X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fparser%2FDictFileParser.java;h=991ed8ad417b29e4941f68fb07e4a1eaf4b4248b;hb=4b1b911b69b63a157189a44124861cf365b12888;hp=b435b4a1ecac0501d5805b603f1e8cc86dde832b;hpb=7d5ada9329d101b59b55691dd2f63ce3e3860011;p=DictionaryPC.git diff --git a/src/com/hughes/android/dictionary/parser/DictFileParser.java b/src/com/hughes/android/dictionary/parser/DictFileParser.java index b435b4a..991ed8a 100644 --- a/src/com/hughes/android/dictionary/parser/DictFileParser.java +++ b/src/com/hughes/android/dictionary/parser/DictFileParser.java @@ -50,17 +50,15 @@ public class DictFileParser implements Parser { public static final Pattern PIPE = Pattern.compile("\\|"); static final Pattern SPACES = Pattern.compile("\\s+"); -// static final Pattern DE_NOUN = Pattern.compile("([^ ]+) *\\{(m|f|n|pl)\\}"); -// static final Pattern EN_VERB = Pattern.compile("^to ([^ ]+)"); static final Pattern BRACKETED = Pattern.compile("\\[([^]]+)\\]"); static final Pattern PARENTHESIZED = Pattern.compile("\\(([^)]+)\\)"); static final Pattern CURLY_BRACED = Pattern.compile("\\{([^}]+)\\}"); - static final Pattern NON_CHAR_DASH = Pattern.compile("[^-'\\p{L}0-9]+"); - public static final Pattern NON_CHAR = Pattern.compile("[^\\p{L}0-9]+"); + static final Pattern NON_CHAR_DASH = Pattern.compile("[^-'\\p{L}\\p{M}\\p{N}]+"); + public static final Pattern NON_CHAR = Pattern.compile("[^\\p{L}\\p{M}\\p{N}]+"); - static final Pattern TRIM_PUNC = Pattern.compile("^[^\\p{L}0-9]+|[^\\p{L}0-9]+$"); + static final Pattern TRIM_PUNC = Pattern.compile("^[^\\p{L}\\p{M}\\p{N}]+|[^\\p{L}\\p{M}\\p{N}]+$"); final Charset charset; final boolean flipCols;