1 | /**********************************************************************
|
---|
2 | *
|
---|
3 | * text_t.h -- a simple 16-bit character string class
|
---|
4 | * Copyright (C) 1999 The New Zealand Digital Library Project
|
---|
5 | *
|
---|
6 | * A component of the Greenstone digital library software
|
---|
7 | * from the New Zealand Digital Library Project at the
|
---|
8 | * University of Waikato, New Zealand.
|
---|
9 | *
|
---|
10 | * This program is free software; you can redistribute it and/or modify
|
---|
11 | * it under the terms of the GNU General Public License as published by
|
---|
12 | * the Free Software Foundation; either version 2 of the License, or
|
---|
13 | * (at your option) any later version.
|
---|
14 | *
|
---|
15 | * This program is distributed in the hope that it will be useful,
|
---|
16 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | * GNU General Public License for more details.
|
---|
19 | *
|
---|
20 | * You should have received a copy of the GNU General Public License
|
---|
21 | * along with this program; if not, write to the Free Software
|
---|
22 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | *
|
---|
24 | * $Id: text_t.h 14342 2007-08-08 03:42:44Z mdewsnip $
|
---|
25 | *
|
---|
26 | *********************************************************************/
|
---|
27 |
|
---|
28 |
|
---|
29 | #ifndef TEXT_T_H
|
---|
30 | #define TEXT_T_H
|
---|
31 |
|
---|
32 | #include "gsdlconf.h"
|
---|
33 |
|
---|
34 | #if defined(GSDL_USE_OBJECTSPACE)
|
---|
35 | # include <ospace\std\vector>
|
---|
36 | # include <ospace\std\list>
|
---|
37 | # include <ospace\std\set>
|
---|
38 | # include <ospace\std\map>
|
---|
39 | #elif defined(GSDL_USE_STL_H)
|
---|
40 | # include <vector.h>
|
---|
41 | # include <list.h>
|
---|
42 | # include <set.h>
|
---|
43 | # include <map.h>
|
---|
44 | #else
|
---|
45 | # include <vector>
|
---|
46 | # include <list>
|
---|
47 | # include <set>
|
---|
48 | # include <map>
|
---|
49 | #endif
|
---|
50 |
|
---|
51 | // use the standard namespace
|
---|
52 | #if !defined (GSDL_NAMESPACE_BROKEN)
|
---|
53 | #if defined(GSDL_USE_OBJECTSPACE)
|
---|
54 | using namespace ospace::std;
|
---|
55 | #else
|
---|
56 | using namespace std;
|
---|
57 | #endif
|
---|
58 | #endif
|
---|
59 |
|
---|
60 | // class prototypes
|
---|
61 | class text_t;
|
---|
62 | class inconvertclass;
|
---|
63 | class outconvertclass;
|
---|
64 |
|
---|
65 |
|
---|
66 | // for those stupid compilers which need it
|
---|
67 | #if defined(GSDL_NEED_DESTROY_USHORT)
|
---|
68 | inline void destroy(unsigned short *) {};
|
---|
69 | inline void destroy(int *) {};
|
---|
70 | #endif
|
---|
71 |
|
---|
72 | typedef vector<unsigned short> usvector;
|
---|
73 | extern const text_t g_EmptyText;
|
---|
74 |
|
---|
75 | // The class text_t can handle long strings which may contain
|
---|
76 | // null characters. It uses unsigned shorts to represent up to
|
---|
77 | // 64K character values.
|
---|
78 | class text_t {
|
---|
79 | public:
|
---|
80 | //type support for ucvector
|
---|
81 | typedef usvector::iterator iterator;
|
---|
82 | typedef usvector::const_iterator const_iterator;
|
---|
83 | typedef usvector::reference reference;
|
---|
84 | typedef usvector::const_reference const_reference;
|
---|
85 | typedef usvector::size_type size_type;
|
---|
86 | typedef usvector::difference_type difference_type;
|
---|
87 | typedef usvector::const_reverse_iterator const_reverse_iterator;
|
---|
88 | typedef usvector::reverse_iterator reverse_iterator;
|
---|
89 |
|
---|
90 | protected:
|
---|
91 | usvector text;
|
---|
92 | unsigned short encoding; // 0 = unicode, 1 = other
|
---|
93 |
|
---|
94 | public:
|
---|
95 | // constructors
|
---|
96 | text_t ();
|
---|
97 | text_t (int i);
|
---|
98 | text_t (const char *s); // assumed to be a normal c string
|
---|
99 | text_t (const char *s, size_type nLength); // support for arrays of chars
|
---|
100 | void setencoding (unsigned short theencoding) {encoding=theencoding;};
|
---|
101 | unsigned short getencoding () {return encoding;};
|
---|
102 |
|
---|
103 | usvector& text_as_usvector() { return text ; };
|
---|
104 | const usvector& text_as_usvector() const { return text ; };
|
---|
105 |
|
---|
106 | // basic container support
|
---|
107 | iterator begin () {return text.begin();}
|
---|
108 | const_iterator begin () const {return text.begin();}
|
---|
109 | iterator end () {return text.end();}
|
---|
110 | const_iterator end () const {return text.end();}
|
---|
111 |
|
---|
112 | void erase(iterator pos) {text.erase(pos);}
|
---|
113 | void erase(iterator first, iterator last) {text.erase(first, last);}
|
---|
114 | void push_back(unsigned short c) {text.push_back(c);}
|
---|
115 | void pop_back() {text.pop_back();}
|
---|
116 | text_t &operator=(const text_t &x) {text=x.text; encoding=x.encoding; return *this;}
|
---|
117 | reference operator[](size_type n) {return text[n];};
|
---|
118 | const_reference operator[](size_type n) const {return text[n];};
|
---|
119 |
|
---|
120 | void reserve (size_type n) {text.reserve(n);}
|
---|
121 | size_type capacity() const { return text.capacity(); }
|
---|
122 |
|
---|
123 | bool empty () const {return text.empty();}
|
---|
124 | size_type size() const {return text.size();}
|
---|
125 | friend inline bool operator!=(const text_t& x, const text_t& y)
|
---|
126 | {return (x.text != y.text);}
|
---|
127 | friend inline bool operator==(const text_t& x, const text_t& y)
|
---|
128 | {return (x.text == y.text);}
|
---|
129 | friend inline bool operator<(const text_t& x, const text_t& y)
|
---|
130 | {return (x.text < y.text);}
|
---|
131 | friend inline bool operator>(const text_t& x, const text_t& y)
|
---|
132 | {return (x.text > y.text);}
|
---|
133 | friend inline bool operator>=(const text_t& x, const text_t& y)
|
---|
134 | {return (x.text >= y.text);}
|
---|
135 | friend inline bool operator<=(const text_t& x, const text_t& y)
|
---|
136 | {return (x.text <= y.text);}
|
---|
137 |
|
---|
138 | // added functionality
|
---|
139 | void clear () {text.erase(text.begin(),text.end());}
|
---|
140 | void append (const text_t &t);
|
---|
141 | void appendrange (iterator first, iterator last);
|
---|
142 | void appendrange (const_iterator first, const_iterator last);
|
---|
143 | text_t &operator+= (const text_t &t) {append(t);return *this;}
|
---|
144 |
|
---|
145 | // support for integers
|
---|
146 | void appendint (int i);
|
---|
147 | void setint (int i) {clear();appendint(i);}
|
---|
148 | text_t &operator=(int i) {setint (i);return *this;}
|
---|
149 | text_t &operator+= (int i) {appendint(i);return *this;}
|
---|
150 | int getint () const;
|
---|
151 |
|
---|
152 | // same as getint but returns an unsigned long
|
---|
153 | unsigned long getulong () const;
|
---|
154 |
|
---|
155 | // support for arrays of chars
|
---|
156 | void appendcarr (const char *s, size_type len);
|
---|
157 | void setcarr (char *s, size_type len) {clear();appendcarr(s,len);}
|
---|
158 |
|
---|
159 | // support for const null-terminated C strings
|
---|
160 | void appendcstr (const char *s);
|
---|
161 | void setcstr (const char *s) {clear();appendcstr(s);}
|
---|
162 | text_t &operator= (const char *s) {setcstr(s);return *this;} // c string
|
---|
163 | text_t &operator+= (const char *s) {appendcstr(s);return *this;} // c string
|
---|
164 |
|
---|
165 | // strings returned from getcarr and getcstr become the callers
|
---|
166 | // responsibility and should be deallocated with "delete []"
|
---|
167 | char *getcarr(size_type &len) const;
|
---|
168 | char *getcstr() const;
|
---|
169 |
|
---|
170 | int replace(text_t toreplace, text_t replacement);
|
---|
171 | };
|
---|
172 |
|
---|
173 | // new stream converter ...
|
---|
174 | ostream& operator<< (ostream &o, const text_t &text);
|
---|
175 |
|
---|
176 | inline text_t operator+(const text_t &t1, const text_t &t2)
|
---|
177 | {
|
---|
178 | text_t tnew = t1;
|
---|
179 | tnew.append(t2);
|
---|
180 | return tnew;
|
---|
181 | }
|
---|
182 |
|
---|
183 | inline text_t operator+(const text_t &t1, int i1)
|
---|
184 | {
|
---|
185 | text_t tnew = t1;
|
---|
186 | tnew.appendint(i1);
|
---|
187 | return tnew;
|
---|
188 | }
|
---|
189 |
|
---|
190 | inline text_t operator+(const text_t &t1, char *s1)
|
---|
191 | {
|
---|
192 | text_t tnew = t1;
|
---|
193 | tnew.appendcstr(s1);
|
---|
194 | return tnew;
|
---|
195 | }
|
---|
196 |
|
---|
197 |
|
---|
198 |
|
---|
199 |
|
---|
200 | struct eqtext_t
|
---|
201 | {
|
---|
202 | bool operator()(const text_t &t1, const text_t &t2) const
|
---|
203 | { return t1 == t2; }
|
---|
204 | };
|
---|
205 |
|
---|
206 | struct lttext_t
|
---|
207 | {
|
---|
208 | bool operator()(const text_t &t1, const text_t &t2) const
|
---|
209 | { return t1 < t2; }
|
---|
210 | };
|
---|
211 |
|
---|
212 |
|
---|
213 | // frequently used derived types
|
---|
214 | typedef set<text_t,lttext_t> text_tset;
|
---|
215 | typedef list<text_t> text_tlist; // more efficient for insertions/deletions
|
---|
216 | typedef vector<text_t> text_tarray; // more space efficient than text_tlist
|
---|
217 | typedef map<text_t, text_t, lttext_t> text_tmap;
|
---|
218 |
|
---|
219 |
|
---|
220 | // general functions which work on text_ts
|
---|
221 |
|
---|
222 | // find a character within a range
|
---|
223 | text_t::const_iterator findchar (text_t::const_iterator first, text_t::const_iterator last,
|
---|
224 | unsigned short c);
|
---|
225 | text_t::iterator findchar (text_t::iterator first, text_t::iterator last,
|
---|
226 | unsigned short c);
|
---|
227 | text_t::iterator findlastchar (text_t::iterator first, text_t::iterator last,
|
---|
228 | unsigned short c);
|
---|
229 | text_t::iterator findword (text_t::iterator first, text_t::iterator last,
|
---|
230 | const text_t &word);
|
---|
231 |
|
---|
232 | // get a string up to the next delimiter (which is skipped)
|
---|
233 | text_t::const_iterator getdelimitstr (text_t::const_iterator first,
|
---|
234 | text_t::const_iterator last,
|
---|
235 | unsigned short c, text_t &outstr);
|
---|
236 | text_t::iterator getdelimitstr (text_t::iterator first, text_t::iterator last,
|
---|
237 | unsigned short c, text_t &outstr);
|
---|
238 |
|
---|
239 | // split a string with a character
|
---|
240 | void splitchar (text_t::const_iterator first, text_t::const_iterator last,
|
---|
241 | unsigned short c, text_tset &outlist);
|
---|
242 | void splitchar (text_t::const_iterator first, text_t::const_iterator last,
|
---|
243 | unsigned short c, text_tlist &outlist);
|
---|
244 | void splitchar (text_t::const_iterator first, text_t::const_iterator last,
|
---|
245 | unsigned short c, text_tarray &outlist);
|
---|
246 |
|
---|
247 | // join a string using a character
|
---|
248 | void joinchar (const text_tset &inlist, unsigned short c, text_t &outtext);
|
---|
249 | void joinchar (const text_tlist &inlist, unsigned short c, text_t &outtext);
|
---|
250 | void joinchar (const text_tarray &inlist, unsigned short c, text_t &outtext);
|
---|
251 | void joinchar (const text_tset &inlist, const text_t &c, text_t &outtext);
|
---|
252 | void joinchar (const text_tlist &inlist, const text_t &c, text_t &outtext);
|
---|
253 | void joinchar (const text_tarray &inlist, const text_t &c, text_t &outtext);
|
---|
254 |
|
---|
255 | // count the occurances of a character within a range
|
---|
256 | int countchar (text_t::const_iterator first, text_t::const_iterator last,
|
---|
257 | unsigned short c);
|
---|
258 |
|
---|
259 | // return a substring of string from first up to but not including last
|
---|
260 | text_t substr (text_t::const_iterator first, text_t::const_iterator last);
|
---|
261 |
|
---|
262 | // convert to lowercase
|
---|
263 | void lc (text_t::iterator first, text_t::iterator last);
|
---|
264 | inline void lc (text_t &t) {lc (t.begin(), t.end());}
|
---|
265 |
|
---|
266 | // convert to uppercase
|
---|
267 | void uc (text_t::iterator first, text_t::iterator last);
|
---|
268 | inline void uc (text_t &t) {uc (t.begin(), t.end());}
|
---|
269 |
|
---|
270 | // checks to see if it is a number (i.e. contains only 0-9)
|
---|
271 | bool is_number (const text_t &text);
|
---|
272 |
|
---|
273 | // checks to see if the text has any letters or digits
|
---|
274 | bool has_unicode_letdig (const text_t &text);
|
---|
275 |
|
---|
276 | // checks to see if a text_t starts with the specified prefix
|
---|
277 | bool starts_with(const text_t& text, const text_t& prefix);
|
---|
278 | // checks to see if a text_t ends with the specified suffix
|
---|
279 | bool ends_with(const text_t& text, const text_t& suffix);
|
---|
280 |
|
---|
281 | // conversion classes used for getting information in to and out of
|
---|
282 | // the text_t class.
|
---|
283 |
|
---|
284 | class convertclass
|
---|
285 | {
|
---|
286 | public:
|
---|
287 | enum status_t {finished, stopped, unfinished};
|
---|
288 |
|
---|
289 | convertclass ();
|
---|
290 | virtual void reset ();
|
---|
291 | };
|
---|
292 |
|
---|
293 |
|
---|
294 |
|
---|
295 | // convert from a char stream to the text_t class
|
---|
296 | // the default version assumes the input is a ascii
|
---|
297 | // character array
|
---|
298 | class inconvertclass : public convertclass
|
---|
299 | {
|
---|
300 | public:
|
---|
301 | inconvertclass ();
|
---|
302 | void reset ();
|
---|
303 | void setinput (char *thestart, size_t thelen);
|
---|
304 |
|
---|
305 | // output will be cleared before the conversion
|
---|
306 | virtual void convert (text_t &output, status_t &status);
|
---|
307 |
|
---|
308 | // will treat the text_t as a 8-bit string and convert
|
---|
309 | // it to a 16-bit string using the about convert method.
|
---|
310 | text_t convert (const text_t &t);
|
---|
311 |
|
---|
312 | protected:
|
---|
313 | char *start;
|
---|
314 | size_t len;
|
---|
315 | };
|
---|
316 |
|
---|
317 | // to get something which will do the conversion
|
---|
318 | // to ascii declare a (non global!) instance like
|
---|
319 | // this
|
---|
320 | // inconvertclass ascii2text_t;
|
---|
321 |
|
---|
322 | #if defined(GSDL_USE_IOS_H)
|
---|
323 | #include <iostream.h> // darwin doesn't have ostream.h...
|
---|
324 | #else
|
---|
325 | #include <ostream>
|
---|
326 | #endif
|
---|
327 |
|
---|
328 | // Convert from a text_t class to a char stream
|
---|
329 | // This default version assumes the output is a ascii
|
---|
330 | // character array. If you set the output stream you
|
---|
331 | // can use this class to output to a stream using the
|
---|
332 | // << operator. The << operator can also be conveniently
|
---|
333 | // used to set the output stream by doing something like
|
---|
334 | //
|
---|
335 | // cout << text_t2ascii << textstr << anothertextstr;
|
---|
336 | //
|
---|
337 | // this class assumes that the input text doesn't change
|
---|
338 | // while the conversion takes place
|
---|
339 | class outconvertclass : public convertclass
|
---|
340 | {
|
---|
341 | public:
|
---|
342 | outconvertclass ();
|
---|
343 | virtual void reset ();
|
---|
344 | virtual void setinput (text_t *theinput);
|
---|
345 | virtual void setdata(text_t *input, text_t::iterator texthere);
|
---|
346 | // note that convert does not null-terminate the
|
---|
347 | // output array of characters
|
---|
348 | virtual void convert (char *output, size_t maxlen,
|
---|
349 | size_t &len, status_t &status);
|
---|
350 |
|
---|
351 | // will convert the 16-bit string to a 8-bit stream
|
---|
352 | // and place the result in a text_t. This method uses
|
---|
353 | // the above convert function.
|
---|
354 | text_t convert (const text_t &t);
|
---|
355 |
|
---|
356 | virtual void setostream (ostream *theouts);
|
---|
357 | ostream *getostream ();
|
---|
358 |
|
---|
359 | protected:
|
---|
360 | text_t *input;
|
---|
361 | text_t::iterator texthere; // only valid if input is valid
|
---|
362 |
|
---|
363 | ostream *outs;
|
---|
364 | };
|
---|
365 |
|
---|
366 | // to get something which will do the conversion
|
---|
367 | // to text_t declare a (non global!) instance like
|
---|
368 | // this
|
---|
369 | // outconvertclass text_t2ascii;
|
---|
370 |
|
---|
371 |
|
---|
372 | // stream operators for the output class
|
---|
373 | outconvertclass &operator<< (ostream &theouts, outconvertclass &outconverter);
|
---|
374 | outconvertclass &operator<< (outconvertclass &outconverter, const text_t &t);
|
---|
375 |
|
---|
376 |
|
---|
377 | #endif
|
---|