"vscode:/vscode.git/clone" did not exist on "9af454142a34536ab1f3c149cc8764b7ab460c0d"
parser.go 3.39 KB
Newer Older
1
2
3
4
package parser

import (
	"bufio"
5
	"bytes"
6
	"errors"
7
	"fmt"
8
	"io"
9
	"log/slog"
10
	"slices"
11
12
13
14
)

type Command struct {
	Name string
15
16
17
18
19
20
	Args string
}

func (c *Command) Reset() {
	c.Name = ""
	c.Args = ""
21
22
23
24
}

func Parse(reader io.Reader) ([]Command, error) {
	var commands []Command
25
	var command, modelCommand Command
26
27

	scanner := bufio.NewScanner(reader)
28
	scanner.Buffer(make([]byte, 0, bufio.MaxScanTokenSize), bufio.MaxScanTokenSize)
29
	scanner.Split(scanModelfile)
30
	for scanner.Scan() {
31
32
33
		line := scanner.Bytes()

		fields := bytes.SplitN(line, []byte(" "), 2)
Michael Yang's avatar
Michael Yang committed
34
		if len(fields) == 0 || len(fields[0]) == 0 {
35
36
37
			continue
		}

38
		switch string(bytes.ToUpper(fields[0])) {
39
40
		case "FROM":
			command.Name = "model"
41
			command.Args = string(bytes.TrimSpace(fields[1]))
42
43
			// copy command for validation
			modelCommand = command
44
45
46
47
		case "ADAPTER":
			command.Name = string(bytes.ToLower(fields[0]))
			command.Args = string(bytes.TrimSpace(fields[1]))
		case "LICENSE", "TEMPLATE", "SYSTEM", "PROMPT":
48
			command.Name = string(bytes.ToLower(fields[0]))
49
			command.Args = string(fields[1])
50
		case "PARAMETER":
51
			fields = bytes.SplitN(fields[1], []byte(" "), 2)
Michael Yang's avatar
Michael Yang committed
52
53
54
55
			if len(fields) < 2 {
				return nil, fmt.Errorf("missing value for %s", fields)
			}

56
			command.Name = string(fields[0])
57
			command.Args = string(bytes.TrimSpace(fields[1]))
58
59
		case "EMBED":
			return nil, fmt.Errorf("deprecated command: EMBED is no longer supported, use the /embed API endpoint instead")
60
61
62
63
64
65
66
67
68
69
		case "MESSAGE":
			command.Name = string(bytes.ToLower(fields[0]))
			fields = bytes.SplitN(fields[1], []byte(" "), 2)
			if len(fields) < 2 {
				return nil, fmt.Errorf("should be in the format <role> <message>")
			}
			if !slices.Contains([]string{"system", "user", "assistant"}, string(bytes.ToLower(fields[0]))) {
				return nil, fmt.Errorf("role must be one of \"system\", \"user\", or \"assistant\"")
			}
			command.Args = fmt.Sprintf("%s: %s", string(bytes.ToLower(fields[0])), string(fields[1]))
70
		default:
Michael Yang's avatar
Michael Yang committed
71
72
			if !bytes.HasPrefix(fields[0], []byte("#")) {
				// log a warning for unknown commands
73
				slog.Warn(fmt.Sprintf("Unknown command: %s", fields[0]))
Michael Yang's avatar
Michael Yang committed
74
			}
75
			continue
76
		}
77
78
79

		commands = append(commands, command)
		command.Reset()
80
81
	}

82
	if modelCommand.Args == "" {
83
		return nil, errors.New("no FROM line for the model was specified")
84
85
86
87
	}

	return commands, scanner.Err()
}
88
89

func scanModelfile(data []byte, atEOF bool) (advance int, token []byte, err error) {
90
91
92
93
94
95
96
97
	advance, token, err = scan([]byte(`"""`), []byte(`"""`), data, atEOF)
	if err != nil {
		return 0, nil, err
	}

	if advance > 0 && token != nil {
		return advance, token, nil
	}
98

99
100
101
102
103
104
105
106
107
108
109
110
111
	advance, token, err = scan([]byte(`"`), []byte(`"`), data, atEOF)
	if err != nil {
		return 0, nil, err
	}

	if advance > 0 && token != nil {
		return advance, token, nil
	}

	return bufio.ScanLines(data, atEOF)
}

func scan(openBytes, closeBytes, data []byte, atEOF bool) (advance int, token []byte, err error) {
112
113
	newline := bytes.IndexByte(data, '\n')

114
115
	if start := bytes.Index(data, openBytes); start >= 0 && start < newline {
		end := bytes.Index(data[start+len(openBytes):], closeBytes)
116
		if end < 0 {
117
			if atEOF {
118
				return 0, nil, fmt.Errorf("unterminated %s: expecting %s", openBytes, closeBytes)
119
120
121
			} else {
				return 0, nil, nil
			}
122
123
		}

124
		n := start + len(openBytes) + end + len(closeBytes)
Michael Yang's avatar
Michael Yang committed
125
126

		newData := data[:start]
127
		newData = append(newData, data[start+len(openBytes):n-len(closeBytes)]...)
Michael Yang's avatar
Michael Yang committed
128
		return n, newData, nil
129
130
	}

131
	return 0, nil, nil
132
}