[fix] removing stopwords index

This commit is contained in:
Al
2015-06-02 12:43:48 -04:00
parent 49816382c1
commit 208366af98
2 changed files with 10 additions and 12 deletions

View File

@@ -62,7 +62,7 @@ ordinal_indicator_template = u'{{{number}, {gender}, "{value}"}}'
stopwords_template = u'"{word}"'
language_template = u'{{"{language}", {rule_index}, {num_rules}, {ordinal_indicator_index}, {num_ordinal_indicators}, {stopword_index}, {num_stopwords}}}'
language_template = u'{{"{language}", {rule_index}, {num_rules}, {ordinal_indicator_index}, {num_ordinal_indicators}}}'
numex_rules_data_template = u'''
numex_rule_source_t numex_rules[] = {{
@@ -148,9 +148,7 @@ def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
rule_index=rule_index,
num_rules=num_rules,
ordinal_indicator_index=ordinal_indicator_index,
num_ordinal_indicators=num_ordinal_indicators,
stopword_index=stopword_index,
num_stopwords=num_stopwords
num_ordinal_indicators=num_ordinal_indicators
)))
out.write(safe_encode(numex_rules_data_template.format(

View File

@@ -1030,12 +1030,12 @@ ordinal_indicator_t ordinal_indicator_rules[] = {
};
numex_language_source_t numex_languages[] = {
{"de", 0, 194, 0, 0, 0, 1},
{"en", 194, 70, 0, 10, 0, 1},
{"es", 264, 133, 10, 20, 0, 1},
{"fr", 397, 83, 30, 20, 0, 1},
{"it", 480, 163, 50, 20, 0, 2},
{"la", 643, 31, 70, 0, 0, 0},
{"pt", 674, 170, 70, 20, 0, 1},
{"ru", 844, 80, 90, 10, 0, 1}
{"de", 0, 194, 0, 0},
{"en", 194, 70, 0, 10},
{"es", 264, 133, 10, 20},
{"fr", 397, 83, 30, 20},
{"it", 480, 163, 50, 20},
{"la", 643, 31, 70, 0},
{"pt", 674, 170, 70, 20},
{"ru", 844, 80, 90, 10}
};