gecko/xpcom/ds/nsPersistentProperties.cpp
2010-04-28 17:55:00 -07:00

752 lines
21 KiB
C++

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* ***** BEGIN LICENSE BLOCK *****
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
*
* The contents of this file are subject to the Mozilla Public License Version
* 1.1 (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
* http://www.mozilla.org/MPL/
*
* Software distributed under the License is distributed on an "AS IS" basis,
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
* for the specific language governing rights and limitations under the
* License.
*
* The Original Code is mozilla.org code.
*
* The Initial Developer of the Original Code is
* Netscape Communications Corporation.
* Portions created by the Initial Developer are Copyright (C) 1998
* the Initial Developer. All Rights Reserved.
*
* Contributor(s):
* Pierre Phaneuf <pp@ludusdesign.com>
*
* Alternatively, the contents of this file may be used under the terms of
* either of the GNU General Public License Version 2 or later (the "GPL"),
* or the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
* in which case the provisions of the GPL or the LGPL are applicable instead
* of those above. If you wish to allow use of your version of this file only
* under the terms of either the GPL or the LGPL, and not to allow others to
* use your version of this file under the terms of the MPL, indicate your
* decision by deleting the provisions above and replace them with the notice
* and other provisions required by the GPL or the LGPL. If you do not delete
* the provisions above, a recipient may use your version of this file under
* the terms of any one of the MPL, the GPL or the LGPL.
*
* ***** END LICENSE BLOCK ***** */
#include "nsID.h"
#include "nsCRT.h"
#include "nsReadableUtils.h"
#include "nsIInputStream.h"
#include "nsUnicharInputStream.h"
#include "pratom.h"
#include "nsEnumeratorUtils.h"
#include "nsReadableUtils.h"
#include "nsPrintfCString.h"
#include "nsDependentString.h"
#define PL_ARENA_CONST_ALIGN_MASK 3
#include "nsPersistentProperties.h"
#include "nsIProperties.h"
#include "nsISupportsArray.h"
#include "nsProperties.h"
struct PropertyTableEntry : public PLDHashEntryHdr
{
// both of these are arena-allocated
const char *mKey;
const PRUnichar *mValue;
};
static PRUnichar*
ArenaStrdup(const nsAFlatString& aString, PLArenaPool* aArena)
{
void *mem;
// add one to include the null terminator
PRInt32 len = (aString.Length()+1) * sizeof(PRUnichar);
PL_ARENA_ALLOCATE(mem, aArena, len);
NS_ASSERTION(mem, "Couldn't allocate space!\n");
if (mem) {
memcpy(mem, aString.get(), len);
}
return static_cast<PRUnichar*>(mem);
}
static char*
ArenaStrdup(const nsAFlatCString& aString, PLArenaPool* aArena)
{
void *mem;
// add one to include the null terminator
PRInt32 len = (aString.Length()+1) * sizeof(char);
PL_ARENA_ALLOCATE(mem, aArena, len);
NS_ASSERTION(mem, "Couldn't allocate space!\n");
if (mem)
memcpy(mem, aString.get(), len);
return static_cast<char*>(mem);
}
static const struct PLDHashTableOps property_HashTableOps = {
PL_DHashAllocTable,
PL_DHashFreeTable,
PL_DHashStringKey,
PL_DHashMatchStringKey,
PL_DHashMoveEntryStub,
PL_DHashClearEntryStub,
PL_DHashFinalizeStub,
nsnull,
};
//
// parser stuff
//
enum EParserState {
eParserState_AwaitingKey,
eParserState_Key,
eParserState_AwaitingValue,
eParserState_Value,
eParserState_Comment
};
enum EParserSpecial {
eParserSpecial_None, // not parsing a special character
eParserSpecial_Escaped, // awaiting a special character
eParserSpecial_Unicode // parsing a \Uxxx value
};
class nsPropertiesParser
{
public:
nsPropertiesParser(nsIPersistentProperties* aProps) :
mHaveMultiLine(PR_FALSE), mState(eParserState_AwaitingKey),
mProps(aProps) {}
void FinishValueState(nsAString& aOldValue) {
static const char trimThese[] = " \t";
mKey.Trim(trimThese, PR_FALSE, PR_TRUE);
// This is really ugly hack but it should be fast
PRUnichar backup_char;
PRUint32 minLength = mMinLength;
if (minLength)
{
backup_char = mValue[minLength-1];
mValue.SetCharAt('x', minLength-1);
}
mValue.Trim(trimThese, PR_FALSE, PR_TRUE);
if (minLength)
mValue.SetCharAt(backup_char, minLength-1);
mProps->SetStringProperty(NS_ConvertUTF16toUTF8(mKey), mValue, aOldValue);
mSpecialState = eParserSpecial_None;
WaitForKey();
}
EParserState GetState() { return mState; }
static NS_METHOD SegmentWriter(nsIUnicharInputStream* aStream,
void* aClosure,
const PRUnichar *aFromSegment,
PRUint32 aToOffset,
PRUint32 aCount,
PRUint32 *aWriteCount);
nsresult ParseBuffer(const PRUnichar* aBuffer, PRUint32 aBufferLength);
private:
PRBool ParseValueCharacter(
PRUnichar c, // character that is just being parsed
const PRUnichar* cur, // pointer to character c in the buffer
const PRUnichar* &tokenStart, // string copying is done in blocks as big as
// possible, tokenStart points to the beginning
// of this block
nsAString& oldValue); // when duplicate property is found, new value
// is stored into hashtable and the old one is
// placed in this variable
void WaitForKey() {
mState = eParserState_AwaitingKey;
}
void EnterKeyState() {
mKey.Truncate();
mState = eParserState_Key;
}
void WaitForValue() {
mState = eParserState_AwaitingValue;
}
void EnterValueState() {
mValue.Truncate();
mMinLength = 0;
mState = eParserState_Value;
mSpecialState = eParserSpecial_None;
}
void EnterCommentState() {
mState = eParserState_Comment;
}
nsAutoString mKey;
nsAutoString mValue;
PRUint32 mUnicodeValuesRead; // should be 4!
PRUnichar mUnicodeValue; // currently parsed unicode value
PRBool mHaveMultiLine; // is TRUE when last processed characters form
// any of following sequences:
// - "\\\r"
// - "\\\n"
// - "\\\r\n"
// - any sequence above followed by any
// combination of ' ' and '\t'
PRBool mMultiLineCanSkipN; // TRUE if "\\\r" was detected
PRUint32 mMinLength; // limit right trimming at the end to not trim
// escaped whitespaces
EParserState mState;
// if we see a '\' then we enter this special state
EParserSpecial mSpecialState;
nsIPersistentProperties* mProps;
};
inline PRBool IsWhiteSpace(PRUnichar aChar)
{
return (aChar == ' ') || (aChar == '\t') ||
(aChar == '\r') || (aChar == '\n');
}
inline PRBool IsEOL(PRUnichar aChar)
{
return (aChar == '\r') || (aChar == '\n');
}
PRBool nsPropertiesParser::ParseValueCharacter(
PRUnichar c, const PRUnichar* cur, const PRUnichar* &tokenStart,
nsAString& oldValue)
{
switch (mSpecialState) {
// the normal state - look for special characters
case eParserSpecial_None:
switch (c) {
case '\\':
if (mHaveMultiLine)
// there is nothing to append to mValue yet
mHaveMultiLine = PR_FALSE;
else
mValue += Substring(tokenStart, cur);
mSpecialState = eParserSpecial_Escaped;
break;
case '\n':
// if we detected multiline and got only "\\\r" ignore next "\n" if any
if (mHaveMultiLine && mMultiLineCanSkipN) {
// but don't allow another '\n' to be skipped
mMultiLineCanSkipN = PR_FALSE;
// Now there is nothing to append to the mValue since we are skipping
// whitespaces at the beginning of the new line of the multiline
// property. Set tokenStart properly to ensure that nothing is appended
// if we find regular line-end or the end of the buffer.
tokenStart = cur+1;
break;
}
// no break
case '\r':
// we're done! We have a key and value
mValue += Substring(tokenStart, cur);
FinishValueState(oldValue);
mHaveMultiLine = PR_FALSE;
break;
default:
// there is nothing to do with normal characters,
// but handle multilines correctly
if (mHaveMultiLine) {
if (c == ' ' || c == '\t') {
// don't allow another '\n' to be skipped
mMultiLineCanSkipN = PR_FALSE;
// Now there is nothing to append to the mValue since we are skipping
// whitespaces at the beginning of the new line of the multiline
// property. Set tokenStart properly to ensure that nothing is appended
// if we find regular line-end or the end of the buffer.
tokenStart = cur+1;
break;
}
mHaveMultiLine = PR_FALSE;
tokenStart = cur;
}
break; // from switch on (c)
}
break; // from switch on (mSpecialState)
// saw a \ character, so parse the character after that
case eParserSpecial_Escaped:
// probably want to start parsing at the next token
// other characters, like 'u' might override this
tokenStart = cur+1;
mSpecialState = eParserSpecial_None;
switch (c) {
// the easy characters - \t, \n, and so forth
case 't':
mValue += PRUnichar('\t');
mMinLength = mValue.Length();
break;
case 'n':
mValue += PRUnichar('\n');
mMinLength = mValue.Length();
break;
case 'r':
mValue += PRUnichar('\r');
mMinLength = mValue.Length();
break;
case '\\':
mValue += PRUnichar('\\');
break;
// switch to unicode mode!
case 'u':
case 'U':
mSpecialState = eParserSpecial_Unicode;
mUnicodeValuesRead = 0;
mUnicodeValue = 0;
break;
// a \ immediately followed by a newline means we're going multiline
case '\r':
case '\n':
mHaveMultiLine = PR_TRUE;
mMultiLineCanSkipN = (c == '\r');
mSpecialState = eParserSpecial_None;
break;
default:
// don't recognize the character, so just append it
mValue += c;
break;
}
break;
// we're in the middle of parsing a 4-character unicode value
// like \u5f39
case eParserSpecial_Unicode:
if(('0' <= c) && (c <= '9'))
mUnicodeValue =
(mUnicodeValue << 4) | (c - '0');
else if(('a' <= c) && (c <= 'f'))
mUnicodeValue =
(mUnicodeValue << 4) | (c - 'a' + 0x0a);
else if(('A' <= c) && (c <= 'F'))
mUnicodeValue =
(mUnicodeValue << 4) | (c - 'A' + 0x0a);
else {
// non-hex character. Append what we have, and move on.
mValue += mUnicodeValue;
mMinLength = mValue.Length();
mSpecialState = eParserSpecial_None;
// leave tokenStart at this unknown character, so it gets appended
tokenStart = cur;
// ensure parsing this non-hex character again
return PR_FALSE;
}
if (++mUnicodeValuesRead >= 4) {
tokenStart = cur+1;
mSpecialState = eParserSpecial_None;
mValue += mUnicodeValue;
mMinLength = mValue.Length();
}
break;
}
return PR_TRUE;
}
NS_METHOD nsPropertiesParser::SegmentWriter(nsIUnicharInputStream* aStream,
void* aClosure,
const PRUnichar *aFromSegment,
PRUint32 aToOffset,
PRUint32 aCount,
PRUint32 *aWriteCount)
{
nsPropertiesParser *parser =
static_cast<nsPropertiesParser *>(aClosure);
parser->ParseBuffer(aFromSegment, aCount);
*aWriteCount = aCount;
return NS_OK;
}
nsresult nsPropertiesParser::ParseBuffer(const PRUnichar* aBuffer,
PRUint32 aBufferLength)
{
const PRUnichar* cur = aBuffer;
const PRUnichar* end = aBuffer + aBufferLength;
// points to the start/end of the current key or value
const PRUnichar* tokenStart = nsnull;
// if we're in the middle of parsing a key or value, make sure
// the current token points to the beginning of the current buffer
if (mState == eParserState_Key ||
mState == eParserState_Value) {
tokenStart = aBuffer;
}
nsAutoString oldValue;
while (cur != end) {
PRUnichar c = *cur;
switch (mState) {
case eParserState_AwaitingKey:
if (c == '#' || c == '!')
EnterCommentState();
else if (!IsWhiteSpace(c)) {
// not a comment, not whitespace, we must have found a key!
EnterKeyState();
tokenStart = cur;
}
break;
case eParserState_Key:
if (c == '=' || c == ':') {
mKey += Substring(tokenStart, cur);
WaitForValue();
}
break;
case eParserState_AwaitingValue:
if (IsEOL(c)) {
// no value at all! mimic the normal value-ending
EnterValueState();
FinishValueState(oldValue);
}
// ignore white space leading up to the value
else if (!IsWhiteSpace(c)) {
tokenStart = cur;
EnterValueState();
// make sure to handle this first character
if (ParseValueCharacter(c, cur, tokenStart, oldValue))
cur++;
// If the character isn't consumed, don't do cur++ and parse
// the character again. This can happen f.e. for char 'X' in sequence
// "\u00X". This character can be control character and must be
// processed again.
continue;
}
break;
case eParserState_Value:
if (ParseValueCharacter(c, cur, tokenStart, oldValue))
cur++;
// See few lines above for reason of doing this
continue;
case eParserState_Comment:
// stay in this state till we hit EOL
if (c == '\r' || c== '\n')
WaitForKey();
break;
}
// finally, advance to the next character
cur++;
}
// if we're still parsing the value and are in eParserSpecial_None, then
// append whatever we have..
if (mState == eParserState_Value && tokenStart &&
mSpecialState == eParserSpecial_None) {
mValue += Substring(tokenStart, cur);
}
// if we're still parsing the key, then append whatever we have..
else if (mState == eParserState_Key && tokenStart) {
mKey += Substring(tokenStart, cur);
}
return NS_OK;
}
nsPersistentProperties::nsPersistentProperties()
: mIn(nsnull)
{
mSubclass = static_cast<nsIPersistentProperties*>(this);
mTable.ops = nsnull;
PL_INIT_ARENA_POOL(&mArena, "PersistentPropertyArena", 2048);
}
nsPersistentProperties::~nsPersistentProperties()
{
PL_FinishArenaPool(&mArena);
if (mTable.ops)
PL_DHashTableFinish(&mTable);
}
nsresult
nsPersistentProperties::Init()
{
if (!PL_DHashTableInit(&mTable, &property_HashTableOps, nsnull,
sizeof(PropertyTableEntry), 20)) {
mTable.ops = nsnull;
return NS_ERROR_OUT_OF_MEMORY;
}
return NS_OK;
}
NS_METHOD
nsPersistentProperties::Create(nsISupports *aOuter, REFNSIID aIID, void **aResult)
{
if (aOuter)
return NS_ERROR_NO_AGGREGATION;
nsPersistentProperties* props = new nsPersistentProperties();
if (props == nsnull)
return NS_ERROR_OUT_OF_MEMORY;
NS_ADDREF(props);
nsresult rv = props->Init();
if (NS_SUCCEEDED(rv))
rv = props->QueryInterface(aIID, aResult);
NS_RELEASE(props);
return rv;
}
NS_IMPL_THREADSAFE_ISUPPORTS2(nsPersistentProperties, nsIPersistentProperties, nsIProperties)
NS_IMETHODIMP
nsPersistentProperties::Load(nsIInputStream *aIn)
{
nsresult rv = nsSimpleUnicharStreamFactory::GetInstance()->
CreateInstanceFromUTF8Stream(aIn, getter_AddRefs(mIn));
if (rv != NS_OK) {
NS_WARNING("Error creating UnicharInputStream");
return NS_ERROR_FAILURE;
}
nsPropertiesParser parser(mSubclass);
PRUint32 nProcessed;
// If this 4096 is changed to some other value, make sure to adjust
// the bug121341.properties test file accordingly.
while (NS_SUCCEEDED(rv = mIn->ReadSegments(nsPropertiesParser::SegmentWriter, &parser, 4096, &nProcessed)) &&
nProcessed != 0);
mIn = nsnull;
if (NS_FAILED(rv))
return rv;
// We may have an unprocessed value at this point
// if the last line did not have a proper line ending.
if (parser.GetState() == eParserState_Value) {
nsAutoString oldValue;
parser.FinishValueState(oldValue);
}
return NS_OK;
}
NS_IMETHODIMP
nsPersistentProperties::SetStringProperty(const nsACString& aKey,
const nsAString& aNewValue,
nsAString& aOldValue)
{
const nsAFlatCString& flatKey = PromiseFlatCString(aKey);
PropertyTableEntry *entry =
static_cast<PropertyTableEntry*>
(PL_DHashTableOperate(&mTable, flatKey.get(), PL_DHASH_ADD));
if (entry->mKey) {
aOldValue = entry->mValue;
NS_WARNING(nsPrintfCString(aKey.Length() + 30,
"the property %s already exists\n",
flatKey.get()).get());
}
else {
aOldValue.Truncate();
}
entry->mKey = ArenaStrdup(flatKey, &mArena);
entry->mValue = ArenaStrdup(PromiseFlatString(aNewValue), &mArena);
return NS_OK;
}
NS_IMETHODIMP
nsPersistentProperties::Save(nsIOutputStream* aOut, const nsACString& aHeader)
{
return NS_ERROR_NOT_IMPLEMENTED;
}
NS_IMETHODIMP
nsPersistentProperties::Subclass(nsIPersistentProperties* aSubclass)
{
if (aSubclass) {
mSubclass = aSubclass;
}
return NS_OK;
}
NS_IMETHODIMP
nsPersistentProperties::GetStringProperty(const nsACString& aKey,
nsAString& aValue)
{
const nsAFlatCString& flatKey = PromiseFlatCString(aKey);
PropertyTableEntry *entry =
static_cast<PropertyTableEntry*>
(PL_DHashTableOperate(&mTable, flatKey.get(), PL_DHASH_LOOKUP));
if (PL_DHASH_ENTRY_IS_FREE(entry))
return NS_ERROR_FAILURE;
aValue = entry->mValue;
return NS_OK;
}
static PLDHashOperator
AddElemToArray(PLDHashTable* table, PLDHashEntryHdr *hdr,
PRUint32 i, void *arg)
{
nsISupportsArray *propArray = (nsISupportsArray *) arg;
PropertyTableEntry* entry =
static_cast<PropertyTableEntry*>(hdr);
nsPropertyElement *element =
new nsPropertyElement(nsDependentCString(entry->mKey),
nsDependentString(entry->mValue));
if (!element)
return PL_DHASH_STOP;
propArray->InsertElementAt(element, i);
return PL_DHASH_NEXT;
}
NS_IMETHODIMP
nsPersistentProperties::Enumerate(nsISimpleEnumerator** aResult)
{
nsCOMPtr<nsISupportsArray> propArray;
nsresult rv = NS_NewISupportsArray(getter_AddRefs(propArray));
if (NS_FAILED(rv))
return rv;
// We know the necessary size; we can avoid growing it while adding elements
if (!propArray->SizeTo(mTable.entryCount))
return NS_ERROR_OUT_OF_MEMORY;
// Step through hash entries populating a transient array
PRUint32 n =
PL_DHashTableEnumerate(&mTable, AddElemToArray, (void *)propArray);
if (n < mTable.entryCount)
return NS_ERROR_OUT_OF_MEMORY;
return NS_NewArrayEnumerator(aResult, propArray);
}
////////////////////////////////////////////////////////////////////////////////
// XXX Some day we'll unify the nsIPersistentProperties interface with
// nsIProperties, but until now...
NS_IMETHODIMP
nsPersistentProperties::Get(const char* prop, const nsIID & uuid, void* *result)
{
return NS_ERROR_NOT_IMPLEMENTED;
}
NS_IMETHODIMP
nsPersistentProperties::Set(const char* prop, nsISupports* value)
{
return NS_ERROR_NOT_IMPLEMENTED;
}
NS_IMETHODIMP
nsPersistentProperties::Undefine(const char* prop)
{
return NS_ERROR_NOT_IMPLEMENTED;
}
NS_IMETHODIMP
nsPersistentProperties::Has(const char* prop, PRBool *result)
{
PropertyTableEntry *entry =
static_cast<PropertyTableEntry*>
(PL_DHashTableOperate(&mTable, prop, PL_DHASH_LOOKUP));
*result = (entry && PL_DHASH_ENTRY_IS_BUSY(entry));
return NS_OK;
}
NS_IMETHODIMP
nsPersistentProperties::GetKeys(PRUint32 *count, char ***keys)
{
return NS_ERROR_NOT_IMPLEMENTED;
}
////////////////////////////////////////////////////////////////////////////////
// PropertyElement
////////////////////////////////////////////////////////////////////////////////
NS_METHOD
nsPropertyElement::Create(nsISupports *aOuter, REFNSIID aIID, void **aResult)
{
if (aOuter)
return NS_ERROR_NO_AGGREGATION;
nsPropertyElement* propElem = new nsPropertyElement();
if (propElem == nsnull)
return NS_ERROR_OUT_OF_MEMORY;
NS_ADDREF(propElem);
nsresult rv = propElem->QueryInterface(aIID, aResult);
NS_RELEASE(propElem);
return rv;
}
NS_IMPL_ISUPPORTS1(nsPropertyElement, nsIPropertyElement)
NS_IMETHODIMP
nsPropertyElement::GetKey(nsACString& aReturnKey)
{
aReturnKey = mKey;
return NS_OK;
}
NS_IMETHODIMP
nsPropertyElement::GetValue(nsAString& aReturnValue)
{
aReturnValue = mValue;
return NS_OK;
}
NS_IMETHODIMP
nsPropertyElement::SetKey(const nsACString& aKey)
{
mKey = aKey;
return NS_OK;
}
NS_IMETHODIMP
nsPropertyElement::SetValue(const nsAString& aValue)
{
mValue = aValue;
return NS_OK;
}
////////////////////////////////////////////////////////////////////////////////