Question

是否可以将Google RE2与流一起使用？我们假设使用正则表达式处理的一些输入文字可能太大而无法保存在内存中。

Answer 1

如果存在最大匹配长度，则可以以至少两倍长度的块读取数据。如果匹配失败，或者从末尾开始少于那么多字符，则剪切当前字符串，然后追加另一个块。

匹配字符串的长度绝不会超过块长度+最大匹配长度。

C＃中的示例：

public static IEnumerable<StreamMatch> MatchesInStream(
        this Regex pattern, TextReader reader,
        int maxMatchLength, int blockLength)
{
    if (maxMatchLength <= 0)
    {
        throw new ArgumentException("Must be positive", "maxMatchLength");
    }
    if (blockLength < maxMatchLength)
    {
        throw new ArgumentException("Must be at least as long as maxMatchLength", "blockLength");
    }

    char[] buffer = new char[blockLength];
    string chunk = "";
    int matchOffset = 0;

    // Read one block, and append to the string
    int charsRead = reader.ReadBlock(buffer, 0, blockLength);
    chunk += new string(buffer, 0, charsRead);

    while (charsRead > 0 && chunk.Length > maxMatchLength)
    {
        int cutPosition = 0;
        foreach (Match match in pattern.Matches(chunk))
        {
            if (match.Index > chunk.Length - maxMatchLength)
            {
                // The match could possibly have matched more characters.
                // Read another block before trying again.
                break;
            }

            yield return new StreamMatch(matchOffset, match);
            cutPosition = match.Index + match.Length;
        }
        cutPosition = Math.Max(cutPosition, chunk.Length - maxMatchLength);
        matchOffset += cutPosition;
        chunk = chunk.Substring(cutPosition);

        charsRead = reader.ReadBlock(buffer, 0, blockLength);
        chunk += new string(buffer, 0, charsRead);
    }

    // Stream has ended. Try to match the last remaining characters.
    foreach (Match match in pattern.Matches(chunk))
    {
        yield return new StreamMatch(matchOffset, match);
    }
}

public class StreamMatch
{
    public int MatchOffset { get; private set; }
    public Match Match { get; private set; }

    public StreamMatch(int matchOffset, Match match)
    {
        MatchOffset = matchOffset;
        Match = match;
    }
}

// One horrible XML parser
var reader = new StreamReader(stream);
var pattern = new Regex(@"<(/?)([\w:-]{1,15})([^<>]{0,50}(?<!/))(/?)>");
foreach (StreamMatch match in pattern.MatchesInStream(reader, 69, 128))
{
    Console.WriteLine(match.Match.Value);
}

流上的RE2正则表达式？

1 个答案: