LLamaSharp/LLama/Grammars/GBNFGrammarParser.cs

399 lines
14 KiB
C#

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using LLama.Exceptions;
using LLama.Native;
namespace LLama.Grammars
{
/// <summary>
/// Source:
/// https://github.com/ggerganov/llama.cpp/blob/6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9/common/grammar-parser.cpp
///
/// The commit hash from URL is the actual commit hash that reflects current C# code.
/// </summary>
internal sealed class GBNFGrammarParser
{
// NOTE: assumes valid utf8 (but checks for overrun)
// copied from llama.cpp
private static uint DecodeUTF8(ref ReadOnlySpan<byte> src)
{
int[] lookup = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
byte firstByte = src[0];
byte highbits = (byte)(firstByte >> 4);
int len = lookup[highbits];
byte mask = (byte)((1 << (8 - len)) - 1);
uint value = (uint)(firstByte & mask);
int end = len;
int pos = 1;
for (; pos < end && pos < src.Length; pos++)
{
value = (uint)((value << 6) + (src[pos] & 0x3F));
}
src = src.Slice(pos);
return value;
}
private static bool IsWordChar(byte c)
{
return ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') || c == '-' || ('0' <= c && c <= '9');
}
private static uint ParseHex(ref ReadOnlySpan<byte> src, int size)
{
int pos = 0;
int end = size;
uint value = 0;
for (; pos < end && pos < src.Length; pos++)
{
value <<= 4;
byte c = src[pos];
if ('a' <= c && c <= 'f')
{
value += (uint)(c - 'a' + 10);
}
else if ('A' <= c && c <= 'F')
{
value += (uint)(c - 'A' + 10);
}
else if ('0' <= c && c <= '9')
{
value += (uint)(c - '0');
}
else
{
break;
}
}
if (pos != end)
{
throw new GrammarUnexpectedHexCharsCount(size, Encoding.UTF8.GetString(src.ToArray()));
}
src = src.Slice(pos);
return value;
}
private static ReadOnlySpan<byte> ParseSpace(ReadOnlySpan<byte> src, bool newlineOk)
{
int pos = 0;
while (pos < src.Length &&
(src[pos] == ' ' || src[pos] == '\t' || src[pos] == '#' ||
(newlineOk && (src[pos] == '\r' || src[pos] == '\n'))))
{
if (src[pos] == '#')
{
while (pos < src.Length && src[pos] != '\r' && src[pos] != '\n')
{
pos++;
}
}
else
{
pos++;
}
}
return src.Slice(pos);
}
private static ReadOnlySpan<byte> ParseName(ReadOnlySpan<byte> src)
{
int pos = 0;
while (pos < src.Length && IsWordChar(src[pos]))
{
pos++;
}
if (pos == 0)
{
throw new GrammarExpectedName(Encoding.UTF8.GetString(src.ToArray()));
}
return src.Slice(pos);
}
private static uint ParseChar(ref ReadOnlySpan<byte> src)
{
if (src[0] == '\\')
{
if (src.Length < 2)
throw new GrammarUnexpectedEndOfInput();
var chr = src[1];
src = src.Slice(2);
return (char)chr switch
{
'x' => ParseHex(ref src, 2),
'u' => ParseHex(ref src, 4),
'U' => ParseHex(ref src, 8),
't' => '\t',
'r' => '\r',
'n' => '\n',
'\\' or '"' or '[' or ']' => chr,
_ => throw new GrammarUnknownEscapeCharacter(Encoding.UTF8.GetString(src.ToArray())),
};
}
if (!src.IsEmpty)
return DecodeUTF8(ref src);
throw new GrammarUnexpectedEndOfInput();
}
private ReadOnlySpan<byte> ParseSequence(
ParseState state,
ReadOnlySpan<byte> pos,
string ruleName,
List<LLamaGrammarElement> outElements,
bool isNested)
{
int lastSymStart = outElements.Count;
while (!pos.IsEmpty)
{
if (pos[0] == '"') // literal string
{
pos = pos.Slice(1);
lastSymStart = outElements.Count;
while (!pos.IsEmpty && pos[0] != '"')
{
var charPair = ParseChar(ref pos);
outElements.Add(new LLamaGrammarElement(LLamaGrammarElementType.CHAR, charPair));
}
pos = ParseSpace(pos.Slice(1), isNested);
}
else if (pos[0] == '[') // char range(s)
{
pos = pos.Slice(1);
var startType = LLamaGrammarElementType.CHAR;
if (pos[0] == '^')
{
pos = pos.Slice(1);
startType = LLamaGrammarElementType.CHAR_NOT;
}
lastSymStart = outElements.Count;
while (!pos.IsEmpty && pos[0] != ']')
{
var charPair = ParseChar(ref pos);
var type = lastSymStart < outElements.Count ? LLamaGrammarElementType.CHAR_ALT : startType;
outElements.Add(new LLamaGrammarElement(type, charPair));
if (pos[0] == '-' && pos[1] != ']')
{
pos = pos.Slice(1);
var endCharPair = ParseChar(ref pos);
outElements.Add(new LLamaGrammarElement(LLamaGrammarElementType.CHAR_RNG_UPPER, endCharPair));
}
}
pos = ParseSpace(pos.Slice(1), isNested);
}
else if (IsWordChar(pos[0])) // rule reference
{
var nameEnd = ParseName(pos);
uint refRuleId = state.GetSymbolId(pos, nameEnd.Length);
pos = ParseSpace(nameEnd, isNested);
lastSymStart = outElements.Count;
outElements.Add(new LLamaGrammarElement(LLamaGrammarElementType.RULE_REF, refRuleId));
}
else if (pos[0] == '(') // grouping
{
// parse nested alternates into synthesized rule
pos = ParseSpace(pos.Slice(1), true);
uint subRuleId = state.GenerateSymbolId(ruleName);
pos = ParseAlternates(state, pos, ruleName, subRuleId, true);
lastSymStart = outElements.Count;
// output reference to synthesized rule
outElements.Add(new LLamaGrammarElement(LLamaGrammarElementType.RULE_REF, subRuleId));
if (pos[0] != ')')
throw new GrammarExpectedNext(")", Encoding.UTF8.GetString(pos.ToArray()));
pos = ParseSpace(pos.Slice(1), isNested);
}
else if (pos[0] == '*' || pos[0] == '+' || pos[0] == '?') // repetition operator
{
if (lastSymStart == outElements.Count)
throw new GrammarExpectedPrevious("*/+/?", Encoding.UTF8.GetString(pos.ToArray()));
// apply transformation to previous symbol (lastSymStart to end) according to
// rewrite rules:
// S* --> S' ::= S S' |
// S+ --> S' ::= S S' | S
// S? --> S' ::= S |
uint subRuleId = state.GenerateSymbolId(ruleName);
List<LLamaGrammarElement> subRule = new List<LLamaGrammarElement>();
// add preceding symbol to generated rule
subRule.AddRange(outElements.GetRange(lastSymStart, outElements.Count - lastSymStart));
if (pos[0] == '*' || pos[0] == '+')
{
// cause generated rule to recurse
subRule.Add(new LLamaGrammarElement(LLamaGrammarElementType.RULE_REF, subRuleId));
}
// mark start of alternate def
subRule.Add(new LLamaGrammarElement(LLamaGrammarElementType.ALT, 0));
if (pos[0] == '+')
{
// add preceding symbol as alternate only for '+' (otherwise empty)
subRule.AddRange(outElements.GetRange(lastSymStart, outElements.Count - lastSymStart));
}
subRule.Add(new LLamaGrammarElement(LLamaGrammarElementType.END, 0));
state.AddRule(subRuleId, subRule);
// in original rule, replace previous symbol with reference to generated rule
outElements.RemoveRange(lastSymStart, outElements.Count - lastSymStart);
outElements.Add(new LLamaGrammarElement(LLamaGrammarElementType.RULE_REF, subRuleId));
pos = ParseSpace(pos.Slice(1), isNested);
}
else
{
break;
}
}
return pos;
}
private ReadOnlySpan<byte> ParseAlternates(
ParseState state,
ReadOnlySpan<byte> src,
string ruleName,
uint ruleId,
bool isNested)
{
var rule = new List<LLamaGrammarElement>();
ReadOnlySpan<byte> pos = ParseSequence(state, src, ruleName, rule, isNested);
while (!pos.IsEmpty && pos[0] == '|')
{
rule.Add(new LLamaGrammarElement(LLamaGrammarElementType.ALT, 0));
pos = ParseSpace(pos.Slice(1), true);
pos = ParseSequence(state, pos, ruleName, rule, isNested);
}
rule.Add(new LLamaGrammarElement(LLamaGrammarElementType.END, 0));
state.AddRule(ruleId, rule);
return pos;
}
private ReadOnlySpan<byte> ParseRule(ParseState state, ReadOnlySpan<byte> src)
{
ReadOnlySpan<byte> nameEnd = ParseName(src);
ReadOnlySpan<byte> pos = ParseSpace(nameEnd, false);
int nameLen = src.Length - nameEnd.Length;
uint ruleId = state.GetSymbolId(src.Slice(0, nameLen), 0);
string name = Encoding.UTF8.GetString(src.Slice(0, nameLen).ToArray());
if (!(pos[0] == ':' && pos[1] == ':' && pos[2] == '='))
throw new GrammarExpectedNext("::=", Encoding.UTF8.GetString(pos.ToArray()));
pos = ParseSpace(pos.Slice(3), true);
pos = ParseAlternates(state, pos, name, ruleId, false);
if (!pos.IsEmpty && pos[0] == '\r')
{
pos = pos.Slice(pos[1] == '\n' ? 2 : 1);
}
else if (!pos.IsEmpty && pos[0] == '\n')
{
pos = pos.Slice(1);
}
else if (!pos.IsEmpty)
{
throw new GrammarExpectedNext("newline or EOF", Encoding.UTF8.GetString(pos.ToArray()));
}
return ParseSpace(pos, true);
}
/// <summary>
/// Parse a string of <a href="https://github.com/ggerganov/llama.cpp/tree/master/grammars">GGML BNF</a>
/// </summary>
/// <param name="input">The string to parse</param>
/// <param name="startRule">The name of the root rule of this grammar</param>
/// <exception cref="GrammarFormatException">Thrown if input is malformed</exception>
/// <returns>A ParseState that can be converted into a grammar for sampling</returns>
public Grammar Parse(string input, string startRule)
{
var byteArray = Encoding.UTF8.GetBytes(input);
var state = new ParseState();
var pos = ParseSpace(byteArray, true);
while (!pos.IsEmpty)
{
pos = ParseRule(state, pos);
}
var names = state.SymbolIds.ToDictionary(a => a.Value, a => a.Key);
var rules = new List<GrammarRule>();
for (var i = 0; i < state.Rules.Count; i++)
{
var elements = state.Rules[i];
var name = names[(uint)i];
rules.Add(new GrammarRule(name, elements));
}
var startRuleIndex = state.SymbolIds[startRule];
return new Grammar(rules, startRuleIndex);
}
private record ParseState
{
public SortedDictionary<string, uint> SymbolIds { get; } = new();
public List<List<LLamaGrammarElement>> Rules { get; } = new();
public uint GetSymbolId(ReadOnlySpan<byte> src, int len)
{
var nextId = (uint)SymbolIds.Count;
var key = Encoding.UTF8.GetString(src.Slice(0, src.Length - len).ToArray());
if (SymbolIds.TryGetValue(key, out uint existingId))
{
return existingId;
}
else
{
SymbolIds[key] = nextId;
return nextId;
}
}
public uint GenerateSymbolId(string baseName)
{
var nextId = (uint)SymbolIds.Count;
var key = $"{baseName}_{nextId}";
SymbolIds[key] = nextId;
return nextId;
}
public void AddRule(uint ruleId, List<LLamaGrammarElement> rule)
{
while (Rules.Count <= ruleId)
{
Rules.Add(new List<LLamaGrammarElement>());
}
Rules[(int)ruleId] = rule;
}
}
}
}