1 // Copyright (C) 2011 The Libphonenumber Authors
3 // Licensed under the Apache License, Version 2.0 (the "License");
4 // you may not use this file except in compliance with the License.
5 // You may obtain a copy of the License at
7 // http://www.apache.org/licenses/LICENSE-2.0
9 // Unless required by applicable law or agreed to in writing, software
10 // distributed under the License is distributed on an "AS IS" BASIS,
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 // See the License for the specific language governing permissions and
13 // limitations under the License.
15 // Author: Lara Rennie
18 // Implementation of a stateful class that finds and extracts telephone numbers
21 #include "phonenumbers/phonenumbermatcher.h"
23 #ifndef I18N_PHONENUMBERS_USE_ICU_REGEXP
24 #error phonenumbermatcher depends on ICU \
25 (i.e. I18N_PHONENUMBERS_USE_ICU_REGEXP must be set)
26 #endif // I18N_PHONENUMBERS_USE_ICU_REGEXP
36 #include <unicode/uchar.h>
38 #include "phonenumbers/alternate_format.h"
39 #include "phonenumbers/base/logging.h"
40 #include "phonenumbers/base/memory/scoped_ptr.h"
41 #include "phonenumbers/base/memory/singleton.h"
42 #include "phonenumbers/callback.h"
43 #include "phonenumbers/default_logger.h"
44 #include "phonenumbers/encoding_utils.h"
45 #include "phonenumbers/normalize_utf8.h"
46 #include "phonenumbers/phonemetadata.pb.h"
47 #include "phonenumbers/phonenumber.pb.h"
48 #include "phonenumbers/phonenumbermatch.h"
49 #include "phonenumbers/phonenumberutil.h"
50 #include "phonenumbers/regexp_adapter.h"
51 #include "phonenumbers/regexp_adapter_icu.h"
52 #include "phonenumbers/stringutil.h"
54 #ifdef I18N_PHONENUMBERS_USE_RE2
55 #include "phonenumbers/regexp_adapter_re2.h"
56 #endif // I18N_PHONENUMBERS_USE_RE2_AND_ICU
60 using std::numeric_limits;
65 namespace phonenumbers {
68 // Returns a regular expression quantifier with an upper and lower limit.
69 string Limit(int lower, int upper) {
72 DCHECK_LT(lower, upper);
73 return StrCat("{", lower, ",", upper, "}");
76 bool IsInvalidPunctuationSymbol(char32 character) {
77 return character == '%' || u_charType(character) == U_CURRENCY_SYMBOL;
80 bool ContainsOnlyValidXChars(const PhoneNumber& number, const string& candidate,
81 const PhoneNumberUtil& util) {
82 // The characters 'x' and 'X' can be (1) a carrier code, in which case they
83 // always precede the national significant number or (2) an extension sign,
84 // in which case they always precede the extension number. We assume a
85 // carrier code is more than 1 digit, so the first case has to have more than
86 // 1 consecutive 'x' or 'X', whereas the second case can only have exactly 1
89 found = candidate.find_first_of("xX");
90 // We ignore the character if 'x' or 'X' appears as the last character of
92 while (found != string::npos && found < candidate.length() - 1) {
93 // We only look for 'x' or 'X' in ASCII form.
94 char next_char = candidate[found + 1];
95 if (next_char == 'x' || next_char == 'X') {
96 // This is the carrier code case, in which the 'X's always precede the
97 // national significant number.
99 if (util.IsNumberMatchWithOneString(
100 number, candidate.substr(found, candidate.length() - found))
101 != PhoneNumberUtil::NSN_MATCH) {
105 string normalized_extension(candidate.substr(found,
106 candidate.length() - found));
107 util.NormalizeDigitsOnly(&normalized_extension);
108 if (normalized_extension != number.extension()) {
112 found = candidate.find_first_of("xX", found + 1);
117 bool AllNumberGroupsRemainGrouped(
118 const PhoneNumberUtil& util,
119 const PhoneNumber& number,
120 const string& normalized_candidate,
121 const vector<string>& formatted_number_groups) {
122 size_t from_index = 0;
123 if (number.country_code_source() != PhoneNumber::FROM_DEFAULT_COUNTRY) {
124 // First skip the country code if the normalized candidate contained it.
125 string country_code = SimpleItoa(number.country_code());
126 from_index = normalized_candidate.find(country_code) + country_code.size();
128 // Check each group of consecutive digits are not broken into separate
129 // groupings in the normalized_candidate string.
130 for (size_t i = 0; i < formatted_number_groups.size(); ++i) {
131 // Fails if the substring of normalized_candidate starting from from_index
132 // doesn't contain the consecutive digits in formatted_number_groups.at(i).
133 from_index = normalized_candidate.find(formatted_number_groups.at(i),
135 if (from_index == string::npos) {
138 // Moves from_index forward.
139 from_index += formatted_number_groups.at(i).length();
140 if (i == 0 && from_index < normalized_candidate.length()) {
141 // We are at the position right after the NDC. We get the region used for
142 // formatting information based on the country code in the phone number,
143 // rather than the number itself, as we do not need to distinguish between
144 // different countries with the same country calling code and this is
147 util.GetRegionCodeForCountryCode(number.country_code(), ®ion);
149 util.GetNddPrefixForRegion(region, true, &ndd_prefix);
150 // Note although normalized_candidate might contain non-ASCII formatting
151 // characters, they won't be treated as ASCII digits when converted to a
153 if (!ndd_prefix.empty() && isdigit(normalized_candidate.at(from_index))) {
154 // This means there is no formatting symbol after the NDC. In this case,
155 // we only accept the number if there is no formatting symbol at all in
156 // the number, except for extensions. This is only important for
157 // countries with national prefixes.
158 string national_significant_number;
159 util.GetNationalSignificantNumber(number, &national_significant_number);
160 return HasPrefixString(normalized_candidate.substr(
161 from_index - formatted_number_groups.at(i).length()),
162 national_significant_number);
166 // The check here makes sure that we haven't mistakenly already used the
167 // extension to match the last group of the subscriber number. Note the
168 // extension cannot have formatting in-between digits.
169 return normalized_candidate.substr(from_index)
170 .find(number.extension()) != string::npos;
173 bool LoadAlternateFormats(PhoneMetadataCollection* alternate_formats) {
174 #if defined(I18N_PHONENUMBERS_USE_ALTERNATE_FORMATS)
175 if (!alternate_formats->ParseFromArray(alternate_format_get(),
176 alternate_format_size())) {
177 LOG(ERROR) << "Could not parse binary data.";
188 class PhoneNumberMatcherRegExps : public Singleton<PhoneNumberMatcherRegExps> {
190 friend class Singleton<PhoneNumberMatcherRegExps>;
192 string opening_parens_;
193 string closing_parens_;
195 // Limit on the number of pairs of brackets in a phone number.
196 string bracket_pair_limit_;
197 // Helper strings for the matching_brackets_ pattern.
198 // An opening bracket at the beginning may not be closed, but subsequent ones
199 // should be. It's also possible that the leading bracket was dropped, so we
200 // shouldn't be surprised if we see a closing bracket first.
201 string leading_maybe_matched_bracket_;
202 string bracket_pairs_;
203 // Limit on the number of leading (plus) characters.
205 // Limit on the number of consecutive punctuation characters.
206 string punctuation_limit_;
207 // The maximum number of digits allowed in a digit-separated block. As we
208 // allow all digits in a single block, this should be set high enough to
209 // accommodate the entire national number and the international country code.
210 int digit_block_limit_;
211 // Limit on the number of blocks separated by punctuation. Uses
212 // kDigitBlockLimit since some formats use spaces to separate each digit.
214 // A punctuation sequence allowing white space.
216 // A digits block without punctuation.
217 string digit_sequence_;
218 // Punctuation that may be at the start of a phone number - brackets and plus
220 string lead_class_chars_;
221 // Same as lead_class_chars_, but enclosed as a character class.
223 // Extra helper strings that form part of pattern_. These are stored
224 // separately since StrCat has a limit of 12 args.
225 string opening_punctuation_;
226 string optional_extn_pattern_;
229 // We use two different reg-ex factories here for performance reasons. RE2 is
230 // much faster for smaller reg-ex patterns, but the main pattern cannot be
231 // handled by RE2 in an efficient way.
232 scoped_ptr<const AbstractRegExpFactory> regexp_factory_for_pattern_;
233 scoped_ptr<const AbstractRegExpFactory> regexp_factory_;
235 // Matches strings that look like publication pages. Example:
236 // Computing Complete Answers to Queries in the Presence of Limited Access
237 // Patterns. Chen Li. VLDB J. 12(3): 211-227 (2003).
239 // The string "211-227 (2003)" is not a telephone number.
240 scoped_ptr<const RegExp> pub_pages_;
241 // Matches strings that look like dates using "/" as a separator. Examples:
242 // 3/10/2011, 31/10/96 or 08/31/95.
243 scoped_ptr<const RegExp> slash_separated_dates_;
244 // Matches timestamps. Examples: "2012-01-02 08:00". Note that the reg-ex does
245 // not include trailing ":\d\d" -- that is covered by time_stamps_suffix_.
246 scoped_ptr<const RegExp> time_stamps_;
247 scoped_ptr<const RegExp> time_stamps_suffix_;
248 // Pattern to check that brackets match. Opening brackets should be closed
249 // within a phone number. This also checks that there is something inside the
250 // brackets. Having no brackets at all is also fine.
251 scoped_ptr<const RegExp> matching_brackets_;
252 // Patterns used to extract phone numbers from a larger phone-number-like
253 // pattern. These are ordered according to specificity. For example,
254 // white-space is last since that is frequently used in numbers, not just to
255 // separate two numbers. We have separate patterns since we don't want to
256 // break up the phone-number-like text on more than one different kind of
257 // symbol at one time, although symbols of the same type (e.g. space) can be
258 // safely grouped together.
260 // Note that if there is a match, we will always check any text found up to
261 // the first match as well.
262 scoped_ptr<vector<const RegExp*> > inner_matches_;
263 scoped_ptr<const RegExp> capture_up_to_second_number_start_pattern_;
264 scoped_ptr<const RegExp> capturing_ascii_digits_pattern_;
265 // Compiled reg-ex representing lead_class_;
266 scoped_ptr<const RegExp> lead_class_pattern_;
267 // Phone number pattern allowing optional punctuation.
268 scoped_ptr<const RegExp> pattern_;
270 PhoneNumberMatcherRegExps()
271 : opening_parens_("(\\[\xEF\xBC\x88\xEF\xBC\xBB" /* "(\\[([" */),
272 closing_parens_(")\\]\xEF\xBC\x89\xEF\xBC\xBD" /* ")\\])]" */),
273 non_parens_(StrCat("[^", opening_parens_, closing_parens_, "]")),
274 bracket_pair_limit_(Limit(0, 3)),
275 leading_maybe_matched_bracket_(StrCat(
276 "(?:[", opening_parens_, "])?",
277 "(?:", non_parens_, "+[", closing_parens_, "])?")),
278 bracket_pairs_(StrCat(
279 "(?:[", opening_parens_, "]", non_parens_, "+",
280 "[", closing_parens_, "])", bracket_pair_limit_)),
281 lead_limit_(Limit(0, 2)),
282 punctuation_limit_(Limit(0, 4)),
283 digit_block_limit_(PhoneNumberUtil::kMaxLengthForNsn +
284 PhoneNumberUtil::kMaxLengthCountryCode),
285 block_limit_(Limit(0, digit_block_limit_)),
286 punctuation_(StrCat("[", PhoneNumberUtil::kValidPunctuation, "]",
287 punctuation_limit_)),
288 digit_sequence_(StrCat("\\p{Nd}", Limit(1, digit_block_limit_))),
289 lead_class_chars_(StrCat(opening_parens_, PhoneNumberUtil::kPlusChars)),
290 lead_class_(StrCat("[", lead_class_chars_, "]")),
291 opening_punctuation_(StrCat("(?:", lead_class_, punctuation_, ")")),
292 optional_extn_pattern_(StrCat(
294 PhoneNumberUtil::GetInstance()->GetExtnPatternsForMatching(),
296 regexp_factory_for_pattern_(new ICURegExpFactory()),
297 #ifdef I18N_PHONENUMBERS_USE_RE2
298 regexp_factory_(new RE2RegExpFactory()),
300 regexp_factory_(new ICURegExpFactory()),
301 #endif // I18N_PHONENUMBERS_USE_RE2
302 pub_pages_(regexp_factory_->CreateRegExp(
303 "\\d{1,5}-+\\d{1,5}\\s{0,4}\\(\\d{1,4}")),
304 slash_separated_dates_(regexp_factory_->CreateRegExp(
305 "(?:(?:[0-3]?\\d/[01]?\\d)|"
306 "(?:[01]?\\d/[0-3]?\\d))/(?:[12]\\d)?\\d{2}")),
307 time_stamps_(regexp_factory_->CreateRegExp(
308 "[12]\\d{3}[-/]?[01]\\d[-/]?[0-3]\\d +[0-2]\\d$")),
309 time_stamps_suffix_(regexp_factory_->CreateRegExp(":[0-5]\\d")),
310 matching_brackets_(regexp_factory_->CreateRegExp(
311 StrCat(leading_maybe_matched_bracket_, non_parens_, "+",
312 bracket_pairs_, non_parens_, "*"))),
313 inner_matches_(new vector<const RegExp*>()),
314 capture_up_to_second_number_start_pattern_(
315 regexp_factory_->CreateRegExp(
316 PhoneNumberUtil::kCaptureUpToSecondNumberStart)),
317 capturing_ascii_digits_pattern_(
318 regexp_factory_->CreateRegExp("(\\d+)")),
319 lead_class_pattern_(regexp_factory_->CreateRegExp(lead_class_)),
320 pattern_(regexp_factory_for_pattern_->CreateRegExp(
321 StrCat("(", opening_punctuation_, lead_limit_,
322 digit_sequence_, "(?:", punctuation_, digit_sequence_, ")",
323 block_limit_, optional_extn_pattern_, ")"))) {
324 inner_matches_->push_back(
325 // Breaks on the slash - e.g. "651-234-2345/332-445-1234"
326 regexp_factory_->CreateRegExp("/+(.*)"));
327 inner_matches_->push_back(
328 // Note that the bracket here is inside the capturing group, since we
329 // consider it part of the phone number. Will match a pattern like
330 // "(650) 223 3345 (754) 223 3321".
331 regexp_factory_->CreateRegExp("(\\([^(]*)"));
332 inner_matches_->push_back(
333 // Breaks on a hyphen - e.g. "12345 - 332-445-1234 is my number." We
334 // require a space on either side of the hyphen for it to be considered
336 regexp_factory_->CreateRegExp("(?:\\p{Z}-|-\\p{Z})\\p{Z}*(.+)"));
337 inner_matches_->push_back(
338 // Various types of wide hyphens. Note we have decided not to enforce a
339 // space here, since it's possible that it's supposed to be used to
340 // break two numbers without spaces, and we haven't seen many instances
341 // of it used within a number.
342 regexp_factory_->CreateRegExp(
343 "[\xE2\x80\x92-\x2D\xE2\x80\x95\xEF\xBC\x8D]" /* "‒-―-" */
345 inner_matches_->push_back(
346 // Breaks on a full stop - e.g. "12345. 332-445-1234 is my number."
347 regexp_factory_->CreateRegExp("\\.+\\p{Z}*([^.]+)"));
348 inner_matches_->push_back(
349 // Breaks on space - e.g. "3324451234 8002341234"
350 regexp_factory_->CreateRegExp("\\p{Z}+(\\P{Z}+)"));
354 DISALLOW_COPY_AND_ASSIGN(PhoneNumberMatcherRegExps);
357 class AlternateFormats : public Singleton<AlternateFormats> {
359 PhoneMetadataCollection format_data_;
361 map<int, const PhoneMetadata*> calling_code_to_alternate_formats_map_;
365 calling_code_to_alternate_formats_map_() {
366 if (!LoadAlternateFormats(&format_data_)) {
367 LOG(DFATAL) << "Could not parse compiled-in metadata.";
370 for (RepeatedPtrField<PhoneMetadata>::const_iterator it =
371 format_data_.metadata().begin();
372 it != format_data_.metadata().end();
374 calling_code_to_alternate_formats_map_.insert(
375 make_pair(it->country_code(), &*it));
379 const PhoneMetadata* GetAlternateFormatsForCountry(int country_calling_code)
381 map<int, const PhoneMetadata*>::const_iterator it =
382 calling_code_to_alternate_formats_map_.find(country_calling_code);
383 if (it != calling_code_to_alternate_formats_map_.end()) {
390 DISALLOW_COPY_AND_ASSIGN(AlternateFormats);
393 PhoneNumberMatcher::PhoneNumberMatcher(const PhoneNumberUtil& util,
395 const string& region_code,
396 PhoneNumberMatcher::Leniency leniency,
398 : reg_exps_(PhoneNumberMatcherRegExps::GetInstance()),
399 alternate_formats_(AlternateFormats::GetInstance()),
402 preferred_region_(region_code),
404 max_tries_(max_tries),
410 PhoneNumberMatcher::PhoneNumberMatcher(const string& text,
411 const string& region_code)
412 : reg_exps_(PhoneNumberMatcherRegExps::GetInstance()),
413 alternate_formats_(NULL), // Not used.
414 phone_util_(*PhoneNumberUtil::GetInstance()),
416 preferred_region_(region_code),
418 max_tries_(numeric_limits<int>::max()),
424 PhoneNumberMatcher::~PhoneNumberMatcher() {
428 bool PhoneNumberMatcher::IsLatinLetter(char32 letter) {
429 // Combining marks are a subset of non-spacing-mark.
430 if (!u_isalpha(letter) && (u_charType(letter) != U_NON_SPACING_MARK)) {
433 UBlockCode block = ublock_getCode(letter);
434 return ((block == UBLOCK_BASIC_LATIN) ||
435 (block == UBLOCK_LATIN_1_SUPPLEMENT) ||
436 (block == UBLOCK_LATIN_EXTENDED_A) ||
437 (block == UBLOCK_LATIN_EXTENDED_ADDITIONAL) ||
438 (block == UBLOCK_LATIN_EXTENDED_B) ||
439 (block == UBLOCK_COMBINING_DIACRITICAL_MARKS));
442 bool PhoneNumberMatcher::ParseAndVerify(const string& candidate, int offset,
443 PhoneNumberMatch* match) {
445 // Check the candidate doesn't contain any formatting which would indicate
446 // that it really isn't a phone number.
447 if (!reg_exps_->matching_brackets_->FullMatch(candidate) ||
448 reg_exps_->pub_pages_->PartialMatch(candidate)) {
452 // If leniency is set to VALID or stricter, we also want to skip numbers that
453 // are surrounded by Latin alphabetic characters, to skip cases like
454 // abc8005001234 or 8005001234def.
455 if (leniency_ >= VALID) {
456 // If the candidate is not at the start of the text, and does not start with
457 // phone-number punctuation, check the previous character.
458 scoped_ptr<RegExpInput> candidate_input(
459 reg_exps_->regexp_factory_->CreateInput(candidate));
461 !reg_exps_->lead_class_pattern_->Consume(candidate_input.get())) {
462 char32 previous_char;
463 const char* previous_char_ptr =
464 EncodingUtils::BackUpOneUTF8Character(text_.c_str(),
465 text_.c_str() + offset);
466 EncodingUtils::DecodeUTF8Char(previous_char_ptr, &previous_char);
467 // We return false if it is a latin letter or an invalid punctuation
469 if (IsInvalidPunctuationSymbol(previous_char) ||
470 IsLatinLetter(previous_char)) {
474 size_t lastCharIndex = offset + candidate.length();
475 if (lastCharIndex < text_.length()) {
477 const char* next_char_ptr =
478 EncodingUtils::AdvanceOneUTF8Character(
479 text_.c_str() + lastCharIndex - 1);
480 EncodingUtils::DecodeUTF8Char(next_char_ptr, &next_char);
481 if (IsInvalidPunctuationSymbol(next_char) || IsLatinLetter(next_char)) {
488 if (phone_util_.ParseAndKeepRawInput(candidate, preferred_region_, &number) !=
489 PhoneNumberUtil::NO_PARSING_ERROR) {
494 // Check Israel * numbers: these are a special case in that they are
495 // four-digit numbers that our library supports, but they can only be dialled
496 // with a leading *. Since we don't actually store or detect the * in our
497 // phone number library, this means in practice we detect most four digit
498 // numbers as being valid for Israel. We are considering moving these numbers
499 // to ShortNumberInfo instead, in which case this problem would go away, but
500 // in the meantime we want to restrict the false matches so we only allow
501 // these numbers if they are preceded by a star. We enforce this for all
502 // leniency levels even though these numbers are technically accepted by
503 // isPossibleNumber and isValidNumber since we consider it to be a deficiency
504 // in those methods that they accept these numbers without the *.
505 // TODO: Remove this or make it significantly less hacky once
506 // we've decided how to handle these short codes going forward in
507 // ShortNumberInfo. We could use the formatting rules for instance, but that
510 phone_util_.GetRegionCodeForCountryCode(number.country_code(), ®ion_code);
511 if (region_code == "IL") {
512 string national_number;
513 phone_util_.GetNationalSignificantNumber(number, &national_number);
514 if (national_number.length() == 4 &&
515 // Just check the previous char, since * is an ASCII character.
516 (offset == 0 || (offset > 0 && text_[offset - 1] != '*'))) {
522 if (VerifyAccordingToLeniency(leniency_, number, candidate)) {
523 match->set_start(offset);
524 match->set_raw_string(candidate);
525 // We used ParseAndKeepRawInput to create this number, but for now we don't
526 // return the extra values parsed. TODO: stop clearing all values here and
527 // switch all users over to using raw_input() rather than the raw_string()
528 // of PhoneNumberMatch.
529 number.clear_country_code_source();
530 number.clear_preferred_domestic_carrier_code();
531 number.clear_raw_input();
532 match->set_number(number);
538 // Helper method to replace the verification method for each enum in the Java
540 bool PhoneNumberMatcher::VerifyAccordingToLeniency(
541 Leniency leniency, const PhoneNumber& number,
542 const string& candidate) const {
544 case PhoneNumberMatcher::POSSIBLE:
545 return phone_util_.IsPossibleNumber(number);
546 case PhoneNumberMatcher::VALID:
547 if (!phone_util_.IsValidNumber(number) ||
548 !ContainsOnlyValidXChars(number, candidate, phone_util_)) {
551 return IsNationalPrefixPresentIfRequired(number);
552 case PhoneNumberMatcher::STRICT_GROUPING: {
553 if (!phone_util_.IsValidNumber(number) ||
554 !ContainsOnlyValidXChars(number, candidate, phone_util_) ||
555 ContainsMoreThanOneSlashInNationalNumber(
556 number, candidate, phone_util_) ||
557 !IsNationalPrefixPresentIfRequired(number)) {
560 ResultCallback4<bool, const PhoneNumberUtil&, const PhoneNumber&,
561 const string&, const vector<string>&>* callback =
562 NewPermanentCallback(&AllNumberGroupsRemainGrouped);
563 bool is_valid = CheckNumberGroupingIsValid(number, candidate, callback);
567 case PhoneNumberMatcher::EXACT_GROUPING: {
568 if (!phone_util_.IsValidNumber(number) ||
569 !ContainsOnlyValidXChars(number, candidate, phone_util_) ||
570 ContainsMoreThanOneSlashInNationalNumber(
571 number, candidate, phone_util_) ||
572 !IsNationalPrefixPresentIfRequired(number)) {
575 ResultCallback4<bool, const PhoneNumberUtil&, const PhoneNumber&,
576 const string&, const vector<string>&>* callback =
577 NewPermanentCallback(
578 this, &PhoneNumberMatcher::AllNumberGroupsAreExactlyPresent);
579 bool is_valid = CheckNumberGroupingIsValid(number, candidate, callback);
584 LOG(ERROR) << "No implementation defined for verification for leniency "
585 << static_cast<int>(leniency);
590 bool PhoneNumberMatcher::ExtractInnerMatch(const string& candidate, int offset,
591 PhoneNumberMatch* match) {
593 for (vector<const RegExp*>::const_iterator regex =
594 reg_exps_->inner_matches_->begin();
595 regex != reg_exps_->inner_matches_->end(); regex++) {
596 scoped_ptr<RegExpInput> candidate_input(
597 reg_exps_->regexp_factory_->CreateInput(candidate));
598 bool is_first_match = true;
600 while ((*regex)->FindAndConsume(candidate_input.get(), &group) &&
602 int group_start_index = candidate.length() -
603 candidate_input->ToString().length() - group.length();
604 if (is_first_match) {
605 // We should handle any group before this one too.
606 string first_group_only = candidate.substr(0, group_start_index);
607 phone_util_.TrimUnwantedEndChars(&first_group_only);
608 bool success = ParseAndVerify(first_group_only, offset, match);
613 is_first_match = false;
615 phone_util_.TrimUnwantedEndChars(&group);
616 bool success = ParseAndVerify(group, offset + group_start_index, match);
626 bool PhoneNumberMatcher::ExtractMatch(const string& candidate, int offset,
627 PhoneNumberMatch* match) {
629 // Skip a match that is more likely to be a date.
630 if (reg_exps_->slash_separated_dates_->PartialMatch(candidate)) {
634 // Skip potential time-stamps.
635 if (reg_exps_->time_stamps_->PartialMatch(candidate)) {
636 scoped_ptr<RegExpInput> following_text(
637 reg_exps_->regexp_factory_->CreateInput(
638 text_.substr(offset + candidate.size())));
639 if (reg_exps_->time_stamps_suffix_->Consume(following_text.get())) {
644 // Try to come up with a valid match given the entire candidate.
645 if (ParseAndVerify(candidate, offset, match)) {
649 // If that failed, try to find an "inner match" - there might be a phone
650 // number within this candidate.
651 return ExtractInnerMatch(candidate, offset, match);
654 bool PhoneNumberMatcher::HasNext() {
655 if (state_ == NOT_READY) {
656 PhoneNumberMatch temp_match;
657 if (!Find(search_index_, &temp_match)) {
660 last_match_.reset(new PhoneNumberMatch(temp_match.start(),
661 temp_match.raw_string(),
662 temp_match.number()));
663 search_index_ = last_match_->end();
667 return state_ == READY;
670 bool PhoneNumberMatcher::Next(PhoneNumberMatch* match) {
672 // Check the state and find the next match as a side-effect if necessary.
676 match->CopyFrom(*last_match_);
678 last_match_.reset(NULL);
682 bool PhoneNumberMatcher::Find(int index, PhoneNumberMatch* match) {
685 scoped_ptr<RegExpInput> text(
686 reg_exps_->regexp_factory_for_pattern_->CreateInput(text_.substr(index)));
688 while ((max_tries_ > 0) &&
689 reg_exps_->pattern_->FindAndConsume(text.get(), &candidate)) {
690 int start = text_.length() - text->ToString().length() - candidate.length();
691 // Check for extra numbers at the end.
692 reg_exps_->capture_up_to_second_number_start_pattern_->
693 PartialMatch(candidate, &candidate);
694 if (ExtractMatch(candidate, start, match)) {
698 index = start + candidate.length();
704 bool PhoneNumberMatcher::CheckNumberGroupingIsValid(
705 const PhoneNumber& phone_number,
706 const string& candidate,
707 ResultCallback4<bool, const PhoneNumberUtil&, const PhoneNumber&,
708 const string&, const vector<string>&>* checker) const {
710 // TODO: Evaluate how this works for other locales (testing has been limited
711 // to NANPA regions) and optimise if necessary.
712 string normalized_candidate =
713 NormalizeUTF8::NormalizeDecimalDigits(candidate);
714 vector<string> formatted_number_groups;
715 GetNationalNumberGroups(phone_number, NULL, // Use default formatting pattern
716 &formatted_number_groups);
717 if (checker->Run(phone_util_, phone_number, normalized_candidate,
718 formatted_number_groups)) {
721 // If this didn't pass, see if there are any alternate formats, and try them
723 const PhoneMetadata* alternate_formats =
724 alternate_formats_->GetAlternateFormatsForCountry(
725 phone_number.country_code());
726 if (alternate_formats) {
727 for (RepeatedPtrField<NumberFormat>::const_iterator it =
728 alternate_formats->number_format().begin();
729 it != alternate_formats->number_format().end(); ++it) {
730 formatted_number_groups.clear();
731 GetNationalNumberGroups(phone_number, &*it, &formatted_number_groups);
732 if (checker->Run(phone_util_, phone_number, normalized_candidate,
733 formatted_number_groups)) {
741 // Helper method to get the national-number part of a number, formatted without
742 // any national prefix, and return it as a set of digit blocks that would be
743 // formatted together.
744 void PhoneNumberMatcher::GetNationalNumberGroups(
745 const PhoneNumber& number,
746 const NumberFormat* formatting_pattern,
747 vector<string>* digit_blocks) const {
748 string rfc3966_format;
749 if (!formatting_pattern) {
750 // This will be in the format +CC-DG;ext=EXT where DG represents groups of
752 phone_util_.Format(number, PhoneNumberUtil::RFC3966, &rfc3966_format);
753 // We remove the extension part from the formatted string before splitting
754 // it into different groups.
755 size_t end_index = rfc3966_format.find(';');
756 if (end_index == string::npos) {
757 end_index = rfc3966_format.length();
759 // The country-code will have a '-' following it.
760 size_t start_index = rfc3966_format.find('-') + 1;
761 SplitStringUsing(rfc3966_format.substr(start_index,
762 end_index - start_index),
765 // We format the NSN only, and split that according to the separator.
766 string national_significant_number;
767 phone_util_.GetNationalSignificantNumber(number,
768 &national_significant_number);
769 phone_util_.FormatNsnUsingPattern(national_significant_number,
771 PhoneNumberUtil::RFC3966,
773 SplitStringUsing(rfc3966_format, "-", digit_blocks);
777 bool PhoneNumberMatcher::IsNationalPrefixPresentIfRequired(
778 const PhoneNumber& number) const {
779 // First, check how we deduced the country code. If it was written in
780 // international format, then the national prefix is not required.
781 if (number.country_code_source() != PhoneNumber::FROM_DEFAULT_COUNTRY) {
784 string phone_number_region;
785 phone_util_.GetRegionCodeForCountryCode(
786 number.country_code(), &phone_number_region);
787 const PhoneMetadata* metadata =
788 phone_util_.GetMetadataForRegion(phone_number_region);
792 // Check if a national prefix should be present when formatting this number.
793 string national_number;
794 phone_util_.GetNationalSignificantNumber(number, &national_number);
795 const NumberFormat* format_rule =
796 phone_util_.ChooseFormattingPatternForNumber(metadata->number_format(),
798 // To do this, we check that a national prefix formatting rule was present and
799 // that it wasn't just the first-group symbol ($1) with punctuation.
800 if (format_rule && !format_rule->national_prefix_formatting_rule().empty()) {
801 if (format_rule->national_prefix_optional_when_formatting()) {
802 // The national-prefix is optional in these cases, so we don't need to
803 // check if it was present.
806 if (phone_util_.FormattingRuleHasFirstGroupOnly(
807 format_rule->national_prefix_formatting_rule())) {
808 // National Prefix not needed for this number.
811 // Normalize the remainder.
812 string raw_input_copy(number.raw_input());
813 // Check if we found a national prefix and/or carrier code at the start of
814 // the raw input, and return the result.
815 phone_util_.NormalizeDigitsOnly(&raw_input_copy);
816 return phone_util_.MaybeStripNationalPrefixAndCarrierCode(
819 NULL); // Don't need to keep the stripped carrier code.
824 bool PhoneNumberMatcher::AllNumberGroupsAreExactlyPresent(
825 const PhoneNumberUtil& util,
826 const PhoneNumber& phone_number,
827 const string& normalized_candidate,
828 const vector<string>& formatted_number_groups) const {
829 const scoped_ptr<RegExpInput> candidate_number(
830 reg_exps_->regexp_factory_->CreateInput(normalized_candidate));
831 vector<string> candidate_groups;
833 while (reg_exps_->capturing_ascii_digits_pattern_->FindAndConsume(
834 candidate_number.get(),
836 candidate_groups.push_back(digit_block);
839 // Set this to the last group, skipping it if the number has an extension.
840 int candidate_number_group_index =
841 phone_number.has_extension() ? candidate_groups.size() - 2
842 : candidate_groups.size() - 1;
843 // First we check if the national significant number is formatted as a block.
844 // We use find and not equals, since the national significant number may be
845 // present with a prefix such as a national number prefix, or the country code
847 string national_significant_number;
848 util.GetNationalSignificantNumber(phone_number,
849 &national_significant_number);
850 if (candidate_groups.size() == 1 ||
851 candidate_groups.at(candidate_number_group_index).find(
852 national_significant_number) != string::npos) {
855 // Starting from the end, go through in reverse, excluding the first group,
856 // and check the candidate and number groups are the same.
857 for (int formatted_number_group_index =
858 (formatted_number_groups.size() - 1);
859 formatted_number_group_index > 0 &&
860 candidate_number_group_index >= 0;
861 --formatted_number_group_index, --candidate_number_group_index) {
862 if (candidate_groups.at(candidate_number_group_index) !=
863 formatted_number_groups.at(formatted_number_group_index)) {
867 // Now check the first group. There may be a national prefix at the start, so
868 // we only check that the candidate group ends with the formatted number
870 return (candidate_number_group_index >= 0 &&
871 HasSuffixString(candidate_groups.at(candidate_number_group_index),
872 formatted_number_groups.at(0)));
876 bool PhoneNumberMatcher::ContainsMoreThanOneSlashInNationalNumber(
877 const PhoneNumber& number,
878 const string& candidate,
879 const PhoneNumberUtil& util) {
880 size_t first_slash_in_body = candidate.find('/');
881 if (first_slash_in_body == string::npos) {
882 // No slashes, this is okay.
885 // Now look for a second one.
886 size_t second_slash_in_body = candidate.find('/', first_slash_in_body + 1);
887 if (second_slash_in_body == string::npos) {
888 // Only one slash, this is okay.
892 // If the first slash is after the country calling code, this is permitted.
893 if (number.country_code_source() == PhoneNumber::FROM_NUMBER_WITH_PLUS_SIGN ||
894 number.country_code_source() ==
895 PhoneNumber::FROM_NUMBER_WITHOUT_PLUS_SIGN) {
896 string normalized_country_code =
897 candidate.substr(0, first_slash_in_body);
898 util.NormalizeDigitsOnly(&normalized_country_code);
899 if (normalized_country_code == SimpleItoa(number.country_code())) {
900 // Any more slashes and this is illegal.
901 return candidate.find('/', second_slash_in_body + 1) != string::npos;
907 } // namespace phonenumbers