51 |
|
|
52 |
#include "pcre_internal.h" |
#include "pcre_internal.h" |
53 |
|
|
54 |
|
/************************************************* |
55 |
|
* Convert any UTF-16 string to host byte order * |
56 |
|
*************************************************/ |
57 |
|
|
58 |
|
/* This function takes an UTF-16 string and converts |
59 |
|
it to host byte order. The length can be explicitly set, |
60 |
|
or autmatically detected for zero terminated strings. |
61 |
|
BOMs can be kept or discarded during the conversion. |
62 |
|
Conversion can be done in place (output == input). |
63 |
|
|
64 |
|
Arguments: |
65 |
|
output the output buffer, its size must be greater |
66 |
|
or equal than the input string |
67 |
|
input any UTF-16 string |
68 |
|
length the number of characters in the input string |
69 |
|
can be less than zero for zero terminated strings |
70 |
|
keep_boms for a non-zero value, the BOM (0xfeff) characters |
71 |
|
are copied as well |
72 |
|
|
73 |
|
Returns: the number of characters placed into the output buffer, |
74 |
|
including the zero-terminator |
75 |
|
*/ |
76 |
|
|
77 |
int |
int |
78 |
pcre16_utf16_to_host_byte_order(PCRE_SCHAR16 *output, PCRE_SPTR16 input, int length, int keep_boms) |
pcre16_utf16_to_host_byte_order(PCRE_SCHAR16 *output, PCRE_SPTR16 input, int length, int keep_boms) |
79 |
{ |
{ |
81 |
/* This function converts any UTF-16 string to host byte order and optionally removes |
/* This function converts any UTF-16 string to host byte order and optionally removes |
82 |
any Byte Order Marks (BOMS). Returns with the remainig length. */ |
any Byte Order Marks (BOMS). Returns with the remainig length. */ |
83 |
BOOL same_bo = TRUE; |
BOOL same_bo = TRUE; |
84 |
PCRE_SPTR16 end = input + length; |
pcre_uchar *optr = (pcre_uchar *)output; |
85 |
|
const pcre_uchar *iptr = (const pcre_uchar *)input; |
86 |
|
const pcre_uchar *end; |
87 |
/* The c variable must be unsigned. */ |
/* The c variable must be unsigned. */ |
88 |
register pcre_uchar c; |
register pcre_uchar c; |
89 |
|
|
90 |
while (input < end) |
if (length < 0) |
91 |
|
length = STRLEN_UC(iptr) + 1; |
92 |
|
end = iptr + length; |
93 |
|
|
94 |
|
while (iptr < end) |
95 |
{ |
{ |
96 |
c = *input++; |
c = *iptr++; |
97 |
if (c == 0xfeff || c == 0xfffe) |
if (c == 0xfeff || c == 0xfffe) |
98 |
{ |
{ |
99 |
/* Detecting the byte order of the machine is unnecessary, it is |
/* Detecting the byte order of the machine is unnecessary, it is |
100 |
enough to know that the UTF-16 string has the same byte order or not. */ |
enough to know that the UTF-16 string has the same byte order or not. */ |
101 |
same_bo = c == 0xfeff; |
same_bo = c == 0xfeff; |
102 |
if (keep_boms != 0) |
if (keep_boms != 0) |
103 |
*output++ = 0xfeff; |
*optr++ = 0xfeff; |
104 |
else |
else |
105 |
length--; |
length--; |
106 |
} |
} |
107 |
else |
else |
108 |
*output++ = same_bo ? c : ((c >> 8) | (c << 8)); /* Flip bytes if needed. */ |
*optr++ = same_bo ? c : ((c >> 8) | (c << 8)); /* Flip bytes if needed. */ |
109 |
} |
} |
110 |
|
|
111 |
#else |
#else |