1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
5 // This file implements utility functions for eliding and formatting UI text.
7 // Note that several of the functions declared in text_elider.h are implemented
8 // in this file using helper classes in an unnamed namespace.
10 #include "ui/gfx/text_elider.h"
15 #include "base/files/file_path.h"
16 #include "base/i18n/break_iterator.h"
17 #include "base/i18n/char_iterator.h"
18 #include "base/i18n/rtl.h"
19 #include "base/memory/scoped_ptr.h"
20 #include "base/strings/string_split.h"
21 #include "base/strings/string_util.h"
22 #include "base/strings/sys_string_conversions.h"
23 #include "base/strings/utf_string_conversions.h"
24 #include "third_party/icu/source/common/unicode/rbbi.h"
25 #include "third_party/icu/source/common/unicode/uloc.h"
26 #include "ui/gfx/font_list.h"
27 #include "ui/gfx/text_utils.h"
29 using base::ASCIIToUTF16
;
30 using base::UTF8ToUTF16
;
31 using base::WideToUTF16
;
36 const char kEllipsis
[] = "\xE2\x80\xA6";
37 const base::char16 kEllipsisUTF16
[] = { 0x2026, 0 };
38 const base::char16 kForwardSlash
= '/';
40 StringSlicer::StringSlicer(const base::string16
& text
,
41 const base::string16
& ellipsis
,
43 bool elide_at_beginning
)
46 elide_in_middle_(elide_in_middle
),
47 elide_at_beginning_(elide_at_beginning
) {
50 base::string16
StringSlicer::CutString(size_t length
, bool insert_ellipsis
) {
51 const base::string16 ellipsis_text
= insert_ellipsis
? ellipsis_
54 if (elide_at_beginning_
)
55 return ellipsis_text
+
56 text_
.substr(FindValidBoundaryBefore(text_
.length() - length
));
58 if (!elide_in_middle_
)
59 return text_
.substr(0, FindValidBoundaryBefore(length
)) + ellipsis_text
;
61 // We put the extra character, if any, before the cut.
62 const size_t half_length
= length
/ 2;
63 const size_t prefix_length
= FindValidBoundaryBefore(length
- half_length
);
64 const size_t suffix_start_guess
= text_
.length() - half_length
;
65 const size_t suffix_start
= FindValidBoundaryAfter(suffix_start_guess
);
66 const size_t suffix_length
=
67 half_length
- (suffix_start_guess
- suffix_start
);
68 return text_
.substr(0, prefix_length
) + ellipsis_text
+
69 text_
.substr(suffix_start
, suffix_length
);
72 size_t StringSlicer::FindValidBoundaryBefore(size_t index
) const {
73 DCHECK_LE(index
, text_
.length());
74 if (index
!= text_
.length())
75 U16_SET_CP_START(text_
.data(), 0, index
);
79 size_t StringSlicer::FindValidBoundaryAfter(size_t index
) const {
80 DCHECK_LE(index
, text_
.length());
81 if (index
!= text_
.length())
82 U16_SET_CP_LIMIT(text_
.data(), 0, index
, text_
.length());
86 base::string16
ElideEmail(const base::string16
& email
,
87 const FontList
& font_list
,
88 float available_pixel_width
) {
89 if (GetStringWidthF(email
, font_list
) <= available_pixel_width
)
92 // Split the email into its local-part (username) and domain-part. The email
93 // spec technically allows for @ symbols in the local-part (username) of the
94 // email under some special requirements. It is guaranteed that there is no @
95 // symbol in the domain part of the email however so splitting at the last @
97 const size_t split_index
= email
.find_last_of('@');
98 DCHECK_NE(split_index
, base::string16::npos
);
99 base::string16 username
= email
.substr(0, split_index
);
100 base::string16 domain
= email
.substr(split_index
+ 1);
101 DCHECK(!username
.empty());
102 DCHECK(!domain
.empty());
104 // Subtract the @ symbol from the available width as it is mandatory.
105 const base::string16 kAtSignUTF16
= ASCIIToUTF16("@");
106 available_pixel_width
-= GetStringWidthF(kAtSignUTF16
, font_list
);
108 // Check whether eliding the domain is necessary: if eliding the username
109 // is sufficient, the domain will not be elided.
110 const float full_username_width
= GetStringWidthF(username
, font_list
);
111 const float available_domain_width
=
112 available_pixel_width
-
113 std::min(full_username_width
,
114 GetStringWidthF(username
.substr(0, 1) + kEllipsisUTF16
,
116 if (GetStringWidthF(domain
, font_list
) > available_domain_width
) {
117 // Elide the domain so that it only takes half of the available width.
118 // Should the username not need all the width available in its half, the
119 // domain will occupy the leftover width.
120 // If |desired_domain_width| is greater than |available_domain_width|: the
121 // minimal username elision allowed by the specifications will not fit; thus
122 // |desired_domain_width| must be <= |available_domain_width| at all cost.
123 const float desired_domain_width
=
124 std::min(available_domain_width
,
125 std::max(available_pixel_width
- full_username_width
,
126 available_pixel_width
/ 2));
127 domain
= ElideText(domain
, font_list
, desired_domain_width
,
129 // Failing to elide the domain such that at least one character remains
130 // (other than the ellipsis itself) remains: return a single ellipsis.
131 if (domain
.length() <= 1U)
132 return base::string16(kEllipsisUTF16
);
135 // Fit the username in the remaining width (at this point the elided username
136 // is guaranteed to fit with at least one character remaining given all the
137 // precautions taken earlier).
138 available_pixel_width
-= GetStringWidthF(domain
, font_list
);
139 username
= ElideText(username
, font_list
, available_pixel_width
,
142 return username
+ kAtSignUTF16
+ domain
;
145 base::string16
ElideFilename(const base::FilePath
& filename
,
146 const FontList
& font_list
,
147 float available_pixel_width
) {
149 base::string16 filename_utf16
= filename
.value();
150 base::string16 extension
= filename
.Extension();
151 base::string16 rootname
= filename
.BaseName().RemoveExtension().value();
152 #elif defined(OS_POSIX)
153 base::string16 filename_utf16
= WideToUTF16(base::SysNativeMBToWide(
155 base::string16 extension
= WideToUTF16(base::SysNativeMBToWide(
156 filename
.Extension()));
157 base::string16 rootname
= WideToUTF16(base::SysNativeMBToWide(
158 filename
.BaseName().RemoveExtension().value()));
161 const float full_width
= GetStringWidthF(filename_utf16
, font_list
);
162 if (full_width
<= available_pixel_width
)
163 return base::i18n::GetDisplayStringInLTRDirectionality(filename_utf16
);
165 if (rootname
.empty() || extension
.empty()) {
166 const base::string16 elided_name
= ElideText(filename_utf16
, font_list
,
167 available_pixel_width
, ELIDE_AT_END
);
168 return base::i18n::GetDisplayStringInLTRDirectionality(elided_name
);
171 const float ext_width
= GetStringWidthF(extension
, font_list
);
172 const float root_width
= GetStringWidthF(rootname
, font_list
);
174 // We may have trimmed the path.
175 if (root_width
+ ext_width
<= available_pixel_width
) {
176 const base::string16 elided_name
= rootname
+ extension
;
177 return base::i18n::GetDisplayStringInLTRDirectionality(elided_name
);
180 if (ext_width
>= available_pixel_width
) {
181 const base::string16 elided_name
= ElideText(
182 rootname
+ extension
, font_list
, available_pixel_width
,
184 return base::i18n::GetDisplayStringInLTRDirectionality(elided_name
);
187 float available_root_width
= available_pixel_width
- ext_width
;
188 base::string16 elided_name
=
189 ElideText(rootname
, font_list
, available_root_width
, ELIDE_AT_END
);
190 elided_name
+= extension
;
191 return base::i18n::GetDisplayStringInLTRDirectionality(elided_name
);
194 base::string16
ElideText(const base::string16
& text
,
195 const FontList
& font_list
,
196 float available_pixel_width
,
197 ElideBehavior elide_behavior
) {
201 const float current_text_pixel_width
= GetStringWidthF(text
, font_list
);
202 const bool elide_in_middle
= (elide_behavior
== ELIDE_IN_MIDDLE
);
203 const bool elide_at_beginning
= (elide_behavior
== ELIDE_AT_BEGINNING
);
204 const bool insert_ellipsis
= (elide_behavior
!= TRUNCATE_AT_END
);
206 const base::string16 ellipsis
= base::string16(kEllipsisUTF16
);
207 StringSlicer
slicer(text
, ellipsis
, elide_in_middle
, elide_at_beginning
);
209 // Pango will return 0 width for absurdly long strings. Cut the string in
210 // half and try again.
211 // This is caused by an int overflow in Pango (specifically, in
212 // pango_glyph_string_extents_range). It's actually more subtle than just
213 // returning 0, since on super absurdly long strings, the int can wrap and
214 // return positive numbers again. Detecting that is probably not worth it
215 // (eliding way too much from a ridiculous string is probably still
216 // ridiculous), but we should check other widths for bogus values as well.
217 if (current_text_pixel_width
<= 0 && !text
.empty()) {
218 const base::string16 cut
=
219 slicer
.CutString(text
.length() / 2, insert_ellipsis
);
220 return ElideText(cut
, font_list
, available_pixel_width
, elide_behavior
);
223 if (current_text_pixel_width
<= available_pixel_width
)
226 if (insert_ellipsis
&&
227 GetStringWidthF(ellipsis
, font_list
) > available_pixel_width
)
228 return base::string16();
230 // Use binary search to compute the elided text.
232 size_t hi
= text
.length() - 1;
234 for (guess
= (lo
+ hi
) / 2; lo
<= hi
; guess
= (lo
+ hi
) / 2) {
235 // We check the width of the whole desired string at once to ensure we
236 // handle kerning/ligatures/etc. correctly.
237 // TODO(skanuj) : Handle directionality of ellipsis based on adjacent
238 // characters. See crbug.com/327963.
239 const base::string16 cut
= slicer
.CutString(guess
, insert_ellipsis
);
240 const float guess_width
= GetStringWidthF(cut
, font_list
);
241 if (guess_width
== available_pixel_width
)
243 if (guess_width
> available_pixel_width
) {
245 // Move back if we are on loop terminating condition, and guess is wider
254 return slicer
.CutString(guess
, insert_ellipsis
);
257 bool ElideString(const base::string16
& input
, int max_len
,
258 base::string16
* output
) {
259 DCHECK_GE(max_len
, 0);
260 if (static_cast<int>(input
.length()) <= max_len
) {
261 output
->assign(input
);
270 output
->assign(input
.substr(0, 1));
273 output
->assign(input
.substr(0, 2));
276 output
->assign(input
.substr(0, 1) + ASCIIToUTF16(".") +
277 input
.substr(input
.length() - 1));
280 output
->assign(input
.substr(0, 1) + ASCIIToUTF16("..") +
281 input
.substr(input
.length() - 1));
284 int rstr_len
= (max_len
- 3) / 2;
285 int lstr_len
= rstr_len
+ ((max_len
- 3) % 2);
286 output
->assign(input
.substr(0, lstr_len
) + ASCIIToUTF16("...") +
287 input
.substr(input
.length() - rstr_len
));
297 // Internal class used to track progress of a rectangular string elide
298 // operation. Exists so the top-level ElideRectangleString() function
299 // can be broken into smaller methods sharing this state.
300 class RectangleString
{
302 RectangleString(size_t max_rows
, size_t max_cols
,
303 bool strict
, base::string16
*output
)
304 : max_rows_(max_rows
),
312 // Perform deferred initializations following creation. Must be called
313 // before any input can be added via AddString().
314 void Init() { output_
->clear(); }
316 // Add an input string, reformatting to fit the desired dimensions.
317 // AddString() may be called multiple times to concatenate together
318 // multiple strings into the region (the current caller doesn't do
320 void AddString(const base::string16
& input
);
322 // Perform any deferred output processing. Must be called after the
323 // last AddString() call has occurred.
327 // Add a line to the rectangular region at the current position,
328 // either by itself or by breaking it into words.
329 void AddLine(const base::string16
& line
);
331 // Add a word to the rectangular region at the current position,
332 // either by itself or by breaking it into characters.
333 void AddWord(const base::string16
& word
);
335 // Add text to the output string if the rectangular boundaries
336 // have not been exceeded, advancing the current position.
337 void Append(const base::string16
& string
);
339 // Set the current position to the beginning of the next line. If
340 // |output| is true, add a newline to the output string if the rectangular
341 // boundaries have not been exceeded. If |output| is false, we assume
342 // some other mechanism will (likely) do similar breaking after the fact.
343 void NewLine(bool output
);
345 // Maximum number of rows allowed in the output string.
348 // Maximum number of characters allowed in the output string.
351 // Current row position, always incremented and may exceed max_rows_
352 // when the input can not fit in the region. We stop appending to
353 // the output string, however, when this condition occurs. In the
354 // future, we may want to expose this value to allow the caller to
355 // determine how many rows would actually be required to hold the
359 // Current character position, should never exceed max_cols_.
362 // True when we do whitespace to newline conversions ourselves.
365 // True when some of the input has been truncated.
368 // String onto which the output is accumulated.
369 base::string16
* output_
;
371 DISALLOW_COPY_AND_ASSIGN(RectangleString
);
374 void RectangleString::AddString(const base::string16
& input
) {
375 base::i18n::BreakIterator
lines(input
,
376 base::i18n::BreakIterator::BREAK_NEWLINE
);
378 while (lines
.Advance())
379 AddLine(lines
.GetString());
381 NOTREACHED() << "BreakIterator (lines) init failed";
385 bool RectangleString::Finalize() {
387 output_
->append(ASCIIToUTF16("..."));
393 void RectangleString::AddLine(const base::string16
& line
) {
394 if (line
.length() < max_cols_
) {
397 base::i18n::BreakIterator
words(line
,
398 base::i18n::BreakIterator::BREAK_SPACE
);
400 while (words
.Advance())
401 AddWord(words
.GetString());
403 NOTREACHED() << "BreakIterator (words) init failed";
406 // Account for naturally-occuring newlines.
411 void RectangleString::AddWord(const base::string16
& word
) {
412 if (word
.length() < max_cols_
) {
413 // Word can be made to fit, no need to fragment it.
414 if (current_col_
+ word
.length() >= max_cols_
)
418 // Word is so big that it must be fragmented.
421 base::i18n::UTF16CharIterator
chars(&word
);
422 while (!chars
.end()) {
423 // When boundary is hit, add as much as will fit on this line.
424 if (current_col_
+ (chars
.char_pos() - char_start
) >= max_cols_
) {
425 Append(word
.substr(array_start
, chars
.array_pos() - array_start
));
427 array_start
= chars
.array_pos();
428 char_start
= chars
.char_pos();
432 // Add the last remaining fragment, if any.
433 if (array_start
!= chars
.array_pos())
434 Append(word
.substr(array_start
, chars
.array_pos() - array_start
));
438 void RectangleString::Append(const base::string16
& string
) {
439 if (current_row_
< max_rows_
)
440 output_
->append(string
);
443 current_col_
+= string
.length();
446 void RectangleString::NewLine(bool output
) {
447 if (current_row_
< max_rows_
) {
449 output_
->append(ASCIIToUTF16("\n"));
457 // Internal class used to track progress of a rectangular text elide
458 // operation. Exists so the top-level ElideRectangleText() function
459 // can be broken into smaller methods sharing this state.
460 class RectangleText
{
462 RectangleText(const FontList
& font_list
,
463 float available_pixel_width
,
464 int available_pixel_height
,
465 WordWrapBehavior wrap_behavior
,
466 std::vector
<base::string16
>* lines
)
467 : font_list_(font_list
),
468 line_height_(font_list
.GetHeight()),
469 available_pixel_width_(available_pixel_width
),
470 available_pixel_height_(available_pixel_height
),
471 wrap_behavior_(wrap_behavior
),
474 last_line_ended_in_lf_(false),
476 insufficient_width_(false),
477 insufficient_height_(false) {}
479 // Perform deferred initializions following creation. Must be called
480 // before any input can be added via AddString().
481 void Init() { lines_
->clear(); }
483 // Add an input string, reformatting to fit the desired dimensions.
484 // AddString() may be called multiple times to concatenate together
485 // multiple strings into the region (the current caller doesn't do
487 void AddString(const base::string16
& input
);
489 // Perform any deferred output processing. Must be called after the last
490 // AddString() call has occured. Returns a combination of
491 // |ReformattingResultFlags| indicating whether the given width or height was
492 // insufficient, leading to elision or truncation.
496 // Add a line to the rectangular region at the current position,
497 // either by itself or by breaking it into words.
498 void AddLine(const base::string16
& line
);
500 // Wrap the specified word across multiple lines.
501 int WrapWord(const base::string16
& word
);
503 // Add a long word - wrapping, eliding or truncating per the wrap behavior.
504 int AddWordOverflow(const base::string16
& word
);
506 // Add a word to the rectangluar region at the current position.
507 int AddWord(const base::string16
& word
);
509 // Append the specified |text| to the current output line, incrementing the
510 // running width by the specified amount. This is an optimization over
511 // |AddToCurrentLine()| when |text_width| is already known.
512 void AddToCurrentLineWithWidth(const base::string16
& text
, float text_width
);
514 // Append the specified |text| to the current output line.
515 void AddToCurrentLine(const base::string16
& text
);
517 // Set the current position to the beginning of the next line.
520 // The font list used for measuring text width.
521 const FontList
& font_list_
;
523 // The height of each line of text.
524 const int line_height_
;
526 // The number of pixels of available width in the rectangle.
527 const float available_pixel_width_
;
529 // The number of pixels of available height in the rectangle.
530 const int available_pixel_height_
;
532 // The wrap behavior for words that are too long to fit on a single line.
533 const WordWrapBehavior wrap_behavior_
;
535 // The current running width.
536 float current_width_
;
538 // The current running height.
541 // The current line of text.
542 base::string16 current_line_
;
544 // Indicates whether the last line ended with \n.
545 bool last_line_ended_in_lf_
;
547 // The output vector of lines.
548 std::vector
<base::string16
>* lines_
;
550 // Indicates whether a word was so long that it had to be truncated or elided
551 // to fit the available width.
552 bool insufficient_width_
;
554 // Indicates whether there were too many lines for the available height.
555 bool insufficient_height_
;
557 DISALLOW_COPY_AND_ASSIGN(RectangleText
);
560 void RectangleText::AddString(const base::string16
& input
) {
561 base::i18n::BreakIterator
lines(input
,
562 base::i18n::BreakIterator::BREAK_NEWLINE
);
564 while (!insufficient_height_
&& lines
.Advance()) {
565 base::string16 line
= lines
.GetString();
566 // The BREAK_NEWLINE iterator will keep the trailing newline character,
567 // except in the case of the last line, which may not have one. Remove
568 // the newline character, if it exists.
569 last_line_ended_in_lf_
= !line
.empty() && line
[line
.length() - 1] == '\n';
570 if (last_line_ended_in_lf_
)
571 line
.resize(line
.length() - 1);
575 NOTREACHED() << "BreakIterator (lines) init failed";
579 int RectangleText::Finalize() {
580 // Remove trailing whitespace from the last line or remove the last line
581 // completely, if it's just whitespace.
582 if (!insufficient_height_
&& !lines_
->empty()) {
583 base::TrimWhitespace(lines_
->back(), base::TRIM_TRAILING
, &lines_
->back());
584 if (lines_
->back().empty() && !last_line_ended_in_lf_
)
587 if (last_line_ended_in_lf_
)
588 lines_
->push_back(base::string16());
589 return (insufficient_width_
? INSUFFICIENT_SPACE_HORIZONTAL
: 0) |
590 (insufficient_height_
? INSUFFICIENT_SPACE_VERTICAL
: 0);
593 void RectangleText::AddLine(const base::string16
& line
) {
594 const float line_width
= GetStringWidthF(line
, font_list_
);
595 if (line_width
<= available_pixel_width_
) {
596 AddToCurrentLineWithWidth(line
, line_width
);
598 // Iterate over positions that are valid to break the line at. In general,
599 // these are word boundaries but after any punctuation following the word.
600 base::i18n::BreakIterator
words(line
,
601 base::i18n::BreakIterator::BREAK_LINE
);
603 while (words
.Advance()) {
604 const bool truncate
= !current_line_
.empty();
605 const base::string16
& word
= words
.GetString();
606 const int lines_added
= AddWord(word
);
609 // Trim trailing whitespace from the line that was added.
610 const int line
= lines_
->size() - lines_added
;
611 base::TrimWhitespace(lines_
->at(line
), base::TRIM_TRAILING
,
614 if (base::ContainsOnlyChars(word
, base::kWhitespaceUTF16
)) {
615 // Skip the first space if the previous line was carried over.
617 current_line_
.clear();
622 NOTREACHED() << "BreakIterator (words) init failed";
625 // Account for naturally-occuring newlines.
629 int RectangleText::WrapWord(const base::string16
& word
) {
630 // Word is so wide that it must be fragmented.
631 base::string16 text
= word
;
633 bool first_fragment
= true;
634 while (!insufficient_height_
&& !text
.empty()) {
635 base::string16 fragment
=
636 ElideText(text
, font_list_
, available_pixel_width_
,
638 // At least one character has to be added at every line, even if the
639 // available space is too small.
641 fragment
= text
.substr(0, 1);
642 if (!first_fragment
&& NewLine())
644 AddToCurrentLine(fragment
);
645 text
= text
.substr(fragment
.length());
646 first_fragment
= false;
651 int RectangleText::AddWordOverflow(const base::string16
& word
) {
654 // Unless this is the very first word, put it on a new line.
655 if (!current_line_
.empty()) {
661 if (wrap_behavior_
== IGNORE_LONG_WORDS
) {
662 current_line_
= word
;
663 current_width_
= available_pixel_width_
;
664 } else if (wrap_behavior_
== WRAP_LONG_WORDS
) {
665 lines_added
+= WrapWord(word
);
667 const ElideBehavior elide_behavior
=
668 (wrap_behavior_
== ELIDE_LONG_WORDS
? ELIDE_AT_END
: TRUNCATE_AT_END
);
669 const base::string16 elided_word
=
670 ElideText(word
, font_list_
, available_pixel_width_
, elide_behavior
);
671 AddToCurrentLine(elided_word
);
672 insufficient_width_
= true;
678 int RectangleText::AddWord(const base::string16
& word
) {
680 base::string16 trimmed
;
681 base::TrimWhitespace(word
, base::TRIM_TRAILING
, &trimmed
);
682 const float trimmed_width
= GetStringWidthF(trimmed
, font_list_
);
683 if (trimmed_width
<= available_pixel_width_
) {
684 // Word can be made to fit, no need to fragment it.
685 if ((current_width_
+ trimmed_width
> available_pixel_width_
) && NewLine())
687 // Append the non-trimmed word, in case more words are added after.
688 AddToCurrentLine(word
);
690 lines_added
= AddWordOverflow(wrap_behavior_
== IGNORE_LONG_WORDS
?
696 void RectangleText::AddToCurrentLine(const base::string16
& text
) {
697 AddToCurrentLineWithWidth(text
, GetStringWidthF(text
, font_list_
));
700 void RectangleText::AddToCurrentLineWithWidth(const base::string16
& text
,
702 if (current_height_
>= available_pixel_height_
) {
703 insufficient_height_
= true;
706 current_line_
.append(text
);
707 current_width_
+= text_width
;
710 bool RectangleText::NewLine() {
711 bool line_added
= false;
712 if (current_height_
< available_pixel_height_
) {
713 lines_
->push_back(current_line_
);
714 current_line_
.clear();
717 insufficient_height_
= true;
719 current_height_
+= line_height_
;
726 bool ElideRectangleString(const base::string16
& input
, size_t max_rows
,
727 size_t max_cols
, bool strict
,
728 base::string16
* output
) {
729 RectangleString
rect(max_rows
, max_cols
, strict
, output
);
731 rect
.AddString(input
);
732 return rect
.Finalize();
735 int ElideRectangleText(const base::string16
& input
,
736 const FontList
& font_list
,
737 float available_pixel_width
,
738 int available_pixel_height
,
739 WordWrapBehavior wrap_behavior
,
740 std::vector
<base::string16
>* lines
) {
741 RectangleText
rect(font_list
,
742 available_pixel_width
,
743 available_pixel_height
,
747 rect
.AddString(input
);
748 return rect
.Finalize();
751 base::string16
TruncateString(const base::string16
& string
, size_t length
) {
752 if (string
.size() <= length
)
753 // String fits, return it.
757 // No room for the elide string, return an empty string.
758 return base::string16();
760 size_t max
= length
- 1;
762 // Added to the end of strings that are too big.
763 static const base::char16 kElideString
[] = { 0x2026, 0 };
766 // Just enough room for the elide string.
769 // Use a line iterator to find the first boundary.
770 UErrorCode status
= U_ZERO_ERROR
;
771 scoped_ptr
<icu::RuleBasedBreakIterator
> bi(
772 static_cast<icu::RuleBasedBreakIterator
*>(
773 icu::RuleBasedBreakIterator::createLineInstance(
774 icu::Locale::getDefault(), status
)));
775 if (U_FAILURE(status
))
776 return string
.substr(0, max
) + kElideString
;
777 bi
->setText(string
.c_str());
778 int32_t index
= bi
->preceding(static_cast<int32_t>(max
));
779 if (index
== icu::BreakIterator::DONE
) {
780 index
= static_cast<int32_t>(max
);
782 // Found a valid break (may be the beginning of the string). Now use
783 // a character iterator to find the previous non-whitespace character.
784 icu::StringCharacterIterator
char_iterator(string
.c_str());
786 // No valid line breaks. Start at the end again. This ensures we break
787 // on a valid character boundary.
788 index
= static_cast<int32_t>(max
);
790 char_iterator
.setIndex(index
);
791 while (char_iterator
.hasPrevious()) {
792 char_iterator
.previous();
793 if (!(u_isspace(char_iterator
.current()) ||
794 u_charType(char_iterator
.current()) == U_CONTROL_CHAR
||
795 u_charType(char_iterator
.current()) == U_NON_SPACING_MARK
)) {
796 // Not a whitespace character. Advance the iterator so that we
797 // include the current character in the truncated string.
798 char_iterator
.next();
802 if (char_iterator
.hasPrevious()) {
803 // Found a valid break point.
804 index
= char_iterator
.getIndex();
806 // String has leading whitespace, return the elide string.
810 return string
.substr(0, index
) + kElideString
;