def load_blog(conn): cursor = conn.cursor() cursor.execute("select a.blogid, a.uid, a.subject, a.dateline, b.message from ihome_blog a left join ihome_blogfield b on a.blogid = b.blogid") rows = cursor.fetchall() for row in rows: parts = [] if row[2]: parts = SegUtil.seg(row[2].encode("utf8")) if row[4]: parts.extend(SegUtil.seg(row[4].encode("utf8"))) for p in parts: if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][1] += 1 cursor.close()
def load_blog(conn): cursor = conn.cursor() cursor.execute("select a.blogid, a.uid, a.subject, a.dateline, b.message from ihome_blog a left join ihome_blogfield b on a.blogid = b.blogid where a.dateline > %d" % start_time) rows = cursor.fetchall() for row in rows: parts = [] if row[2]: parts = SegUtil.seg(row[2].encode("utf8")) if row[4]: parts.extend(SegUtil.seg(row[4].encode("utf8"))) for p in parts: if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if len(p[0]) <= 0: continue if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][1] += 1 cursor.close()
def load_complain(conn,date): cursor = conn.cursor() cursor.execute("select doid,uid,message,datatime from ihome_complain where datatime="+date) rows = cursor.fetchall() for row in rows: parts = [] if row[2]: parts = SegUtil.seg(row[2].encode("utf8")) for p in parts: if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][2] += 1 cursor.close()
def load_complain(conn): cursor = conn.cursor() cursor.execute("select doid,uid,message,addtime from ihome_complain ") rows = cursor.fetchall() for row in rows: parts = [] if row[2]: parts = SegUtil.seg(row[2].encode("utf8")) for p in parts: if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][2] += 1 cursor.close()
def load_doing(conn): cursor = conn.cursor() cursor.execute( "select a.doid, a.uid, a.message, a.dateline from ihome_doing a") rows = cursor.fetchall() for row in rows: parts = [] if row[2]: parts = SegUtil.seg(row[2].encode("utf8")) for p in parts: if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][2] += 1 cursor.close()
def load_doing(conn): cursor = conn.cursor() cursor.execute("select a.doid, a.uid, a.message, a.dateline from ihome_doing a") rows = cursor.fetchall() for row in rows: parts = [] if row[2]: parts = SegUtil.seg(row[2].encode("utf8")) for p in parts: if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][2] += 1 cursor.close()
if p[1] == 'ns' or p[1] == 'nr' or p[1] == 'nt' or p[1] == 'n': if p[0] not in tags: tags[p[0]] = [0, 0, 0] tags[p[0]][0] += 1 tags[p[0]][2] += 1 cursor.close() if __name__ == '__main__': conn = MySQLdb.connect(host="localhost", user='******', passwd='nameLR9969', db='ihome', port=3306, charset='utf8') SegUtil.init('.') load_complain(conn) h = [] for k,v in tags.iteritems(): if len(h) < 100: heapq.heappush(h, (v, k)) else: heapq.heappushpop(h, (v, k)) cursor = conn.cursor() lastid = -1 results = [] while len(h) > 0: num, word = heapq.heappop(h)