Source code

001package org.hl7.fhir.dstu3.utils;
002
003import org.hl7.fhir.dstu3.model.ExpressionNode;
004import org.hl7.fhir.dstu3.model.ExpressionNode.SourceLocation;
005import org.hl7.fhir.exceptions.FHIRException;
006import org.hl7.fhir.utilities.Utilities;
007
008// shared lexer for concrete syntaxes 
009// - FluentPath
010// - Mapping language
011
012public class FHIRLexer {
013  public class FHIRLexerException extends FHIRException {
014
015    public FHIRLexerException() {
016      super();
017    }
018
019    public FHIRLexerException(String message, Throwable cause) {
020      super(message, cause);
021    }
022
023    public FHIRLexerException(String message) {
024      super(message);
025    }
026
027    public FHIRLexerException(Throwable cause) {
028      super(cause);
029    }
030
031  }
032  private String source;
033  private int cursor;
034  private int currentStart;
035  private String current;
036  private SourceLocation currentLocation;
037  private SourceLocation currentStartLocation;
038  private int id;
039
040  public FHIRLexer(String source) throws FHIRLexerException {
041    this.source = source;
042    currentLocation = new SourceLocation(1, 1);
043    next();
044  }
045  public String getCurrent() {
046    return current;
047  }
048  public SourceLocation getCurrentLocation() {
049    return currentLocation;
050  }
051
052  public boolean isConstant(boolean incDoubleQuotes) {
053    return current.charAt(0) == '\'' || (incDoubleQuotes && current.charAt(0) == '"') || current.charAt(0) == '@' || current.charAt(0) == '%' || 
054        current.charAt(0) == '-' || current.charAt(0) == '+' || (current.charAt(0) >= '0' && current.charAt(0) <= '9') || 
055        current.equals("true") || current.equals("false") || current.equals("{}");
056  }
057
058  public boolean isStringConstant() {
059    return current.charAt(0) == '\'' || current.charAt(0) == '"';
060  }
061
062  public String take() throws FHIRLexerException {
063    String s = current;
064    next();
065    return s;
066  }
067
068  public int takeInt() throws FHIRLexerException {
069    String s = current;
070    if (!Utilities.isInteger(s))
071      throw error("Found "+current+" expecting an integer");
072    next();
073    return Integer.parseInt(s);
074  }
075
076  public boolean isToken() {
077    if (Utilities.noString(current))
078      return false;
079
080    if (current.startsWith("$"))
081      return true;
082
083    if (current.equals("*") || current.equals("**"))
084      return true;
085
086    if ((current.charAt(0) >= 'A' && current.charAt(0) <= 'Z') || (current.charAt(0) >= 'a' && current.charAt(0) <= 'z')) {
087      for (int i = 1; i < current.length(); i++) 
088        if (!( (current.charAt(1) >= 'A' && current.charAt(1) <= 'Z') || (current.charAt(1) >= 'a' && current.charAt(1) <= 'z') ||
089            (current.charAt(1) >= '0' && current.charAt(1) <= '9')))
090          return false;
091      return true;
092    }
093    return false;
094  }
095
096  public FHIRLexerException error(String msg) {
097    return error(msg, currentLocation.toString());
098  }
099
100  public FHIRLexerException error(String msg, String location) {
101    return new FHIRLexerException("Error at "+location+": "+msg);
102  }
103
104  public void next() throws FHIRLexerException {
105    current = null;
106    boolean last13 = false;
107    while (cursor < source.length() && Character.isWhitespace(source.charAt(cursor))) {
108      if (source.charAt(cursor) == '\r') {
109        currentLocation.setLine(currentLocation.getLine() + 1);
110        currentLocation.setColumn(1);
111        last13 = true;
112      } else if (!last13 && (source.charAt(cursor) == '\n')) {
113        currentLocation.setLine(currentLocation.getLine() + 1);
114        currentLocation.setColumn(1);
115        last13 = false;
116      } else {
117        last13 = false;
118        currentLocation.setColumn(currentLocation.getColumn() + 1);
119      }
120      cursor++;
121    }
122    currentStart = cursor;
123    currentStartLocation = currentLocation;
124    if (cursor < source.length()) {
125      char ch = source.charAt(cursor);
126      if (ch == '!' || ch == '>' || ch == '<' || ch == ':' || ch == '-' || ch == '=')  {
127        cursor++;
128        if (cursor < source.length() && (source.charAt(cursor) == '=' || source.charAt(cursor) == '~' || source.charAt(cursor) == '-')) 
129          cursor++;
130        current = source.substring(currentStart, cursor);
131      } else if (ch == '.' ) {
132        cursor++;
133        if (cursor < source.length() && (source.charAt(cursor) == '.')) 
134          cursor++;
135        current = source.substring(currentStart, cursor);
136      } else if (ch >= '0' && ch <= '9') {
137          cursor++;
138        boolean dotted = false;
139        while (cursor < source.length() && ((source.charAt(cursor) >= '0' && source.charAt(cursor) <= '9') || (source.charAt(cursor) == '.') && !dotted)) {
140          if (source.charAt(cursor) == '.')
141            dotted = true;
142          cursor++;
143        }
144        if (source.charAt(cursor-1) == '.')
145          cursor--;
146        current = source.substring(currentStart, cursor);
147      }  else if ((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')) {
148        while (cursor < source.length() && ((source.charAt(cursor) >= 'A' && source.charAt(cursor) <= 'Z') || (source.charAt(cursor) >= 'a' && source.charAt(cursor) <= 'z') || 
149            (source.charAt(cursor) >= '0' && source.charAt(cursor) <= '9') || source.charAt(cursor) == '_')) 
150          cursor++;
151        current = source.substring(currentStart, cursor);
152      } else if (ch == '%') {
153        cursor++;
154        if (cursor < source.length() && (source.charAt(cursor) == '"')) {
155          cursor++;
156          while (cursor < source.length() && (source.charAt(cursor) != '"'))
157            cursor++;
158          cursor++;
159        } else
160        while (cursor < source.length() && ((source.charAt(cursor) >= 'A' && source.charAt(cursor) <= 'Z') || (source.charAt(cursor) >= 'a' && source.charAt(cursor) <= 'z') || 
161            (source.charAt(cursor) >= '0' && source.charAt(cursor) <= '9') || source.charAt(cursor) == ':' || source.charAt(cursor) == '-'))
162          cursor++;
163        current = source.substring(currentStart, cursor);
164      } else if (ch == '/') {
165        cursor++;
166        if (cursor < source.length() && (source.charAt(cursor) == '/')) {
167          cursor++;
168          while (cursor < source.length() && !((source.charAt(cursor) == '\r') || source.charAt(cursor) == '\n')) 
169            cursor++;
170        }
171        current = source.substring(currentStart, cursor);
172      } else if (ch == '$') {
173        cursor++;
174        while (cursor < source.length() && (source.charAt(cursor) >= 'a' && source.charAt(cursor) <= 'z'))
175          cursor++;
176        current = source.substring(currentStart, cursor);
177      } else if (ch == '{') {
178        cursor++;
179        ch = source.charAt(cursor);
180        if (ch == '}')
181          cursor++;
182        current = source.substring(currentStart, cursor);
183      } else if (ch == '"'){
184        cursor++;
185        boolean escape = false;
186        while (cursor < source.length() && (escape || source.charAt(cursor) != '"')) {
187          if (escape)
188            escape = false;
189          else 
190            escape = (source.charAt(cursor) == '\\');
191          cursor++;
192        }
193        if (cursor == source.length())
194          throw error("Unterminated string");
195        cursor++;
196        current = "\""+source.substring(currentStart+1, cursor-1)+"\"";
197      } else if (ch == '\''){
198        cursor++;
199        char ech = ch;
200        boolean escape = false;
201        while (cursor < source.length() && (escape || source.charAt(cursor) != ech)) {
202          if (escape)
203            escape = false;
204          else 
205            escape = (source.charAt(cursor) == '\\');
206          cursor++;
207        }
208        if (cursor == source.length())
209          throw error("Unterminated string");
210        cursor++;
211        current = source.substring(currentStart, cursor);
212        if (ech == '\'')
213          current = "\'"+current.substring(1, current.length() - 1)+"\'";
214      } else if (ch == '@'){
215        cursor++;
216        while (cursor < source.length() && isDateChar(source.charAt(cursor)))
217          cursor++;          
218        current = source.substring(currentStart, cursor);
219      } else { // if CharInSet(ch, ['.', ',', '(', ')', '=', '$']) then
220        cursor++;
221        current = source.substring(currentStart, cursor);
222      }
223    }
224  }
225
226
227  private boolean isDateChar(char ch) {
228    return ch == '-' || ch == ':' || ch == 'T' || ch == '+' || ch == 'Z' || Character.isDigit(ch);
229  }
230  public boolean isOp() {
231    return ExpressionNode.Operation.fromCode(current) != null;
232  }
233  public boolean done() {
234    return currentStart >= source.length();
235  }
236  public int nextId() {
237    id++;
238    return id;
239  }
240  public SourceLocation getCurrentStartLocation() {
241    return currentStartLocation;
242  }
243  
244  // special case use
245  public void setCurrent(String current) {
246    this.current = current;
247  }
248
249  public boolean hasComment() {
250    return !done() && current.startsWith("//");
251  }
252  public boolean hasToken(String kw) {
253    return !done() && kw.equals(current);
254  }
255  public boolean hasToken(String... names) {
256    if (done()) 
257      return false;
258    for (String s : names)
259      if (s.equals(current))
260        return true;
261    return false;
262  }
263  
264  public void token(String kw) throws FHIRLexerException {
265    if (!kw.equals(current)) 
266      throw error("Found \""+current+"\" expecting \""+kw+"\"");
267    next();
268  }
269  
270  public String readConstant(String desc) throws FHIRLexerException {
271    if (!isStringConstant())
272      throw error("Found "+current+" expecting \"["+desc+"]\"");
273
274    return processConstant(take());
275  }
276
277  public String processConstant(String s) throws FHIRLexerException {
278    StringBuilder b = new StringBuilder();
279    int i = 1;
280    while (i < s.length()-1) {
281      char ch = s.charAt(i);
282      if (ch == '\\') {
283        i++;
284        switch (s.charAt(i)) {
285        case 't': 
286          b.append('\t');
287          break;
288        case 'r':
289          b.append('\r');
290          break;
291        case 'n': 
292          b.append('\n');
293          break;
294        case 'f': 
295          b.append('\f');
296          break;
297        case '\'':
298          b.append('\'');
299          break;
300        case '\\': 
301          b.append('\\');
302          break;
303        case '/': 
304          b.append('\\');
305          break;
306        case 'u':
307          i++;
308          int uc = Integer.parseInt(s.substring(i, i+4), 16);
309          b.append((char) uc);
310          i = i + 4;
311          break;
312        default:
313          throw new FHIRLexerException("Unknown character escape \\"+s.charAt(i));
314        }
315      } else {
316        b.append(ch);
317        i++;
318      }
319    }
320    return b.toString();
321
322  }
323  public void skipToken(String token) throws FHIRLexerException {
324    if (getCurrent().equals(token))
325      next();
326    
327  }
328  public String takeDottedToken() throws FHIRLexerException {
329    StringBuilder b = new StringBuilder();
330    b.append(take());
331    while (!done() && getCurrent().equals(".")) {
332      b.append(take());
333      b.append(take());
334    }
335    return b.toString();
336  }
337  
338  void skipComments() throws FHIRLexerException {
339    while (!done() && hasComment())
340      next();
341  }
342
343}