Commit | Line | Data |
---|---|---|
ec1f3444 NR |
1 | package be.nikiroo.utils; |
2 | ||
ec1f3444 | 3 | import java.io.ByteArrayInputStream; |
ec1f3444 | 4 | import java.io.IOException; |
ec1f3444 NR |
5 | import java.security.MessageDigest; |
6 | import java.security.NoSuchAlgorithmException; | |
7 | import java.text.Normalizer; | |
8 | import java.text.Normalizer.Form; | |
9 | import java.text.ParseException; | |
10 | import java.text.SimpleDateFormat; | |
ec1f3444 | 11 | import java.util.Date; |
db31c358 | 12 | import java.util.Scanner; |
ec1f3444 NR |
13 | import java.util.regex.Pattern; |
14 | ||
ec1f3444 NR |
15 | import org.unbescape.html.HtmlEscape; |
16 | import org.unbescape.html.HtmlEscapeLevel; | |
17 | import org.unbescape.html.HtmlEscapeType; | |
18 | ||
19 | /** | |
20 | * This class offer some utilities based around {@link String}s. | |
21 | * | |
22 | * @author niki | |
23 | */ | |
24 | public class StringUtils { | |
25 | /** | |
26 | * This enum type will decide the alignment of a {@link String} when padding | |
27 | * is applied or if there is enough horizontal space for it to be aligned. | |
28 | */ | |
29 | public enum Alignment { | |
30 | /** Aligned at left. */ | |
31 | Beginning, | |
32 | /** Centered. */ | |
33 | Center, | |
34 | /** Aligned at right. */ | |
35 | End | |
36 | } | |
37 | ||
38 | static private Pattern marks = Pattern | |
39 | .compile("[\\p{InCombiningDiacriticalMarks}\\p{IsLm}\\p{IsSk}]+"); | |
40 | ||
41 | /** | |
42 | * Fix the size of the given {@link String} either with space-padding or by | |
43 | * shortening it. | |
44 | * | |
45 | * @param text | |
46 | * the {@link String} to fix | |
47 | * @param width | |
48 | * the size of the resulting {@link String} or -1 for a noop | |
49 | * | |
50 | * @return the resulting {@link String} of size <i>size</i> | |
51 | */ | |
52 | static public String padString(String text, int width) { | |
53 | return padString(text, width, true, Alignment.Beginning); | |
54 | } | |
55 | ||
56 | /** | |
57 | * Fix the size of the given {@link String} either with space-padding or by | |
58 | * optionally shortening it. | |
59 | * | |
60 | * @param text | |
61 | * the {@link String} to fix | |
62 | * @param width | |
63 | * the size of the resulting {@link String} if the text fits or | |
64 | * if cut is TRUE or -1 for a noop | |
65 | * @param cut | |
66 | * cut the {@link String} shorter if needed | |
67 | * @param align | |
68 | * align the {@link String} in this position if we have enough | |
69 | * space | |
70 | * | |
71 | * @return the resulting {@link String} of size <i>size</i> minimum | |
72 | */ | |
73 | static public String padString(String text, int width, boolean cut, | |
74 | Alignment align) { | |
75 | ||
76 | if (width >= 0) { | |
77 | if (text == null) | |
78 | text = ""; | |
79 | ||
80 | int diff = width - text.length(); | |
81 | ||
82 | if (diff < 0) { | |
83 | if (cut) | |
84 | text = text.substring(0, width); | |
85 | } else if (diff > 0) { | |
86 | if (diff < 2 && align != Alignment.End) | |
87 | align = Alignment.Beginning; | |
88 | ||
89 | switch (align) { | |
90 | case Beginning: | |
91 | text = text + new String(new char[diff]).replace('\0', ' '); | |
92 | break; | |
93 | case End: | |
94 | text = new String(new char[diff]).replace('\0', ' ') + text; | |
95 | break; | |
96 | case Center: | |
97 | default: | |
98 | int pad1 = (diff) / 2; | |
99 | int pad2 = (diff + 1) / 2; | |
100 | text = new String(new char[pad1]).replace('\0', ' ') + text | |
101 | + new String(new char[pad2]).replace('\0', ' '); | |
102 | break; | |
103 | } | |
104 | } | |
105 | } | |
106 | ||
107 | return text; | |
108 | } | |
109 | ||
110 | /** | |
111 | * Sanitise the given input to make it more Terminal-friendly by removing | |
112 | * combining characters. | |
113 | * | |
114 | * @param input | |
115 | * the input to sanitise | |
116 | * @param allowUnicode | |
117 | * allow Unicode or only allow ASCII Latin characters | |
118 | * | |
119 | * @return the sanitised {@link String} | |
120 | */ | |
121 | static public String sanitize(String input, boolean allowUnicode) { | |
122 | return sanitize(input, allowUnicode, !allowUnicode); | |
123 | } | |
124 | ||
125 | /** | |
126 | * Sanitise the given input to make it more Terminal-friendly by removing | |
127 | * combining characters. | |
128 | * | |
129 | * @param input | |
130 | * the input to sanitise | |
131 | * @param allowUnicode | |
132 | * allow Unicode or only allow ASCII Latin characters | |
133 | * @param removeAllAccents | |
134 | * TRUE to replace all accentuated characters by their non | |
135 | * accentuated counter-parts | |
136 | * | |
137 | * @return the sanitised {@link String} | |
138 | */ | |
139 | static public String sanitize(String input, boolean allowUnicode, | |
140 | boolean removeAllAccents) { | |
141 | ||
142 | if (removeAllAccents) { | |
143 | input = Normalizer.normalize(input, Form.NFKD); | |
144 | input = marks.matcher(input).replaceAll(""); | |
145 | } | |
146 | ||
147 | input = Normalizer.normalize(input, Form.NFKC); | |
148 | ||
149 | if (!allowUnicode) { | |
150 | StringBuilder builder = new StringBuilder(); | |
151 | for (int index = 0; index < input.length(); index++) { | |
152 | char car = input.charAt(index); | |
153 | // displayable chars in ASCII are in the range 32<->255, | |
154 | // except DEL (127) | |
155 | if (car >= 32 && car <= 255 && car != 127) { | |
156 | builder.append(car); | |
157 | } | |
158 | } | |
159 | input = builder.toString(); | |
160 | } | |
161 | ||
162 | return input; | |
163 | } | |
164 | ||
165 | /** | |
166 | * Convert between time in milliseconds to {@link String} in a "static" way | |
167 | * (to exchange data over the wire, for instance). | |
168 | * | |
169 | * @param time | |
170 | * the time in milliseconds | |
171 | * | |
172 | * @return the time as a {@link String} | |
173 | */ | |
174 | static public String fromTime(long time) { | |
175 | SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); | |
176 | return sdf.format(new Date(time)); | |
177 | } | |
178 | ||
179 | /** | |
180 | * Convert between time as a {@link String} to milliseconds in a "static" | |
181 | * way (to exchange data over the wire, for instance). | |
182 | * | |
db31c358 | 183 | * @param displayTime |
ec1f3444 NR |
184 | * the time as a {@link String} |
185 | * | |
186 | * @return the time in milliseconds | |
187 | */ | |
db31c358 | 188 | static public long toTime(String displayTime) { |
ec1f3444 NR |
189 | SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); |
190 | try { | |
db31c358 | 191 | return sdf.parse(displayTime).getTime(); |
ec1f3444 NR |
192 | } catch (ParseException e) { |
193 | return -1; | |
194 | } | |
195 | } | |
196 | ||
ec1f3444 NR |
197 | /** |
198 | * Return a hash of the given {@link String}. | |
199 | * | |
200 | * @param input | |
201 | * the input data | |
202 | * | |
203 | * @return the hash | |
204 | */ | |
b771aed5 | 205 | static public String getMd5Hash(String input) { |
ec1f3444 NR |
206 | try { |
207 | MessageDigest md = MessageDigest.getInstance("MD5"); | |
208 | md.update(input.getBytes()); | |
209 | byte byteData[] = md.digest(); | |
210 | ||
211 | StringBuffer hexString = new StringBuffer(); | |
212 | for (int i = 0; i < byteData.length; i++) { | |
213 | String hex = Integer.toHexString(0xff & byteData[i]); | |
214 | if (hex.length() == 1) | |
215 | hexString.append('0'); | |
216 | hexString.append(hex); | |
217 | } | |
218 | ||
219 | return hexString.toString(); | |
220 | } catch (NoSuchAlgorithmException e) { | |
221 | return input; | |
222 | } | |
223 | } | |
224 | ||
ec1f3444 NR |
225 | /** |
226 | * Remove the HTML content from the given input, and un-html-ize the rest. | |
227 | * | |
228 | * @param html | |
229 | * the HTML-encoded content | |
230 | * | |
231 | * @return the HTML-free equivalent content | |
232 | */ | |
233 | public static String unhtml(String html) { | |
234 | StringBuilder builder = new StringBuilder(); | |
235 | ||
236 | int inTag = 0; | |
237 | for (char car : html.toCharArray()) { | |
238 | if (car == '<') { | |
239 | inTag++; | |
240 | } else if (car == '>') { | |
241 | inTag--; | |
242 | } else if (inTag <= 0) { | |
243 | builder.append(car); | |
244 | } | |
245 | } | |
246 | ||
7ee9568b NR |
247 | char nbsp = ' '; // non-breakable space (a special char) |
248 | char space = ' '; | |
249 | return HtmlEscape.unescapeHtml(builder.toString()).replace(nbsp, space); | |
ec1f3444 NR |
250 | } |
251 | ||
252 | /** | |
253 | * Escape the given {@link String} so it can be used in XML, as content. | |
254 | * | |
255 | * @param input | |
256 | * the input {@link String} | |
257 | * | |
258 | * @return the escaped {@link String} | |
259 | */ | |
260 | public static String xmlEscape(String input) { | |
261 | if (input == null) { | |
262 | return ""; | |
263 | } | |
264 | ||
265 | return HtmlEscape.escapeHtml(input, | |
266 | HtmlEscapeType.HTML4_NAMED_REFERENCES_DEFAULT_TO_HEXA, | |
267 | HtmlEscapeLevel.LEVEL_1_ONLY_MARKUP_SIGNIFICANT); | |
268 | } | |
269 | ||
270 | /** | |
271 | * Escape the given {@link String} so it can be used in XML, as text content | |
272 | * inside double-quotes. | |
273 | * | |
274 | * @param input | |
275 | * the input {@link String} | |
276 | * | |
277 | * @return the escaped {@link String} | |
278 | */ | |
279 | public static String xmlEscapeQuote(String input) { | |
280 | if (input == null) { | |
281 | return ""; | |
282 | } | |
283 | ||
284 | return HtmlEscape.escapeHtml(input, | |
285 | HtmlEscapeType.HTML4_NAMED_REFERENCES_DEFAULT_TO_HEXA, | |
286 | HtmlEscapeLevel.LEVEL_1_ONLY_MARKUP_SIGNIFICANT); | |
287 | } | |
db31c358 NR |
288 | |
289 | public static String zip64(String data) { | |
290 | try { | |
291 | return Base64.encodeBytes(data.getBytes(), Base64.GZIP); | |
292 | } catch (IOException e) { | |
293 | e.printStackTrace(); | |
294 | return null; | |
295 | } | |
296 | } | |
297 | ||
298 | public static String unzip64(String data) throws IOException { | |
299 | ByteArrayInputStream in = new ByteArrayInputStream(Base64.decode(data, | |
300 | Base64.GZIP)); | |
301 | ||
302 | Scanner scan = new Scanner(in); | |
303 | scan.useDelimiter("\\A"); | |
304 | try { | |
305 | return scan.next(); | |
306 | } finally { | |
307 | scan.close(); | |
308 | } | |
309 | } | |
ec1f3444 | 310 | } |