def countPages(extractor, limit=-1):
page_idx = 0
printerval = 1000
togo = printerval
t = Timer()
for ex in extractor:
page_idx = page_idx + 1
if togo<=0:
togo = printerval
print "Counted " + str(page_idx) + " pages"
togo = togo - 1
if limit>0 and page_idx>limit:
break
t.printTime()
def scan(extractor_filename, methods=[], startIdx=0, skipevery=1, offset=0):
logging.basicConfig(format = '%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
page_idx = 0
dbconn = dbconnect.MySQLDatabaseConnection.connectionWithConfiguration('local')
extractor = wikiutils.WikiTextExtractor(extractor_filename)
commit_idx=0
commit_ceil=1000
commit_timer = Timer()
ignore_namespaces = 'Wikipedia Category File Portal Template MediaWiki User Help Book Draft'.split()
thisdir = "/".join(os.path.realpath(__file__).split("/")[:-1]) + "/"
id2word = gensim.corpora.Dictionary.load_from_text(thisdir + 'lda/results_wordids.txt.bz2')
lda = gensim.models.ldamodel.LdaModel.load(thisdir + 'lda/lda_1000')
ctx = ScanContext(extractor, dbconn, id2word, lda)
for ex in extractor:
title = extractor.titleForCurrentPage()
skip_scan = False
skip_scan = skip_scan or (page_idx < startIdx)
skip_scan = skip_scan or (page_idx+offset)%skipevery != 0
skip_scan = skip_scan or any([title.startswith(ignore + ":") for ignore in ignore_namespaces])
if not (skip_scan):
try:
for m in methods:
functionDict[m](ctx)
except Exception as e:
print e
page_idx += 1
commit_idx += 1
if commit_idx >= commit_ceil:
commit_idx=0
dbconn.connection.commit()
commit_timer.printTime()
commit_timer = Timer()
if offset==0:
print "\tScanning page %d: %s" % (page_idx, extractor.titleForCurrentPage())
dbconn.connection.commit()
print "Scan complete!"
def findIambsForPages(ptext, pageID):
iambic_runs = []
paragraphs = ptext.split("\n")
t = Timer()
for paragraph in paragraphs:
t.begin("makeblob")
blob = textblob.TextBlob(paragraph)
t.end("makeblob")
t.begin("find iambs")
for sen in blob.sentences:
## Get the runs for each sentence
## Each sentence is just a raw, nasty string of the code-stripped sentence
iambic_runs = iambic_runs + wordutils.extract_iambic_pentameter(sen.string)
t.end("find iambs")
# t.printTime()
return iambic_runs
def extract_iambic_pentameter(sentence):
t = Timer()
iambic_runs = []
run_start_index = 0;
run_end_index = 0;
# [x[1] for x in parse(sentence, chunks=False, tokenize=False).split()[0]]
# City of Tshwane; Metropolitan Municipality official website
## Get just the parts of speech of just the first sentence
sentence = " ".join(sentence.strip().split())
words = sentence.split();
if len(words) < 2:
return iambic_runs
has_parsed_sentence = False
for w in range(len(words)):
raw_word = words[w]
safe_word = make_safe(raw_word)
run_end_index = w+1;
if safe_word in d:
t.begin("safety")
raw_previous_words = words[run_start_index:run_end_index]
safe_words = map(make_safe, raw_previous_words)
t.end("safety")
if is_iambic(safe_words):
if line_sylcount(safe_words) >= 10:
if line_sylcount(safe_words) == 10:
## Optimization: Don't do the expensive sentence parsing until you need to
t.begin("parse")
if not has_parsed_sentence:
t.begin("parse-itself")
pos = parse(sentence).split()[0]
t.end("parse-itself")
pos = filter(lambda x: re.match('^[\w-]+', x[1]) is not None, pos)
chnk = [x[2] for x in pos]
pos = [x[1] for x in pos]
pos = [p if p not in pos_map else pos_map[p] for p in pos]
has_parsed_sentence = True
if (len(words) != len(pos)):
# print("Skipping sentence ''" + sentence + "''")
return iambic_runs
t.end("parse")
t.begin("other")
hasVerb = u'VB' in pos[run_start_index:run_end_index]
leadChunk = None
if run_start_index > 0:
if chnk[run_start_index] == chnk[run_start_index-1]:
leadChunk = chnk[run_start_index]
lagChunk = None
if run_end_index < len(pos)-1:
if chnk[run_end_index] == chnk[run_end_index+1]:
lagChunk = chnk[run_end_index]
t.end("other")
t.begin("make-options")
options = optionsForWords(words, pos, chnk, run_start_index, run_end_index)
t.end("make-options")
t.begin("append")
newrun = PoemLine(" ".join(raw_previous_words), pos[run_start_index:run_end_index], options=options)
iambic_runs.append(newrun)
t.end("append")
run_start_index += 1
t.begin("advance")
while run_end_index > run_end_index:
raw_previous_words = words[run_start_index:run_end_index]
if is_iambic(map(make_safe, raw_previous_words)):
t.end("advance")
break
run_start_index += 1
t.end("advance")
else:
run_start_index = run_end_index
else:
run_start_index = run_end_index
# if iambic_runs:
# t.printTime()
return iambic_runs
import argparse
from util.benchmarking import Timer
import wikibard.wikibard as wikibard
import db.dbreader as dbreader
import db.dbconnect as dbconnect
import random
import codecs
parser = argparse.ArgumentParser(description="Write a bunch of poems, see how long it takes")
parser.add_argument('count', type=int, help="number of poems to write")
parser.add_argument('--output', type=str, default=None, help="output file")
if __name__ == '__main__':
args = parser.parse_args()
t = Timer()
t.begin("poems")
outf = None
if args.output:
outf = codecs.open(args.output, 'w', 'utf-8')
dbconfig = dbconnect.MySQLDatabaseConnection.dbconfigForName('local')
dbconn = dbconnect.MySQLDatabaseConnection.connectionWithConfiguration('local')
cursor = dbconn.connection.cursor()
query = (
"""SELECT view_counts.id FROM view_counts INNER JOIN page_categories"""
""" ON page_categories.id = view_counts.id WHERE view_counts.count < 202917"""
""" ORDER BY view_counts.count DESC LIMIT %s;"""
)
values = (10000, )
cursor.execute(query, values)
res = cursor.fetchall()
for i in range(args.count):
