Skip to content

Commit 002d166

Browse files
rollratrollrat
authored
Update ScannerGenerator.cs
1 parent f33098d commit 002d166

File tree

1 file changed

+77
-21
lines changed

1 file changed

+77
-21
lines changed

‎ScannerGenerator.cs‎

Lines changed: 77 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
/*
1+
/*
22
33
Copyright (C) 2019. rollrat All Rights Reserved.
44
@@ -24,7 +24,9 @@ public class SimpleRegex
2424
public SimpleRegex() { }
2525
public List<string> build_errors = new List<string>();
2626
public diagram Diagram;
27-
public const char e_closure = (char)0;
27+
public const char e_closure = (char)0xFFFF;
28+
29+
public const int byte_size = 256;
2830

2931
public class transition_node
3032
{
@@ -225,6 +227,14 @@ private diagram make_nfa(string pattern)
225227
case 'r':
226228
ch_list.Add('\t');
227229
break;
230+
case 'x':
231+
char ch2;
232+
ch2 = (char)(pattern[i + 1] >= 'A' ? (pattern[i + 1] - 'A' + 10) : pattern[i + 1] - '0');
233+
ch2 <<= 4;
234+
ch2 |= (char)(pattern[i + 2] >= 'A' ? (pattern[i + 2] - 'A' + 10) : pattern[i + 2] - '0');
235+
i += 2;
236+
ch_list.Add(ch2);
237+
break;
228238

229239
default:
230240
build_errors.Add($"{pattern[i]} escape character not found!");
@@ -245,11 +255,11 @@ private diagram make_nfa(string pattern)
245255
var ends_point2 = new transition_node { index = index_count++, transition = new List<Tuple<char, transition_node>>() };
246256
if (inverse)
247257
{
248-
var set = new bool[128];
258+
var set = new bool[byte_size];
249259
var nch_list = new List<char>();
250260
foreach (var ch2 in ch_list)
251261
set[ch2] = true;
252-
for (int j = 0; j < 128; j++)
262+
for (int j = 0; j < byte_size; j++)
253263
if (!set[j])
254264
nch_list.Add((char)j);
255265
ch_list.Clear();
@@ -270,7 +280,7 @@ private diagram make_nfa(string pattern)
270280

271281
case '.':
272282
var ends_point3 = new transition_node { index = index_count++, transition = new List<Tuple<char, transition_node>>() };
273-
for( int i2 = 0; i2 < 128; i2++)
283+
for( int i2 = 0; i2 < byte_size; i2++)
274284
{
275285
cur.transition.Add(new Tuple<char, transition_node>((char)i2, ends_point3));
276286
}
@@ -293,7 +303,7 @@ private diagram make_nfa(string pattern)
293303
ch = pattern[i];
294304
else
295305
{
296-
switch (ch)
306+
switch (pattern[i])
297307
{
298308
case 'n':
299309
ch = '\n';
@@ -304,9 +314,16 @@ private diagram make_nfa(string pattern)
304314
case 'r':
305315
ch = '\r';
306316
break;
317+
case 'x':
318+
ch = (char)(pattern[i + 1] >= 'A' ? (pattern[i + 1] - 'A' + 10) : pattern[i + 1] - '0');
319+
ch <<= 4;
320+
ch |= (char)(pattern[i + 2] >= 'A' ? (pattern[i + 2] - 'A' + 10) : pattern[i + 2] - '0');
321+
i += 2;
322+
break;
307323

308324
default:
309325
build_errors.Add($"{pattern[i]} escape character not found!");
326+
ch = pattern[i];
310327
break;
311328
}
312329

@@ -557,7 +574,7 @@ private bool opt_nfa(diagram dia)
557574
check[tn.index] = true;
558575

559576
// Delete unnecessary e-closure with pull left
560-
if (tn.transition.Count == 1 && tn.transition[0].Item1 == 0)
577+
if (tn.transition.Count == 1 && tn.transition[0].Item1 == e_closure)
561578
{
562579
var index_left = tn.index;
563580
var index_right = tn.transition[0].Item2.index;
@@ -576,14 +593,14 @@ private bool opt_nfa(diagram dia)
576593

577594
// Delete recursive e-closure
578595
for (int i = 0; i < tn.transition.Count; i++)
579-
if (tn.transition[i].Item1 == 0 && tn.transition[i].Item2.index == tn.index)
596+
if (tn.transition[i].Item1 == e_closure && tn.transition[i].Item2.index == tn.index)
580597
tn.transition.RemoveAt(i--);
581598

582599
// Merge rounding e-closure
583600
for (int i = 0; i < tn.transition.Count; i++)
584-
if (tn.transition[i].Item1 == 0)
601+
if (tn.transition[i].Item1 == e_closure)
585602
for (int j = 0; j < tn.transition[i].Item2.transition.Count; j++)
586-
if (tn.transition[i].Item2.transition[j].Item1 == 0 && tn.transition[i].Item2.transition[j].Item2.index == tn.index)
603+
if (tn.transition[i].Item2.transition[j].Item1 == e_closure && tn.transition[i].Item2.transition[j].Item2.index == tn.index)
587604
{
588605
var index_left = tn.index;
589606
var index_right = tn.transition[i].Item2.index;
@@ -615,7 +632,7 @@ private bool opt_nfa(diagram dia)
615632
var index_right = tn.index;
616633

617634
for (int i = 0; i < dia.nodes[index_left].transition.Count; i++)
618-
if (dia.nodes[index_left].transition[i].Item2.index == dia.nodes[index_right].index && dia.nodes[index_left].transition[i].Item1 == 0)
635+
if (dia.nodes[index_left].transition[i].Item2.index == dia.nodes[index_right].index && dia.nodes[index_left].transition[i].Item1 == e_closure)
619636
{
620637
if (dia.nodes[index_left].transition[i].Item2.is_acceptable)
621638
{
@@ -647,7 +664,7 @@ private bool opt_nfa(diagram dia)
647664
dia.nodes[top].is_acceptable = true;
648665
if (inverse_transition.ContainsKey(top))
649666
foreach (var inv in inverse_transition[top])
650-
if (dia.nodes[inv].transition.Where(x => x.Item2.index == top).First().Item1 == 0)
667+
if (dia.nodes[inv].transition.Where(x => x.Item2.index == top).First().Item1 == e_closure)
651668
acc_nodes.Enqueue(inv);
652669
}
653670

@@ -698,11 +715,11 @@ private diagram nfa2dfa(diagram dia)
698715
var check = new List<bool>(dia.count_of_vertex);
699716
check.AddRange(Enumerable.Repeat(false, dia.count_of_vertex));
700717
var e_q = new Queue<int>();
701-
d_q.ToList().Where(qe => qe.Item1 == 0).ToList().ForEach(qee => { e_q.Enqueue(qee.Item2); });
718+
d_q.ToList().Where(qe => qe.Item1 == e_closure).ToList().ForEach(qee => { e_q.Enqueue(qee.Item2); });
702719

703720
foreach (var qe in d_q)
704721
{
705-
if (qe.Item1 == 0)
722+
if (qe.Item1 == e_closure)
706723
e_q.Enqueue(qe.Item2);
707724
else
708725
check[qe.Item2] = true;
@@ -714,7 +731,7 @@ private diagram nfa2dfa(diagram dia)
714731
if (check[d]) continue;
715732
check[d] = true;
716733
foreach (var tns in dia.nodes[d].transition)
717-
if (tns.Item1 == 0)
734+
if (tns.Item1 == e_closure)
718735
e_q.Enqueue(tns.Item2.index);
719736
else
720737
d_q.Enqueue(new Tuple<char, int>(tns.Item1, tns.Item2.index));
@@ -725,13 +742,13 @@ private diagram nfa2dfa(diagram dia)
725742
while (d_q.Count != 0)
726743
{
727744
var dd = d_q.Dequeue();
728-
if (dd.Item1 == 0) continue;
745+
if (dd.Item1 == e_closure) continue;
729746
if (dic.ContainsKey(dd.Item1))
730747
dic[dd.Item1].Add(dd.Item2);
731748
else
732749
dic.Add(dd.Item1, new HashSet<int> { dd.Item2 });
733750
foreach (var node in dia.nodes[dd.Item2].transition)
734-
if (node.Item1 == 0)
751+
if (node.Item1 == e_closure)
735752
dic[dd.Item1].Add(node.Item2.index);
736753
}
737754

@@ -1056,8 +1073,8 @@ public Scanner CreateScannerInstance(string delimiter = "\n\r ")
10561073
var accept_table = new string[diagram.count_of_vertex];
10571074
for (int i = 0; i < table.Length; i++)
10581075
{
1059-
table[i] = new int[255];
1060-
for (int j = 0; j < 255; j++)
1076+
table[i] = new int[SimpleRegex.byte_size];
1077+
for (int j = 0; j < SimpleRegex.byte_size; j++)
10611078
table[i][j] = -1;
10621079
}
10631080

@@ -1080,7 +1097,9 @@ public Scanner CreateScannerInstance(string delimiter = "\n\r ")
10801097
/// </summary>
10811098
public class Scanner
10821099
{
1100+
[JsonProperty]
10831101
int[][] transition_table;
1102+
[JsonProperty]
10841103
string[] accept_table;
10851104
string target;
10861105
int pos = 0;
@@ -1155,7 +1174,7 @@ public Tuple<string, string, int, int> Next()
11551174
err_pos.Add(pos);
11561175
continue;
11571176
}
1158-
return new Tuple<string, string, int, int> (accept_table[node_pos], builder.ToString(), cur_line + 1, cur_column + 1);
1177+
return new Tuple<string, string, int, int>(accept_table[node_pos], builder.ToString(), cur_line + 1, cur_column + 1);
11591178

11601179
default:
11611180
if (target[pos] == '\n') { current_line++; current_column = 1; } else current_column++;
@@ -1167,7 +1186,7 @@ public Tuple<string, string, int, int> Next()
11671186
}
11681187
if (accept_table[node_pos] == null)
11691188
throw new Exception($"[SCANNER] Pattern not found! L:{cur_line}, C:{cur_column}, D:'{builder.ToString()}'");
1170-
return new Tuple<string, string, int, int> (accept_table[node_pos], builder.ToString(), cur_line + 1, cur_column + 1);
1189+
return new Tuple<string, string, int, int>(accept_table[node_pos], builder.ToString(), cur_line + 1, cur_column + 1);
11711190
}
11721191

11731192
public Tuple<string, string, int, int> Lookahead()
@@ -1177,5 +1196,42 @@ public Tuple<string, string, int, int> Lookahead()
11771196
pos = npos;
11781197
return result;
11791198
}
1199+
1200+
public static Scanner FromString(string json)
1201+
=> JsonConvert.DeserializeObject<Scanner>(json);
1202+
public override string ToString()
1203+
=> JsonConvert.SerializeObject(this, Formatting.None);
1204+
public string ToCSCode(string class_name)
1205+
{
1206+
var builder = new StringBuilder();
1207+
var indent = "";
1208+
Action up_indent = () => { indent += " "; };
1209+
Action down_indent = () => { if (indent.Length > 0) indent = indent.Substring(4); };
1210+
Action<string> append = (string s) => { builder.Append($"{indent}{s}\r\n"); };
1211+
append("public class " + class_name);
1212+
append("{");
1213+
up_indent();
1214+
1215+
///////////////////
1216+
append("int[][] transition_table = new int[][] {");
1217+
up_indent();
1218+
foreach (var gt in transition_table)
1219+
append("new int[] {" + string.Join(",", gt.Select(x => x.ToString().PadLeft(4))) + " },");
1220+
down_indent();
1221+
append("};");
1222+
append("");
1223+
1224+
///////////////////
1225+
append("string[] accept_table = new string[] {");
1226+
up_indent();
1227+
append(string.Join(",", accept_table.Select(x => x != null ? $"\"{x.ToString().PadLeft(4)}\"" : "null")));
1228+
down_indent();
1229+
append("};");
1230+
append("");
1231+
1232+
down_indent();
1233+
append("}");
1234+
return builder.ToString();
1235+
}
11801236
}
11811237
}

0 commit comments

Comments
 (0)