def leerPGN(self, fichero, dlTmp):
erroneos = duplicados = importados = n = 0
t1 = time.time() - 0.7 # para que empiece enseguida
next_n = random.randint(100, 200)
codec = Util.file_encoding(fichero)
sicodec = codec not in ("utf-8", "ascii")
liRegs = []
nRegs = 0
conexion = self._conexion
cursor = self._cursor
cursor.execute("SELECT FEN FROM GAMES")
liRows = cursor.fetchall()
stRegs = set(row[0] for row in liRows)
sql = "insert into GAMES (FEN,EVENT,SITE,DATE,WHITE,BLACK,RESULT,XPV,PGN,PLIES) values (?,?,?,?,?,?,?,?,?,?);"
liCabs = self.liCamposBase[:-1] # all except PLIES PGN, TAGS
liCabs.append("PLYCOUNT")
with LCEngine.PGNreader(fichero, 0) as fpgn:
for n, (pgn, pv, dCab, raw, liFens) in enumerate(fpgn, 1):
if "FEN" not in dCab:
erroneos += 1
else:
fen = dCab["FEN"]
if fen in stRegs:
dup = True
else:
cursor.execute(
"SELECT COUNT(*) FROM GAMES WHERE FEN = ?",
(fen, ))
num = cursor.fetchone()[0]
dup = num > 0
if dup:
duplicados += 1
else:
stRegs.add(fen)
if sicodec:
for k, v in dCab.iteritems():
dCab[k] = unicode(v,
encoding=codec,
errors="ignore")
if pgn:
pgn = unicode(pgn,
encoding=codec,
errors="ignore")
event = dCab.get("EVENT", "")
site = dCab.get("SITE", "")
date = dCab.get("DATE", "")
white = dCab.get("WHITE", "")
black = dCab.get("BLACK", "")
result = dCab.get("RESULT", "")
plies = (pv.count(" ") + 1) if pv else 0
if pgn:
pgn = Util.var2blob(pgn)
xpv = pv2xpv(pv)
reg = (fen, event, site, date, white, black, result,
xpv, pgn, plies)
liRegs.append(reg)
nRegs += 1
importados += 1
if nRegs == 10000:
cursor.executemany(sql, liRegs)
liRegs = []
stRegs = set()
conexion.commit()
if n == next_n:
if time.time() - t1 > 0.8:
if not dlTmp.actualiza(
erroneos + duplicados + importados, erroneos,
duplicados, importados):
break
t1 = time.time()
next_n = n + random.randint(100, 500)
if liRegs:
cursor.executemany(sql, liRegs)
conexion.commit()
dlTmp.actualiza(erroneos + duplicados + importados, erroneos,
duplicados, importados)
dlTmp.ponSaving()
conexion.commit()
dlTmp.ponContinuar()
self.lee_rowids()
def leerPGNs(self, ficheros, dlTmp):
erroneos = duplicados = importados = n = 0
t1 = time.time()-0.7 # para que empiece enseguida
if self.with_dbSTAT:
self.dbSTAT.massive_append_set(True)
def write_logs(fich, pgn):
with open(fich, "ab") as ferr:
ferr.write(pgn)
ferr.write("\n")
codec = Util.file_encoding(ficheros[0])
sicodec = codec not in ("utf-8", "ascii")
liRegs = []
stRegs = set()
nRegs = 0
conexion = self._conexion
cursor = self._cursor
sql = "insert into games (XPV,EVENT,SITE,DATE,WHITE,BLACK,RESULT,ECO,WHITEELO,BLACKELO,PGN,PLIES) values (?,?,?,?,?,?,?,?,?,?,?,?);"
liCabs = self.liCamposBase[:-1] # all except PLIES PGN, TAGS
liCabs.append("PLYCOUNT")
for fichero in ficheros:
nomfichero = os.path.basename(fichero)
fich_erroneos = os.path.join(VarGen.configuracion.carpetaTemporal(), nomfichero[:-3] + "errors.pgn")
fich_duplicados = os.path.join(VarGen.configuracion.carpetaTemporal(), nomfichero[:-3] + "duplicates.pgn")
dlTmp.pon_titulo(nomfichero)
next_n = random.randint(100, 200)
with LCEngine.PGNreader(fichero, self.depthStat()) as fpgn:
for n, (pgn, pv, dCab, raw, liFens) in enumerate(fpgn, 1):
if not pv:
erroneos += 1
write_logs(fich_erroneos, pgn)
else:
fen = dCab.get("FEN", None)
if fen and fen != ControlPosicion.FEN_INICIAL:
erroneos += 1
else:
xpv = pv2xpv(pv)
if xpv in stRegs:
dup = True
else:
cursor.execute("SELECT COUNT(*) FROM games WHERE XPV = ?", (xpv,))
num = cursor.fetchone()[0]
dup = num > 0
if dup:
duplicados += 1
write_logs(fich_duplicados, pgn)
else:
stRegs.add(xpv)
if sicodec:
for k, v in dCab.iteritems():
dCab[k] = unicode(v, encoding=codec, errors="ignore")
if pgn:
pgn = unicode(pgn, encoding=codec, errors="ignore")
if raw: # si no tiene variantes ni comentarios, se graba solo las tags que faltan
liRTags = [(k,v) for k, v in dCab.iteritems() if k not in liCabs] # k is always upper
if liRTags:
pgn = {}
pgn["RTAGS"] = liRTags
else:
pgn = None
event = dCab.get("EVENT", "")
site = dCab.get("SITE", "")
date = dCab.get("DATE", "")
white = dCab.get("WHITE", "")
black = dCab.get("BLACK", "")
result = dCab.get("RESULT", "")
eco = dCab.get("ECO", "")
whiteelo = dCab.get("WHITEELO", "")
blackelo = dCab.get("BLACKELO", "")
plies = (pv.count(" ")+1) if pv else 0
if pgn:
pgn = Util.var2blob(pgn)
reg = (xpv, event, site, date, white, black, result, eco, whiteelo, blackelo, pgn, plies)
if self.with_dbSTAT:
self.dbSTAT.append_fen(pv, result, liFens)
liRegs.append(reg)
nRegs += 1
importados += 1
if nRegs == 10000:
nRegs = 0
cursor.executemany(sql, liRegs)
liRegs = []
stRegs = set()
conexion.commit()
if self.with_dbSTAT:
self.dbSTAT.massive_append_set(False)
self.dbSTAT.commit()
self.dbSTAT.massive_append_set(True)
if n == next_n:
if time.time()-t1> 0.8:
if not dlTmp.actualiza(erroneos+duplicados+importados, erroneos, duplicados, importados):
break
t1 = time.time()
next_n = n + random.randint(100, 500)
if liRegs:
cursor.executemany(sql, liRegs)
conexion.commit()
dlTmp.actualiza(erroneos+duplicados+importados, erroneos, duplicados, importados)
dlTmp.ponSaving()
if self.with_dbSTAT:
self.dbSTAT.massive_append_set(False)
self.dbSTAT.commit()
conexion.commit()
dlTmp.ponContinuar()
