py/gadict_headwords.py
author Oleksandr Gavenko <gavenkoa@gmail.com>
Fri, 11 Nov 2016 00:19:47 +0200
changeset 667 5f69f0776c37
parent 643 c2c32f45dde6
child 720 b5a4b476eddf
permissions -rw-r--r--
Add missing imports.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
643
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     1
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     2
import sys
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     3
import codecs
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     4
import io
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     5
import regex
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     6
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     7
FINAME = None
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     8
FONAME = None
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
     9
if len(sys.argv) >= 2:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    10
    FINAME = sys.argv[1]
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    11
if len(sys.argv) >= 3:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    12
    FONAME = sys.argv[2]
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    13
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    14
FIN = io.open(FINAME, mode='r', buffering=1, encoding="utf-8")
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    15
if FONAME is None:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    16
    FOUT = sys.stdout
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    17
else:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    18
    FOUT = codecs.open(FONAME, "w", "utf-8")
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    19
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    20
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    21
class GadictParser:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    22
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    23
    SEPARATOR_RE = regex.compile(u"^__$")
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    24
    EMPTY_RE = regex.compile( u"^$" )
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    25
    HEADWORD_RE = regex.compile( u"^(\\p{L}.*)$" )
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    26
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    27
    def __init__(self, stream):
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    28
        self.stream = stream
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    29
        self.lineno = 0
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    30
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    31
    def parse(self):
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    32
        wlist = []
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    33
        while True:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    34
            line = self.stream.readline()
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    35
            if len(line) == 0:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    36
                break
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    37
            self.lineno += 1
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    38
            m = self.SEPARATOR_RE.match(line)
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    39
            if not m:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    40
                continue
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    41
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    42
            line = self.stream.readline()
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    43
            if len(line) == 0:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    44
                break
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    45
            self.lineno += 1
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    46
            m = self.EMPTY_RE.match(line)
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    47
            if not m:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    48
                raise Exception("Line {:d}: '{:s}' is not empty line\n".format(self.lineno, line))
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    49
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    50
            line = self.stream.readline()
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    51
            if len(line) == 0:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    52
                break
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    53
            line = line.strip()
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    54
            self.lineno += 1
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    55
            m = self.HEADWORD_RE.match(line)
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    56
            if not m:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    57
                raise Exception("Line {:d}: '{:s}' is not a headword\n".format(self.lineno, line))
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    58
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    59
            wlist.append(line)
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    60
        return wlist
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    61
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    62
try:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    63
    parser = GadictParser(FIN)
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    64
    for headword in parser.parse():
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    65
        FOUT.write(headword)
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    66
        FOUT.write("\n")
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    67
finally:
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    68
    FIN.close()
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    69
    FOUT.close()
c2c32f45dde6 Search rare words in gadict.
Oleksandr Gavenko <gavenkoa@gmail.com>
parents:
diff changeset
    70