresholdSize):
print "looking for original articles smaller than %d bytes" % thresholdSize
count = 0
countNoComma = 0
totalSizeNoComma = 0
articles = []
for article in wikipediasql.iterWikipediaArticles(fileName,
None,
fUseCache=True,
fRecreateCache=False):
if article.fRedirect():
continue
body = article.getText()
if len(body) < thresholdSize:
#print "size: %d, title: '%s'" % (len(body),article.getTitle())
articles.append(article)
if -1 == body.find(","):
countNoComma += 1
totalSizeNoComma += len(body)
count += 1
if count % 20000 == 0:
print "processed %d articles, found %d small" % (count,
len(articles))
print "Articles without comma in orig: %d" % countNoComma
avgSize = float(totalSizeNoComma) / float(countNoComma)
print "Average size: %.2f" % avgSize
return articles
def findOrigArticleNoRedirect(fileName,titleToFind):
titleToFind = titleToFind.lower()
titleToFind = titleToFind.replace(" ", "_")
print "looking for article with title %s" % titleToFind
count = 0
for article in wikipediasql.iterWikipediaArticles(fileName,None,fUseCache=True,fRecreateCache=False):
title = article.getTitle().lower()
if title == titleToFind:
return article
#if count % 50000 == 0:
# print "processed %d articles, last title %s" % (count,title)
count += 1
return None
def getRandomArticle(fileName):
articleNum = 0
randomArticle = None
for article in wikipediasql.iterWikipediaArticles(fileName,None,fUseCache=True,fRecreateCache=False):
if article.fRedirect():
continue
articleNum += 1
# How likely is it that this is the last line of the file ?
# kjk note: I don't get it
if random.uniform(0,articleNum)<1:
randomArticle = article
if articleNum % 30000 == 0:
print "processed %d articles" % articleNum
return randomArticle
def findOrigArticleNoRedirect(fileName, titleToFind):
titleToFind = titleToFind.lower()
titleToFind = titleToFind.replace(" ", "_")
print "looking for article with title %s" % titleToFind
count = 0
for article in wikipediasql.iterWikipediaArticles(fileName,
None,
fUseCache=True,
fRecreateCache=False):
title = article.getTitle().lower()
if title == titleToFind:
return article
#if count % 50000 == 0:
# print "processed %d articles, last title %s" % (count,title)
count += 1
return None
def getRandomArticle(fileName):
articleNum = 0
randomArticle = None
for article in wikipediasql.iterWikipediaArticles(fileName,
None,
fUseCache=True,
fRecreateCache=False):
if article.fRedirect():
continue
articleNum += 1
# How likely is it that this is the last line of the file ?
# kjk note: I don't get it
if random.uniform(0, articleNum) < 1:
randomArticle = article
if articleNum % 30000 == 0:
print "processed %d articles" % articleNum
return randomArticle
def findOrigArticlesUnderThreshold(fileName,thresholdSize):
print "looking for original articles smaller than %d bytes" % thresholdSize
count = 0
countNoComma = 0
totalSizeNoComma = 0
articles = []
for article in wikipediasql.iterWikipediaArticles(fileName,None,fUseCache=True,fRecreateCache=False):
if article.fRedirect():
continue
body = article.getText()
if len(body)<thresholdSize:
#print "size: %d, title: '%s'" % (len(body),article.getTitle())
articles.append(article)
if -1 == body.find(","):
countNoComma += 1
totalSizeNoComma += len(body)
count += 1
if count % 20000 == 0:
print "processed %d articles, found %d small" % (count,len(articles))
print "Articles without comma in orig: %d" % countNoComma
avgSize = float(totalSizeNoComma)/float(countNoComma)
print "Average size: %.2f" % avgSize
return articles
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()
