| 1 |
/* |
| 2 |
* gchardet.cpp |
| 3 |
* Copyright (C) 2009 Adrian Perez <aperez@igalia.com> |
| 4 |
*/ |
| 5 |
|
| 6 |
#include "gchardet.h" |
| 7 |
#include "nscore.h" |
| 8 |
#include "nsUniversalDetector.h" |
| 9 |
#include <string.h> |
| 10 |
|
| 11 |
|
| 12 |
class CharsetDetector : public nsUniversalDetector |
| 13 |
{ |
| 14 |
public: |
| 15 |
CharsetDetector (PRUint32 aLanguageFilter) |
| 16 |
: nsUniversalDetector (aLanguageFilter) {} |
| 17 |
bool GetDone () const { return mDone; } |
| 18 |
const char* GetCharset () const { return mDetectedCharset; } |
| 19 |
protected: |
| 20 |
virtual void Report(const char* aCharset); |
| 21 |
}; |
| 22 |
|
| 23 |
|
| 24 |
void CharsetDetector::Report(const char* aCharset) |
| 25 |
{ |
| 26 |
(void) aCharset; /* Keep compiler happy */ |
| 27 |
} |
| 28 |
|
| 29 |
|
| 30 |
|
| 31 |
g_chardet_t* |
| 32 |
g_chardet_new (guint32 lang_filter) |
| 33 |
{ |
| 34 |
return reinterpret_cast <void*> (new CharsetDetector (lang_filter)); |
| 35 |
} |
| 36 |
|
| 37 |
|
| 38 |
gboolean |
| 39 |
g_chardet_handle (g_chardet_t *cd, |
| 40 |
const gchar *buf, |
| 41 |
guint32 len) |
| 42 |
{ |
| 43 |
nsresult rv; |
| 44 |
g_return_val_if_fail (cd != NULL, 0); |
| 45 |
|
| 46 |
CharsetDetector *impl = reinterpret_cast <CharsetDetector*> (cd); |
| 47 |
|
| 48 |
/* |
| 49 |
* If the charset was already found, there is no need to keep supplying |
| 50 |
* data: report that properly to our caller. |
| 51 |
*/ |
| 52 |
if (impl -> GetDone ()) |
| 53 |
return false; |
| 54 |
|
| 55 |
rv = impl -> HandleData (buf, len); |
| 56 |
|
| 57 |
/* |
| 58 |
* In theory, the only error so far we can get out of the character set |
| 59 |
* detector is NS_ERROR_OUT_OF_MEMORY. This is done only for some sanity |
| 60 |
* checking, because OOM conditions are are likely to continue happening, |
| 61 |
* so if we force the caller to continue supplygin data and there is free |
| 62 |
* memory in later invocations, things should work. |
| 63 |
*/ |
| 64 |
if (rv != NS_OK) |
| 65 |
g_return_val_if_fail (rv == NS_ERROR_OUT_OF_MEMORY, true); |
| 66 |
|
| 67 |
return ! impl -> GetDone (); |
| 68 |
} |
| 69 |
|
| 70 |
|
| 71 |
void |
| 72 |
g_chardet_free (g_chardet_t *cd) |
| 73 |
{ |
| 74 |
g_return_if_fail (cd != NULL); |
| 75 |
CharsetDetector *impl = reinterpret_cast <CharsetDetector*> (cd); |
| 76 |
delete impl; |
| 77 |
} |
| 78 |
|
| 79 |
const gchar* |
| 80 |
g_chardet_charset (g_chardet_t *cd) |
| 81 |
{ |
| 82 |
g_return_val_if_fail (cd != NULL, NULL); |
| 83 |
CharsetDetector *impl = reinterpret_cast <CharsetDetector*> (cd); |
| 84 |
return (impl -> GetDone()) ? impl -> GetCharset() : NULL; |
| 85 |
} |
| 86 |
|
| 87 |
|
| 88 |
gboolean |
| 89 |
g_chardet_found (g_chardet_t *cd) |
| 90 |
{ |
| 91 |
g_return_val_if_fail (cd != NULL, FALSE); |
| 92 |
CharsetDetector *impl = reinterpret_cast <CharsetDetector*> (cd); |
| 93 |
return impl -> GetDone(); |
| 94 |
} |
| 95 |
|
| 96 |
void |
| 97 |
g_chardet_finish (g_chardet_t *cd) |
| 98 |
{ |
| 99 |
g_return_if_fail (cd != NULL); |
| 100 |
CharsetDetector *impl = reinterpret_cast <CharsetDetector*> (cd); |
| 101 |
impl -> DataEnd (); |
| 102 |
} |
| 103 |
|
| 104 |
|
| 105 |
gchar* |
| 106 |
g_chardet_detect (const gchar *string, guint32 lang_filter) |
| 107 |
{ |
| 108 |
g_return_val_if_fail (string != NULL, NULL); |
| 109 |
CharsetDetector impl (lang_filter); |
| 110 |
impl.HandleData (string, strlen (string)); |
| 111 |
return g_strdup (impl.GetCharset ()); |
| 112 |
} |
