Difference between revisions of "Combined Library"

From Second Life Wiki
Jump to navigation Jump to search
m (→‎Replace: Link to the native LSL function introduced in simulator release 2023-01-27.577942)
 
(36 intermediate revisions by 9 users not shown)
Line 5: Line 5:
The library is still being worked on so only some of the compiled functions will be posted at this time. All functions in the library are hand optimized. To understand the logic, it is best to pull the code apart but be wary of LSL's strict order of operations.
The library is still being worked on so only some of the compiled functions will be posted at this time. All functions in the library are hand optimized. To understand the logic, it is best to pull the code apart but be wary of LSL's strict order of operations.


You can download the latest release from here: [http://mailerdaemon.home.comcast.net/CombinedLibrary.zip CombinedLibrary.zip]
You can download the latest release from here: [http://www.megaupload.com/?d=LV0334YT CombinedLibrary.zip]
 
'''Note: This project has not been substantially updated since 2008.'''


<div id="box">
<div id="box">
Line 11: Line 13:
<div style="padding: 0.5em;" id="str_replace">
<div style="padding: 0.5em;" id="str_replace">
=== Replace ===
=== Replace ===
As of simulator release 2023-01-27.577942, there is the native LSL function [[llReplaceSubString]].
==== Older, user-defined functions ====
The originally posted function was largely superseded by this version posted below by [[User:Haravikk Mistral|Haravikk Mistral]], it is much faster (under both [[LSL]] and [[Mono]]) while having identical behaviour, and in the majority of usages is more memory efficient:
The originally posted function was largely superseded by this version posted below by [[User:Haravikk Mistral|Haravikk Mistral]], it is much faster (under both [[LSL]] and [[Mono]]) while having identical behaviour, and in the majority of usages is more memory efficient:


<lsl>string strReplace(string str, string search, string replace) {
<source lang="lsl2">string strReplace(string str, string search, string replace) {
     return llDumpList2String(llParseStringKeepNulls((str = "") + str, [search], []), replace);
     return llDumpList2String(llParseStringKeepNulls((str = "") + str, [search], []), replace);
}</lsl>
}</source>


Here is the original version of the function, while slower it is more memory efficient in LSO when the input string requires many parts to be replaced. For example when replacing ever space in a paragraph with an underscore. In LSO the cost of dividing the string into a list balloons the size of the string considerably.
Here is the original version of the function, while slower it is more memory efficient in LSO when the input string requires a large number of parts to be replaced. For example when replacing every space in a paragraph with an underscore. In LSO the cost of dividing the string into a list balloons the size of the string considerably.


<lsl>string str_replace(string src, string from, string to)
<source lang="lsl2">string str_replace(string src, string from, string to)
{//replaces all occurrences of 'from' with 'to' in 'src'.
{//replaces all occurrences of 'from' with 'to' in 'src'.
     integer len = (~-(llStringLength(from)));
     integer len = (~-(llStringLength(from)));
Line 25: Line 30:
     {
     {
         string  buffer = src;
         string  buffer = src;
         integer b_pos = -1;
         integer b_pos = ERR_GENERIC;
         integer to_len = (~-(llStringLength(to)));
         integer to_len = (~-(llStringLength(to)));
         @loop;//instead of a while loop, saves 5 bytes (and run faster).
         @loop;//instead of a while loop, saves 5 bytes (and runs faster).
         integer to_pos = ~llSubStringIndex(buffer, from);
         integer to_pos = ~llSubStringIndex(buffer, from);
         if(to_pos)
         if(to_pos)
Line 40: Line 45:
     }
     }
     return src;
     return src;
}</lsl>
}</source>


=== Trim ===
=== Trim ===
<lsl>string TrimRight(string src, string chrs)//Mono Unsafe, LSO Safe
<source lang="lsl2">string TrimRight(string src, string chrs)//LSLEditor Unsafe, LSL Safe
{
{
     integer i = llStringLength(src);
     integer i = llStringLength(src);
Line 50: Line 55:
}
}


string TrimLeft(string src, string chrs)//Mono Unsafe, LSO Safe
string TrimLeft(string src, string chrs)//LSLEditor Unsafe, LSL Safe
{
{
     integer i = ~llStringLength(src);
     integer i = ~llStringLength(src);
Line 57: Line 62:
}
}


string TrimBoth(string src, string chrs)//Mono Unsafe, LSO Safe
string TrimBoth(string src, string chrs)//LSLEditor Unsafe, LSL Safe
{
{
     integer i = ~llStringLength(src);
     integer i = ~llStringLength(src);
Line 64: Line 69:
     do ; while(~llSubStringIndex(chrs, llGetSubString(src, i = ~-i, i)) && i);
     do ; while(~llSubStringIndex(chrs, llGetSubString(src, i = ~-i, i)) && i);
     return llDeleteSubString(src, -~(i), 0x7FFFFFF0);
     return llDeleteSubString(src, -~(i), 0x7FFFFFF0);
}</lsl>
}</source>
 
=== Unescape ===
=== Unescape ===
This is not the version found in Combined Library, it does not support the \h, \u or \U escape codes which are supported by the library version. Both support these escape codes: \\, \", \n, \t, \s
This is not the version found in Combined Library, it does not support the \h, \u or \U escape codes which are supported by the library version. Both support these escape codes: \\, \", \n, \t, \s


Like [[User:Haravikk Mistral|Haravikk's]] [[#Replace|Replace]] function, this has the potential to use a huge amount of memory, consider using the library version instead if memory usage becomes a problem.
Like [[User:Haravikk Mistral|Haravikk's]] [[#Replace|Replace]] function, this has the potential to use a huge amount of memory, consider using the library version instead if memory usage becomes a problem. If you do not need the extra features and do not have memory limitations this is a great alternative.


<lsl>string Unescape(string in)
<source lang="lsl2">string Unescape(string in)
{
{
     list search = ["\\n","\\\"","\\t","\\s"];
     list search = ["\\n","\\\"","\\t","\\s"];
     list replace = ["\n","\"","\t"," "];
     list replace = ["\n","\"","\t"," "];
     list split = llParseStringKeepNulls((in = "") + in, ["\\\\"], []);
     list split = llParseStringKeepNulls((in = "") + in, ["\\\\"], []);//"
     integer pos = ([] != split);
     integer pos = ([] != split);
     do
     do
Line 83: Line 89:
             item = llDumpList2String(llParseStringKeepNulls((item = "") + item, llList2List(search, qos, qos), []), llList2String(replace, qos));
             item = llDumpList2String(llParseStringKeepNulls((item = "") + item, llList2List(search, qos, qos), []), llList2String(replace, qos));
         while(qos = -~qos);
         while(qos = -~qos);
         split = llListReplaceList((split = []) + split, [item], pos, pos);
         split = llListReplaceList((split = []) + split, [(item = "") + item], pos, pos);
     }
     }
     while(pos = -~pos);
     while(pos = -~pos);
     return llDumpList2String(split, "\\");
     return llDumpList2String(split, "\\");
}</lsl>
}</source>
</div></div>
</div></div>


Line 98: Line 104:


=== UTF8 to Unicode Integer ===
=== UTF8 to Unicode Integer ===
Converts a character into it an integer.
Converts a character into an integer.
<lsl>
<source lang="lsl2">
integer UTF8ToUnicodeInteger(string input)//Mono Safe, LSO Safe
integer UTF8ToUnicodeInteger(string input)//Mono Safe, LSO Safe
{
{
Line 116: Line 122:
     return result >> 24;
     return result >> 24;
}
}
</lsl>
</source>


=== Unicode Integer to UTF8 ===
=== Unicode Integer to UTF8 ===
Convert an integer into a character.
Converts an integer into a character.
<lsl>
<source lang="lsl2">string UnicodeIntegerToUTF8(integer input)//Mono Safe, LSLEditor Safe, LSO Incomplete
string UnicodeIntegerToUTF8(integer input)//Mono Safe, LSO Safe
{//LSO allows for the older UTF-8 range, this function only supports the new UTF-16 range.
{
    if(input > 0)
     integer bytes = llCeil(llLog(input) / 0.69314718055994530941723212145818);
     {
    bytes = (input >= 0x80) * (bytes + ~(((1 << bytes) - input) > 0)) / 5;//adjust
        if(input <= 0x7FF)
    string result = "%" + byte2hex((input >> (6 * bytes)) | ((0x3F80 >> bytes) << !bytes));
        {
     while (bytes)
            if(input <= 0x7F)
         result += "%" + byte2hex((((input >> (6 * (bytes = ~-bytes))) | 0x80) & 0xBF));
                return llBase64ToString(llIntegerToBase64(input << 24));
     return llUnescapeURL(result);
            return llBase64ToString(llIntegerToBase64(0xC0800000 | ((input << 18) & 0x1F000000) | ((input << 16) & 0x3F0000)));here.
        }
        if(input <= 0xFFFF)
            return llBase64ToString(llIntegerToBase64(0xE0808000 | ((input << 12) & 0x0F000000) | ((input << 10) & 0x3F0000) | ((input << 8) & 0x3F00)));
        if(input <= 0x10FFFF)
            return llBase64ToString(llIntegerToBase64(0xF0808080 | ((input << 06) & 0x07000000) | ((input << 04) & 0x3F0000) | ((input << 2) & 0x3F00) | (input & 0x3F)));
    }
    return "";
}
</source>
 
=== Byte Length of UTF-8 Encoded String ===
 
Returns the number of bytes that would be used if the string were encoded in UTF-8.
 
<source lang="lsl2">integer StringUTF8Size(string str) {
    return (3 * llSubStringIndex(llStringToBase64(str)+"=", str = "=")) >> 2;
}</source>
 
=== Truncate To Length ===
 
Keep the first n-bytes, discard the rest. If a character is truncated, the damaged portion is discarded.
 
<source lang="lsl2">
string TruncateToLength(string str, integer bytes) {
    string temp = llStringToBase64(str);
     if(llStringLength(temp) <= (bytes * 8) / 6)
         return str;
    temp = llBase64ToString(llGetSubString(temp+"==", (bytes * -2) % 3, (bytes * 4 - 1) / 3));
    integer i = llStringLength(temp) - 1;
    if(llGetSubString(temp, i, i) != llGetSubString(str, i, i))
        return llDeleteSubString(temp, i, i);//last char was multi-byte and was truncated.
     return temp;
}
}
</source>


string byte2hex(integer x)//Mono Safe, LSO Safe
=== Split To Length ===
{//Helper function for use with unicode characters.
    integer y = (x >> 4) & 0xF;
    return llGetSubString(hexc, y, y) + llGetSubString(hexc, x & 0xF, x & 0xF);
}//This function would benifit greatly from the DUP opcode, it would remove 19 bytes.


string hexc="0123456789ABCDEF";
Splits a string so at most only a specified number of bytes appear in each substring. If an entire character cannot fit in the current stride, it is pushed into the next.
</lsl>
 
<source lang="lsl2">
list SplitToLength(string in, integer bytes) {
    list out = [];
    while(in) {
        string str = llDeleteSubString(in, bytes, 0x7ffffff0);
        string temp = llStringToBase64(str);
        if(llStringLength(temp) <= (bytes * 8) / 6) {
            out += str;
            in = llDeleteSubString(in, 0, bytes);
        } else {
            temp = llBase64ToString(llGetSubString(temp + "==", (bytes * -2) % 3, (bytes * 4 - 1) / 3));
            integer i = llStringLength(temp) - 1;
            if(llGetSubString(temp, i, i) != llGetSubString(str, i, i)) {
                if(!i)
                    return out + ["", in, -2];
                out += llDeleteSubString(temp, i, i);//last char was multi-byte and was truncated.
                in = llGetSubString(in, i, 0x7ffffff0);
            } else {
                out += temp;
                in = llDeleteSubString(in, 0, i);
            }
        }
    }
    return out;
}
</source>
</div></div>
</div></div>
<div id="box">
<div id="box">


Line 149: Line 209:


The way it works is it keeps an Unsearched Buffer (UB) which is a subset of the Input Buffer (IB) and it records the Position of the UB in the IB (P). Each iteration it searches the UB and adds that resulting index to the P, then it uses P as the index to replace that section in the IB, finally it recalculates the new P, then is uses IB with the new P to update UB.
The way it works is it keeps an Unsearched Buffer (UB) which is a subset of the Input Buffer (IB) and it records the Position of the UB in the IB (P). Each iteration it searches the UB and adds that resulting index to the P, then it uses P as the index to replace that section in the IB, finally it recalculates the new P, then is uses IB with the new P to update UB.
<lsl>
<source lang="lsl2">
list ListReplace(list src, list from, list to)
list ListReplace(list src, list from, list to)
{//replaces all occurrences of 'from' with 'to' in 'src'.
{//replaces all occurrences of 'from' with 'to' in 'src'.
Line 156: Line 216:
     {
     {
         list  buffer = src;
         list  buffer = src;
         integer b_pos = -1;
         integer b_pos = ERR_GENERIC;
         integer to_len = ~([] != to);
         integer to_len = ~([] != to);
         @loop;//instead of a while loop, saves 5 bytes (and run faster).
         @loop;//instead of a while loop, saves 5 bytes (and runs faster).
         integer to_pos = ~llListFindList(buffer, from);
         integer to_pos = ~llListFindList(buffer, from);
         if(to_pos)
         if(to_pos)
Line 172: Line 232:
     return src;
     return src;
}
}
</lsl>
</source>


=== Compare ===
=== Compare ===
Line 178: Line 238:


'''Note:''' It will ignore the sign on zero (0.0 == -0.0). If the sign on zero is important to you, follow the instructions in the comments on how to enable checking for this. Currently no functions in LSL treat negative zero differently then positive zero.
'''Note:''' It will ignore the sign on zero (0.0 == -0.0). If the sign on zero is important to you, follow the instructions in the comments on how to enable checking for this. Currently no functions in LSL treat negative zero differently then positive zero.
<lsl>
<source lang="lsl2">
integer ListCompare(list input_a, list input_b)
integer ListCompare(list input_a, list input_b)
{
{
Line 212: Line 272:
     return !b;
     return !b;
}
}
</lsl>
</source>


If you do not care about positive/negative zero, then a better way to perform the comparison is:
If you do not care about positive/negative zero, then a better way to perform the comparison is:
<lsl>
<source lang="lsl2">
integer ListCompare(list a, list b) {
integer ListCompare(list a, list b) {
     integer aL = a != [];
     integer aL = a != [];
Line 223: Line 283:
     return llListFindList((a = []) + a, (b = []) + b) == 0;
     return llListFindList((a = []) + a, (b = []) + b) == 0;
}
}
</lsl>
</source>
If you just want to compare the length of the lists use <code>(a == b)</code>:
<source lang="lsl2">
integer ListCompare(list a, list b){
    return (a==b);
}
</source>
 
 
</div></div>
</div></div>
<div id="box">
<div id="box">
Line 231: Line 299:
===Parameter & Return Based===
===Parameter & Return Based===
Binary functions that use parameters and returns
Binary functions that use parameters and returns
<lsl>
<source lang="lsl2">
integer ReadBase64Integer(string data, integer index)
integer ReadBase64Integer(string data, integer index)
{
{
Line 373: Line 441:
     list out;
     list out;
     integer int;
     integer int;
     while(PPi(i) <= len)
     while((i = -~i) <= len)
         if((int = ReadBase64Integer(a, i)) || (i ^ len))
         if((int = ReadBase64Integer(a, i)) || (i ^ len))
             out += int;
             out += int;
Line 390: Line 458:


string hexc="0123456789ABCDEF";
string hexc="0123456789ABCDEF";
</lsl>
</source>


===Global Buffers===
===Global Buffers===
These versions of the functions depend upon global buffers. Not all the functions have been ported to this model (because it is very complicated to build functions with macros so they can be flipped from one mode to the other).
These versions of the functions depend upon global buffers. Not all the functions have been ported to this model (because it is very complicated to build functions with macros so they can be flipped from one mode to the other).
<lsl>
<source lang="lsl2">
integer v0;
integer v0;
integer v1;
integer v1;
Line 469: Line 537:
     integer g;
     integer g;
     do{
     do{
         v0 = ((integer)("0x"+(llGetSubString(a, g, g + 7)))); v1 = ((integer)("0x"+(llGetSubString(a, g + 8, g + 15)))); WriteBase64IntegerPair(g >> 3);
         v0 = ((integer)("0x"+(llGetSubString(a, g, g + 7))));
        v1 = ((integer)("0x"+(llGetSubString(a, g + 8, g + 15))));
        WriteBase64IntegerPair(g >> 3);
     }while((g += 16) < e);
     }while((g += 16) < e);


Line 504: Line 574:
     do
     do
     {
     {
         ReadBase64Integer(i); if(v0 || (i ^ len))
         ReadBase64Integer(i);
        if(v0 || (i ^ len))
             out += v0;
             out += v0;
     }while((i = -~i) <= len);
     }while((i = -~i) <= len);
Line 528: Line 599:


string hexc="0123456789ABCDEF";
string hexc="0123456789ABCDEF";
</lsl>
</source>
</div></div>
</div></div>


Line 535: Line 606:
== Float Union Integer ==
== Float Union Integer ==
<div style="padding: 0.5em;">
<div style="padding: 0.5em;">
The Float Union Integer functions allow for safely storing floats in integers and provide for retreival. All real numbers are supported.
The Float Union Integer functions allow for safely storing floats in integers and provide for retrieval. All real numbers are supported.


{{JIRA|SVC-1377}} breaks these functions for Mono but once it is fixed (it has been imported) then they should run.
For versions that support Infinity and NaN see [[User:Strife_Onizuka/Float_Functions]] (the two versions have not been reconciled yet).


===General===
===General===
These versions will work well in LSO and Mono.  
These versions will work well in LSO and Mono.  


<lsl>integer fui(float input)//Mono Safe, LSO Safe, Doubles Unsupported, LSLEditor Unsafe
<source lang="lsl2">integer fui(float input)//Mono Safe, LSO Safe, Doubles Unsupported, LSLEditor Unsafe
{//union float to integer
{//union float to integer
     if((input) != 0.0){//is it non zero?
     if((input) != 0.0){//is it non zero?
         integer sign = (input < 0) << 31;//the sign, but later this variable is reused to store the shift
         integer sign = (input < 0) << 31;//the sign, but later this variable is reused to store the shift
         if((input = llFabs(input)) < 2.3509887016445750159374730744445e-38)//Denormalized range check & last stirde of normalized range
         if((input = llFabs(input)) < 2.3509887016445750159374730744445e-38)//Denormalized range check & last stride of normalized range
             return sign | (integer)(input / 1.4012984643248170709237295832899e-45);//the math overlaps; saves cpu time.
             return sign | (integer)(input / 1.4012984643248170709237295832899e-45);//the math overlaps; saves cpu time.
         integer exp = llFloor((llLog(input) / 0.69314718055994530941723212145818));//extremes will error towards extremes. following yuch corrects it.
         integer exp = llFloor((llLog(input) / 0.69314718055994530941723212145818));//extremes will error towards extremes. following yuch corrects it.
Line 557: Line 628:
{//union integer to float
{//union integer to float
     return llPow(2.0, (input | !input) - 150) * (((!!(input = (0xff & (input >> 23)))) << 23) | ((input & 0x7fffff))) * (1 | (input >> 31));
     return llPow(2.0, (input | !input) - 150) * (((!!(input = (0xff & (input >> 23)))) << 23) | ((input & 0x7fffff))) * (1 | (input >> 31));
}//will crash if the raw exponent == 0xff; reason for crash deviates from float standard; though a crash is warented.</lsl>
}//will crash if the raw exponent == 0xff; reason for crash deviates from float standard; though a crash is warented.</source>


===LSLEditor Safe===
===LSLEditor Safe===
These versions will run in LSLEditor (as well as LSO and Mono). Doubles are not supported and anything outside the float range will become signed infinity or zero respectively.
These versions will run in LSLEditor (as well as LSO and Mono). Doubles are not supported and anything outside the float range will become signed infinity or zero respectively.


<lsl>integer fui(float input)//Mono Safe, LSO Safe, Doubles Unsupported, LSLEditor Safe
<source lang="lsl2">integer fui(float input)//Mono Safe, LSO Safe, Doubles Unsupported, LSLEditor Safe
{//union float to integer
{//union float to integer
     if((input) != 0.0){//is it non zero?
     if((input) != 0.0){//is it non zero?
         integer sign = (integer)(input < 0) << 31;//the sign, but later this variable is reused to store the shift
         integer sign = (integer)(input < 0) << 31;//the sign, but later this variable is reused to store the shift
         if((input = llFabs(input)) < 2.3509887016445750159374730744445e-38)//Denormalized range check & last stirde of normalized range
         if((input = llFabs(input)) < 2.3509887016445750159374730744445e-38)//Denormalized range check & last stride of normalized range
             return sign | (integer)(input / 1.4012984643248170709237295832899e-45);//the math overlaps; saves cpu time.
             return sign | (integer)(input / 1.4012984643248170709237295832899e-45);//the math overlaps; saves cpu time.
         if(input > 3.4028234663852885981170418348452e+38)//infinity generation for doubles x_x
         if(input > 3.4028234663852885981170418348452e+38)//infinity generation for doubles x_x
Line 584: Line 655:
     integer exp = 0xff & (input >> 23);
     integer exp = 0xff & (input >> 23);
     return llPow(2.0, (exp | !exp) - 150) * (((!!exp) << 23) | ((input & 0x7fffff))) * (1 | (input >> 31));
     return llPow(2.0, (exp | !exp) - 150) * (((!!exp) << 23) | ((input & 0x7fffff))) * (1 | (input >> 31));
}//will crash if the raw exponent == 0xff; reason for crash deviates from float standard; though a crash is warented.</lsl>
}//will crash if the raw exponent == 0xff; reason for crash deviates from float standard; though a crash is warented.</source>


</div>
</div>
</div>
</div>
== Change Log ==
==== 1.0 ====
*New Macros for hiding implementation details of optimizations.
**These macros can also be disabled so the resulting LSL is readable.
*New Macros for VM specific Order-of-Execution compiling.
**More pointless parentheses are now generated. Too lazy at this time to create extra macros keep this from happening.
**Only operators that are commutative are supported at this time.
***Be warned: It won't stop you if you use a non-commutative operator and your code will be wrong.
**This means that you can write optimized code and not have to worry about which VM is executing it.
*Many functions have been updated to use the new macros.
==== Unversioned ====
* Changes before 1.0 were never recorded.
{{LSLC|Library|Combined Library}}
{{LSLC|Library|Combined Library}}
[[Category:LSL_User-Defined_Functions]]

Latest revision as of 00:42, 6 February 2023

The Combined Library is comprised of about 55 functions all of which are released under CC-by v3.0 license.

The library is still being worked on so only some of the compiled functions will be posted at this time. All functions in the library are hand optimized. To understand the logic, it is best to pull the code apart but be wary of LSL's strict order of operations.

You can download the latest release from here: CombinedLibrary.zip

Note: This project has not been substantially updated since 2008.

String Functions

Replace

As of simulator release 2023-01-27.577942, there is the native LSL function llReplaceSubString.

Older, user-defined functions

The originally posted function was largely superseded by this version posted below by Haravikk Mistral, it is much faster (under both LSL and Mono) while having identical behaviour, and in the majority of usages is more memory efficient: 

string strReplace(string str, string search, string replace) {
    return llDumpList2String(llParseStringKeepNulls((str = "") + str, [search], []), replace);
}

Here is the original version of the function, while slower it is more memory efficient in LSO when the input string requires a large number of parts to be replaced. For example when replacing every space in a paragraph with an underscore. In LSO the cost of dividing the string into a list balloons the size of the string considerably. 

string str_replace(string src, string from, string to)
{//replaces all occurrences of 'from' with 'to' in 'src'.
    integer len = (~-(llStringLength(from)));
    if(~len)
    {
        string  buffer = src;
        integer b_pos = ERR_GENERIC;
        integer to_len = (~-(llStringLength(to)));
        @loop;//instead of a while loop, saves 5 bytes (and runs faster).
        integer to_pos = ~llSubStringIndex(buffer, from);
        if(to_pos)
        {
//            b_pos -= to_pos;
//            src = llInsertString(llDeleteSubString(src, b_pos, b_pos + len), b_pos, to);
//            b_pos += to_len;
//            buffer = llGetSubString(src, (-~(b_pos)), 0x8000);
            buffer = llGetSubString(src = llInsertString(llDeleteSubString(src, b_pos -= to_pos, b_pos + len), b_pos, to), (-~(b_pos += to_len)), 0x8000);
            jump loop;
        }
    }
    return src;
}

Trim

string TrimRight(string src, string chrs)//LSLEditor Unsafe, LSL Safe
{
    integer i = llStringLength(src);
    do ; while(~llSubStringIndex(chrs, llGetSubString(src, i = ~-i, i)) && i);
    return llDeleteSubString(src, -~(i), 0x7FFFFFF0);
}

string TrimLeft(string src, string chrs)//LSLEditor Unsafe, LSL Safe
{
    integer i = ~llStringLength(src);
    do ; while(i && ~llSubStringIndex(chrs, llGetSubString(src, i = -~i, i)));
    return llDeleteSubString(src, 0x8000000F, ~-(i));
}

string TrimBoth(string src, string chrs)//LSLEditor Unsafe, LSL Safe
{
    integer i = ~llStringLength(src);
    do ; while(i && ~llSubStringIndex(chrs, llGetSubString(src, i = -~i, i)));
    i = llStringLength(src = llDeleteSubString(src, 0x8000000F, ~-(i)));
    do ; while(~llSubStringIndex(chrs, llGetSubString(src, i = ~-i, i)) && i);
    return llDeleteSubString(src, -~(i), 0x7FFFFFF0);
}

Unescape

This is not the version found in Combined Library, it does not support the \h, \u or \U escape codes which are supported by the library version. Both support these escape codes: \\, \", \n, \t, \s

Like Haravikk's Replace function, this has the potential to use a huge amount of memory, consider using the library version instead if memory usage becomes a problem. If you do not need the extra features and do not have memory limitations this is a great alternative. 

string Unescape(string in)
{
    list search = ["\\n","\\\"","\\t","\\s"];
    list replace = ["\n","\"","\t"," "];
    list split = llParseStringKeepNulls((in = "") + in, ["\\\\"], []);//"
    integer pos = ([] != split);
    do
    {
        integer qos = -4;
        string item = llList2String(split, pos);
        do
            item = llDumpList2String(llParseStringKeepNulls((item = "") + item, llList2List(search, qos, qos), []), llList2String(replace, qos));
        while(qos = -~qos);
        split = llListReplaceList((split = []) + split, [(item = "") + item], pos, pos);
    }
    while(pos = -~pos);
    return llDumpList2String(split, "\\");
}

Unicode functions

LSL uses UTF-8 to as the base format for strings. UTF-8 is an encoding system for storing Unicode characters, each Unicode character has a number. These functions allow for the conversion between the string form and the integer form. They may not be pretty but they are tight and get the job done quickly.

UTF8 to Unicode Integer

Converts a character into an integer. 

integer UTF8ToUnicodeInteger(string input)//Mono Safe, LSO Safe
{
    integer result = llBase64ToInteger(llStringToBase64(input = llGetSubString(input,0,0)));
    if(result & 0x80000000){//multibyte, continuing to use base64 is impractical because it requires smart shifting.
        integer end = (integer)("0x"+llGetSubString(input = (string)llParseString2List(llEscapeURL(input),(list)"%",[]),-8,-1));
        integer begin = (integer)("0x"+llDeleteSubString(input,-8,-1));
        return  (   (  0x0000003f &  end       ) |
                    (( 0x00003f00 &  end) >> 2 ) | 
                    (( 0x003f0000 &  end) >> 4 ) | 
                    (( 0x3f000000 &  end) >> 6 ) |
                    (( 0x0000003f &  begin) << 24) |
                    (( 0x00000100 &  begin) << 22) ) & 
                    (0x7FFFFFFF >> (5 * ((integer)(llLog(~result) / 0.69314718055994530941723212145818) - 25)));
    }
    return result >> 24;
}

Unicode Integer to UTF8

Converts an integer into a character. 

string UnicodeIntegerToUTF8(integer input)//Mono Safe, LSLEditor Safe, LSO Incomplete
{//LSO allows for the older UTF-8 range, this function only supports the new UTF-16 range.
    if(input > 0)
    {
        if(input <= 0x7FF)
        {
            if(input <= 0x7F)
                return llBase64ToString(llIntegerToBase64(input << 24));
            return llBase64ToString(llIntegerToBase64(0xC0800000 | ((input << 18) & 0x1F000000) | ((input << 16) & 0x3F0000)));here.
        }
        if(input <= 0xFFFF)
            return llBase64ToString(llIntegerToBase64(0xE0808000 | ((input << 12) & 0x0F000000) | ((input << 10) & 0x3F0000) | ((input << 8) & 0x3F00)));
        if(input <= 0x10FFFF)
            return llBase64ToString(llIntegerToBase64(0xF0808080 | ((input << 06) & 0x07000000) | ((input << 04) & 0x3F0000) | ((input << 2) & 0x3F00) | (input & 0x3F)));
    }
    return "";
}

Byte Length of UTF-8 Encoded String

Returns the number of bytes that would be used if the string were encoded in UTF-8. 

integer StringUTF8Size(string str) {
    return (3 * llSubStringIndex(llStringToBase64(str)+"=", str = "=")) >> 2;
}

Truncate To Length

Keep the first n-bytes, discard the rest. If a character is truncated, the damaged portion is discarded. 

string TruncateToLength(string str, integer bytes) {
    string temp = llStringToBase64(str);
    if(llStringLength(temp) <= (bytes * 8) / 6)
        return str;
    temp = llBase64ToString(llGetSubString(temp+"==", (bytes * -2) % 3, (bytes * 4 - 1) / 3));
    integer i = llStringLength(temp) - 1;
    if(llGetSubString(temp, i, i) != llGetSubString(str, i, i))
        return llDeleteSubString(temp, i, i);//last char was multi-byte and was truncated.
    return temp;
}

Split To Length

Splits a string so at most only a specified number of bytes appear in each substring. If an entire character cannot fit in the current stride, it is pushed into the next. 

list SplitToLength(string in, integer bytes) {
    list out = [];
    while(in) {
        string str = llDeleteSubString(in, bytes, 0x7ffffff0);
        string temp = llStringToBase64(str);
        if(llStringLength(temp) <= (bytes * 8) / 6) {
            out += str;
            in = llDeleteSubString(in, 0, bytes);
        } else {
            temp = llBase64ToString(llGetSubString(temp + "==", (bytes * -2) % 3, (bytes * 4 - 1) / 3));
            integer i = llStringLength(temp) - 1;
            if(llGetSubString(temp, i, i) != llGetSubString(str, i, i)) {
                if(!i)
                    return out + ["", in, -2];
                out += llDeleteSubString(temp, i, i);//last char was multi-byte and was truncated.
                in = llGetSubString(in, i, 0x7ffffff0);
            } else {
                out += temp;
                in = llDeleteSubString(in, 0, i);
            }
        }
    }
    return out;
}

List Functions

Replace

The design of the logic had to overcome two hurdles. The first was keeping it from searching previous replacements (otherwise you could fall into an infinite loop or infinitely grow the memory). The second was so that it could do null replacements. Both of these hurdles were overcome but at the cost of some readability.

The way it works is it keeps an Unsearched Buffer (UB) which is a subset of the Input Buffer (IB) and it records the Position of the UB in the IB (P). Each iteration it searches the UB and adds that resulting index to the P, then it uses P as the index to replace that section in the IB, finally it recalculates the new P, then is uses IB with the new P to update UB. 

list ListReplace(list src, list from, list to)
{//replaces all occurrences of 'from' with 'to' in 'src'.
    integer len = ~([] != from);
    if(~len)
    {
        list  buffer = src;
        integer b_pos = ERR_GENERIC;
        integer to_len = ~([] != to);
        @loop;//instead of a while loop, saves 5 bytes (and runs faster).
        integer to_pos = ~llListFindList(buffer, from);
        if(to_pos)
        {
//            b_pos -= to_pos;
//            src = llListReplaceList(src, to, b_pos, b_pos + len);
//            b_pos += to_len;
//            buffer = llList2List(src, (-~(b_pos)), 0x4000);
            buffer = llList2List(src = llListReplaceList(src, to, b_pos -= to_pos, b_pos + len), (-~(b_pos += to_len)), 0x4000);
            jump loop;
        }
    }
    return src;
}

Compare

Compares two lists, returns true if identical.

Note: It will ignore the sign on zero (0.0 == -0.0). If the sign on zero is important to you, follow the instructions in the comments on how to enable checking for this. Currently no functions in LSL treat negative zero differently then positive zero. 

integer ListCompare(list input_a, list input_b)
{
    integer b = input_a != input_b;
    if(!b)
    {
        if((b = [] != input_a))
        {
            integer counter = b;
            do
            {
                if((b = llGetListEntryType(input_a,counter)) - llGetListEntryType(input_b,counter))
                    jump end_a;
                //Don't need to bother with TYPE_KEY
                if(b == TYPE_FLOAT) {
                    if((b = (llList2Float(input_a,counter) != llList2Float(input_b,counter))))
                        jump end_b;
                } else if(b == TYPE_VECTOR) {//it costs more to update b and jump out then to just return
                    if((b = (llList2Vector(input_a,counter) != llList2Vector(input_b,counter))))//so just return.
                        jump end_c;
                } else if(b == TYPE_ROTATION) {
                    if((b = (llList2Rot(input_a,counter) != llList2Rot(input_b,counter))))
                        jump end_d;
                }
                else//comment this line out if you care about the sign on zero.
                    if((b = (llList2String(input_a,counter) != llList2String(input_b,counter))))
                        jump end_e;
            }while((counter = -~counter));
            //if you get here, b equals zero, so we don't even have to change it's value for the return.
        }
    }
    @end_a;@end_b;@end_c;@end_d;@end_e;
    return !b;
}

If you do not care about positive/negative zero, then a better way to perform the comparison is: 

integer ListCompare(list a, list b) {
    integer aL = a != [];
    if (aL != (b != [])) return 0;
    else if (aL == 0)    return 1;
    
    return llListFindList((a = []) + a, (b = []) + b) == 0;
}

If you just want to compare the length of the lists use (a == b): 

integer ListCompare(list a, list b){
    return (a==b);
}


Base64 & Hex Encoding

Parameter & Return Based

Binary functions that use parameters and returns 

integer ReadBase64Integer(string data, integer index)
{
    integer S = (index << 5) % 6;
    index = llBase64ToInteger(llGetSubString((data = llGetSubString(data, index = ((index << 5) / 6), index+6)) + "AAAAAA", 0, 5)) << S;
    if(S)
        index = index | (llBase64ToInteger(llGetSubString("A" + (llDeleteSubString(data, 0, 1)) + "AAAAA", 0, 5)) >> (6 - S));
    return index;
}

list ReadBase64IntegerPair(string data, integer index)
{
    integer S = (index << 5) % 6;
    index = llBase64ToInteger(llGetSubString("A" + (llDeleteSubString(data = llGetSubString(data, index = ((index << 5) / 6), index + 11), 0, 2)) + "AAAAA", 0, 5));
    return [
        (llBase64ToInteger(llGetSubString((data) + "AAAAAA", 0, 5)) << S) | (index >> (12 - S)),
        ((llBase64ToInteger(llGetSubString((llDeleteSubString(data,0,5)) + "AAAAAA", 0, 5)) >> (4 - S)) & ~(0xF0000000 << S)) | (index << (20 + S))
    ];
}

string WriteBase64Integer(string data, integer index, integer value)
{

    integer S = 12 - ((index % 3) << 1);
    return  llDeleteSubString(
        llInsertString(
        data,
        index = ((index << 4) / 3), 	
            llInsertString(
                llIntegerToBase64(
                    (llBase64ToInteger(llGetSubString((data = llGetSubString(data, index, index+7)) + "AAAAAA", 0, 5)) & (0xFFF00000 << S)) | 
                    ((value >> (12 - S)) & ~(0xFFF00000 << S))
                ), 2, 
                llIntegerToBase64(
                    (llBase64ToInteger(llGetSubString((llDeleteSubString(data, 0, 1)) + "AAAAAA", 0, 5)) & ~(0xFFFFFFFF << S)) | 
                    (value << S)
        )	)	), index+7, index + 22);//insert it then remove the old and the extra.
}

string WriteBase64IntegerPair(string data, integer index, integer low, integer high)
{

    integer S = 10 - ((index << 5) % 6);
    return  llDeleteSubString(
        llInsertString(
            data,
            index = ((index << 5) / 6),
            llInsertString(
                llInsertString(
                    llIntegerToBase64(
                        (llBase64ToInteger(llGetSubString((data = llGetSubString(data, index, index + 12)) + "AAAAAA", 0, 5)) & (0xFFC00000 << S)) | 
                        ((low >> (10 - S)) & ~(0xFFC00000 << S))
                    ), 3,
                    llIntegerToBase64(
                        ((high >> (24 - S)) & ~(0xFFFFFF00 << S)) | 
                        (low << (8 + S))
                )   ), 7,
                llIntegerToBase64(
                    (llBase64ToInteger(llGetSubString((llDeleteSubString(data, 0, 6)) + "AAAAAA", 0, 5)) & ~(0xFFFFFFFF << S)) | 
                    (high << S)
        )   )   ), index + 12, index + 35);//insert it then remove the old and the extra.
}

string DwordList2Hex(list in)
{
    string out = ""; integer len = ~(in != []);

    while((len = -~len))
    {
        integer int = llList2Integer(in, len);
        integer j = 8;
        string mout = "";
        do
        {
            mout = llGetSubString(hexc, int & 15, int & 15) + mout;
            int = int >> 4;
        }while((j = ~-j));
        out += mout;
    }
    return out;

}

string HexToBase64(string a)
{
    string d = "";
    integer e = (llStringLength(a += "0000000000000000") - 9) & 0xFFFFFFF8;
    integer g;
    do{
        d = WriteBase64IntegerPair(d, g >> 3, ((integer)("0x"+(llGetSubString(a, g, g + 7)))), ((integer)("0x"+(llGetSubString(a, g + 8, g + 15)))));
    }while((g += 16) < e);
    return d;
}

list Base64ToHex(string a)
{
    list out = [];
    integer len = ((llStringLength(a = TrimRight(a,"A=")) + 4) * 6) >> 5;
    integer i = -1;
    integer int;

    while((i = -~i) <= len)
    {
        if((int = ReadBase64Integer(a, i)) || (i ^ len))
        {
            integer j = 8;
            string mout = "";
            do
            {
                mout = llGetSubString(hexc, int & 15, int & 15) + mout;
                int = int >> 4;
            }while((j = ~-j));
            out += mout;
        }
    }
    return out;

}

integer ReadBase64Byte(string data, integer index)
{
    return 0xFF & (
        llBase64ToInteger(llGetSubString((llGetSubString(data, (index << 3) / 6, (-~((index << 3) / 6)))) + "AAAAAA", 0, 5)) >> 
        (24 - ((index << 3) % 6)));
}

string WriteBase64Byte(string data, integer index, integer value)
{
    integer S = 24 - ((index = (index << 3)) % 6);
    return llDeleteSubString(llInsertString(data, index /= 6,
        llIntegerToBase64(
            (llBase64ToInteger(llGetSubString((llGetSubString(data, index, (-~(index)))) + "AAAAAA", 0, 5)) & ~(0xFF << S)) | 
            ((value & 0xFF) << S)
    )   ), index + 2, index + 9);//insert it then remove the old and the extra.
}

list Base64ToDwordList(string a)
{
    integer len = (6 * (llStringLength(a = TrimRight(a,"A=")) + 4)) >> 5;
    integer i = -1;
    list out;
    integer int;
    while((i = -~i) <= len)
        if((int = ReadBase64Integer(a, i)) || (i ^ len))
            out += int;
    return out;
}

string DwordListToBase64(list a)
{
    integer len = (a != []);
    integer i = -1;
    string out;
    while((i = -~i) < len)
        out = WriteBase64Integer(out, i, llList2Integer(a, i));
    return TrimRight(out,"A");
}

string hexc="0123456789ABCDEF";

Global Buffers

These versions of the functions depend upon global buffers. Not all the functions have been ported to this model (because it is very complicated to build functions with macros so they can be flipped from one mode to the other). 

integer v0;
integer v1;
string v;

ReadBase64Integer(integer index)
{
    integer S = (index << 5) % 6;
    string T = llGetSubString(v, index = ((index << 5) / 6), index+6);
    v0 = llBase64ToInteger(llGetSubString((T) + "AAAAAA", 0, 5)) << S;
    if(S)
        v0 = v0 | (llBase64ToInteger(llGetSubString("A" + (llDeleteSubString(T, 0, 1)) + "AAAAA", 0, 5)) >> (6 - S));

}

ReadBase64IntegerPair(integer index)
{
    integer S = (index << 5) % 6;
    string buf = llGetSubString(v, index = ((index << 5) / 6), index + 11);
    index = llBase64ToInteger(llGetSubString("A" + (llDeleteSubString(buf, 0, 2)) + "AAAAA", 0, 5));

        v0 = (llBase64ToInteger(llGetSubString((buf) + "AAAAAA", 0, 5)) << S) | (index >> (12 - S));
        v1 = ((llBase64ToInteger(llGetSubString((llDeleteSubString(buf,0,5)) + "AAAAAA", 0, 5)) >> (4 - S)) & ~(0xF0000000 << S)) | (index << (20 + S));

}

WriteBase64Integer(integer index)
{

    integer S = 12 - ((index % 3) << 1);
    v =  llDeleteSubString(
        llInsertString(
        v,
        index = ((index << 4) / 3), 	
            llInsertString(
                llIntegerToBase64(
                    (llBase64ToInteger(llGetSubString((v = llGetSubString(v, index, index+7)) + "AAAAAA", 0, 5)) & (0xFFF00000 << S)) | 
                    ((v0 >> (12 - S)) & ~(0xFFF00000 << S))
                ), 2, 
                llIntegerToBase64(
                    (llBase64ToInteger(llGetSubString((llDeleteSubString(v, 0, 1)) + "AAAAAA", 0, 5)) & ~(0xFFFFFFFF << S)) | 
                    (v0 << S)
        )	)	), index+7, index + 22);//insert it then remove the old and the extra.
}

WriteBase64IntegerPair(integer index)
{

    integer S = 10 - ((index << 5) % 6);
    v =  llDeleteSubString(
        llInsertString(
            v,
            index = ((index << 5) / 6),
            llInsertString(
                llInsertString(
                    llIntegerToBase64(
                        (llBase64ToInteger(llGetSubString((v = llGetSubString(v, index, index + 12)) + "AAAAAA", 0, 5)) & (0xFFC00000 << S)) | 
                        ((v0 >> (10 - S)) & ~(0xFFC00000 << S))
                    ), 3,
                    llIntegerToBase64(
                        ((v1 >> (24 - S)) & ~(0xFFFFFF00 << S)) | 
                        (v0 << (8 + S))
                )   ), 7,
                llIntegerToBase64(
                    (llBase64ToInteger(llGetSubString((llDeleteSubString(v, 0, 6)) + "AAAAAA", 0, 5)) & ~(0xFFFFFFFF << S)) | 
                    (v1 << S)
        )   )   ), index + 12, index + 35);//insert it then remove the old and the extra.
}

HexToBase64(string a)
{
    v = "";
    integer e = (llStringLength(a += "0000000000000000") - 9) & 0xFFFFFFF8;
    integer g;
    do{
        v0 = ((integer)("0x"+(llGetSubString(a, g, g + 7))));
        v1 = ((integer)("0x"+(llGetSubString(a, g + 8, g + 15))));
        WriteBase64IntegerPair(g >> 3);
    }while((g += 16) < e);

}

list Base64ToHex()
{
    list out = [];
    integer len = ((llStringLength(v = TrimRight(v,"A=")) + 4) * 6) >> 5;
    integer i = 0;

    do{
        ReadBase64Integer(i); if(v0 || (i ^ len))
        {
            integer j = 8;
            string mout = "";
            do{
                mout = llGetSubString(hexc, v0 & 15, v0 & 15) + mout;
                v0 = v0 >> 4;
            }while((j = ~-j));
            out += mout;
        }
    }while((i = -~i) <= len);
    return out;

}

list Base64ToDwordList()
{
    integer len = (6 * (llStringLength(v = TrimRight(v,"A=")) + 4)) >> 5;
    integer i = 0;
    list out = [];

    do
    {
        ReadBase64Integer(i);
        if(v0 || (i ^ len))
            out += v0;
    }while((i = -~i) <= len);
    return out;
}

ReadBase64Byte(integer index)
{
    v0 = 0xFF & (
        llBase64ToInteger(llGetSubString((llGetSubString(v, (index << 3) / 6, (-~((index << 3) / 6)))) + "AAAAAA", 0, 5)) >> 
        (24 - ((index << 3) % 6)));
}

WriteBase64Byte(integer index)
{
    integer S = 24 - ((index = (index << 3)) % 6);
    v = llDeleteSubString(llInsertString(v, index /= 6,
        llIntegerToBase64(
            (llBase64ToInteger(llGetSubString((llGetSubString(v, index, (-~(index)))) + "AAAAAA", 0, 5)) & ~(0xFF << S)) | 
            ((v0 & 0xFF) << S)
    )   ), index + 2, index + 9);//insert it then remove the old and the extra.
}

string hexc="0123456789ABCDEF";

Float Union Integer

The Float Union Integer functions allow for safely storing floats in integers and provide for retrieval. All real numbers are supported.

For versions that support Infinity and NaN see User:Strife_Onizuka/Float_Functions (the two versions have not been reconciled yet).

General

These versions will work well in LSO and Mono. 

integer fui(float input)//Mono Safe, LSO Safe, Doubles Unsupported, LSLEditor Unsafe
{//union float to integer
    if((input) != 0.0){//is it non zero?
        integer sign = (input < 0) << 31;//the sign, but later this variable is reused to store the shift
        if((input = llFabs(input)) < 2.3509887016445750159374730744445e-38)//Denormalized range check & last stride of normalized range
            return sign | (integer)(input / 1.4012984643248170709237295832899e-45);//the math overlaps; saves cpu time.
        integer exp = llFloor((llLog(input) / 0.69314718055994530941723212145818));//extremes will error towards extremes. following yuch corrects it.
        return (0x7FFFFF & (integer)(input * (0x1000000 >> sign))) | (((exp + 126 + (sign = ((integer)input - (3 <= (input /= (float)("0x1p"+(string)(exp -= ((exp >> 31) | 1)))))))) << 23 ) | sign);
    }//for grins, detect the sign on zero. it's not pretty but it works. the previous requires alot of unwinding to understand it.
    return ((string)input == (string)(-0.0)) << 31;
}

float iuf(integer input)//LSLEditor Unsafe, LSO Safe, Mono Safe
{//union integer to float
    return llPow(2.0, (input | !input) - 150) * (((!!(input = (0xff & (input >> 23)))) << 23) | ((input & 0x7fffff))) * (1 | (input >> 31));
}//will crash if the raw exponent == 0xff; reason for crash deviates from float standard; though a crash is warented.

LSLEditor Safe

These versions will run in LSLEditor (as well as LSO and Mono). Doubles are not supported and anything outside the float range will become signed infinity or zero respectively. 

integer fui(float input)//Mono Safe, LSO Safe, Doubles Unsupported, LSLEditor Safe
{//union float to integer
    if((input) != 0.0){//is it non zero?
        integer sign = (integer)(input < 0) << 31;//the sign, but later this variable is reused to store the shift
        if((input = llFabs(input)) < 2.3509887016445750159374730744445e-38)//Denormalized range check & last stride of normalized range
            return sign | (integer)(input / 1.4012984643248170709237295832899e-45);//the math overlaps; saves cpu time.
        if(input > 3.4028234663852885981170418348452e+38)//infinity generation for doubles x_x
            return sign | 0x7F800000;//return signed infinity
        integer exp = llFloor((llLog(input) / 0.69314718055994530941723212145818));//extremes will error towards extremes. following yuch corrects it.
        input /= llPow(2.0, exp -= ((exp >> 31) | 1));
        integer d = (integer)input - (3 <= input);
        return (0x7FFFFF & (integer)(input * (0x1000000 >> d))) | (((exp + 126 + d) << 23 ) | sign);
    }//for grins, detect the sign on zero. it's not pretty but it works. the previous requires alot of unwinding to understand it.
    return (integer)((string)input == (string)(-0.0)) << 31;
}

float iuf(integer input)//LSLEditor Safe, LSO Safe, Mono Safe
{//union integer to float
    if((input & 0x7FFFFFFF) == 0x7F800000)//Infinity Check
        return ((input >> 31) | 1) / 0.0;
    integer exp = 0xff & (input >> 23);
    return llPow(2.0, (exp | !exp) - 150) * (((!!exp) << 23) | ((input & 0x7fffff))) * (1 | (input >> 31));
}//will crash if the raw exponent == 0xff; reason for crash deviates from float standard; though a crash is warented.

Change Log

1.0

  • New Macros for hiding implementation details of optimizations.
    • These macros can also be disabled so the resulting LSL is readable.
  • New Macros for VM specific Order-of-Execution compiling.
    • More pointless parentheses are now generated. Too lazy at this time to create extra macros keep this from happening.
    • Only operators that are commutative are supported at this time.
      • Be warned: It won't stop you if you use a non-commutative operator and your code will be wrong.
    • This means that you can write optimized code and not have to worry about which VM is executing it.
  • Many functions have been updated to use the new macros.

Unversioned

  • Changes before 1.0 were never recorded.
Retrieved from "https://wiki.secondlife.com/w/index.php?title=Combined_Library&oldid=1213295"