|
1
4 package org.jruby;
5
6 import java.util.regex.Matcher  ;
7 import java.util.regex.Pattern ;
8 import java.util.regex.PatternSyntaxException ;
9
10 import org.jruby.parser.ReOptions;
11
12 public class RegexpTranslator {
13
14 private static final Pattern SHARP_IN_CHARACTER_CLASS_PATTERN = Pattern.compile("(\\[[^]]*)#(.*?])");
15 private static final Pattern SPACE_IN_CHARACTER_CLASS_PATTERN = Pattern.compile("(\\[[^]]*) (.*?])");
16 private static final Pattern COMMENT_PATTERN = Pattern.compile("\\(\\?#[^)]*\\)");
17 private static final Pattern HEX_SINGLE_DIGIT_PATTERN = Pattern.compile("\\\\x(\\p{XDigit})(?!\\p{XDigit})");
18 private static final Pattern OCTAL_SINGLE_ZERO_PATTERN = Pattern.compile("\\\\(0)(?![0-7])");
19 private static final Pattern OCTAL_MISSING_ZERO_PATTERN = Pattern.compile("\\\\([1-7][0-7]{1,2})");
20 private static final Pattern POSIX_NAME = Pattern.compile("\\[:(\\w+):\\]");
21
22 public Pattern translate(String regex, int options, int javaRegexFlags) throws PatternSyntaxException {
23 javaRegexFlags |= translateFlags(options);
24 regex = translatePattern(regex, (javaRegexFlags & Pattern.COMMENTS) != 0);
25
26 return Pattern.compile(regex, javaRegexFlags);
27 }
28
29 private String translatePosixPattern(String regex) {
31 for (Matcher matcher = POSIX_NAME.matcher(regex); matcher.find(); matcher = POSIX_NAME.matcher(regex)) {
32 String value = matcher.group(1);
33
34 if ("alnum".equals(value)) {
35 regex = matcher.replaceFirst("\\\\p{Alnum}");
36 } else if ("alpha".equals(value)) {
37 regex = matcher.replaceFirst("\\\\p{Alpha}");
38 } else if ("blank".equals(value)) {
39 regex = matcher.replaceFirst("\\\\p{Blank}");
40 } else if ("cntrl".equals(value)) {
41 regex = matcher.replaceFirst("\\\\p{Cntrl}");
42 } else if ("digit".equals(value)) {
43 regex = matcher.replaceFirst("\\\\p{Digit}");
44 } else if ("graph".equals(value)) {
45 regex = matcher.replaceFirst("\\\\p{Graph}");
46 } else if ("lower".equals(value)) {
47 regex = matcher.replaceFirst("\\\\p{Lower}");
48 } else if ("print".equals(value)) {
49 regex = matcher.replaceFirst("\\\\p{Print}");
50 } else if ("punct".equals(value)) {
51 regex = matcher.replaceFirst("\\\\p{Punct}");
52 } else if ("space".equals(value)) {
53 regex = matcher.replaceFirst("\\\\p{Space}");
54 } else if ("upper".equals(value)) {
55 regex = matcher.replaceFirst("\\\\p{Upper}");
56 } else if ("xdigit".equals(value)) {
57 regex = matcher.replaceFirst("\\\\p{XDigit}");
58 } else {
59 regex = matcher.replaceFirst("\\\\[:" + value + ":\\\\]");
60 }
61 }
62 return regex;
63 }
64
65 String translatePattern(String regex, boolean commentsAllowed) {
66 regex = COMMENT_PATTERN.matcher(regex).replaceAll("");
67 regex = translatePosixPattern(regex);
68 regex = HEX_SINGLE_DIGIT_PATTERN.matcher(regex).replaceAll("\\\\"+"x0$1");
69 regex = OCTAL_SINGLE_ZERO_PATTERN.matcher(regex).replaceAll("\\\\"+"0$1");
70 regex = OCTAL_MISSING_ZERO_PATTERN.matcher(regex).replaceAll("\\\\"+"0$1");
71 if (commentsAllowed) {
72 regex = SPACE_IN_CHARACTER_CLASS_PATTERN.matcher(regex).replaceAll("$1\\\\x20$2");
73 regex = SHARP_IN_CHARACTER_CLASS_PATTERN.matcher(regex).replaceAll("$1\\\\x23$2");
74 }
75
76 return regex;
77 }
78
79 int translateFlags(int options) {
80 int flags = Pattern.MULTILINE;
81 if ((options & ReOptions.RE_OPTION_IGNORECASE) > 0) {
82 flags |= Pattern.CASE_INSENSITIVE;
83 }
84 if ((options & ReOptions.RE_OPTION_EXTENDED) > 0) {
85 flags |= Pattern.COMMENTS;
86 }
87 if ((options & ReOptions.RE_OPTION_MULTILINE) > 0) {
88 flags |= Pattern.DOTALL;
89 }
90 return flags;
91 }
92
93 }
94 |
Popular Tags |
Java API By Example, From Geeks To Geeks.