String8.h revision 497a92cc5ba2176b8a8484b0a7da040eac0e887b 
1/*
2 * Copyright (C) 2005 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 *      http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#ifndef ANDROID_STRING8_H
18#define ANDROID_STRING8_H
19
20#include <utils/Errors.h>
21
22// Need this for the char16_t type; String8.h should not
23// be depedent on the String16 class.
24#include <utils/String16.h>
25
26#include <stdint.h>
27#include <string.h>
28#include <sys/types.h>
29
30// ---------------------------------------------------------------------------
31
32extern "C" {
33
34typedef uint32_t char32_t;
35
36size_t strlen32(const char32_t *);
37size_t strnlen32(const char32_t *, size_t);
38
39/*
40 * Returns the length of "src" when "src" is valid UTF-8 string.
41 * Returns 0 if src is NULL, 0-length string or non UTF-8 string.
42 * This function should be used to determine whether "src" is valid UTF-8
43 * characters with valid unicode codepoints. "src" must be null-terminated.
44 *
45 * If you are going to use other GetUtf... functions defined in this header
46 * with string which may not be valid UTF-8 with valid codepoint (form 0 to
47 * 0x10FFFF), you should use this function before calling others, since the
48 * other functions do not check whether the string is valid UTF-8 or not.
49 *
50 * If you do not care whether "src" is valid UTF-8 or not, you should use
51 * strlen() as usual, which should be much faster.
52 */
53size_t utf8_length(const char *src);
54
55/*
56 * Returns the UTF-32 length of "src".
57 */
58size_t utf32_length(const char *src, size_t src_len);
59
60/*
61 * Returns the UTF-8 length of "src".
62 */
63size_t utf8_length_from_utf16(const char16_t *src, size_t src_len);
64
65/*
66 * Returns the UTF-8 length of "src".
67 */
68size_t utf8_length_from_utf32(const char32_t *src, size_t src_len);
69
70/*
71 * Returns the unicode value at "index".
72 * Returns -1 when the index is invalid (equals to or more than "src_len").
73 * If returned value is positive, it is able to be converted to char32_t, which
74 * is unsigned. Then, if "next_index" is not NULL, the next index to be used is
75 * stored in "next_index". "next_index" can be NULL.
76 */
77int32_t utf32_at(const char *src, size_t src_len,
78                 size_t index, size_t *next_index);
79
80/*
81 * Stores a UTF-32 string converted from "src" in "dst", if "dst_length" is not
82 * large enough to store the string, the part of the "src" string is stored
83 * into "dst".
84 * Returns the size actually used for storing the string.
85 * "dst" is not null-terminated when dst_len is fully used (like strncpy).
86 */
87size_t utf8_to_utf32(const char* src, size_t src_len,
88                     char32_t* dst, size_t dst_len);
89
90/*
91 * Stores a UTF-8 string converted from "src" in "dst", if "dst_length" is not
92 * large enough to store the string, the part of the "src" string is stored
93 * into "dst" as much as possible. See the examples for more detail.
94 * Returns the size actually used for storing the string.
95 * dst" is not null-terminated when dst_len is fully used (like strncpy).
96 *
97 * Example 1
98 * "src" == \u3042\u3044 (\xE3\x81\x82\xE3\x81\x84)
99 * "src_len" == 2
100 * "dst_len" >= 7
101 * ->
102 * Returned value == 6
103 * "dst" becomes \xE3\x81\x82\xE3\x81\x84\0
104 * (note that "dst" is null-terminated)
105 *
106 * Example 2
107 * "src" == \u3042\u3044 (\xE3\x81\x82\xE3\x81\x84)
108 * "src_len" == 2
109 * "dst_len" == 5
110 * ->
111 * Returned value == 3
112 * "dst" becomes \xE3\x81\x82\0
113 * (note that "dst" is null-terminated, but \u3044 is not stored in "dst"
114 * since "dst" does not have enough size to store the character)
115 *
116 * Example 3
117 * "src" == \u3042\u3044 (\xE3\x81\x82\xE3\x81\x84)
118 * "src_len" == 2
119 * "dst_len" == 6
120 * ->
121 * Returned value == 6
122 * "dst" becomes \xE3\x81\x82\xE3\x81\x84
123 * (note that "dst" is NOT null-terminated, like strncpy)
124 */
125size_t utf32_to_utf8(const char32_t* src, size_t src_len,
126                     char* dst, size_t dst_len);
127
128size_t utf16_to_utf8(const char16_t* src, size_t src_len,
129                     char* dst, size_t dst_len);
130
131}
132
133// ---------------------------------------------------------------------------
134
135namespace android {
136
137class TextOutput;
138
139//! This is a string holding UTF-8 characters. Does not allow the value more
140// than 0x10FFFF, which is not valid unicode codepoint.
141class String8
142{
143public:
144                                String8();
145                                String8(const String8& o);
146    explicit                    String8(const char* o);
147    explicit                    String8(const char* o, size_t numChars);
148
149    explicit                    String8(const String16& o);
150    explicit                    String8(const char16_t* o);
151    explicit                    String8(const char16_t* o, size_t numChars);
152    explicit                    String8(const char32_t* o);
153    explicit                    String8(const char32_t* o, size_t numChars);
154                                ~String8();
155
156    inline  const char*         string() const;
157    inline  size_t              size() const;
158    inline  size_t              length() const;
159    inline  size_t              bytes() const;
160    inline  bool                isEmpty() const;
161
162    inline  const SharedBuffer* sharedBuffer() const;
163
164            void                clear();
165
166            void                setTo(const String8& other);
167            status_t            setTo(const char* other);
168            status_t            setTo(const char* other, size_t numChars);
169            status_t            setTo(const char16_t* other, size_t numChars);
170            status_t            setTo(const char32_t* other,
171                                      size_t length);
172
173            status_t            append(const String8& other);
174            status_t            append(const char* other);
175            status_t            append(const char* other, size_t numChars);
176
177            status_t            appendFormat(const char* fmt, ...)
178                    __attribute__((format (printf, 2, 3)));
179
180            // Note that this function takes O(N) time to calculate the value.
181            // No cache value is stored.
182            size_t              getUtf32Length() const;
183            int32_t             getUtf32At(size_t index,
184                                           size_t *next_index) const;
185            size_t              getUtf32(char32_t* dst, size_t dst_len) const;
186
187    inline  String8&            operator=(const String8& other);
188    inline  String8&            operator=(const char* other);
189
190    inline  String8&            operator+=(const String8& other);
191    inline  String8             operator+(const String8& other) const;
192
193    inline  String8&            operator+=(const char* other);
194    inline  String8             operator+(const char* other) const;
195
196    inline  int                 compare(const String8& other) const;
197
198    inline  bool                operator<(const String8& other) const;
199    inline  bool                operator<=(const String8& other) const;
200    inline  bool                operator==(const String8& other) const;
201    inline  bool                operator!=(const String8& other) const;
202    inline  bool                operator>=(const String8& other) const;
203    inline  bool                operator>(const String8& other) const;
204
205    inline  bool                operator<(const char* other) const;
206    inline  bool                operator<=(const char* other) const;
207    inline  bool                operator==(const char* other) const;
208    inline  bool                operator!=(const char* other) const;
209    inline  bool                operator>=(const char* other) const;
210    inline  bool                operator>(const char* other) const;
211
212    inline                      operator const char*() const;
213
214            char*               lockBuffer(size_t size);
215            void                unlockBuffer();
216            status_t            unlockBuffer(size_t size);
217
218            // return the index of the first byte of other in this at or after
219            // start, or -1 if not found
220            ssize_t             find(const char* other, size_t start = 0) const;
221
222            void                toLower();
223            void                toLower(size_t start, size_t numChars);
224            void                toUpper();
225            void                toUpper(size_t start, size_t numChars);
226
227    /*
228     * These methods operate on the string as if it were a path name.
229     */
230
231    /*
232     * Set the filename field to a specific value.
233     *
234     * Normalizes the filename, removing a trailing '/' if present.
235     */
236    void setPathName(const char* name);
237    void setPathName(const char* name, size_t numChars);
238
239    /*
240     * Get just the filename component.
241     *
242     * "/tmp/foo/bar.c" --> "bar.c"
243     */
244    String8 getPathLeaf(void) const;
245
246    /*
247     * Remove the last (file name) component, leaving just the directory
248     * name.
249     *
250     * "/tmp/foo/bar.c" --> "/tmp/foo"
251     * "/tmp" --> "" // ????? shouldn't this be "/" ???? XXX
252     * "bar.c" --> ""
253     */
254    String8 getPathDir(void) const;
255
256    /*
257     * Retrieve the front (root dir) component.  Optionally also return the
258     * remaining components.
259     *
260     * "/tmp/foo/bar.c" --> "tmp" (remain = "foo/bar.c")
261     * "/tmp" --> "tmp" (remain = "")
262     * "bar.c" --> "bar.c" (remain = "")
263     */
264    String8 walkPath(String8* outRemains = NULL) const;
265
266    /*
267     * Return the filename extension.  This is the last '.' and up to
268     * four characters that follow it.  The '.' is included in case we
269     * decide to expand our definition of what constitutes an extension.
270     *
271     * "/tmp/foo/bar.c" --> ".c"
272     * "/tmp" --> ""
273     * "/tmp/foo.bar/baz" --> ""
274     * "foo.jpeg" --> ".jpeg"
275     * "foo." --> ""
276     */
277    String8 getPathExtension(void) const;
278
279    /*
280     * Return the path without the extension.  Rules for what constitutes
281     * an extension are described in the comment for getPathExtension().
282     *
283     * "/tmp/foo/bar.c" --> "/tmp/foo/bar"
284     */
285    String8 getBasePath(void) const;
286
287    /*
288     * Add a component to the pathname.  We guarantee that there is
289     * exactly one path separator between the old path and the new.
290     * If there is no existing name, we just copy the new name in.
291     *
292     * If leaf is a fully qualified path (i.e. starts with '/', it
293     * replaces whatever was there before.
294     */
295    String8& appendPath(const char* leaf);
296    String8& appendPath(const String8& leaf)  { return appendPath(leaf.string()); }
297
298    /*
299     * Like appendPath(), but does not affect this string.  Returns a new one instead.
300     */
301    String8 appendPathCopy(const char* leaf) const
302                                             { String8 p(*this); p.appendPath(leaf); return p; }
303    String8 appendPathCopy(const String8& leaf) const { return appendPathCopy(leaf.string()); }
304
305    /*
306     * Converts all separators in this string to /, the default path separator.
307     *
308     * If the default OS separator is backslash, this converts all
309     * backslashes to slashes, in-place. Otherwise it does nothing.
310     * Returns self.
311     */
312    String8& convertToResPath();
313
314private:
315            status_t            real_append(const char* other, size_t numChars);
316            char*               find_extension(void) const;
317
318            const char* mString;
319};
320
321TextOutput& operator<<(TextOutput& to, const String16& val);
322
323// ---------------------------------------------------------------------------
324// No user servicable parts below.
325
326inline int compare_type(const String8& lhs, const String8& rhs)
327{
328    return lhs.compare(rhs);
329}
330
331inline int strictly_order_type(const String8& lhs, const String8& rhs)
332{
333    return compare_type(lhs, rhs) < 0;
334}
335
336inline const char* String8::string() const
337{
338    return mString;
339}
340
341inline size_t String8::length() const
342{
343    return SharedBuffer::sizeFromData(mString)-1;
344}
345
346inline size_t String8::size() const
347{
348    return length();
349}
350
351inline bool String8::isEmpty() const
352{
353    return length() == 0;
354}
355
356inline size_t String8::bytes() const
357{
358    return SharedBuffer::sizeFromData(mString)-1;
359}
360
361inline const SharedBuffer* String8::sharedBuffer() const
362{
363    return SharedBuffer::bufferFromData(mString);
364}
365
366inline String8& String8::operator=(const String8& other)
367{
368    setTo(other);
369    return *this;
370}
371
372inline String8& String8::operator=(const char* other)
373{
374    setTo(other);
375    return *this;
376}
377
378inline String8& String8::operator+=(const String8& other)
379{
380    append(other);
381    return *this;
382}
383
384inline String8 String8::operator+(const String8& other) const
385{
386    String8 tmp(*this);
387    tmp += other;
388    return tmp;
389}
390
391inline String8& String8::operator+=(const char* other)
392{
393    append(other);
394    return *this;
395}
396
397inline String8 String8::operator+(const char* other) const
398{
399    String8 tmp(*this);
400    tmp += other;
401    return tmp;
402}
403
404inline int String8::compare(const String8& other) const
405{
406    return strcmp(mString, other.mString);
407}
408
409inline bool String8::operator<(const String8& other) const
410{
411    return strcmp(mString, other.mString) < 0;
412}
413
414inline bool String8::operator<=(const String8& other) const
415{
416    return strcmp(mString, other.mString) <= 0;
417}
418
419inline bool String8::operator==(const String8& other) const
420{
421    return strcmp(mString, other.mString) == 0;
422}
423
424inline bool String8::operator!=(const String8& other) const
425{
426    return strcmp(mString, other.mString) != 0;
427}
428
429inline bool String8::operator>=(const String8& other) const
430{
431    return strcmp(mString, other.mString) >= 0;
432}
433
434inline bool String8::operator>(const String8& other) const
435{
436    return strcmp(mString, other.mString) > 0;
437}
438
439inline bool String8::operator<(const char* other) const
440{
441    return strcmp(mString, other) < 0;
442}
443
444inline bool String8::operator<=(const char* other) const
445{
446    return strcmp(mString, other) <= 0;
447}
448
449inline bool String8::operator==(const char* other) const
450{
451    return strcmp(mString, other) == 0;
452}
453
454inline bool String8::operator!=(const char* other) const
455{
456    return strcmp(mString, other) != 0;
457}
458
459inline bool String8::operator>=(const char* other) const
460{
461    return strcmp(mString, other) >= 0;
462}
463
464inline bool String8::operator>(const char* other) const
465{
466    return strcmp(mString, other) > 0;
467}
468
469inline String8::operator const char*() const
470{
471    return mString;
472}
473
474}  // namespace android
475
476// ---------------------------------------------------------------------------
477
478#endif // ANDROID_STRING8_H
479