diff options
author | philip.liard@gmail.com <philip.liard@gmail.com@ee073f10-1060-11df-b6a4-87a95322a99c> | 2011-09-21 17:43:54 +0000 |
---|---|---|
committer | philip.liard@gmail.com <philip.liard@gmail.com@ee073f10-1060-11df-b6a4-87a95322a99c> | 2011-09-21 17:43:54 +0000 |
commit | 6a0a07f4f0066eba2dc9bb81465f4e67d670c7b6 (patch) | |
tree | 0ae8d76ac1c43f0aada6fcf6b2c4449b8340e1da | |
parent | 70942011411fc60b15cddf0a3032b1614777dbc4 (diff) | |
download | phonenumbers-6a0a07f4f0066eba2dc9bb81465f4e67d670c7b6.tar.gz |
CPP: Add phonenumbermatcher.
git-svn-id: http://libphonenumber.googlecode.com/svn/trunk/cpp/src/phonenumbers@356 ee073f10-1060-11df-b6a4-87a95322a99c
-rw-r--r-- | encoding_utils.h | 11 | ||||
-rw-r--r-- | phonenumbermatch.cc | 91 | ||||
-rw-r--r-- | phonenumbermatch.h | 125 | ||||
-rw-r--r-- | phonenumbermatcher.cc | 626 | ||||
-rw-r--r-- | phonenumbermatcher.h | 158 | ||||
-rw-r--r-- | phonenumberutil.h | 3 | ||||
-rw-r--r-- | stringutil.cc | 58 | ||||
-rw-r--r-- | stringutil.h | 19 |
8 files changed, 1091 insertions, 0 deletions
diff --git a/encoding_utils.h b/encoding_utils.h index 85819a5..415ce62 100644 --- a/encoding_utils.h +++ b/encoding_utils.h @@ -16,6 +16,7 @@ #define I18N_PHONENUMBERS_ENCODING_UTILS_H_ #include "base/basictypes.h" +#include "phonenumbers/utf/unilib.h" #include "phonenumbers/utf/utf.h" namespace i18n { @@ -32,6 +33,16 @@ class EncodingUtils { *out = r; return len; } + + static const char* AdvanceOneUTF8Character(const char* buf_utf8) { + return buf_utf8 + UniLib::OneCharLen(buf_utf8); + } + + static const char* BackUpOneUTF8Character(const char* start, + const char* end) { + while (start < end && UniLib::IsTrailByte(*--end)) {} + return end; + } }; } // namespace phonenumbers diff --git a/phonenumbermatch.cc b/phonenumbermatch.cc new file mode 100644 index 0000000..bcf5efb --- /dev/null +++ b/phonenumbermatch.cc @@ -0,0 +1,91 @@ +// Copyright (C) 2011 The Libphonenumber Authors +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// Author: Tao Huang +// +// Implementation of a mutable match of a phone number within a piece of +// text. Matches may be found using PhoneNumberUtil::FindNumbers. + +#include "phonenumbers/phonenumbermatch.h" + +#include <string> + +#include "phonenumbers/phonenumber.h" +#include "phonenumbers/phonenumber.pb.h" +#include "phonenumbers/stringutil.h" + +namespace i18n { +namespace phonenumbers { + +PhoneNumberMatch::PhoneNumberMatch(int start, + const string& raw_string, + const PhoneNumber& number) + : start_(start), raw_string_(raw_string), number_(number) { +} + +PhoneNumberMatch::PhoneNumberMatch() + : start_(-1), raw_string_(""), number_(PhoneNumber::default_instance()) { +} + +const PhoneNumber& PhoneNumberMatch::number() const { + return number_; +} + +int PhoneNumberMatch::start() const { + return start_; +} + +int PhoneNumberMatch::end() const { + return start_ + raw_string_.length(); +} + +int PhoneNumberMatch::length() const { + return raw_string_.length(); +} + +const string& PhoneNumberMatch::raw_string() const { + return raw_string_; +} + +void PhoneNumberMatch::set_start(int start) { + start_ = start; +} + +void PhoneNumberMatch::set_raw_string(const string& raw_string) { + raw_string_ = raw_string; +} + +void PhoneNumberMatch::set_number(const PhoneNumber& number) { + number_.CopyFrom(number); +} + +string PhoneNumberMatch::ToString() const { + return StrCat("PhoneNumberMatch [", start(), ",", end(), ") ", + raw_string_.c_str()); +} + +bool PhoneNumberMatch::Equals(const PhoneNumberMatch& match) const { + return ExactlySameAs(match.number_, number_) && + match.raw_string_.compare(raw_string_) == 0 && + match.start_ == start_; +} + +void PhoneNumberMatch::CopyFrom(const PhoneNumberMatch& match) { + raw_string_ = match.raw_string(); + start_ = match.start(); + number_ = match.number(); +} + +} // namespace phonenumbers +} // namespace i18n diff --git a/phonenumbermatch.h b/phonenumbermatch.h new file mode 100644 index 0000000..5ebfd9d --- /dev/null +++ b/phonenumbermatch.h @@ -0,0 +1,125 @@ +// Copyright (C) 2011 The Libphonenumber Authors +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// Author: Tao Huang +// +// A mutable match of a phone number within a piece of text. +// Matches may be found using PhoneNumberUtil::FindNumbers. +// +// A match consists of the phone number as well as the start and end offsets of +// the corresponding subsequence of the searched text. Use raw_string() to +// obtain a copy of the matched subsequence. +// +// The following annotated example clarifies the relationship between the +// searched text, the match offsets, and the parsed number: +// +// string text = "Call me at +1 425 882-8080 for details."; +// const string country = "US"; +// +// // Find the first phone number match: +// PhoneNumberMatcher matcher(text, country); +// if (matcher.HasNext()) { +// PhoneNumberMatch match; +// matcher.Next(&match); +// } +// +// // raw_string() contains the phone number as it appears in the text. +// "+1 425 882-8080" == match.raw_string(); +// +// // start() and end() define the range of the matched subsequence. +// string subsequence = text.substr(match.start(), match.end()); +// "+1 425 882-8080" == subsequence; +// +// // number() returns the the same result as PhoneNumberUtil::Parse() +// // invoked on raw_string(). +// const PhoneNumberUtil& util = *PhoneNumberUtil::GetInstance(); +// util.Parse(match.raw_string(), country).Equals(match.number()); +// +// This class is a port of PhoneNumberMatch.java + +#ifndef I18N_PHONENUMBERS_PHONENUMBERMATCH_H_ +#define I18N_PHONENUMBERS_PHONENUMBERMATCH_H_ + +#include <string> + +#include "base/basictypes.h" +#include "phonenumbers/phonenumber.pb.h" + +namespace i18n { +namespace phonenumbers { + +using std::string; + +class PhoneNumberMatch { + public: + // Creates a new match. + // - start is the index into the target text. + // - match is the matched string of the target text. + // - number is the matched phone number. + PhoneNumberMatch(int start, + const string& raw_string, + const PhoneNumber& number); + + // Default constructor. + PhoneNumberMatch(); + + ~PhoneNumberMatch() {} + + // Returns the phone number matched by the receiver. + const PhoneNumber& number() const; + + // Returns the start index of the matched phone number within the searched + // text. + int start() const; + + // Returns the exclusive end index of the matched phone number within the + // searched text. + int end() const; + + // Returns the length of the text matched in the searched text. + int length() const; + + // Returns the raw string matched as a phone number in the searched text. + const string& raw_string() const; + + // Returns a string containing debug information. + string ToString() const; + + void set_start(int start); + + void set_raw_string(const string& raw_string); + + void set_number(const PhoneNumber& number); + + bool Equals(const PhoneNumberMatch& number) const; + + void CopyFrom(const PhoneNumberMatch& number); + + private: + // The start index into the text. + int start_; + + // The raw substring matched. + string raw_string_; + + // The matched phone number. + PhoneNumber number_; + + DISALLOW_COPY_AND_ASSIGN(PhoneNumberMatch); +}; + +} // namespace phonenumbers +} // namespace i18n + +#endif // I18N_PHONENUMBERS_PHONENUMBERMATCH_H_ diff --git a/phonenumbermatcher.cc b/phonenumbermatcher.cc new file mode 100644 index 0000000..a3d528f --- /dev/null +++ b/phonenumbermatcher.cc @@ -0,0 +1,626 @@ +// Copyright (C) 2011 The Libphonenumber Authors +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// Author: Lara Rennie +// Author: Tao Huang +// +// Implementation of a stateful class that finds and extracts telephone numbers +// from text. + +#include "phonenumbers/phonenumbermatcher.h" + +#ifndef USE_ICU_REGEXP +#error phonenumbermatcher depends on ICU (i.e. USE_ICU_REGEXP must be set) +#endif // USE_ICU_REGEXP + +#include <limits> +#include <string> +#include <unicode/uchar.h> + +#include "base/logging.h" +#include "base/memory/scoped_ptr.h" +#include "base/memory/singleton.h" +#include "phonenumbers/default_logger.h" +#include "phonenumbers/encoding_utils.h" +#include "phonenumbers/normalize_utf8.h" +#include "phonenumbers/phonenumber.pb.h" +#include "phonenumbers/phonenumbermatch.h" +#include "phonenumbers/phonenumberutil.h" +#include "phonenumbers/regexp_adapter.h" +#include "phonenumbers/regexp_adapter_icu.h" +#include "phonenumbers/stringutil.h" + +#ifdef USE_RE2 +#include "phonenumbers/regexp_adapter_re2.h" +#endif // USE_RE2_AND_ICU + +using std::numeric_limits; +using std::string; +using std::vector; + +namespace i18n { +namespace phonenumbers { + +namespace { +// Returns a regular expression quantifier with an upper and lower limit. +string Limit(int lower, int upper) { + DCHECK_GE(lower, 0); + DCHECK_GT(upper, 0); + DCHECK_LT(lower, upper); + return StrCat("{", lower, ",", upper, "}"); +} + +bool IsCurrencySymbol(char32 character) { + return (u_charType(character) == U_CURRENCY_SYMBOL); +} + +// Helper method to get the national-number part of a number, formatted without +// any national prefix, and return it as a set of digit blocks that would be +// formatted together. +void GetNationalNumberGroups(const PhoneNumberUtil& util, + const PhoneNumber& number, + vector<string>* digit_blocks) { + // This will be in the format +CC-DG;ext=EXT where DG represents groups of + // digits. + string rfc3966_format; + util.Format(number, PhoneNumberUtil::RFC3966, &rfc3966_format); + // We remove the extension part from the formatted string before splitting it + // into different groups. + size_t end_index = rfc3966_format.find(';'); + if (end_index == string::npos) { + end_index = rfc3966_format.length(); + } + // The country-code will have a '-' following it. + size_t start_index = rfc3966_format.find('-') + 1; + SplitStringUsing(rfc3966_format.substr(start_index, end_index - start_index), + "-", digit_blocks); +} + +bool ContainsOnlyValidXChars(const PhoneNumber& number, const string& candidate, + const PhoneNumberUtil& util) { + // The characters 'x' and 'X' can be (1) a carrier code, in which case they + // always precede the national significant number or (2) an extension sign, + // in which case they always precede the extension number. We assume a + // carrier code is more than 1 digit, so the first case has to have more than + // 1 consecutive 'x' or 'X', whereas the second case can only have exactly 1 + // 'x' or 'X'. + size_t found; + found = candidate.find_first_of("xX"); + // We ignore the character if 'x' or 'X' appears as the last character of + // the string. + while (found != string::npos && found < candidate.length() - 1) { + // We only look for 'x' or 'X' in ASCII form. + char next_char = candidate[found + 1]; + if (next_char == 'x' || next_char == 'X') { + // This is the carrier code case, in which the 'X's always precede the + // national significant number. + ++found; + if (util.IsNumberMatchWithOneString( + number, candidate.substr(found, candidate.length() - found)) + != PhoneNumberUtil::NSN_MATCH) { + return false; + } + } else { + string normalized_extension(candidate.substr(found, + candidate.length() - found)); + util.NormalizeDigitsOnly(&normalized_extension); + if (normalized_extension != number.extension()) { + return false; + } + } + found = candidate.find_first_of("xX", found + 1); + } + return true; +} +} // namespace + +#ifdef USE_GOOGLE_BASE +class PhoneNumberMatcherRegExps { + friend struct DefaultSingletonTraits<PhoneNumberMatcherRegExps>; +#else +class PhoneNumberMatcherRegExps : public Singleton<PhoneNumberMatcherRegExps> { + friend class Singleton<PhoneNumberMatcherRegExps>; +#endif // USE_GOOGLE_BASE + private: + string opening_parens_; + string closing_parens_; + string non_parens_; + // Limit on the number of pairs of brackets in a phone number. + string bracket_pair_limit_; + // Helper strings for the matching_brackets_ pattern. + // An opening bracket at the beginning may not be closed, but subsequent ones + // should be. It's also possible that the leading bracket was dropped, so we + // shouldn't be surprised if we see a closing bracket first. + string leading_maybe_matched_bracket_; + string bracket_pairs_; + // Limit on the number of leading (plus) characters. + string lead_limit_; + // Limit on the number of consecutive punctuation characters. + string punctuation_limit_; + // The maximum number of digits allowed in a digit-separated block. As we + // allow all digits in a single block, this should be set high enough to + // accommodate the entire national number and the international country code. + int digit_block_limit_; + // Limit on the number of blocks separated by punctuation. Uses + // kDigitBlockLimit since some formats use spaces to separate each digit. + string block_limit_; + // A punctuation sequence allowing white space. + string punctuation_; + // A digits block without punctuation. + string digit_sequence_; + // Punctuation that may be at the start of a phone number - brackets and plus + // signs. + string lead_class_chars_; + // Same as lead_class_chars_, but enclosed as a character class. + string lead_class_; + // Extra helper strings that form part of pattern_. These are stored + // separately since StrCat has a limit of 12 args. + string opening_punctuation_; + string optional_extn_pattern_; + + public: + // We use two different reg-ex factories here for performance reasons. RE2 is + // much faster for smaller reg-ex patterns, but the main pattern cannot be + // handled by RE2 in an efficient way. + scoped_ptr<const AbstractRegExpFactory> regexp_factory_for_pattern_; + scoped_ptr<const AbstractRegExpFactory> regexp_factory_; + + // Matches strings that look like publication pages. Example: + // Computing Complete Answers to Queries in the Presence of Limited Access + // Patterns. Chen Li. VLDB J. 12(3): 211-227 (2003). + // + // The string "211-227 (2003)" is not a telephone number. + scoped_ptr<const RegExp> pub_pages_; + // Matches strings that look like dates using "/" as a separator. Examples: + // 3/10/2011, 31/10/96 or 08/31/95. + scoped_ptr<const RegExp> slash_separated_dates_; + // Pattern to check that brackets match. Opening brackets should be closed + // within a phone number. This also checks that there is something inside the + // brackets. Having no brackets at all is also fine. + scoped_ptr<const RegExp> matching_brackets_; + // Matches white-space, which may indicate the end of a phone number and the + // start of something else (such as a neighbouring zip-code). If white-space + // is found, continues to match all characters that are not typically used to + // start a phone number. + scoped_ptr<const RegExp> group_separator_; + scoped_ptr<const RegExp> capture_up_to_second_number_start_pattern_; + scoped_ptr<const RegExp> capturing_ascii_digits_pattern_; + // Compiled reg-ex representing lead_class_; + scoped_ptr<const RegExp> lead_class_pattern_; + // Phone number pattern allowing optional punctuation. + scoped_ptr<const RegExp> pattern_; + +#ifdef USE_GOOGLE_BASE + PhoneNumberMatcherRegExps* PhoneNumberMatcherRegExps::GetInstance() { + return Singleton<PhoneNumberMatcherRegExps>::get(); + } +#endif // USE_GOOGLE_BASE + + PhoneNumberMatcherRegExps() + : opening_parens_("(\\[\xEF\xBC\x88\xEF\xBC\xBB" /* "(\\[([" */), + closing_parens_(")\\]\xEF\xBC\x89\xEF\xBC\xBD" /* ")\\])]" */), + non_parens_(StrCat("[^", opening_parens_, closing_parens_, "]")), + bracket_pair_limit_(Limit(0, 3)), + leading_maybe_matched_bracket_(StrCat( + "(?:[", opening_parens_, "])?", + "(?:", non_parens_, "+[", closing_parens_, "])?")), + bracket_pairs_(StrCat( + "(?:[", opening_parens_, "]", non_parens_, "+", + "[", closing_parens_, "])", bracket_pair_limit_)), + lead_limit_(Limit(0, 2)), + punctuation_limit_(Limit(0, 4)), + digit_block_limit_(PhoneNumberUtil::kMaxLengthForNsn + + PhoneNumberUtil::kMaxLengthCountryCode), + block_limit_(Limit(0, digit_block_limit_)), + punctuation_(StrCat("[", PhoneNumberUtil::kValidPunctuation, "]", + punctuation_limit_)), + digit_sequence_(StrCat("\\p{Nd}", Limit(1, digit_block_limit_))), + lead_class_chars_(StrCat(opening_parens_, PhoneNumberUtil::kPlusChars)), + lead_class_(StrCat("[", lead_class_chars_, "]")), + opening_punctuation_(StrCat("(?:", lead_class_, punctuation_, ")")), + optional_extn_pattern_(StrCat( + "(?i)(?:", + PhoneNumberUtil::GetInstance()->GetExtnPatternsForMatching(), + ")?")), + regexp_factory_for_pattern_(new ICURegExpFactory()), +#ifdef USE_RE2 + regexp_factory_(new RE2RegExpFactory()), +#else + regexp_factory_(new ICURegExpFactory()), +#endif // USE_RE2 + pub_pages_(regexp_factory_->CreateRegExp( + "\\d{1,5}-+\\d{1,5}\\s{0,4}\\(\\d{1,4}")), + slash_separated_dates_(regexp_factory_->CreateRegExp( + "(?:(?:[0-3]?\\d/[01]?\\d)|" + "(?:[01]?\\d/[0-3]?\\d))/(?:[12]\\d)?\\d{2}")), + matching_brackets_(regexp_factory_->CreateRegExp( + StrCat(leading_maybe_matched_bracket_, non_parens_, "+", + bracket_pairs_, non_parens_, "*"))), + group_separator_(regexp_factory_->CreateRegExp( + StrCat("\\p{Z}", "[^", lead_class_chars_, "\\p{Nd}]*"))), + capture_up_to_second_number_start_pattern_( + regexp_factory_->CreateRegExp( + PhoneNumberUtil::kCaptureUpToSecondNumberStart)), + capturing_ascii_digits_pattern_( + regexp_factory_->CreateRegExp("(\\d+)")), + lead_class_pattern_(regexp_factory_->CreateRegExp(lead_class_)), + pattern_(regexp_factory_for_pattern_->CreateRegExp( + StrCat("(", opening_punctuation_, lead_limit_, + digit_sequence_, "(?:", punctuation_, digit_sequence_, ")", + block_limit_, optional_extn_pattern_, ")"))) { + } + + private: + DISALLOW_COPY_AND_ASSIGN(PhoneNumberMatcherRegExps); +}; + +PhoneNumberMatcher::PhoneNumberMatcher(const PhoneNumberUtil& util, + const string& text, + const string& region_code, + PhoneNumberMatcher::Leniency leniency, + int max_tries) + : reg_exps_(PhoneNumberMatcherRegExps::GetInstance()), + phone_util_(util), + text_(text), + preferred_region_(region_code), + leniency_(leniency), + max_tries_(max_tries), + state_(NOT_READY), + last_match_(NULL), + search_index_(0) { +} + +PhoneNumberMatcher::PhoneNumberMatcher(const string& text, + const string& region_code) + : reg_exps_(PhoneNumberMatcherRegExps::GetInstance()), + phone_util_(*PhoneNumberUtil::GetInstance()), + text_(text), + preferred_region_(region_code), + leniency_(VALID), + max_tries_(numeric_limits<int>::max()), + state_(NOT_READY), + last_match_(NULL), + search_index_(0) { +} + +PhoneNumberMatcher::~PhoneNumberMatcher() { +} + +// static +bool PhoneNumberMatcher::IsLatinLetter(char32 letter) { + // Combining marks are a subset of non-spacing-mark. + if (!u_isalpha(letter) && (u_charType(letter) != U_NON_SPACING_MARK)) { + return false; + } + UBlockCode block = ublock_getCode(letter); + return ((block == UBLOCK_BASIC_LATIN) || + (block == UBLOCK_LATIN_1_SUPPLEMENT) || + (block == UBLOCK_LATIN_EXTENDED_A) || + (block == UBLOCK_LATIN_EXTENDED_ADDITIONAL) || + (block == UBLOCK_LATIN_EXTENDED_B) || + (block == UBLOCK_COMBINING_DIACRITICAL_MARKS)); +} + +bool PhoneNumberMatcher::ParseAndVerify(const string& candidate, int offset, + PhoneNumberMatch* match) { + DCHECK(match); + // Check the candidate doesn't contain any formatting which would indicate + // that it really isn't a phone number. + if (!reg_exps_->matching_brackets_->FullMatch(candidate)) { + return false; + } + + // If leniency is set to VALID or stricter, we also want to skip numbers that + // are surrounded by Latin alphabetic characters, to skip cases like + // abc8005001234 or 8005001234def. + if (leniency_ >= VALID) { + // If the candidate is not at the start of the text, and does not start with + // phone-number punctuation, check the previous character. + scoped_ptr<RegExpInput> candidate_input( + reg_exps_->regexp_factory_->CreateInput(candidate)); + if (offset > 0 && + !reg_exps_->lead_class_pattern_->Consume(candidate_input.get())) { + char32 previous_char; + const char* previous_char_ptr = + EncodingUtils::BackUpOneUTF8Character(text_.c_str(), + text_.c_str() + offset); + EncodingUtils::DecodeUTF8Char(previous_char_ptr, &previous_char); + // We return false if it is a latin letter or a currency symbol. + if (IsCurrencySymbol(previous_char) || IsLatinLetter(previous_char)) { + return false; + } + } + size_t lastCharIndex = offset + candidate.length(); + if (lastCharIndex < text_.length()) { + char32 next_char; + const char* next_char_ptr = + EncodingUtils::AdvanceOneUTF8Character( + text_.c_str() + lastCharIndex - 1); + EncodingUtils::DecodeUTF8Char(next_char_ptr, &next_char); + if (IsCurrencySymbol(next_char) || IsLatinLetter(next_char)) { + return false; + } + } + } + + PhoneNumber number; + if (phone_util_.Parse(candidate, preferred_region_, &number) != + PhoneNumberUtil::NO_PARSING_ERROR) { + return false; + } + if (VerifyAccordingToLeniency(leniency_, number, candidate)) { + match->set_start(offset); + match->set_raw_string(candidate); + match->set_number(number); + return true; + } + return false; +} + +// Helper method to replace the verification method for each enum in the Java +// version. +bool PhoneNumberMatcher::VerifyAccordingToLeniency( + Leniency leniency, const PhoneNumber& number, + const string& candidate) const { + switch (leniency) { + case PhoneNumberMatcher::POSSIBLE: + return phone_util_.IsPossibleNumber(number); + case PhoneNumberMatcher::VALID: + if (!phone_util_.IsValidNumber(number)) { + return false; + } + return ContainsOnlyValidXChars(number, candidate, phone_util_); + case PhoneNumberMatcher::STRICT_GROUPING: { + if (!phone_util_.IsValidNumber(number) || + !ContainsOnlyValidXChars(number, candidate, phone_util_) || + // Two or more slashes were present. + FindNth(candidate, '/', 2) != string::npos) { + return false; + } + // TODO(lararennie,shaopengjia): Evaluate how this works for other locales + // (testing has been limited to NANPA regions) and optimise if necessary. + string normalized_candidate = + NormalizeUTF8::NormalizeDecimalDigits(candidate); + vector<string> formatted_number_groups; + GetNationalNumberGroups(phone_util_, number, &formatted_number_groups); + size_t from_index = 0; + // Check each group of consecutive digits are not broken into separate + // groups in the normalized_candidate string. + for (size_t i = 0; i < formatted_number_groups.size(); ++i) { + // Fails if the substring of normalized_candidate starting from + // from_index doesn't contain the consecutive digits in digit_group. + from_index = normalized_candidate.find(formatted_number_groups.at(i), + from_index); + if (from_index == string::npos) { + return false; + } + // Moves from_index forward. + from_index += formatted_number_groups.at(i).length(); + if (i == 0 && from_index < normalized_candidate.length()) { + // We are at the position right after the NDC. Note although + // normalized_candidate might contain non-ASCII formatting characters, + // they won't be treated as ASCII digits when converted to a char. + if (isdigit(normalized_candidate.at(from_index))) { + // This means there is no formatting symbol after the NDC. In this + // case, we only accept the number if there is no formatting + // symbol at all in the number, except for extensions. + string national_significant_number; + phone_util_.GetNationalSignificantNumber( + number, &national_significant_number); + return HasPrefixString( + normalized_candidate.substr( + from_index - formatted_number_groups.at(i).length()), + national_significant_number); + } + } + } + // The check here makes sure that we haven't mistakenly already used the + // extension to match the last group of the subscriber number. Note the + // extension cannot have formatting in-between digits. + return + normalized_candidate.substr(from_index).find(number.extension()) != + string::npos; + } + case PhoneNumberMatcher::EXACT_GROUPING: { + if (!phone_util_.IsValidNumber(number) || + !ContainsOnlyValidXChars(number, candidate, phone_util_) || + // Two or more slashes were present. + FindNth(candidate, '/', 2) != string::npos) { + return false; + } + // TODO(lararennie,shaopengjia): Evaluate how this works for other locales + // (testing has been limited to NANPA regions) and optimise if necessary. + vector<string> candidate_groups; + string normalized_candidate = + NormalizeUTF8::NormalizeDecimalDigits(candidate); + const scoped_ptr<RegExpInput> candidate_number( + reg_exps_->regexp_factory_->CreateInput(normalized_candidate)); + string digit_block; + while (reg_exps_->capturing_ascii_digits_pattern_->FindAndConsume( + candidate_number.get(), + &digit_block)) { + candidate_groups.push_back(digit_block); + } + + // Set this to the last group, skipping it if the number has an extension. + int candidate_number_group_index = + number.has_extension() ? candidate_groups.size() - 2 + : candidate_groups.size() - 1; + // First we check if the national significant number is formatted as a + // block. We use contains and not equals, since the national significant + // number may be present with a prefix such as a national number prefix, + // or the country code itself. + string national_significant_number; + phone_util_.GetNationalSignificantNumber(number, + &national_significant_number); + if (candidate_groups.size() == 1 || + candidate_groups.at(candidate_number_group_index).find( + national_significant_number) != string::npos) { + return true; + } + vector<string> formatted_number_groups; + GetNationalNumberGroups(phone_util_, number, &formatted_number_groups); + // Starting from the end, go through in reverse, excluding the first + // group, and check the candidate and number groups are the same. + for (int formatted_number_group_index = + (formatted_number_groups.size() - 1); + formatted_number_group_index > 0 && + candidate_number_group_index >= 0; + --formatted_number_group_index, --candidate_number_group_index) { + if (candidate_groups.at(candidate_number_group_index) != + formatted_number_groups.at(formatted_number_group_index)) { + return false; + } + } + // Now check the first group. There may be a national prefix at the start, + // so we only check that the candidate group ends with the formatted + // number group. + return (candidate_number_group_index >= 0 && + HasSuffixString(candidate_groups.at(candidate_number_group_index), + formatted_number_groups.at(0))); + } + default: + LOG(ERROR) << "No implementation defined for verification for leniency " + << static_cast<int>(leniency); + return false; + } +} + +bool PhoneNumberMatcher::ExtractInnerMatch(const string& candidate, int offset, + PhoneNumberMatch* match) { + DCHECK(match); + // Try removing either the first or last "group" in the number and see if this + // gives a result. We consider white space to be a possible indication of + // the start or end of the phone number. + scoped_ptr<RegExpInput> candidate_input( + reg_exps_->regexp_factory_->CreateInput(candidate)); + if (reg_exps_->group_separator_->FindAndConsume(candidate_input.get(), + NULL)) { + // Try the first group by itself. + int group_start_index = + candidate.length() - candidate_input->ToString().length(); + string first_group_only = candidate.substr(0, group_start_index); + phone_util_.TrimUnwantedEndChars(&first_group_only); + bool success = ParseAndVerify(first_group_only, offset, match); + if (success) { + return true; + } + --max_tries_; + + // Try the rest of the candidate without the first group. + string without_first_group(candidate_input->ToString()); + phone_util_.TrimUnwantedEndChars(&without_first_group); + success = + ParseAndVerify(without_first_group, offset + group_start_index, match); + if (success) { + return true; + } + --max_tries_; + + if (max_tries_ > 0) { + while (reg_exps_->group_separator_->FindAndConsume(candidate_input.get(), + NULL)) { + // Find the last group. + } + int last_group_start = + candidate.length() - candidate_input->ToString().length(); + string without_last_group = candidate.substr(0, last_group_start); + phone_util_.TrimUnwantedEndChars(&without_last_group); + if (without_last_group == first_group_only) { + // If there are only two groups, then the group "without the last group" + // is the same as the first group. In these cases, we don't want to + // re-check the number group, so we exit already. + return false; + } + success = ParseAndVerify(without_last_group, offset, match); + if (success) { + return true; + } + --max_tries_; + } + } + return false; +} + +bool PhoneNumberMatcher::ExtractMatch(const string& candidate, int offset, + PhoneNumberMatch* match) { + DCHECK(match); + // Skip a match that is more likely a publication page reference or a date. + if (reg_exps_->pub_pages_->PartialMatch(candidate) || + reg_exps_->slash_separated_dates_->PartialMatch(candidate)) { + return false; + } + + // Try to come up with a valid match given the entire candidate. + if (ParseAndVerify(candidate, offset, match)) { + return true; + } + + // If that failed, try to find an "inner match" - there might be a phone + // number within this candidate. + return ExtractInnerMatch(candidate, offset, match); +} + +bool PhoneNumberMatcher::HasNext() { + if (state_ == NOT_READY) { + PhoneNumberMatch temp_match; + if (!Find(search_index_, &temp_match)) { + state_ = DONE; + } else { + last_match_.reset(new PhoneNumberMatch(temp_match.start(), + temp_match.raw_string(), + temp_match.number())); + search_index_ = last_match_->end(); + state_ = READY; + } + } + return state_ == READY; +} + +bool PhoneNumberMatcher::Next(PhoneNumberMatch* match) { + DCHECK(match); + // Check the state and find the next match as a side-effect if necessary. + if (!HasNext()) { + return false; + } + match->CopyFrom(*last_match_); + state_ = NOT_READY; + last_match_.reset(NULL); + return true; +} + +bool PhoneNumberMatcher::Find(int index, PhoneNumberMatch* match) { + DCHECK(match); + + scoped_ptr<RegExpInput> text( + reg_exps_->regexp_factory_for_pattern_->CreateInput(text_.substr(index))); + string candidate; + while ((max_tries_ > 0) && + reg_exps_->pattern_->FindAndConsume(text.get(), &candidate)) { + int start = text_.length() - text->ToString().length() - candidate.length(); + // Check for extra numbers at the end. + reg_exps_->capture_up_to_second_number_start_pattern_-> + PartialMatch(candidate, &candidate); + if (ExtractMatch(candidate, start, match)) { + return true; + } + + index = start + candidate.length(); + --max_tries_; + } + return false; +} + +} // namespace phonenumbers +} // namespace i18n diff --git a/phonenumbermatcher.h b/phonenumbermatcher.h new file mode 100644 index 0000000..0eb013d --- /dev/null +++ b/phonenumbermatcher.h @@ -0,0 +1,158 @@ +// Copyright (C) 2011 The Libphonenumber Authors +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// Author: Lara Rennie +// Author: Tao Huang +// +// This is a direct port from PhoneNumberMatcher.java. +// Changes to this class should also happen to the Java version, whenever it +// makes sense. + +#ifndef I18N_PHONENUMBERS_PHONENUMBERMATCHER_H_ +#define I18N_PHONENUMBERS_PHONENUMBERMATCHER_H_ + +#include <string> + +#include "base/basictypes.h" +#include "base/memory/scoped_ptr.h" +#include "phonenumbers/regexp_adapter.h" + +namespace i18n { +namespace phonenumbers { + +using std::string; + +class PhoneNumber; +class PhoneNumberMatch; +class PhoneNumberMatcherRegExps; +class PhoneNumberUtil; + +class PhoneNumberMatcher { + friend class PhoneNumberMatcherTest; + public: + // Leniency when finding potential phone numbers in text segments. The levels + // here are ordered in increasing strictness. + enum Leniency { + // Phone numbers accepted are possible, but not necessarily valid. + POSSIBLE, + // Phone numbers accepted are possible and valid. + VALID, + // Phone numbers accepted are valid and are grouped in a possible way for + // this locale. For example, a US number written as "65 02 53 00 00" is not + // accepted at this leniency level, whereas "650 253 0000" or "6502530000" + // are. Numbers with more than one '/' symbol are also dropped at this + // level. + // Warning: This and the next level might result in lower coverage + // especially for regions outside of country code "+1". + STRICT_GROUPING, + // Phone numbers accepted are valid and are grouped in the same way that we + // would have formatted it, or as a single block. For example, a US number + // written as "650 2530000" is not accepted at this leniency level, whereas + // "650 253 0000" or "6502530000" are. + EXACT_GROUPING, + }; + + // Constructs a phone number matcher. + PhoneNumberMatcher(const PhoneNumberUtil& util, + const string& text, + const string& region_code, + Leniency leniency, + int max_tries); + + // Wrapper to construct a phone number matcher, with no limitation on the + // number of retries and VALID Leniency. + PhoneNumberMatcher(const string& text, + const string& region_code); + + ~PhoneNumberMatcher(); + + // Returns true if the text sequence has another match. + bool HasNext(); + + // Gets next match from text sequence. + bool Next(PhoneNumberMatch* match); + + private: + // The potential states of a PhoneNumberMatcher. + enum State { + NOT_READY, + READY, + DONE, + }; + + // Attempts to extract a match from a candidate string. Returns true if a + // match is found, otherwise returns false. The value "offset" refers to the + // start index of the candidate string within the overall text. + bool Find(int index, PhoneNumberMatch* match); + + // Attempts to extract a match from candidate. Returns true if the match was + // found, otherwise returns false. + bool ExtractMatch(const string& candidate, int offset, + PhoneNumberMatch* match); + + // Attempts to extract a match from a candidate string if the whole candidate + // does not qualify as a match. Returns true if a match is found, otherwise + // returns false. + bool ExtractInnerMatch(const string& candidate, int offset, + PhoneNumberMatch* match); + + // Parses a phone number from the candidate using PhoneNumberUtil::Parse() and + // verifies it matches the requested leniency. If parsing and verification + // succeed, returns true, otherwise this method returns false; + bool ParseAndVerify(const string& candidate, int offset, + PhoneNumberMatch* match); + + bool VerifyAccordingToLeniency(Leniency leniency, const PhoneNumber& number, + const string& candidate) const; + + // Helper method to determine if a character is a Latin-script letter or not. + // For our purposes, combining marks should also return true since we assume + // they have been added to a preceding Latin character. + static bool IsLatinLetter(char32 letter); + + // Helper class holding useful regular expressions. + const PhoneNumberMatcherRegExps* reg_exps_; + + // The phone number utility; + const PhoneNumberUtil& phone_util_; + + // The text searched for phone numbers; + const string text_; + + // The region(country) to assume for phone numbers without an international + // prefix. + const string preferred_region_; + + // The degree of validation requested. + Leniency leniency_; + + // The maximum number of retries after matching an invalid number. + int max_tries_; + + // The iteration tristate. + State state_; + + // The last successful match, NULL unless in State.READY. + scoped_ptr<PhoneNumberMatch> last_match_; + + // The next index to start searching at. Undefined in State.DONE. + int search_index_; + + DISALLOW_COPY_AND_ASSIGN(PhoneNumberMatcher); +}; + +} // namespace phonenumbers +} // namespace i18n + +#endif // I18N_PHONENUMBERS_PHONENUMBERMATCHER_H_ diff --git a/phonenumberutil.h b/phonenumberutil.h index bd9cfb7..6bc8022 100644 --- a/phonenumberutil.h +++ b/phonenumberutil.h @@ -67,6 +67,9 @@ class PhoneNumberUtil : public Singleton<PhoneNumberUtil> { friend class Singleton<PhoneNumberUtil>; #endif friend class AsYouTypeFormatter; + friend class PhoneNumberMatcher; + friend class PhoneNumberMatcherRegExps; + friend class PhoneNumberMatcherTest; friend class PhoneNumberUtilTest; public: ~PhoneNumberUtil(); diff --git a/stringutil.cc b/stringutil.cc index e9d7a88..8d021a4 100644 --- a/stringutil.cc +++ b/stringutil.cc @@ -14,6 +14,7 @@ // Author: Philippe Liard +#include <algorithm> #include <cassert> #include <cstring> #include <sstream> @@ -23,6 +24,7 @@ namespace i18n { namespace phonenumbers { +using std::equal; using std::stringstream; string operator+(const string& s, int n) { @@ -54,6 +56,43 @@ string SimpleItoa(uint64 n) { return GenericSimpleItoa(n); } +bool HasPrefixString(const string& s, const string& prefix) { + return s.size() >= prefix.size() && + equal(s.begin(), s.begin() + prefix.size(), prefix.begin()); +} + +size_t FindNth(const string& s, char c, int n) { + size_t pos = string::npos; + + for (int i = 0; i < n; ++i) { + pos = s.find_first_of(c, pos + 1); + if (pos == string::npos) { + break; + } + } + return pos; +} + +void SplitStringUsing(const string& s, const string& delimiter, + vector<string>* result) { + assert(result); + size_t start_pos = 0; + size_t find_pos = string::npos; + if (delimiter.empty()) { + return; + } + while ((find_pos = s.find(delimiter, start_pos)) != string::npos) { + const string substring = s.substr(start_pos, find_pos - start_pos); + if (!substring.empty()) { + result->push_back(substring); + } + start_pos = find_pos + delimiter.length(); + } + if (start_pos != s.length()) { + result->push_back(s.substr(start_pos)); + } +} + void StripString(string* s, const char* remove, char replacewith) { const char* str_start = s->c_str(); const char* str = str_start; @@ -255,6 +294,25 @@ string StrCat(const StringHolder& s1, const StringHolder& s2, string StrCat(const StringHolder& s1, const StringHolder& s2, const StringHolder& s3, const StringHolder& s4, const StringHolder& s5, const StringHolder& s6, + const StringHolder& s7, const StringHolder& s8) { + string result; + result.reserve(s1.Length() + s2.Length() + s3.Length() + s4.Length() + + s5.Length() + s6.Length() + s7.Length() + s8.Length() + 1); + result += s1; + result += s2; + result += s3; + result += s4; + result += s5; + result += s6; + result += s7; + result += s8; + + return result; +} + +string StrCat(const StringHolder& s1, const StringHolder& s2, + const StringHolder& s3, const StringHolder& s4, + const StringHolder& s5, const StringHolder& s6, const StringHolder& s7, const StringHolder& s8, const StringHolder& s9) { string result; diff --git a/stringutil.h b/stringutil.h index ee46bb2..6d90d00 100644 --- a/stringutil.h +++ b/stringutil.h @@ -19,6 +19,7 @@ #include <cstddef> #include <string> +#include <vector> #include "base/basictypes.h" @@ -26,6 +27,7 @@ namespace i18n { namespace phonenumbers { using std::string; +using std::vector; // Supports string("hello") + 10. string operator+(const string& s, int n); @@ -34,6 +36,18 @@ string operator+(const string& s, int n); string SimpleItoa(uint64 n); string SimpleItoa(int n); +// Returns whether the provided string starts with the supplied prefix. +bool HasPrefixString(const string& s, const string& prefix); + +// Returns the index of the nth occurence of c in s or string::npos if less than +// n occurrences are present. +size_t FindNth(const string& s, char c, int n); + +// Splits a string using a character delimiter. Appends the components to the +// provided vector. Note that empty tokens are ignored. +void SplitStringUsing(const string& s, const string& delimiter, + vector<string>* result); + // Replaces any occurrence of the character 'remove' (or the characters // in 'remove') with the character 'replacewith'. void StripString(string* s, const char* remove, char replacewith); @@ -116,6 +130,11 @@ string StrCat(const StringHolder& s1, const StringHolder& s2, string StrCat(const StringHolder& s1, const StringHolder& s2, const StringHolder& s3, const StringHolder& s4, const StringHolder& s5, const StringHolder& s6, + const StringHolder& s7, const StringHolder& s8); + +string StrCat(const StringHolder& s1, const StringHolder& s2, + const StringHolder& s3, const StringHolder& s4, + const StringHolder& s5, const StringHolder& s6, const StringHolder& s7, const StringHolder& s8, const StringHolder& s9); |