PostgreSQL Source Code  git master
euc_cn_and_mic.c
Go to the documentation of this file.
1 /*-------------------------------------------------------------------------
2  *
3  * EUC_CN and MULE_INTERNAL
4  *
5  * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group
6  * Portions Copyright (c) 1994, Regents of the University of California
7  *
8  * IDENTIFICATION
9  * src/backend/utils/mb/conversion_procs/euc_cn_and_mic/euc_cn_and_mic.c
10  *
11  *-------------------------------------------------------------------------
12  */
13 
14 #include "postgres.h"
15 #include "fmgr.h"
16 #include "mb/pg_wchar.h"
17 
19 
22 
23 /* ----------
24  * conv_proc(
25  * INTEGER, -- source encoding id
26  * INTEGER, -- destination encoding id
27  * CSTRING, -- source string (null terminated C string)
28  * CSTRING, -- destination string (null terminated C string)
29  * INTEGER, -- source string length
30  * BOOL -- if true, don't throw an error if conversion fails
31  * ) returns INTEGER;
32  *
33  * Returns the number of bytes successfully converted.
34  * ----------
35  */
36 
37 static int euc_cn2mic(const unsigned char *euc, unsigned char *p, int len, bool noError);
38 static int mic2euc_cn(const unsigned char *mic, unsigned char *p, int len, bool noError);
39 
40 Datum
42 {
43  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
44  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
45  int len = PG_GETARG_INT32(4);
46  bool noError = PG_GETARG_BOOL(5);
47  int converted;
48 
50 
51  converted = euc_cn2mic(src, dest, len, noError);
52 
53  PG_RETURN_INT32(converted);
54 }
55 
56 Datum
58 {
59  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
60  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
61  int len = PG_GETARG_INT32(4);
62  bool noError = PG_GETARG_BOOL(5);
63  int converted;
64 
66 
67  converted = mic2euc_cn(src, dest, len, noError);
68 
69  PG_RETURN_INT32(converted);
70 }
71 
72 /*
73  * EUC_CN ---> MIC
74  */
75 static int
76 euc_cn2mic(const unsigned char *euc, unsigned char *p, int len, bool noError)
77 {
78  const unsigned char *start = euc;
79  int c1;
80 
81  while (len > 0)
82  {
83  c1 = *euc;
84  if (IS_HIGHBIT_SET(c1))
85  {
86  if (len < 2 || !IS_HIGHBIT_SET(euc[1]))
87  {
88  if (noError)
89  break;
90  report_invalid_encoding(PG_EUC_CN, (const char *) euc, len);
91  }
92  *p++ = LC_GB2312_80;
93  *p++ = c1;
94  *p++ = euc[1];
95  euc += 2;
96  len -= 2;
97  }
98  else
99  { /* should be ASCII */
100  if (c1 == 0)
101  {
102  if (noError)
103  break;
104  report_invalid_encoding(PG_EUC_CN, (const char *) euc, len);
105  }
106  *p++ = c1;
107  euc++;
108  len--;
109  }
110  }
111  *p = '\0';
112 
113  return euc - start;
114 }
115 
116 /*
117  * MIC ---> EUC_CN
118  */
119 static int
120 mic2euc_cn(const unsigned char *mic, unsigned char *p, int len, bool noError)
121 {
122  const unsigned char *start = mic;
123  int c1;
124 
125  while (len > 0)
126  {
127  c1 = *mic;
128  if (IS_HIGHBIT_SET(c1))
129  {
130  if (c1 != LC_GB2312_80)
131  {
132  if (noError)
133  break;
135  (const char *) mic, len);
136  }
137  if (len < 3 || !IS_HIGHBIT_SET(mic[1]) || !IS_HIGHBIT_SET(mic[2]))
138  {
139  if (noError)
140  break;
142  (const char *) mic, len);
143  }
144  mic++;
145  *p++ = *mic++;
146  *p++ = *mic++;
147  len -= 3;
148  }
149  else
150  { /* should be ASCII */
151  if (c1 == 0)
152  {
153  if (noError)
154  break;
156  (const char *) mic, len);
157  }
158  *p++ = c1;
159  mic++;
160  len--;
161  }
162  }
163  *p = '\0';
164 
165  return mic - start;
166 }
#define PG_GETARG_INT32(n)
Definition: fmgr.h:269
void report_invalid_encoding(int encoding, const char *mbstr, int len)
Definition: mbutils.c:1647
#define PG_RETURN_INT32(x)
Definition: fmgr.h:354
#define PG_GETARG_BOOL(n)
Definition: fmgr.h:274
Datum mic_to_euc_cn(PG_FUNCTION_ARGS)
#define IS_HIGHBIT_SET(ch)
Definition: c.h:1156
#define LC_GB2312_80
Definition: pg_wchar.h:130
static int euc_cn2mic(const unsigned char *euc, unsigned char *p, int len, bool noError)
uintptr_t Datum
Definition: postgres.h:411
#define CHECK_ENCODING_CONVERSION_ARGS(srcencoding, destencoding)
Definition: pg_wchar.h:527
void report_untranslatable_char(int src_encoding, int dest_encoding, const char *mbstr, int len)
Definition: mbutils.c:1679
PG_MODULE_MAGIC
Datum euc_cn_to_mic(PG_FUNCTION_ARGS)
PG_FUNCTION_INFO_V1(euc_cn_to_mic)
static int mic2euc_cn(const unsigned char *mic, unsigned char *p, int len, bool noError)
#define PG_GETARG_CSTRING(n)
Definition: fmgr.h:277
#define PG_FUNCTION_ARGS
Definition: fmgr.h:193