mirror of
https://gitlab.freedesktop.org/uchardet/uchardet.git
synced 2025-12-06 16:56:40 +08:00
script: fix BuildLangModel.py.
This commit is contained in:
parent
598fe90c91
commit
362086bf56
@ -414,7 +414,7 @@ elif lang.alphabet is not None:
|
|||||||
logfd.write("\n[{:2}] Char {}: {} %".format(order, chr(char), ratio * 100))
|
logfd.write("\n[{:2}] Char {}: {} %".format(order, chr(char), ratio * 100))
|
||||||
accumulated_ratios += ratio
|
accumulated_ratios += ratio
|
||||||
freq_count += 1
|
freq_count += 1
|
||||||
if accumulated_ratios < 0.4:
|
if very_freq_ratio < 0.4:
|
||||||
very_freq_count += 1
|
very_freq_count += 1
|
||||||
very_freq_ratio += ratio
|
very_freq_ratio += ratio
|
||||||
else:
|
else:
|
||||||
@ -440,14 +440,16 @@ elif lang.frequent_ranges is not None:
|
|||||||
logfd.write("\n[{:2}] Char {}: {} %".format(order, chr(char), ratio * 100))
|
logfd.write("\n[{:2}] Char {}: {} %".format(order, chr(char), ratio * 100))
|
||||||
frequent_ranges_size -= 1
|
frequent_ranges_size -= 1
|
||||||
break
|
break
|
||||||
if accumulated_ratios < 0.4:
|
|
||||||
very_freq_count += 1
|
|
||||||
very_freq_ratio += ratio
|
|
||||||
else:
|
else:
|
||||||
# A frequent character in the non-frequent range.
|
# A frequent character in the non-frequent range.
|
||||||
logfd.write("\n[{:2}] Char {}: {} %".format(order, chr(char), ratio * 100))
|
logfd.write("\n[{:2}] Char {}: {} %".format(order, chr(char), ratio * 100))
|
||||||
freq_count += 1
|
freq_count += 1
|
||||||
accumulated_ratios += ratio
|
accumulated_ratios += ratio
|
||||||
|
|
||||||
|
if very_freq_ratio < 0.4:
|
||||||
|
very_freq_count += 1
|
||||||
|
very_freq_ratio += ratio
|
||||||
|
|
||||||
if frequent_ranges_size <= 0:
|
if frequent_ranges_size <= 0:
|
||||||
break
|
break
|
||||||
|
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user