"<"
+ str(id)
+ ","
+ str(int(matches.groups()[1]) + 1)
+ ","
+ str(in_title[term.lower()])
+ ">",
)
except Exception, e:
print "80:", e
return inverted_index.replace(
matches.groups()[0], "<" + str(id) + "," + str(int(matches.groups()[1]) + 1) + ",0>"
)
else:
try:
return uummuuWord.add_to_index(inverted_index, id, 1, in_title[term.lower()])
except Exception, e:
print "86:", e
return uummuuWord.add_to_index(inverted_index, id, 1, 0)
else:
matches = re.search("(<" + str(id) + ",(\d)*,([0|1])>)", inverted_index)
if len(matches.groups()) == 0:
return inverted_index
else:
if int(matches.groups()[1]) > 1:
try:
return inverted_index.replace(
matches.groups()[0],
"<"
+ str(id)
styled_objs, no_style_objs = main_style_parse(parser.styled_objects, parser.full_style, True);
for key in styled_objs.keys():
print key + ' ==> ' + str(styled_objs[key]);
def insert_into_db(cursor, doc_id, word, weight, freq, in_title):
word = word.replace("'", "");
if len(word) == 0:
return;
cursor.execute("SELECT invert_list FROM sites_index WHERE word = '%s';" %word);
results = cursor.fetchall();
present = False;
if(len(word) > 124):
word = word[0:124];
try:
entry = results[0][0];
present = True;
except Exception, e:
entry = "";
present = False;
entry = add_to_index(entry, doc_id, weight, freq, in_title);
if not present:
cursor.execute("INSERT into sites_index(word, invert_list, freq, length, num_docs) VALUES ('%s', '%s', %d, %d, 1);" %(word, entry, freq, len(word)));
else:
cursor.execute("UPDATE sites_index set invert_list = '%s', num_docs = num_docs + 1, freq = freq + %d where word = '%s';" %(entry, freq, word));
try:
cursor.execute("INSERT INTO sites_worddoc(word, doc_id, occurrence, weight, in_title) VALUES ('%s', %d, %d, %d, %d);" %(word, doc_id, freq, weight, in_title));
except Exception, e:
print "word was:", word, 'length is:', len(word);
print 'I quit because:', e;
