protected internal override object GetInstanceIdValue(int id)
{
int shifted = id - base.MaxInstanceId;
if (1 <= shifted && shifted <= MAX_INSTANCE_ID)
{
RegExpImpl impl = Impl;
object stringResult;
switch (shifted)
{
case Id_multiline:
case Id_STAR:
return(impl.multiline);
case Id_input:
case Id_UNDERSCORE:
stringResult = impl.input;
break;
case Id_lastMatch:
case Id_AMPERSAND:
stringResult = impl.lastMatch;
break;
case Id_lastParen:
case Id_PLUS:
stringResult = impl.lastParen;
break;
case Id_leftContext:
case Id_BACK_QUOTE:
stringResult = impl.leftContext;
break;
case Id_rightContext:
case Id_QUOTE:
stringResult = impl.rightContext;
break;
default: {
// Must be one of $1..$9, convert to 0..8
int substring_number = shifted - DOLLAR_ID_BASE - 1;
stringResult = impl.getParenSubString(substring_number);
break;
}
}
return((stringResult == null) ? "" : stringResult.ToString());
}
return(base.GetInstanceIdValue(id));
}
/// <summary> Analog of do_replace in jsstr.c</summary>
private static void do_replace(GlobData rdata, Context cx, RegExpImpl regExpImpl)
{
System.Text.StringBuilder charBuf = rdata.charBuf;
int cp = 0;
string da = rdata.repstr;
int dp = rdata.dollar;
if (dp != -1)
{
int [] skip = new int [1];
do
{
int len = dp - cp;
charBuf.Append(da.Substring(cp, (dp) - (cp)));
cp = dp;
SubString sub = interpretDollar(cx, regExpImpl, da, dp, skip);
if (sub != null)
{
len = sub.length;
if (len > 0)
{
charBuf.Append(sub.charArray, sub.index, len);
}
cp += skip [0];
dp += skip [0];
}
else
{
++dp;
}
dp = da.IndexOf((char)'$', dp);
}while (dp >= 0);
}
int daL = da.Length;
if (daL > cp)
{
charBuf.Append(da.Substring(cp, (daL) - (cp)));
}
}
/// <summary> Analog of C match_or_replace.</summary>
private static object matchOrReplace (Context cx, IScriptable scope, IScriptable thisObj, object [] args, RegExpImpl reImpl, GlobData data, bool forceFlat)
{
BuiltinRegExp re;
string str = ScriptConvert.ToString (thisObj);
data.str = str;
IScriptable topScope = ScriptableObject.GetTopLevelScope (scope);
if (args.Length == 0) {
object compiled = BuiltinRegExp.compileRE ("", "", false);
re = new BuiltinRegExp (topScope, compiled);
}
else if (args [0] is BuiltinRegExp) {
re = (BuiltinRegExp)args [0];
}
else {
string src = ScriptConvert.ToString (args [0]);
string opt;
if (data.optarg < args.Length) {
args [0] = src;
opt = ScriptConvert.ToString (args [data.optarg]);
}
else {
opt = null;
}
object compiled = BuiltinRegExp.compileRE (src, opt, forceFlat);
re = new BuiltinRegExp (topScope, compiled);
}
data.regexp = re;
data.global = (re.Flags & BuiltinRegExp.JSREG_GLOB) != 0;
int [] indexp = new int [] { 0 };
object result = null;
if (data.mode == EcmaScript.NET.RegExpActions.Search) {
result = re.executeRegExp (cx, scope, reImpl, str, indexp, BuiltinRegExp.TEST);
if (result != null && result.Equals (true))
result = (int)reImpl.leftContext.length;
else
result = -1;
}
else if (data.global) {
re.lastIndex = 0;
for (int count = 0; indexp [0] <= str.Length; count++) {
result = re.executeRegExp (cx, scope, reImpl, str, indexp, BuiltinRegExp.TEST);
if (result == null || !result.Equals (true))
break;
if (data.mode == EcmaScript.NET.RegExpActions.Match) {
match_glob (data, cx, scope, count, reImpl);
}
else {
if (data.mode != EcmaScript.NET.RegExpActions.Replace)
Context.CodeBug ();
SubString lastMatch = reImpl.lastMatch;
int leftIndex = data.leftIndex;
int leftlen = lastMatch.index - leftIndex;
data.leftIndex = lastMatch.index + lastMatch.length;
replace_glob (data, cx, scope, reImpl, leftIndex, leftlen);
}
if (reImpl.lastMatch.length == 0) {
if (indexp [0] == str.Length)
break;
indexp [0]++;
}
}
}
else {
result = re.executeRegExp (cx, scope, reImpl, str, indexp, ((data.mode == EcmaScript.NET.RegExpActions.Replace) ? BuiltinRegExp.TEST : BuiltinRegExp.MATCH));
}
return result;
}
/// <summary> Analog of do_replace in jsstr.c</summary>
private static void do_replace (GlobData rdata, Context cx, RegExpImpl regExpImpl)
{
System.Text.StringBuilder charBuf = rdata.charBuf;
int cp = 0;
string da = rdata.repstr;
int dp = rdata.dollar;
if (dp != -1) {
int [] skip = new int [1];
do {
int len = dp - cp;
charBuf.Append (da.Substring (cp, (dp) - (cp)));
cp = dp;
SubString sub = interpretDollar (cx, regExpImpl, da, dp, skip);
if (sub != null) {
len = sub.length;
if (len > 0) {
charBuf.Append (sub.charArray, sub.index, len);
}
cp += skip [0];
dp += skip [0];
}
else {
++dp;
}
dp = da.IndexOf ((char)'$', dp);
}
while (dp >= 0);
}
int daL = da.Length;
if (daL > cp) {
charBuf.Append (da.Substring (cp, (daL) - (cp)));
}
}
private static SubString interpretDollar (Context cx, RegExpImpl res, string da, int dp, int [] skip)
{
char dc;
int num, tmp;
if (da [dp] != '$')
Context.CodeBug ();
/* Allow a real backslash (literal "\\") to escape "$1" etc. */
Context.Versions version = cx.Version;
if (version != Context.Versions.Default && version <= Context.Versions.JS1_4) {
if (dp > 0 && da [dp - 1] == '\\')
return null;
}
int daL = da.Length;
if (dp + 1 >= daL)
return null;
/* Interpret all Perl match-induced dollar variables. */
dc = da [dp + 1];
if (BuiltinRegExp.isDigit (dc)) {
int cp;
if (version != Context.Versions.Default && version <= Context.Versions.JS1_4) {
if (dc == '0')
return null;
/* Check for overflow to avoid gobbling arbitrary decimal digits. */
num = 0;
cp = dp;
while (++cp < daL && BuiltinRegExp.isDigit (dc = da [cp])) {
tmp = 10 * num + (dc - '0');
if (tmp < num)
break;
num = tmp;
}
}
else {
/* ECMA 3, 1-9 or 01-99 */
int parenCount = (res.parens == null) ? 0 : res.parens.Length;
num = dc - '0';
if (num > parenCount)
return null;
cp = dp + 2;
if ((dp + 2) < daL) {
dc = da [dp + 2];
if (BuiltinRegExp.isDigit (dc)) {
tmp = 10 * num + (dc - '0');
if (tmp <= parenCount) {
cp++;
num = tmp;
}
}
}
if (num == 0)
return null; /* $0 or $00 is not valid */
}
/* Adjust num from 1 $n-origin to 0 array-index-origin. */
num--;
skip [0] = cp - dp;
return res.getParenSubString (num);
}
skip [0] = 2;
switch (dc) {
case '$':
return new SubString ("$");
case '&':
return res.lastMatch;
case '+':
return res.lastParen;
case '`':
if (version == Context.Versions.JS1_2) {
/*
* JS1.2 imitated the Perl4 bug where left context at each step
* in an iterative use of a global regexp started from last match,
* not from the start of the target string. But Perl4 does start
* $` at the beginning of the target string when it is used in a
* substitution, so we emulate that special case here.
*/
res.leftContext.index = 0;
res.leftContext.length = res.lastMatch.index;
}
return res.leftContext;
case '\'':
return res.rightContext;
}
return null;
}
/*
* Analog of replace_glob() in jsstr.c
*/
private static void replace_glob (GlobData rdata, Context cx, IScriptable scope, RegExpImpl reImpl, int leftIndex, int leftlen)
{
int replen;
string lambdaStr;
if (rdata.lambda != null) {
// invoke lambda function with args lastMatch, $1, $2, ... $n,
// leftContext.length, whole string.
SubString [] parens = reImpl.parens;
int parenCount = (parens == null) ? 0 : parens.Length;
object [] args = new object [parenCount + 3];
args [0] = reImpl.lastMatch.ToString ();
for (int i = 0; i < parenCount; i++) {
SubString sub = parens [i];
if (sub != null) {
args [i + 1] = sub.ToString ();
}
else {
args [i + 1] = Undefined.Value;
}
}
args [parenCount + 1] = (int)reImpl.leftContext.length;
args [parenCount + 2] = rdata.str;
// This is a hack to prevent expose of reImpl data to
// JS function which can run new regexps modifing
// regexp that are used later by the engine.
// TODO: redesign is necessary
if (reImpl != cx.RegExpProxy)
Context.CodeBug ();
RegExpImpl re2 = new RegExpImpl ();
re2.multiline = reImpl.multiline;
re2.input = reImpl.input;
cx.RegExpProxy = re2;
try {
IScriptable parent = ScriptableObject.GetTopLevelScope (scope);
object result = rdata.lambda.Call (cx, parent, parent, args);
lambdaStr = ScriptConvert.ToString (result);
}
finally {
cx.RegExpProxy = reImpl;
}
replen = lambdaStr.Length;
}
else {
lambdaStr = null;
replen = rdata.repstr.Length;
if (rdata.dollar >= 0) {
int [] skip = new int [1];
int dp = rdata.dollar;
do {
SubString sub = interpretDollar (cx, reImpl, rdata.repstr, dp, skip);
if (sub != null) {
replen += sub.length - skip [0];
dp += skip [0];
}
else {
++dp;
}
dp = rdata.repstr.IndexOf ((char)'$', dp);
}
while (dp >= 0);
}
}
int growth = leftlen + replen + reImpl.rightContext.length;
System.Text.StringBuilder charBuf = rdata.charBuf;
if (charBuf == null) {
charBuf = new System.Text.StringBuilder (growth);
rdata.charBuf = charBuf;
}
else {
charBuf.EnsureCapacity (rdata.charBuf.Length + growth);
}
charBuf.Append (reImpl.leftContext.charArray, leftIndex, leftlen);
if (rdata.lambda != null) {
charBuf.Append (lambdaStr);
}
else {
do_replace (rdata, cx, reImpl);
}
}
/*
* Analog of match_glob() in jsstr.c
*/
private static void match_glob (GlobData mdata, Context cx, IScriptable scope, int count, RegExpImpl reImpl)
{
if (mdata.arrayobj == null) {
IScriptable s = ScriptableObject.GetTopLevelScope (scope);
mdata.arrayobj = ScriptRuntime.NewObject (cx, s, "Array", null);
}
SubString matchsub = reImpl.lastMatch;
string matchstr = matchsub.ToString ();
mdata.arrayobj.Put (count, mdata.arrayobj, matchstr);
}
/*
* indexp is assumed to be an array of length 1
*/
internal virtual object executeRegExp(Context cx, IScriptable scopeObj, RegExpImpl res, string str, int [] indexp, int matchType)
{
REGlobalData gData = new REGlobalData ();
int start = indexp [0];
char [] charArray = str.ToCharArray ();
int end = charArray.Length;
if (start > end)
start = end;
//
// Call the recursive matcher to do the real work.
//
bool matches = matchRegExp (gData, re, charArray, start, end, res.multiline);
if (!matches) {
if (matchType != PREFIX)
return null;
return Undefined.Value;
}
int index = gData.cp;
int i = index;
indexp [0] = i;
int matchlen = i - (start + gData.skipped);
int ep = index;
index -= matchlen;
object result;
IScriptable obj;
if (matchType == TEST) {
/*
* Testing for a match and updating cx.regExpImpl: don't allocate
* an array object, do return true.
*/
result = true;
obj = null;
}
else {
/*
* The array returned on match has element 0 bound to the matched
* string, elements 1 through re.parenCount bound to the paren
* matches, an index property telling the length of the left context,
* and an input property referring to the input string.
*/
IScriptable scope = GetTopLevelScope (scopeObj);
result = ScriptRuntime.NewObject (cx, scope, "Array", null);
obj = (IScriptable)result;
string matchstr = new string (charArray, index, matchlen);
obj.Put (0, obj, matchstr);
}
if (re.parenCount == 0) {
res.parens = null;
res.lastParen = SubString.EmptySubString;
}
else {
SubString parsub = null;
int num;
res.parens = new SubString [re.parenCount];
for (num = 0; num < re.parenCount; num++) {
int cap_index = gData.parens_index (num);
string parstr;
if (cap_index != -1) {
int cap_length = gData.parens_length (num);
parsub = new SubString (charArray, cap_index, cap_length);
res.parens [num] = parsub;
if (matchType == TEST)
continue;
parstr = parsub.ToString ();
obj.Put (num + 1, obj, parstr);
}
else {
if (matchType != TEST)
obj.Put (num + 1, obj, Undefined.Value);
}
}
res.lastParen = parsub;
}
if (!(matchType == TEST)) {
/*
* Define the index and input properties last for better for/in loop
* order (so they come after the elements).
*/
obj.Put ("index", obj, (object)(start + gData.skipped));
obj.Put ("input", obj, str);
}
if (res.lastMatch == null) {
res.lastMatch = new SubString ();
res.leftContext = new SubString ();
res.rightContext = new SubString ();
}
res.lastMatch.charArray = charArray;
res.lastMatch.index = index;
res.lastMatch.length = matchlen;
res.leftContext.charArray = charArray;
if (cx.Version == Context.Versions.JS1_2) {
/*
* JS1.2 emulated Perl4.0.1.8 (patch level 36) for global regexps used
* in scalar contexts, and unintentionally for the string.match "list"
* psuedo-context. On "hi there bye", the following would result:
*
* Language while(/ /g){print("$`");} s/ /$`/g
* perl4.036 "hi", "there" "hihitherehi therebye"
* perl5 "hi", "hi there" "hihitherehi therebye"
* js1.2 "hi", "there" "hihitheretherebye"
*
* Insofar as JS1.2 always defined $` as "left context from the last
* match" for global regexps, it was more consistent than perl4.
*/
res.leftContext.index = start;
res.leftContext.length = gData.skipped;
}
else {
/*
* For JS1.3 and ECMAv2, emulate Perl5 exactly:
*
* js1.3 "hi", "hi there" "hihitherehi therebye"
*/
res.leftContext.index = 0;
res.leftContext.length = start + gData.skipped;
}
res.rightContext.charArray = charArray;
res.rightContext.index = ep;
res.rightContext.length = end - ep;
return result;
}
private static SubString interpretDollar(Context cx, RegExpImpl res, string da, int dp, int [] skip)
{
char dc;
int num, tmp;
if (da [dp] != '$')
{
Context.CodeBug();
}
/* Allow a real backslash (literal "\\") to escape "$1" etc. */
Context.Versions version = cx.Version;
if (version != Context.Versions.Default && version <= Context.Versions.JS1_4)
{
if (dp > 0 && da [dp - 1] == '\\')
{
return(null);
}
}
int daL = da.Length;
if (dp + 1 >= daL)
{
return(null);
}
/* Interpret all Perl match-induced dollar variables. */
dc = da [dp + 1];
if (BuiltinRegExp.isDigit(dc))
{
int cp;
if (version != Context.Versions.Default && version <= Context.Versions.JS1_4)
{
if (dc == '0')
{
return(null);
}
/* Check for overflow to avoid gobbling arbitrary decimal digits. */
num = 0;
cp = dp;
while (++cp < daL && BuiltinRegExp.isDigit(dc = da [cp]))
{
tmp = 10 * num + (dc - '0');
if (tmp < num)
{
break;
}
num = tmp;
}
}
else
{
/* ECMA 3, 1-9 or 01-99 */
int parenCount = (res.parens == null) ? 0 : res.parens.Length;
num = dc - '0';
if (num > parenCount)
{
return(null);
}
cp = dp + 2;
if ((dp + 2) < daL)
{
dc = da [dp + 2];
if (BuiltinRegExp.isDigit(dc))
{
tmp = 10 * num + (dc - '0');
if (tmp <= parenCount)
{
cp++;
num = tmp;
}
}
}
if (num == 0)
{
return(null); /* $0 or $00 is not valid */
}
}
/* Adjust num from 1 $n-origin to 0 array-index-origin. */
num--;
skip [0] = cp - dp;
return(res.getParenSubString(num));
}
skip [0] = 2;
switch (dc)
{
case '$':
return(new SubString("$"));
case '&':
return(res.lastMatch);
case '+':
return(res.lastParen);
case '`':
if (version == Context.Versions.JS1_2)
{
/*
* JS1.2 imitated the Perl4 bug where left context at each step
* in an iterative use of a global regexp started from last match,
* not from the start of the target string. But Perl4 does start
* $` at the beginning of the target string when it is used in a
* substitution, so we emulate that special case here.
*/
res.leftContext.index = 0;
res.leftContext.length = res.lastMatch.index;
}
return(res.leftContext);
case '\'':
return(res.rightContext);
}
return(null);
}
/*
* Analog of replace_glob() in jsstr.c
*/
private static void replace_glob(GlobData rdata, Context cx, IScriptable scope, RegExpImpl reImpl, int leftIndex, int leftlen)
{
int replen;
string lambdaStr;
if (rdata.lambda != null)
{
// invoke lambda function with args lastMatch, $1, $2, ... $n,
// leftContext.length, whole string.
SubString [] parens = reImpl.parens;
int parenCount = (parens == null) ? 0 : parens.Length;
object [] args = new object [parenCount + 3];
args [0] = reImpl.lastMatch.ToString();
for (int i = 0; i < parenCount; i++)
{
SubString sub = parens [i];
if (sub != null)
{
args [i + 1] = sub.ToString();
}
else
{
args [i + 1] = Undefined.Value;
}
}
args [parenCount + 1] = (int)reImpl.leftContext.length;
args [parenCount + 2] = rdata.str;
// This is a hack to prevent expose of reImpl data to
// JS function which can run new regexps modifing
// regexp that are used later by the engine.
// TODO: redesign is necessary
if (reImpl != cx.RegExpProxy)
{
Context.CodeBug();
}
RegExpImpl re2 = new RegExpImpl();
re2.multiline = reImpl.multiline;
re2.input = reImpl.input;
cx.RegExpProxy = re2;
try {
IScriptable parent = ScriptableObject.GetTopLevelScope(scope);
object result = rdata.lambda.Call(cx, parent, parent, args);
lambdaStr = ScriptConvert.ToString(result);
}
finally {
cx.RegExpProxy = reImpl;
}
replen = lambdaStr.Length;
}
else
{
lambdaStr = null;
replen = rdata.repstr.Length;
if (rdata.dollar >= 0)
{
int [] skip = new int [1];
int dp = rdata.dollar;
do
{
SubString sub = interpretDollar(cx, reImpl, rdata.repstr, dp, skip);
if (sub != null)
{
replen += sub.length - skip [0];
dp += skip [0];
}
else
{
++dp;
}
dp = rdata.repstr.IndexOf((char)'$', dp);
}while (dp >= 0);
}
}
int growth = leftlen + replen + reImpl.rightContext.length;
System.Text.StringBuilder charBuf = rdata.charBuf;
if (charBuf == null)
{
charBuf = new System.Text.StringBuilder(growth);
rdata.charBuf = charBuf;
}
else
{
charBuf.EnsureCapacity(rdata.charBuf.Length + growth);
}
charBuf.Append(reImpl.leftContext.charArray, leftIndex, leftlen);
if (rdata.lambda != null)
{
charBuf.Append(lambdaStr);
}
else
{
do_replace(rdata, cx, reImpl);
}
}
/*
* Analog of match_glob() in jsstr.c
*/
private static void match_glob(GlobData mdata, Context cx, IScriptable scope, int count, RegExpImpl reImpl)
{
if (mdata.arrayobj == null)
{
IScriptable s = ScriptableObject.GetTopLevelScope(scope);
mdata.arrayobj = ScriptRuntime.NewObject(cx, s, "Array", null);
}
SubString matchsub = reImpl.lastMatch;
string matchstr = matchsub.ToString();
mdata.arrayobj.Put(count, mdata.arrayobj, matchstr);
}
/// <summary> Analog of C match_or_replace.</summary>
private static object matchOrReplace(Context cx, IScriptable scope, IScriptable thisObj, object [] args, RegExpImpl reImpl, GlobData data, bool forceFlat)
{
BuiltinRegExp re;
string str = ScriptConvert.ToString(thisObj);
data.str = str;
IScriptable topScope = ScriptableObject.GetTopLevelScope(scope);
if (args.Length == 0)
{
object compiled = BuiltinRegExp.compileRE("", "", false);
re = new BuiltinRegExp(topScope, compiled);
}
else if (args [0] is BuiltinRegExp)
{
re = (BuiltinRegExp)args [0];
}
else
{
string src = ScriptConvert.ToString(args [0]);
string opt;
if (data.optarg < args.Length)
{
args [0] = src;
opt = ScriptConvert.ToString(args [data.optarg]);
}
else
{
opt = null;
}
object compiled = BuiltinRegExp.compileRE(src, opt, forceFlat);
re = new BuiltinRegExp(topScope, compiled);
}
data.regexp = re;
data.global = (re.Flags & BuiltinRegExp.JSREG_GLOB) != 0;
int [] indexp = new int [] { 0 };
object result = null;
if (data.mode == EcmaScript.NET.RegExpActions.Search)
{
result = re.executeRegExp(cx, scope, reImpl, str, indexp, BuiltinRegExp.TEST);
if (result != null && result.Equals(true))
{
result = (int)reImpl.leftContext.length;
}
else
{
result = -1;
}
}
else if (data.global)
{
re.lastIndex = 0;
for (int count = 0; indexp [0] <= str.Length; count++)
{
result = re.executeRegExp(cx, scope, reImpl, str, indexp, BuiltinRegExp.TEST);
if (result == null || !result.Equals(true))
{
break;
}
if (data.mode == EcmaScript.NET.RegExpActions.Match)
{
match_glob(data, cx, scope, count, reImpl);
}
else
{
if (data.mode != EcmaScript.NET.RegExpActions.Replace)
{
Context.CodeBug();
}
SubString lastMatch = reImpl.lastMatch;
int leftIndex = data.leftIndex;
int leftlen = lastMatch.index - leftIndex;
data.leftIndex = lastMatch.index + lastMatch.length;
replace_glob(data, cx, scope, reImpl, leftIndex, leftlen);
}
if (reImpl.lastMatch.length == 0)
{
if (indexp [0] == str.Length)
{
break;
}
indexp [0]++;
}
}
}
else
{
result = re.executeRegExp(cx, scope, reImpl, str, indexp, ((data.mode == EcmaScript.NET.RegExpActions.Replace) ? BuiltinRegExp.TEST : BuiltinRegExp.MATCH));
}
return(result);
}
