| @@ -1,125 +1,45 | |||
| 
             | 
        1 | 1 | using System; | 
| 
             | 
        2 | 2 | using Implab.Automaton.RegularExpressions; | 
| 
             | 
        3 | 3 | using Implab.Automaton; | 
| 
             | 
        4 | using System.Diagnostics; | |
| 
             | 
        4 | 5 | |
| 
             | 
        5 | 6 | namespace Implab.Formats { | 
| 
             | 
        6 | 7 | public struct BufferScanner<TTag> { | 
| 
             | 
        7 | char[] m_buffer; | |
| 
             | 
        8 | int m_offset; | |
| 
             | 
        9 | int m_position; | |
| 
             | 
        10 | int m_hi; | |
| 
             | 
        11 | ||
| 
             | 
        12 | readonly int m_chunk; | |
| 
             | 
        13 | readonly int m_limit; | |
| 
             | 
        14 | ||
| 
             | 
        15 | 8 | readonly DFAStateDescriptor<TTag>[] m_dfa; | 
| 
             | 
        16 | 9 | int m_state; | 
| 
             | 
        10 | int m_pos; | |
| 
             | 
        17 | 11 | |
| 
             | 
        18 | 
            
                     public BufferScanner(DFAStateDescriptor<TTag>[] dfa, int initialState | 
    |
| 
             | 
        12 | public BufferScanner(DFAStateDescriptor<TTag>[] dfa, int initialState) { | |
| 
             | 
        19 | 13 | m_dfa = dfa; | 
| 
             | 
        20 | 14 | m_state = initialState; | 
| 
             | 
        21 | m_chunk = chunk; | |
| 
             | 
        22 | m_limit = limit; | |
| 
             | 
        23 | m_buffer = null; | |
| 
             | 
        24 | m_offset = 0; | |
| 
             | 
        25 | m_position = 0; | |
| 
             | 
        26 | m_hi = 0; | |
| 
             | 
        27 | } | |
| 
             | 
        28 | ||
| 
             | 
        29 | public char[] Buffer { | |
| 
             | 
        30 | get { | |
| 
             | 
        31 | return m_buffer; | |
| 
             | 
        32 | } | |
| 
             | 
        33 | } | |
| 
             | 
        34 | ||
| 
             | 
        35 | public int HiMark { | |
| 
             | 
        36 | get { | |
| 
             | 
        37 | return m_hi; | |
| 
             | 
        38 | } | |
| 
             | 
        39 | 15 | } | 
| 
             | 
        40 | 16 | |
| 
             | 
        41 | 17 | public int Position { | 
| 
             | 
        42 | get { | |
| 
             | 
        43 | return m_position; | |
| 
             | 
        44 | } | |
| 
             | 
        45 | } | |
| 
             | 
        46 | ||
| 
             | 
        47 | public int Length { | |
| 
             | 
        48 | get { | |
| 
             | 
        49 | return m_hi - m_position; | |
| 
             | 
        50 | } | |
| 
             | 
        51 | } | |
| 
             | 
        52 | ||
| 
             | 
        53 | public int TokenOffset { | |
| 
             | 
        54 | get { | |
| 
             | 
        55 | return m_offset; | |
| 
             | 
        56 | } | |
| 
             | 
        57 | } | |
| 
             | 
        58 | ||
| 
             | 
        59 | public int TokenLength { | |
| 
             | 
        60 | get { | |
| 
             | 
        61 | return m_position - m_offset; | |
| 
             | 
        62 | } | |
| 
             | 
        63 | } | |
| 
             | 
        64 | ||
| 
             | 
        65 | public void Init(char[] buffer, int position, int length) { | |
| 
             | 
        66 | m_buffer = buffer; | |
| 
             | 
        67 | m_position = position; | |
| 
             | 
        68 | m_offset = position; | |
| 
             | 
        69 | m_hi = position + length; | |
| 
             | 
        70 | } | |
| 
             | 
        71 | ||
| 
             | 
        72 | public int Extend() { | |
| 
             | 
        73 | // free space | |
| 
             | 
        74 | var free = m_buffer.Length - m_hi; | |
| 
             | 
        75 | ||
| 
             | 
        76 | // if the buffer have enough free space | |
| 
             | 
        77 | if (free > 0) | |
| 
             | 
        78 | return free; | |
| 
             | 
        79 | ||
| 
             | 
        80 | // effective size of the buffer | |
| 
             | 
        81 | var size = m_buffer.Length - m_offset; | |
| 
             | 
        82 | ||
| 
             | 
        83 | // calculate the new size | |
| 
             | 
        84 | int grow = Math.Min(m_limit - size, m_chunk); | |
| 
             | 
        85 | if (grow <= 0) | |
| 
             | 
        86 | throw new ParserException(String.Format("Input buffer {0} bytes limit exceeded", m_limit)); | |
| 
             | 
        87 | ||
| 
             | 
        88 | var temp = new char[size + grow]; | |
| 
             | 
        89 | Array.Copy(m_buffer, m_offset, temp, 0, m_hi - m_offset); | |
| 
             | 
        90 | m_position -= m_offset; | |
| 
             | 
        91 | m_hi -= m_offset; | |
| 
             | 
        92 | m_offset = 0; | |
| 
             | 
        93 | m_buffer = temp; | |
| 
             | 
        94 | ||
| 
             | 
        95 | return free + grow; | |
| 
             | 
        96 | } | |
| 
             | 
        97 | ||
| 
             | 
        98 | public void RaiseMark(int size) { | |
| 
             | 
        99 | m_hi += size; | |
| 
             | 
        18 | get { return m_pos; } | |
| 
             | 
        100 | 19 | } | 
| 
             | 
        101 | 20 | |
| 
             | 
        102 | 21 | /// <summary> | 
| 
             | 
        103 | 22 | /// Scan this instance. | 
| 
             | 
        104 | 23 | /// </summary> | 
| 
             | 
        105 | 24 | /// <returns><c>true</c> - additional data required</returns> | 
| 
             | 
        106 | public bool Scan() { | |
| 
             | 
        107 | 
            
                          | 
    |
| 
             | 
        108 | 
            
                          | 
    |
| 
             | 
        109 | var next = m_dfa[m_state].transitions[(int)ch]; | |
| 
             | 
        25 | public bool Scan(int[] buffer, int position, int length) { | |
| 
             | 
        26 | var hi = position + length; | |
| 
             | 
        27 | m_pos = position; | |
| 
             | 
        28 | ||
| 
             | 
        29 | while (position < hi) { | |
| 
             | 
        30 | var next = m_dfa[m_state].transitions[buffer[position]]; | |
| 
             | 
        110 | 31 | if (next == DFAConst.UNREACHABLE_STATE) { | 
| 
             | 
        111 | 32 | if (m_dfa[m_state].final) | 
| 
             | 
        112 | 33 | return false; | 
| 
             | 
        113 | 34 | |
| 
             | 
        114 | 35 | throw new ParserException( | 
| 
             | 
        115 | 36 | String.Format( | 
| 
             | 
        116 | 
            
                                         "Unexpected  | 
    |
| 
             | 
        117 | new string(m_buffer, m_offset, m_position - m_offset) | |
| 
             | 
        37 | "Unexpected symbol" | |
| 
             | 
        118 | 38 | ) | 
| 
             | 
        119 | 39 | ); | 
| 
             | 
        120 | 40 | } | 
| 
             | 
        41 | m_pos++; | |
| 
             | 
        121 | 42 | m_state = next; | 
| 
             | 
        122 | m_position++; | |
| 
             | 
        123 | 43 | } | 
| 
             | 
        124 | 44 | |
| 
             | 
        125 | 45 | return true; | 
| @@ -129,8 +49,7 namespace Implab.Formats { | |||
| 
             | 
        129 | 49 | if (!m_dfa[m_state].final) | 
| 
             | 
        130 | 50 | throw new ParserException( | 
| 
             | 
        131 | 51 | String.Format( | 
| 
             | 
        132 | 
            
                                     "Unexpected  | 
    |
| 
             | 
        133 | new string(m_buffer, m_offset, m_position - m_offset) | |
| 
             | 
        52 | "Unexpected EOF" | |
| 
             | 
        134 | 53 | ) | 
| 
             | 
        135 | 54 | ); | 
| 
             | 
        136 | 55 | } | 
| @@ -4,11 +4,11 using Implab.Components; | |||
| 
             | 
        4 | 4 | namespace Implab.Formats { | 
| 
             | 
        5 | 5 | public abstract class TextScanner<TTag> : Disposable { | 
| 
             | 
        6 | 6 | |
| 
             | 
        7 | 
            
                      | 
    |
| 
             | 
        8 | 
            
                     int m_ | 
    |
| 
             | 
        9 | 
            
                     int m_ | 
    |
| 
             | 
        10 | int m_tokenOffset; | |
| 
             | 
        7 | readonly int[] m_buffer; | |
| 
             | 
        8 | int m_bufferOffset; | |
| 
             | 
        9 | int m_dataLength; | |
| 
             | 
        11 | 10 | int m_tokenLength; | 
| 
             | 
        11 | ||
| 
             | 
        12 | 12 | TTag[] m_tags; | 
| 
             | 
        13 | 13 | |
| 
             | 
        14 | 14 | BufferScanner<TTag> m_scanner; | 
| @@ -17,29 +17,36 namespace Implab.Formats { | |||
| 
             | 
        17 | 17 | if (EOF) | 
| 
             | 
        18 | 18 | return false; | 
| 
             | 
        19 | 19 | |
| 
             | 
        20 | 
            
                         // create a new scanner from template (scanners are  | 
    |
| 
             | 
        20 | // create a new scanner from template (scanners are value types) | |
| 
             | 
        21 | 21 | var inst = m_scanner; | 
| 
             | 
        22 | 22 | |
| 
             | 
        23 | // initialize the scanner | |
| 
             | 
        24 | inst.Init(m_buffer, m_offset, m_length); | |
| 
             | 
        23 | m_tokenLength = 0; | |
| 
             | 
        24 | ||
| 
             | 
        25 | while (inst.Scan(m_buffer, m_bufferOffset, m_dataLength)) { | |
| 
             | 
        26 | m_tokenLength += m_dataLength; | |
| 
             | 
        27 | ||
| 
             | 
        28 | var actual = Read(m_buffer, 0, m_buffer.Length); | |
| 
             | 
        29 | ||
| 
             | 
        30 | m_bufferOffset = 0; | |
| 
             | 
        31 | m_dataLength = actual; | |
| 
             | 
        25 | 32 | |
| 
             | 
        26 | // do work | |
| 
             | 
        27 | 
            
                          | 
    |
| 
             | 
        28 | 
            
                              | 
    |
| 
             | 
        33 | if (actual == 0) { | |
| 
             | 
        34 | inst.Eof(); | |
| 
             | 
        35 | break; | |
| 
             | 
        36 | } | |
| 
             | 
        37 | } | |
| 
             | 
        38 | ||
| 
             | 
        39 | var len = inst.Position - m_bufferOffset; | |
| 
             | 
        40 | m_tokenLength += len; | |
| 
             | 
        41 | m_dataLength -= len; | |
| 
             | 
        42 | m_bufferOffset = inst.Position; | |
| 
             | 
        29 | 43 | |
| 
             | 
        30 | 44 | // save result; | 
| 
             | 
        31 | m_buffer = inst.Buffer; | |
| 
             | 
        32 | m_length = inst.Length; | |
| 
             | 
        33 | m_offset = inst.Position; | |
| 
             | 
        34 | m_tokenOffset = inst.TokenOffset; | |
| 
             | 
        35 | m_tokenLength = inst.TokenLength; | |
| 
             | 
        36 | 45 | |
| 
             | 
        37 | 46 | m_tags = inst.GetTokenTags(); | 
| 
             | 
        38 | 47 | } | 
| 
             | 
        39 | 48 | |
| 
             | 
        40 | protected string GetToken() { | |
| 
             | 
        41 | return new String(m_buffer, m_tokenOffset, m_tokenLength); | |
| 
             | 
        42 | } | |
| 
             | 
        49 | protected abstract int Read(int[] buffer, int offset, int size); | |
| 
             | 
        43 | 50 | |
| 
             | 
        44 | 51 | protected TTag[] Tags { | 
| 
             | 
        45 | 52 | get { | 
| @@ -47,24 +54,6 namespace Implab.Formats { | |||
| 
             | 
        47 | 54 | } | 
| 
             | 
        48 | 55 | } | 
| 
             | 
        49 | 56 | |
| 
             | 
        50 | /// <summary> | |
| 
             | 
        51 | /// Feed the specified scanner. | |
| 
             | 
        52 | /// </summary> | |
| 
             | 
        53 | /// <param name="scanner">Scanner.</param> | |
| 
             | 
        54 | /// <example> | |
| 
             | 
        55 | /// protected override void Feed(ref BufferScanner<TTag> scanner) { | |
| 
             | 
        56 | /// var size = scanner.Extend(); | |
| 
             | 
        57 | /// var actual = m_reader.Read(scanner.Buffer, scanner.HiMark, size); | |
| 
             | 
        58 | /// if (actual == 0) { | |
| 
             | 
        59 | /// m_eof = true; | |
| 
             | 
        60 | /// scanner.Eof(); | |
| 
             | 
        61 | /// } else { | |
| 
             | 
        62 | /// scanner.RaiseHiMark(actual); | |
| 
             | 
        63 | /// } | |
| 
             | 
        64 | /// } | |
| 
             | 
        65 | /// </example> | |
| 
             | 
        66 | protected abstract void Feed(ref BufferScanner<TTag> scanner); | |
| 
             | 
        67 | ||
| 
             | 
        68 | 57 | public abstract bool EOF { get; } | 
| 
             | 
        69 | 58 | |
| 
             | 
        70 | 59 | } | 
        
        General Comments 0
    
    
  
  
                      You need to be logged in to leave comments.
                      Login now
                    
                