| line |
true |
false |
branch |
|
124
|
1 |
0 |
if defined(@_ > 0) |
|
125
|
0 |
1 |
@_ ? : |
|
127
|
0 |
1 |
if $$self{'verbose'} |
|
128
|
0 |
1 |
if $$self{'verbose'} |
|
156
|
0 |
1 |
unless open VOCAB, ">$VOCAB_FILE" |
|
157
|
0 |
1 |
unless open SNT, ">$SNT_FILE" |
|
160
|
0 |
1 |
if ($NONTOKEN_FILE ne '') |
|
170
|
0 |
1 |
unless open CORPUS, $_ |
|
179
|
18 |
4 |
if (not exists $vocab_hash{$token}) { } |
|
188
|
0 |
3 |
if $new_line |
|
223
|
1 |
0 |
if (-e $TOKEN_FILE) { } |
|
224
|
0 |
1 |
unless open TOKEN, $TOKEN_FILE |
|
228
|
0 |
1 |
if (length $_ <= 0) |
|
229
|
0 |
1 |
unless (m[^/] and m[/$]) |
|
246
|
0 |
1 |
if (length $tokenizerRegex > 0) |
|
256
|
0 |
1 |
if ($#tokenRegex < 0) |
|
272
|
0 |
0 |
if ($NONTOKEN_FILE) { } |
|
275
|
0 |
0 |
unless open NOTOK, $NONTOKEN_FILE |
|
282
|
0 |
0 |
if (/^\s*$/) |
|
284
|
0 |
0 |
unless (m[^/]) |
|
288
|
0 |
0 |
unless (m[/$]) |
|
301
|
0 |
0 |
if (length $nontokenizerRegex <= 0) |
|
325
|
0 |
0 |
unless open FILE, $file |
|
333
|
0 |
0 |
if (/^\s*$/) |
|
336
|
0 |
0 |
unless (m[^/]) |
|
340
|
0 |
0 |
unless (m[/$]) |
|
353
|
0 |
0 |
if (length $stop_regex <= 0) |
|
371
|
0 |
1 |
unless open VOCAB, $VOCAB_FILE |
|
515
|
1 |
0 |
if $window_size == 0 |
|
519
|
0 |
1 |
if $window_size > $ngram_size |
|
537
|
0 |
1 |
unless open SNT, $SNT_FILE |
|
567
|
4 |
84 |
if ($i + $ngram_size > $N + 1) |
|
570
|
42 |
42 |
if ($bits[0] != 1) |
|
573
|
21 |
21 |
if ($size != $ngram_size) |
|
598
|
0 |
0 |
if (vec($win, $j, $win_bit) == 1) |
|
613
|
0 |
1 |
unless open SNTNGRAM, ">$SNTNGRAM_FILE" |
|
617
|
1 |
0 |
unless (@vocab_array) |
|
626
|
42 |
0 |
if vec($win, $_, $win_bit) == 1 |
|
629
|
0 |
21 |
if ($ngram[$#ngram] == 0) |
|
632
|
1 |
20 |
if ($#prev == -1) { } |
|
|
20 |
0 |
elsif (join(' ', @ngram) ne join(' ', @prev)) { } |
|
642
|
20 |
0 |
if ($return_value == 1) |
|
651
|
1 |
0 |
if ($return_value == 1) |
|
669
|
0 |
21 |
if ($stop_flag) |
|
672
|
0 |
0 |
if ($stop_mode =~ /OR|or/) { } |
|
676
|
0 |
0 |
if ($token_ngram[$i] =~ /$stop_regex/) |
|
|
0 |
0 |
if ($stop_mode =~ /OR|or/) { } |
|
679
|
0 |
0 |
unless ($token_ngram[$i] =~ /$stop_regex/) |
|
682
|
0 |
0 |
if ($doStop and $marginals) |
|
684
|
0 |
0 |
if (exists $remove_hash{$_ . ':' . $ngram[$_]}) { } |
|
693
|
0 |
0 |
if ($_ == 1) |
|
|
0 |
21 |
if ($new_line) |
|
698
|
21 |
0 |
if ($doStop == 0 and $line == 0) |
|
699
|
21 |
0 |
if ($remove <= $freq) { } |
|
701
|
21 |
0 |
if ($frequency <= $freq) |
|
705
|
0 |
0 |
if (exists $remove_hash{$_ . ':' . $ngram[$_]}) { } |
|
721
|
0 |
1 |
unless open SNTNGRAM, $SNTNGRAM_FILE |
|
724
|
0 |
1 |
unless open NGRAM, ">$NGRAM_FILE" |
|
727
|
0 |
1 |
unless (@vocab_array) |
|
738
|
21 |
0 |
if ($marginals) |
|
757
|
21 |
21 |
if ($i == 0) |
|
758
|
0 |
21 |
if ($_[$i] == vec($corpus, $N, $bit)) |
|
760
|
21 |
21 |
if ($i == $#_) |
|
761
|
0 |
21 |
if ($_[$i] == vec($corpus, 0, $bit)) |
|
764
|
0 |
42 |
if ($stop_flag or $remove > 0) |
|
765
|
0 |
0 |
if (exists $remove_hash{$i . ':' . $_[$i]}) |
|
782
|
120 |
0 |
if (vec($a, $i, $win_bit) == 1) |
|
783
|
120 |
0 |
if (vec($b, $i, $win_bit) == 1) |
|
787
|
115 |
5 |
if ($a_array[$z] != $b_array[$z]) |
|
789
|
30 |
0 |
$a_array[$x] < $b_array[$x] ? : |
|
|
30 |
30 |
$a_array[$x] > $b_array[$x] ? : |