Qore Programming Language  1.9.0
QoreEncoding.h
Go to the documentation of this file.
1 /* -*- mode: c++; indent-tabs-mode: nil -*- */
2 /*
3  QoreEncoding.h
4 
5  Qore Programming Language
6 
7  Copyright (C) 2003 - 2022 Qore Technologies, s.r.o.
8 
9  Permission is hereby granted, free of charge, to any person obtaining a
10  copy of this software and associated documentation files (the "Software"),
11  to deal in the Software without restriction, including without limitation
12  the rights to use, copy, modify, merge, publish, distribute, sublicense,
13  and/or sell copies of the Software, and to permit persons to whom the
14  Software is furnished to do so, subject to the following conditions:
15 
16  The above copyright notice and this permission notice shall be included in
17  all copies or substantial portions of the Software.
18 
19  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
20  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
21  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
22  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
23  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
24  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
25  DEALINGS IN THE SOFTWARE.
26 
27  Note that the Qore library is released under a choice of three open-source
28  licenses: MIT (as above), LGPL 2+, or GPL 2+; see README-LICENSE for more
29  information.
30 */
31 
32 #ifndef _QORE_ENCODING_H
33 
34 #define _QORE_ENCODING_H
35 
41 #include <qore/common.h>
42 #include <qore/QoreThreadLock.h>
43 
44 #include <cstring>
45 #include <map>
46 #include <string>
47 #include <strings.h>
48 
50 typedef size_t (*mbcs_length_t)(const char* str, const char* end, bool &invalid);
51 
53 typedef size_t (*mbcs_end_t)(const char* str, const char* end, size_t num_chars, bool &invalid);
54 
56 typedef size_t (*mbcs_pos_t)(const char* str, const char* ptr, bool &invalid);
57 
59 
63 typedef qore_offset_t (*mbcs_charlen_t)(const char* str, size_t valid_len);
64 
66 typedef unsigned (*mbcs_get_unicode_t)(const char* p);
67 
68 // private implementation of the QoreEncoding class
69 hashdecl qore_encoding_private;
70 
72 
83 class QoreEncoding {
84  friend hashdecl qore_encoding_private;
85 
86 protected:
87  qore_encoding_private* priv;
88 
89 public:
90  DLLLOCAL QoreEncoding(const char* code, const char* desc = nullptr, unsigned char minwidth = 1,
91  unsigned char maxwidth = 1, mbcs_length_t l = nullptr, mbcs_end_t e = nullptr, mbcs_pos_t p = nullptr,
92  mbcs_charlen_t c = nullptr, mbcs_get_unicode_t gu = nullptr, bool ascii_compat = true);
93 
94  DLLLOCAL ~QoreEncoding();
95 
97 
103  DLLEXPORT size_t getLength(const char* p, const char* end, bool& invalid) const;
104 
106 
112  DLLEXPORT size_t getLength(const char* p, const char* end, ExceptionSink* xsink) const;
113 
115 
122  DLLEXPORT size_t getByteLen(const char* p, const char* end, size_t c, bool& invalid) const;
123 
125 
132  DLLEXPORT size_t getByteLen(const char* p, const char* end, size_t c, ExceptionSink* xsink) const;
133 
135 
141  DLLEXPORT size_t getCharPos(const char* p, const char* end, bool& invalid) const;
142 
144 
150  DLLEXPORT size_t getCharPos(const char* p, const char* end, ExceptionSink* xsink) const;
151 
153 
159  DLLEXPORT qore_offset_t getCharLen(const char* p, size_t valid_len) const;
160 
162  DLLEXPORT bool isMultiByte() const;
163 
165  DLLEXPORT const char* getCode() const;
166 
168  DLLEXPORT const char* getDesc() const;
169 
171  DLLEXPORT int getMaxCharWidth() const;
172 
174 
176  DLLEXPORT unsigned getMinCharWidth() const;
177 
179 
181  DLLEXPORT bool isAsciiCompat() const;
182 
184 
193  DLLEXPORT int getUnicode(const char* p, const char* end, unsigned& clen, ExceptionSink* xsink) const;
194 };
195 
196 // case-insensitive maps for encodings
197 typedef std::map<const char*, QoreEncoding*, ltcstrcase> encoding_map_t;
198 typedef std::map<const char*, const QoreEncoding*, ltcstrcase> const_encoding_map_t;
199 
200 class QoreString;
201 
203 
206 private:
207  DLLLOCAL static encoding_map_t emap;
208  DLLLOCAL static const_encoding_map_t amap;
209  DLLLOCAL static QoreThreadLock mutex;
210 
211  DLLLOCAL static const QoreEncoding* addUnlocked(const char* n_code, const char* n_desc = 0, unsigned char n_minwidth = 1, unsigned char n_maxwidth = 1, mbcs_length_t l = 0, mbcs_end_t e = 0, mbcs_pos_t p = 0, mbcs_charlen_t c = 0, mbcs_get_unicode_t gu = 0, bool n_ascii_compat = true);
212  DLLLOCAL static const QoreEncoding* findUnlocked(const char* name);
213 
214 public:
216  DLLEXPORT static void addAlias(const QoreEncoding* qcs, const char* alias);
217 
219  DLLEXPORT static const QoreEncoding* findCreate(const char* name);
220 
222  DLLEXPORT static const QoreEncoding* findCreate(const QoreString* str);
223 
225  DLLEXPORT static void showEncodings();
226 
228  DLLEXPORT static void showAliases();
229 
231  DLLEXPORT static const QoreEncoding* add(const char* code, const char* desc = 0, unsigned char maxwidth = 1, mbcs_length_t l = 0, mbcs_end_t e = 0, mbcs_pos_t p = 0, mbcs_charlen_t = 0);
232 
233  DLLLOCAL static void init(const char* def);
234  DLLLOCAL QoreEncodingManager();
235  DLLLOCAL ~QoreEncodingManager();
236 };
237 
238 DLLEXPORT size_t q_get_byte_len(const QoreEncoding* enc, const char* p, const char* end, size_t c, ExceptionSink* xsink);
239 DLLEXPORT qore_offset_t q_get_char_len(const QoreEncoding* enc, const char* p, size_t valid_len, ExceptionSink* xsink);
240 
242 DLLEXPORT extern QoreEncodingManager QEM;
243 
244 // builtin character encodings
245 DLLEXPORT extern const QoreEncoding* QCS_DEFAULT,
269 
270 #endif // _QORE_ENCODING_H
DLLEXPORT const QoreEncoding * QCS_UTF16
UTF-16 (only UTF-8 and UTF-16* are multi-byte encodings)
Definition: QoreEncoding.h:248
DLLEXPORT const QoreEncoding * QCS_KOI7
Russian: Kod Obmena Informatsiey, 7 bit characters.
Definition: QoreEncoding.h:268
size_t(* mbcs_pos_t)(const char *str, const char *ptr, bool &invalid)
for multi-byte character set encodings: gives the character position of the ptr
Definition: QoreEncoding.h:56
DLLEXPORT const QoreEncoding * QCS_KOI8_R
Russian: Kod Obmena Informatsiey, 8 bit.
Definition: QoreEncoding.h:266
DLLEXPORT const QoreEncoding * QCS_KOI8_U
Ukrainian: Kod Obmena Informatsiey, 8 bit.
Definition: QoreEncoding.h:267
DLLEXPORT const QoreEncoding * QCS_ISO_8859_4
latin-4, Northern European character set
Definition: QoreEncoding.h:254
DLLEXPORT const QoreEncoding * QCS_ISO_8859_11
Thai character set.
Definition: QoreEncoding.h:261
DLLEXPORT const QoreEncoding * QCS_DEFAULT
the default encoding for the Qore library
DLLEXPORT const QoreEncoding * QCS_ISO_8859_15
latin-9, Western European with euro symbol
Definition: QoreEncoding.h:264
DLLEXPORT const QoreEncoding * QCS_UTF16LE
UTF-16LE (only UTF-8 and UTF-16* are multi-byte encodings)
Definition: QoreEncoding.h:250
DLLEXPORT const QoreEncoding * QCS_ISO_8859_6
Arabic character set.
Definition: QoreEncoding.h:256
DLLEXPORT const QoreEncoding * QCS_ISO_8859_8
Hebrew character set.
Definition: QoreEncoding.h:258
DLLEXPORT const QoreEncoding * QCS_UTF16BE
UTF-16BE (only UTF-8 and UTF-16* are multi-byte encodings)
Definition: QoreEncoding.h:249
DLLEXPORT const QoreEncoding * QCS_ISO_8859_9
latin-5, Turkish character set
Definition: QoreEncoding.h:259
DLLEXPORT const QoreEncoding * QCS_ISO_8859_7
Greek character set.
Definition: QoreEncoding.h:257
DLLEXPORT const QoreEncoding * QCS_ISO_8859_16
latin-10, Southeast European character set
Definition: QoreEncoding.h:265
qore_offset_t(* mbcs_charlen_t)(const char *str, size_t valid_len)
for multi-byte encodings: gives the number of total bytes for the character given one or more charact...
Definition: QoreEncoding.h:63
DLLEXPORT const QoreEncoding * QCS_UTF8
UTF-8 multi-byte encoding (only UTF-8 and UTF-16 are multi-byte encodings)
Definition: QoreEncoding.h:247
DLLEXPORT const QoreEncoding * QCS_ISO_8859_2
latin-2, Central European encoding
Definition: QoreEncoding.h:252
DLLEXPORT const QoreEncoding * QCS_ISO_8859_10
latin-6, Nordic character set
Definition: QoreEncoding.h:260
DLLEXPORT const QoreEncoding * QCS_ISO_8859_1
latin-1, Western European encoding
Definition: QoreEncoding.h:251
DLLEXPORT const QoreEncoding * QCS_ISO_8859_5
Cyrillic character set.
Definition: QoreEncoding.h:255
size_t(* mbcs_end_t)(const char *str, const char *end, size_t num_chars, bool &invalid)
for multi-byte character set encodings: gives the number of bytes for the number of chars
Definition: QoreEncoding.h:53
DLLEXPORT QoreEncodingManager QEM
the QoreEncodingManager object
DLLEXPORT const QoreEncoding * QCS_ISO_8859_3
latin-3, Southern European character set
Definition: QoreEncoding.h:253
size_t(* mbcs_length_t)(const char *str, const char *end, bool &invalid)
for multi-byte character set encodings: gives the length of the string in characters
Definition: QoreEncoding.h:50
DLLEXPORT const QoreEncoding * QCS_USASCII
ascii encoding
Definition: QoreEncoding.h:246
unsigned(* mbcs_get_unicode_t)(const char *p)
returns the unicode code point for the given character, assumes there is enough data for the characte...
Definition: QoreEncoding.h:66
DLLEXPORT const QoreEncoding * QCS_ISO_8859_14
latin-8, Celtic character set
Definition: QoreEncoding.h:263
DLLEXPORT const QoreEncoding * QCS_ISO_8859_13
latin-7, Baltic rim character set
Definition: QoreEncoding.h:262
container for holding Qore-language exception information and also for registering a "thread_exit" ca...
Definition: ExceptionSink.h:48
defines string encoding functions in Qore
Definition: QoreEncoding.h:83
DLLEXPORT bool isMultiByte() const
returns true if the encoding is a multi-byte encoding
DLLEXPORT int getUnicode(const char *p, const char *end, unsigned &clen, ExceptionSink *xsink) const
returns the unicode code point for the given character; if there are any errors (invalid character,...
DLLEXPORT int getMaxCharWidth() const
returns the maximum character width in bytes for the encoding
DLLEXPORT unsigned getMinCharWidth() const
returns the minimum character width in bytes for the encoding
DLLEXPORT size_t getCharPos(const char *p, const char *end, bool &invalid) const
gives the character position (number of characters) starting from the first pointer to the second
DLLEXPORT const char * getCode() const
returns the string code (ex: "UTF-8") for the encoding
DLLEXPORT const char * getDesc() const
returns the description for the encoding
DLLEXPORT size_t getByteLen(const char *p, const char *end, size_t c, bool &invalid) const
gives the number of bytes for the number of chars in the string or up to the end of the string
DLLEXPORT bool isAsciiCompat() const
returns true if the character encoding is backwards-compatible with ASCII
DLLEXPORT qore_offset_t getCharLen(const char *p, size_t valid_len) const
gives the number of total bytes for the next character at the given pointer
DLLEXPORT size_t getLength(const char *p, const char *end, bool &invalid) const
gives the length of the string in characters
manages encodings in Qore
Definition: QoreEncoding.h:205
static DLLEXPORT const QoreEncoding * add(const char *code, const char *desc=0, unsigned char maxwidth=1, mbcs_length_t l=0, mbcs_end_t e=0, mbcs_pos_t p=0, mbcs_charlen_t=0)
adds a new encoding to the list
static DLLEXPORT void showAliases()
prints out all aliases to stdout
static DLLEXPORT const QoreEncoding * findCreate(const char *name)
finds an encoding if it exists (also looks up against alias names) and creates a new one if it doesn'...
static DLLEXPORT void showEncodings()
prints out all valid encodings to stdout
static DLLEXPORT void addAlias(const QoreEncoding *qcs, const char *alias)
adds an alias for an encoding
Qore's string type supported by the QoreEncoding class.
Definition: QoreString.h:93
provides a mutually-exclusive thread lock
Definition: QoreThreadLock.h:49
intptr_t qore_offset_t
used for offsets that could be negative
Definition: common.h:76