src/edu/berkeley/sbp/tib/Tib.java

   1 // Copyright 2005 the Contributors, as shown in the revision logs.
   2 // Licensed under the Apache Public Source License 2.0 ("the License").
   3 // You may not use this file except in compliance with the License.
   4
   5 package edu.berkeley.sbp.tib;
   6 import edu.berkeley.sbp.*;
   7 import edu.berkeley.sbp.misc.*;
   8 import java.util.*;
   9 import java.io.*;
  10
  11 // TODO: multiple {{ }} for superquotation
  12 // TODO: strings
  13 // TODO: comments
  14
  15 /**
  16  *   A slow, ugly, inefficient, inelegant, ad-hoc parser for TIB files.
  17  *
  18  *   Despite being inelegant, this parser keeps all internal state on
  19  *   the stack (the only heap objects created are those which are
  20  *   returned).
  21  *
  22  *   This was written as an ad-hoc parser to facilitate
  23  *   experimentation with the TIB spec.  Once the spec is finalized it
  24  *   should probably be rewritten.
  25  */
  26 public class Tib implements Token.Stream<CharToken> {
  27
  28     public Tib(String s) throws IOException, Invalid { this(new StringReader(s)); }
  29     public Tib(Reader r) throws IOException, Invalid { this(new BufferedReader(r)); }
  30     public Tib(InputStream is) throws IOException, Invalid { this(new BufferedReader(new InputStreamReader(is))); }
  31     public Tib(BufferedReader br) throws IOException, Invalid {
  32         this.br = br;
  33         istack.add(-1);
  34         //cur = parse(br);
  35         //System.out.println("\rparsing: \"" + cur.toString(0, -1) + "\"");
  36     }
  37
  38     private Block cur;
  39     private String s = "";
  40     int pos = 0;
  41     int spos = 0;
  42
  43     int _row = 1;
  44     int _col = 0;
  45     public Token.Location getLocation() { return new CharToken.CartesianLocation(_row, _col); }
  46     private BufferedReader br;
  47
  48     boolean waiting = false;
  49     char waitingChar = ' ';
  50     boolean indenting = true;
  51     int indentation = 0;
  52     private ArrayList<Integer> istack = new ArrayList<Integer>();
  53     public CharToken next(int numstates) throws IOException {
  54         CharToken ret = nextc(numstates);
  55         if      (ret==CharToken.left)  System.out.print("\033[31m{\033[0m");
  56         else if (ret==CharToken.right) System.out.print("\033[31m}\033[0m");
  57         else if (ret==null) return null;
  58         else System.out.print(ret.c);
  59         return ret;
  60     }
  61
  62     CharToken waitingBrace = null;
  63     public CharToken nextc(int numstates) throws IOException {
  64         char c;
  65         if (waitingBrace != null) {
  66             CharToken ret = waitingBrace;
  67             waitingBrace = null;
  68             return ret;
  69         }
  70         if (waiting) {
  71             waiting = false;
  72             c = waitingChar;
  73         } else {
  74             int i = br.read();
  75             if (i==-1) {
  76                 if (istack.size() > 1) {
  77                     istack.remove(istack.size()-1);
  78                     return CharToken.right;
  79                 }
  80                 return null;
  81             }
  82             c = (char)i;
  83         }
  84         if (c=='\n') { _row++; _col=0; }
  85         else         _col++;
  86         if (indenting) {
  87             if (c==' ') { indentation++; return done(c); }
  88             if (c=='\n') { indentation = 0; if (blank) return nextc(numstates); blank = true; waiting = true; waitingChar='\n'; return new CharToken('\n'); }
  89             int last = istack.size()==0 ? -1 : istack.get(istack.size()-1);
  90             if (indentation==last) {
  91                 if (blank) {
  92                     indenting = false;
  93                     waitingChar = c;
  94                     waiting = true;
  95                     waitingBrace = CharToken.left;
  96                     return CharToken.right;
  97                     //return nextc(numstates);
  98                 }
  99                 blank = false;
 100                 indenting = false;
 101                 return done(c);
 102             }
 103             blank = false;
 104             waitingChar = c;
 105             waiting = true;
 106             if (indentation > last) {
 107                 indenting = false;
 108                 istack.add(indentation);
 109                 System.out.print("\033[31m+"+indentation+"+\033[0m");
 110                 return CharToken.left;
 111             } else /*if (indentation < last)*/ {
 112                 istack.remove(istack.size()-1);
 113                 System.out.print("\033[31m-"+last+"-\033[0m");
 114                 blank = true;
 115                 return CharToken.right;
 116             }
 117         } else {
 118             blank = false;
 119             if (c=='\n') { indenting=true; indentation = 0; }
 120             return done(c);
 121         }
 122     }
 123     public CharToken done(char c) {
 124         switch(c) {
 125             case '{': return CharToken.left;
 126             case '}': return CharToken.right;
 127             default: return new CharToken(c);
 128         }
 129     }
 130     boolean blank = false;
 131     /*
 132     public CharToken next(int numstates) throws IOException {
 133         if (cur==null) return null;
 134         if (s != null) {
 135             if (spos < s.length()) {
 136                 char c = s.charAt(spos++);
 137                 if (c=='\n') { _row++; _col = 0; }
 138                 else _col++;
 139                 return new CharToken(c);
 140             }
 141             s = null;
 142         }
 143         if (pos >= cur.size()) {
 144             pos = cur.iip+1;
 145             _row = cur.endrow;
 146             _col = cur.endcol;
 147             cur = cur.parent;
 148             if (cur==null) return null;
 149             return CharToken.right;
 150         }
 151         Object o = cur.child(pos++);
 152         if (o instanceof String) {
 153             spos = 0;
 154             s = (String)o;
 155             return next(numstates);
 156         }
 157         if (o instanceof Block) {
 158             Block b = (Block)o;
 159             _row = b.row;
 160             _col = b.col;
 161         }
 162         if (((Block)o).isLiteral()) {
 163             spos = 0;
 164             s = ((Block.Literal)o).text();
 165             return next(numstates);
 166         }
 167         cur = (Block)o;
 168         pos = 0;
 169         return CharToken.left;
 170     }
 171     */
 172     public static Block parse(BufferedReader br) throws Invalid, IOException {
 173         int row=0, col=0;
 174         try {
 175             boolean blankLine = false;
 176             Block top = new Block.Root();
 177             for(String s = br.readLine(); s != null; s = br.readLine()) {
 178                 row++;
 179                 col=0;
 180                 while (s.length() > 0 &&
 181                        s.charAt(0) == ' ' &&
 182                        (!(top instanceof Block.Literal) || col < top.col)) { col++; s = s.substring(1); }
 183                 if ((top instanceof Block.Literal) && col >= top.col) { top.add(s); continue; }
 184                 if (s.length()==0)                                    { blankLine = true; continue; }
 185                 while (col < top.col) {
 186                     if (s.startsWith("{}") && top instanceof Block.Literal && ((Block.Literal)top).braceCol == col) break;
 187                     blankLine = false;
 188                     top.endrow = row;
 189                     top.endcol = col;
 190                     top = top.closeIndent();
 191                 }
 192                 if (s.startsWith("{}")) {
 193                     int bc = col;
 194                     boolean append = top instanceof Block.Literal && ((Block.Literal)top).braceCol == bc;
 195                     s = s.substring(2);
 196                     col += 2;
 197                     while (s.length() > 0 && s.charAt(0) == ' ' && !(append && col >= top.col) ) { col++; s = s.substring(1); }
 198                     if (append) top.add(s); else (top = new Block.Literal(top, row, col, bc)).add(s);
 199                     continue;
 200                 }
 201                 while (s.length() > 0 && s.charAt(s.length()-1)==' ') { s = s.substring(0, s.length()-1); }
 202                 if (col > top.col) top = new Block.Indent(top, row, col);
 203                 else if (blankLine) { top.endrow=row; top.endcol=col; top = top.closeIndent(); top = new Block.Indent(top, row, col); }
 204                 blankLine = false;
 205                 for(int i=0; i<s.length(); i++) {
 206                     top.add(s.charAt(i));
 207                     switch(s.charAt(i)) {
 208                         case '{':  top = new Block.Brace(top, row, col);   break;
 209                         case '}':  top.endrow=row; top.endcol=col; top = top.closeBrace();                 break;
 210                     }
 211                 }
 212                 top.add('\n');
 213                 top.finishWord();
 214             }
 215             // FIXME
 216             Block ret = top;
 217             while (ret.parent != null) ret = ret.parent;
 218             return ret;
 219         } catch (InternalException ie) {
 220             throw new Invalid(ie, row, col);
 221         }
 222     }
 223
 224     public static class Block {
 225                       Block  parent;
 226         public  final int    row;
 227         public  final int    col;
 228         public   int    endrow;
 229         public   int    endcol;
 230         public final int iip;
 231         private final Vector children = new Vector();
 232         private       String pending  = "";
 233
 234         public int    size() { return children.size(); }
 235         public Object child(int i) { return children.elementAt(i); }
 236         public boolean isLiteral() {  return false; }
 237
 238         protected Block(int row, int col)                {
 239             this.row=row;
 240             this.col=col;
 241             this.iip = -1;
 242             parent = null;
 243         }
 244         public    Block(Block parent, int row, int col)  {
 245             this.row=row;
 246             this.col=col;
 247             this.iip = parent.size();
 248             (this.parent = parent).add(this);
 249         }
 250
 251         public void    add(String s)        { children.addElement(s); }
 252         public void    add(char c)          {
 253             if (c == '{' || c == '}') { finishWord(); return; }
 254             if (c != ' ') { pending += c; return; }
 255             if (pending.length() > 0) { finishWord(); add(" "); return; }
 256             if (size()==0) return;
 257             if (child(size()-1).equals(" ")) return;
 258             add(" ");
 259             return;
 260         }
 261
 262         public void    add(Block  b)        { children.addElement(b); }
 263         public Block   promote()            { parent.parent.replaceLast(this); return close(); }
 264         public Object  lastChild()          { return children.lastElement(); }
 265         public Block   lastChildAsBlock()   { return (Block)lastChild(); }
 266         public void    replaceLast(Block b) { children.setElementAt(b, children.size()-1); b.parent = this; }
 267
 268         public void    finishWord()         { if (pending.length() > 0) { add(pending); pending = ""; } }
 269
 270         public Block   closeBrace()         { throw new InternalException("attempt to closeBrace() a "+getClass().getName()); }
 271         public Block   closeIndent()        { throw new InternalException("attempt to closeIndent() a "+getClass().getName()); }
 272         public Block   close() {
 273             while(size() > 0 && child(size()-1).equals(" ")) children.setSize(children.size()-1);
 274             if (size()==0) throw new InternalException("PARSER BUG: attempt to close an empty block (should never happen)");
 275             if (size() > 1 || !(lastChild() instanceof Block)) return parent;
 276             return lastChildAsBlock().promote();
 277         }
 278         public String toString() { return toString(80); }
 279         public String toString(int justificationLimit) { return toString(0, 80); }
 280         protected String toString(int indent, int justificationLimit) {
 281             StringBuffer ret = new StringBuffer();
 282             StringBuffer line = new StringBuffer();
 283             for(int i=0; i<children.size(); i++) {
 284                 Object o = children.elementAt(i);
 285                 if (i>0 && children.elementAt(i-1) instanceof Block && justificationLimit!=-1) ret.append("\n");
 286                 if (o instanceof Block) {
 287                     ret.append(justify(line.toString(), indent, justificationLimit));
 288                     line.setLength(0);
 289                     if (justificationLimit==-1) {
 290                         ret.append("{");
 291                         ret.append(((Block)o).toString(indent+2, justificationLimit));
 292                         ret.append("}");
 293                     } else {
 294                         ret.append(((Block)o).toString(indent+2, justificationLimit));
 295                     }
 296                 } else {
 297                     line.append(o.toString());
 298                 }
 299             }
 300             ret.append(justify(line.toString(), indent, justificationLimit));
 301             return ret.toString();
 302         }
 303
 304         private static class Root extends Block {
 305             public Root() { super(0, Integer.MIN_VALUE); }
 306             public Block close() { throw new InternalException("attempted to close top block"); }
 307             public String toString(int justificationLimit) { return toString(-2, justificationLimit); }
 308         }
 309
 310         private static class Brace  extends Block {
 311             public Brace(Block parent, int row, int col) { super(parent, row, col); }
 312             public Block closeBrace() { return super.close(); }
 313         }
 314
 315         private static class Indent extends Block {
 316             public Indent(Block parent, int row, int col) { super(parent, row, col); }
 317             public Block closeIndent() { return super.close(); }
 318         }
 319
 320         private static class Literal extends Block {
 321             private StringBuffer content = new StringBuffer();
 322             public final int braceCol;
 323             public    Literal(Block parent, int row, int col, int braceCol) { super(parent,row,col); this.braceCol = braceCol; }
 324             public    boolean isLiteral() {  return true; }
 325             public    int size() { return 1; }
 326             public    Object child(int i) { return i==0 ? content.toString() : null; }
 327             public    Block  close() { return parent; }
 328             public    Block  closeIndent() { return close(); }
 329             public    void   add(String s) { if (content.length()>0) content.append('\n'); content.append(s); }
 330             public    String text() { return content.toString(); }
 331             protected String toString(int indent, int justificationLimit) {
 332                 StringBuffer ret = new StringBuffer();
 333                 String s = content.toString();
 334                 while(s.length() > 0) {
 335                     int nl = s.indexOf('\n');
 336                     if (nl==-1) nl = s.length();
 337                     ret.append(spaces(indent));
 338                     ret.append("{}  ");
 339                     ret.append(s.substring(0, nl));
 340                     s = s.substring(Math.min(s.length(),nl+1));
 341                     ret.append('\n');
 342                 }
 343                 return ret.toString();
 344             }
 345         }
 346     }
 347
 348
 349     // Exceptions //////////////////////////////////////////////////////////////////////////////
 350
 351     private static class InternalException extends RuntimeException { public InternalException(String s) { super(s); } }
 352     public static class Invalid extends /*IOException*/RuntimeException {
 353         public Invalid(InternalException ie, int row, int col) {
 354             super(ie.getMessage() + " at " + row + ":" + col);
 355         }
 356     }
 357
 358     // Testing //////////////////////////////////////////////////////////////////////////////
 359
 360     public static void main(String[] s) throws Exception {
 361         System.out.println(parse(new BufferedReader(new InputStreamReader(System.in))).toString(-1)); }
 362
 363     // Utilities //////////////////////////////////////////////////////////////////////////////
 364
 365     public static String spaces(int i) { if (i<=0) return ""; return " " + spaces(i-1); }
 366
 367     private static String justify(String s, int indent, int justificationLimit) {
 368         if (s.length() == 0) return "";
 369         if (justificationLimit==-1) return s;
 370         StringBuffer ret = new StringBuffer();
 371         while(s.length() > 0) {
 372             if (s.charAt(0) == ' ') { s = s.substring(1); continue; }
 373             ret.append(spaces(indent));
 374             int i = s.indexOf(' ');
 375             if (i==-1) i = s.length();
 376             while(s.indexOf(' ', i+1) != -1 && s.indexOf(' ', i+1) < justificationLimit-indent) i = s.indexOf(' ', i+1);
 377             if (s.length() + indent < justificationLimit) i = s.length();
 378             ret.append(s.substring(0, i));
 379             s = s.substring(i);
 380             ret.append('\n');
 381         }
 382         return ret.toString();
 383     }
 384
 385     // Grammar //////////////////////////////////////////////////////////////////////////////
 386
 387     public static class Grammar extends MetaGrammar {
 388         private int anon = 0;
 389         private final Element ws = Repeat.maximal0(nonTerminal("w"));
 390         public Grammar() { dropAll.add(ws); }
 391         public Object walk(Tree<String> tree) {
 392             String head = tree.head();
 393             if (tree.numChildren()==0) return super.walk(tree);
 394             if ("{".equals(head)) {
 395                 String s = "braced"+(anon++);
 396                 Union u = nonTerminal(s);
 397                 Union u2 = ((PreSequence)walk(tree, 0)).sparse(ws).buildUnion();
 398                 u2.add(Sequence.singleton(new Element[] { u }, 0, null, null));
 399                 return nonTerminal(s,
 400                                    new PreSequence[][] {
 401                                        new PreSequence[] {
 402                                            new PreSequence(new Element[] { CharToken.leftBrace,
 403                                                                            ws,
 404                                                                            u2,
 405                                                                            ws,
 406                                                                            CharToken.rightBrace
 407                                            })
 408                                        }
 409                                    },
 410                                    false,
 411                                    false);
 412             }
 413             return super.walk(tree);
 414         }
 415     }
 416
 417 }
 418