icuSources/i18n/unicode/search.h

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 2001-2003 IBM and others. All rights reserved.
   4 **********************************************************************
   5 *   Date        Name        Description
   6 *  03/22/2000   helena      Creation.
   7 **********************************************************************
   8 */
   9
  10 #ifndef SEARCH_H
  11 #define SEARCH_H
  12
  13 #include "unicode/utypes.h"
  14
  15 #if !UCONFIG_NO_COLLATION
  16
  17 #include "unicode/uobject.h"
  18 #include "unicode/unistr.h"
  19 #include "unicode/chariter.h"
  20 #include "unicode/brkiter.h"
  21 #include "unicode/usearch.h"
  22
  23 /**
  24 * @stable ICU 2.0
  25 */
  26 struct USearch;
  27 /**
  28 * @stable ICU 2.0
  29 */
  30 typedef struct USearch USearch;
  31
  32 U_NAMESPACE_BEGIN
  33
  34 /**
  35  * <tt>SearchIterator</tt> is an abstract base class that provides
  36  * methods to search for a pattern within a text string. Instances of
  37  * <tt>SearchIterator</tt> maintain a current position and scans over the
  38  * target text, returning the indices the pattern is matched and the length
  39  * of each match.
  40  * <p>
  41  * <tt>SearchIterator</tt> defines a protocol for text searching.
  42  * Subclasses provide concrete implementations of various search algorithms.
  43  * For example, <tt>StringSearch</tt> implements language-sensitive pattern
  44  * matching based on the comparison rules defined in a
  45  * <tt>RuleBasedCollator</tt> object.
  46  * <p>
  47  * Other options for searching includes using a BreakIterator to restrict
  48  * the points at which matches are detected.
  49  * <p>
  50  * <tt>SearchIterator</tt> provides an API that is similar to that of
  51  * other text iteration classes such as <tt>BreakIterator</tt>. Using
  52  * this class, it is easy to scan through text looking for all occurances of
  53  * a given pattern. The following example uses a <tt>StringSearch</tt>
  54  * object to find all instances of "fox" in the target string. Any other
  55  * subclass of <tt>SearchIterator</tt> can be used in an identical
  56  * manner.
  57  * <pre><code>
  58  * UnicodeString target("The quick brown fox jumped over the lazy fox");
  59  * UnicodeString pattern("fox");
  60  *
  61  * SearchIterator *iter  = new StringSearch(pattern, target);
  62  * UErrorCode      error = U_ZERO_ERROR;
  63  * for (int pos = iter->first(error); pos != USEARCH_DONE;
  64  *                               pos = iter->next(error)) {
  65  *     printf("Found match at %d pos, length is %d\n", pos,
  66  *                                             iter.getMatchLength());
  67  * }
  68  * </code></pre>
  69  *
  70  * @see StringSearch
  71  * @see RuleBasedCollator
  72  */
  73 class U_I18N_API SearchIterator : public UObject {
  74
  75 public:
  76
  77     // public constructors and destructors -------------------------------
  78
  79     /**
  80     * Copy constructor that creates a SearchIterator instance with the same
  81     * behavior, and iterating over the same text.
  82     * @param other the SearchIterator instance to be copied.
  83     * @stable ICU 2.0
  84     */
  85     SearchIterator(const SearchIterator &other);
  86
  87     /**
  88      * Destructor. Cleans up the search iterator data struct.
  89      * @stable ICU 2.0
  90      */
  91     virtual ~SearchIterator();
  92
  93     // public get and set methods ----------------------------------------
  94
  95     /**
  96      * Sets the index to point to the given position, and clears any state
  97      * that's affected.
  98      * <p>
  99      * This method takes the argument index and sets the position in the text
 100      * string accordingly without checking if the index is pointing to a
 101      * valid starting point to begin searching.
 102      * @param position within the text to be set. If position is less
 103      *             than or greater than the text range for searching,
 104      *          an U_INDEX_OUTOFBOUNDS_ERROR will be returned
 105      * @param status for errors if it occurs
 106      * @stable ICU 2.0
 107      */
 108     virtual void setOffset(int32_t position, UErrorCode &status) = 0;
 109
 110     /**
 111      * Return the current index in the text being searched.
 112      * If the iteration has gone past the end of the text
 113      * (or past the beginning for a backwards search), USEARCH_DONE
 114      * is returned.
 115      * @return current index in the text being searched.
 116      * @stable ICU 2.0
 117      */
 118     virtual int32_t getOffset(void) const = 0;
 119
 120     /**
 121     * Sets the text searching attributes located in the enum
 122     * USearchAttribute with values from the enum USearchAttributeValue.
 123     * USEARCH_DEFAULT can be used for all attributes for resetting.
 124     * @param attribute text attribute (enum USearchAttribute) to be set
 125     * @param value text attribute value
 126     * @param status for errors if it occurs
 127     * @stable ICU 2.0
 128     */
 129     void setAttribute(USearchAttribute       attribute,
 130                       USearchAttributeValue  value,
 131                       UErrorCode            &status);
 132
 133     /**
 134     * Gets the text searching attributes
 135     * @param attribute text attribute (enum USearchAttribute) to be retrieve
 136     * @return text attribute value
 137     * @stable ICU 2.0
 138     */
 139     USearchAttributeValue getAttribute(USearchAttribute  attribute) const;
 140
 141     /**
 142     * Returns the index to the match in the text string that was searched.
 143     * This call returns a valid result only after a successful call to
 144     * <tt>first</tt>, <tt>next</tt>, <tt>previous</tt>, or <tt>last</tt>.
 145     * Just after construction, or after a searching method returns
 146     * <tt>USEARCH_DONE</tt>, this method will return <tt>USEARCH_DONE</tt>.
 147     * <p>
 148     * Use getMatchedLength to get the matched string length.
 149     * @return index of a substring within the text string that is being
 150     *         searched.
 151     * @see #first
 152     * @see #next
 153     * @see #previous
 154     * @see #last
 155     * @stable ICU 2.0
 156     */
 157     int32_t getMatchedStart(void) const;
 158
 159     /**
 160      * Returns the length of text in the string which matches the search
 161      * pattern. This call returns a valid result only after a successful call
 162      * to <tt>first</tt>, <tt>next</tt>, <tt>previous</tt>, or <tt>last</tt>.
 163      * Just after construction, or after a searching method returns
 164      * <tt>USEARCH_DONE</tt>, this method will return 0.
 165      * @return The length of the match in the target text, or 0 if there
 166      *         is no match currently.
 167      * @see #first
 168      * @see #next
 169      * @see #previous
 170      * @see #last
 171      * @stable ICU 2.0
 172      */
 173     int32_t getMatchedLength(void) const;
 174
 175     /**
 176      * Returns the text that was matched by the most recent call to
 177      * <tt>first</tt>, <tt>next</tt>, <tt>previous</tt>, or <tt>last</tt>.
 178      * If the iterator is not pointing at a valid match (e.g. just after
 179      * construction or after <tt>USEARCH_DONE</tt> has been returned,
 180      * returns an empty string.
 181      * @param result stores the matched string or an empty string if a match
 182      *        is not found.
 183      * @see #first
 184      * @see #next
 185      * @see #previous
 186      * @see #last
 187      * @stable ICU 2.0
 188      */
 189     void getMatchedText(UnicodeString &result) const;
 190
 191     /**
 192      * Set the BreakIterator that will be used to restrict the points
 193      * at which matches are detected. The user is responsible for deleting
 194      * the breakiterator.
 195      * @param breakiter A BreakIterator that will be used to restrict the
 196      *                points at which matches are detected. If a match is
 197      *                found, but the match's start or end index is not a
 198      *                boundary as determined by the <tt>BreakIterator</tt>,
 199      *                the match will be rejected and another will be searched
 200      *                for. If this parameter is <tt>NULL</tt>, no break
 201      *                detection is attempted.
 202      * @param status for errors if it occurs
 203      * @see BreakIterator
 204      * @stable ICU 2.0
 205      */
 206     void setBreakIterator(BreakIterator *breakiter, UErrorCode &status);
 207
 208     /**
 209      * Returns the BreakIterator that is used to restrict the points at
 210      * which matches are detected.  This will be the same object that was
 211      * passed to the constructor or to <tt>setBreakIterator</tt>.
 212      * Note that <tt>NULL</tt> is a legal value; it means that break
 213      * detection should not be attempted.
 214      * @return BreakIterator used to restrict matchings.
 215      * @see #setBreakIterator
 216      * @stable ICU 2.0
 217      */
 218     const BreakIterator * getBreakIterator(void) const;
 219
 220     /**
 221      * Set the string text to be searched. Text iteration will hence begin at
 222      * the start of the text string. This method is useful if you want to
 223      * re-use an iterator to search for the same pattern within a different
 224      * body of text. The user is responsible for deleting the text.
 225      * @param text string to be searched.
 226      * @param status for errors. If the text length is 0,
 227      *        an U_ILLEGAL_ARGUMENT_ERROR is returned.
 228      * @stable ICU 2.0
 229      */
 230     virtual void setText(const UnicodeString &text, UErrorCode &status);
 231
 232     /**
 233      * Set the string text to be searched. Text iteration will hence begin at
 234      * the start of the text string. This method is useful if you want to
 235      * re-use an iterator to search for the same pattern within a different
 236      * body of text.
 237      * <p>
 238      * Note: No parsing of the text within the <tt>CharacterIterator</tt>
 239      * will be done during searching for this version. The block of text
 240      * in <tt>CharacterIterator</tt> will be used as it is.
 241      * The user is responsible for deleting the text.
 242      * @param text string iterator to be searched.
 243      * @param status for errors if any. If the text length is 0 then an
 244      *        U_ILLEGAL_ARGUMENT_ERROR is returned.
 245      * @stable ICU 2.0
 246      */
 247     virtual void setText(CharacterIterator &text, UErrorCode &status);
 248
 249     /**
 250      * Return the string text to be searched.
 251      * @return text string to be searched.
 252      * @stable ICU 2.0
 253      */
 254     const UnicodeString & getText(void) const;
 255
 256     // operator overloading ----------------------------------------------
 257
 258     /**
 259      * Equality operator.
 260      * @param that SearchIterator instance to be compared.
 261      * @return TRUE if both BreakIterators are of the same class, have the
 262      *         same behavior, terates over the same text and have the same
 263      *         attributes. FALSE otherwise.
 264      * @stable ICU 2.0
 265      */
 266     virtual UBool operator==(const SearchIterator &that) const;
 267
 268     /**
 269      * Not-equal operator.
 270      * @param that SearchIterator instance to be compared.
 271      * @return FALSE if operator== returns TRUE, and vice versa.
 272      * @stable ICU 2.0
 273      */
 274     UBool operator!=(const SearchIterator &that) const;
 275
 276     // public methods ----------------------------------------------------
 277
 278     /**
 279      * Returns a copy of SearchIterator with the same behavior, and
 280      * iterating over the same text, as this one. Note that all data will be
 281      * replicated, except for the text string to be searched.
 282      * @return cloned object
 283      * @stable ICU 2.0
 284      */
 285     virtual SearchIterator* safeClone(void) const = 0;
 286
 287     /**
 288      * Returns the first index at which the string text matches the search
 289      * pattern. The iterator is adjusted so that its current index (as
 290      * returned by <tt>getOffset</tt>) is the match position if one
 291      * was found.
 292      * If a match is not found, <tt>USEARCH_DONE</tt> will be returned and
 293      * the iterator will be adjusted to the index USEARCH_DONE
 294      * @param  status for errors if it occurs
 295      * @return The character index of the first match, or
 296      *         <tt>USEARCH_DONE</tt> if there are no matches.
 297      * @see #getOffset
 298      * @stable ICU 2.0
 299      */
 300     int32_t first(UErrorCode &status);
 301
 302     /**
 303      * Returns the first index greater than <tt>position</tt> at which the
 304      * string text matches the search pattern. The iterator is adjusted so
 305      * that its current index (as returned by <tt>getOffset</tt>) is the
 306      * match position if one was found. If a match is not found,
 307      * <tt>USEARCH_DONE</tt> will be returned and the iterator will be
 308      * adjusted to the index USEARCH_DONE
 309      * @param  position where search if to start from. If position is less
 310      *             than or greater than the text range for searching,
 311      *          an U_INDEX_OUTOFBOUNDS_ERROR will be returned
 312      * @param  status for errors if it occurs
 313      * @return The character index of the first match following
 314      *         <tt>position</tt>, or <tt>USEARCH_DONE</tt> if there are no
 315      *         matches.
 316      * @see #getOffset
 317      * @stable ICU 2.0
 318      */
 319     int32_t following(int32_t position, UErrorCode &status);
 320
 321     /**
 322      * Returns the last index in the target text at which it matches the
 323      * search pattern. The iterator is adjusted so that its current index
 324      * (as returned by <tt>getOffset</tt>) is the match position if one was
 325      * found.
 326      * If a match is not found, <tt>USEARCH_DONE</tt> will be returned and
 327      * the iterator will be adjusted to the index USEARCH_DONE.
 328      * @param  status for errors if it occurs
 329      * @return The index of the first match, or <tt>USEARCH_DONE</tt> if
 330      *         there are no matches.
 331      * @see #getOffset
 332      * @stable ICU 2.0
 333      */
 334     int32_t last(UErrorCode &status);
 335
 336     /**
 337      * Returns the first index less than <tt>position</tt> at which the string
 338      * text matches the search pattern. The iterator is adjusted so that its
 339      * current index (as returned by <tt>getOffset</tt>) is the match
 340      * position if one was found. If a match is not found,
 341      * <tt>USEARCH_DONE</tt> will be returned and the iterator will be
 342      * adjusted to the index USEARCH_DONE
 343      * @param  position where search is to start from. If position is less
 344      *             than or greater than the text range for searching,
 345      *          an U_INDEX_OUTOFBOUNDS_ERROR will be returned
 346      * @param  status for errors if it occurs
 347      * @return The character index of the first match preceding
 348      *         <tt>position</tt>, or <tt>USEARCH_DONE</tt> if there are
 349      *         no matches.
 350      * @see #getOffset
 351      * @stable ICU 2.0
 352      */
 353     int32_t preceding(int32_t position, UErrorCode &status);
 354
 355     /**
 356      * Returns the index of the next point at which the text matches the
 357      * search pattern, starting from the current position
 358      * The iterator is adjusted so that its current index (as returned by
 359      * <tt>getOffset</tt>) is the match position if one was found.
 360      * If a match is not found, <tt>USEARCH_DONE</tt> will be returned and
 361      * the iterator will be adjusted to a position after the end of the text
 362      * string.
 363      * @param  status for errors if it occurs
 364      * @return The index of the next match after the current position,
 365      *          or <tt>USEARCH_DONE</tt> if there are no more matches.
 366      * @see #getOffset
 367      * @stable ICU 2.0
 368      */
 369      int32_t next(UErrorCode &status);
 370
 371     /**
 372      * Returns the index of the previous point at which the string text
 373      * matches the search pattern, starting at the current position.
 374      * The iterator is adjusted so that its current index (as returned by
 375      * <tt>getOffset</tt>) is the match position if one was found.
 376      * If a match is not found, <tt>USEARCH_DONE</tt> will be returned and
 377      * the iterator will be adjusted to the index USEARCH_DONE
 378      * @param  status for errors if it occurs
 379      * @return The index of the previous match before the current position,
 380      *          or <tt>USEARCH_DONE</tt> if there are no more matches.
 381      * @see #getOffset
 382      * @stable ICU 2.0
 383      */
 384     int32_t previous(UErrorCode &status);
 385
 386     /**
 387     * Resets the iteration.
 388     * Search will begin at the start of the text string if a forward
 389     * iteration is initiated before a backwards iteration. Otherwise if a
 390     * backwards iteration is initiated before a forwards iteration, the
 391     * search will begin at the end of the text string.
 392     * @stable ICU 2.0
 393     */
 394     virtual void reset();
 395
 396 protected:
 397     // protected data members ---------------------------------------------
 398
 399     /**
 400     * C search data struct
 401     * @stable ICU 2.0
 402     */
 403     USearch *m_search_;
 404
 405     /**
 406     * Break iterator.
 407     * Currently the C++ breakiterator does not have getRules etc to reproduce
 408     * another in C. Hence we keep the original around and do the verification
 409     * at the end of the match. The user is responsible for deleting this
 410     * break iterator.
 411     * @stable ICU 2.0
 412     */
 413     BreakIterator *m_breakiterator_;
 414
 415     /**
 416     * Unicode string version of the search text
 417     * @stable ICU 2.0
 418     */
 419     UnicodeString  m_text_;
 420
 421     // protected constructors and destructors -----------------------------
 422
 423     /**
 424     * Default constructor.
 425     * Initializes data to the default values.
 426     * @stable ICU 2.0
 427     */
 428     SearchIterator();
 429
 430     /**
 431      * Constructor for use by subclasses.
 432      * @param text The target text to be searched.
 433      * @param breakiter A {@link BreakIterator} that is used to restrict the
 434      *                points at which matches are detected. If
 435      *                <tt>handleNext</tt> or <tt>handlePrev</tt> finds a
 436      *                match, but the match's start or end index is not a
 437      *                boundary as determined by the <tt>BreakIterator</tt>,
 438      *                the match is rejected and <tt>handleNext</tt> or
 439      *                <tt>handlePrev</tt> is called again. If this parameter
 440      *                is <tt>NULL</tt>, no break detection is attempted.
 441      * @see #handleNext
 442      * @see #handlePrev
 443      * @stable ICU 2.0
 444      */
 445     SearchIterator(const UnicodeString &text,
 446                          BreakIterator *breakiter = NULL);
 447
 448     /**
 449      * Constructor for use by subclasses.
 450      * <p>
 451      * Note: No parsing of the text within the <tt>CharacterIterator</tt>
 452      * will be done during searching for this version. The block of text
 453      * in <tt>CharacterIterator</tt> will be used as it is.
 454      * @param text The target text to be searched.
 455      * @param breakiter A {@link BreakIterator} that is used to restrict the
 456      *                points at which matches are detected. If
 457      *                <tt>handleNext</tt> or <tt>handlePrev</tt> finds a
 458      *                match, but the match's start or end index is not a
 459      *                boundary as determined by the <tt>BreakIterator</tt>,
 460      *                the match is rejected and <tt>handleNext</tt> or
 461      *                <tt>handlePrev</tt> is called again. If this parameter
 462      *                is <tt>NULL</tt>, no break detection is attempted.
 463      * @see #handleNext
 464      * @see #handlePrev
 465      * @stable ICU 2.0
 466      */
 467     SearchIterator(CharacterIterator &text, BreakIterator *breakiter = NULL);
 468
 469     // protected methods --------------------------------------------------
 470
 471     /**
 472      * Assignment operator. Sets this iterator to have the same behavior,
 473      * and iterate over the same text, as the one passed in.
 474      * @param that instance to be copied.
 475      * @stable ICU 2.0
 476      */
 477     SearchIterator & operator=(const SearchIterator &that);
 478
 479     /**
 480      * Abstract method which subclasses override to provide the mechanism
 481      * for finding the next match in the target text. This allows different
 482      * subclasses to provide different search algorithms.
 483      * <p>
 484      * If a match is found, the implementation should return the index at
 485      * which the match starts and should call
 486      * <tt>setMatchLength</tt> with the number of characters
 487      * in the target text that make up the match. If no match is found, the
 488      * method should return USEARCH_DONE.
 489      * <p>
 490      * @param position The index in the target text at which the search
 491      *                 should start.
 492      * @param status for error codes if it occurs.
 493      * @return index at which the match starts, else if match is not found
 494      *         USEARCH_DONE is returned
 495      * @see #setMatchLength
 496      * @stable ICU 2.0
 497      */
 498     virtual int32_t handleNext(int32_t position, UErrorCode &status)
 499                                                                          = 0;
 500
 501     /**
 502      * Abstract method which subclasses override to provide the mechanism for
 503      * finding the previous match in the target text. This allows different
 504      * subclasses to provide different search algorithms.
 505      * <p>
 506      * If a match is found, the implementation should return the index at
 507      * which the match starts and should call
 508      * <tt>setMatchLength</tt> with the number of characters
 509      * in the target text that make up the match. If no match is found, the
 510      * method should return USEARCH_DONE.
 511      * <p>
 512      * @param position The index in the target text at which the search
 513      *                 should start.
 514      * @param status for error codes if it occurs.
 515      * @return index at which the match starts, else if match is not found
 516      *         USEARCH_DONE is returned
 517      * @see #setMatchLength
 518      * @stable ICU 2.0
 519      */
 520      virtual int32_t handlePrev(int32_t position, UErrorCode &status)
 521                                                                          = 0;
 522
 523     /**
 524      * Sets the length of the currently matched string in the text string to
 525      * be searched.
 526      * Subclasses' <tt>handleNext</tt> and <tt>handlePrev</tt>
 527      * methods should call this when they find a match in the target text.
 528      * @param length length of the matched text.
 529      * @see #handleNext
 530      * @see #handlePrev
 531      * @stable ICU 2.0
 532      */
 533     virtual void setMatchLength(int32_t length);
 534
 535     /**
 536      * Sets the offset of the currently matched string in the text string to
 537      * be searched.
 538      * Subclasses' <tt>handleNext</tt> and <tt>handlePrev</tt>
 539      * methods should call this when they find a match in the target text.
 540      * @param position start offset of the matched text.
 541      * @see #handleNext
 542      * @see #handlePrev
 543      * @stable ICU 2.0
 544      */
 545     virtual void setMatchStart(int32_t position);
 546
 547     /**
 548     * sets match not found
 549     * @stable ICU 2.0
 550     */
 551     void setMatchNotFound();
 552 };
 553
 554 inline UBool SearchIterator::operator!=(const SearchIterator &that) const
 555 {
 556    return !operator==(that);
 557 }
 558 U_NAMESPACE_END
 559
 560 #endif /* #if !UCONFIG_NO_COLLATION */
 561
 562 #endif
 563