| 1 | /* This file is part of the KDE libraries |
|---|
| 2 | SPDX-FileCopyrightText: 2006 Jacob R Rideout <kde@jacobrideout.net> |
|---|
| 3 | |
|---|
| 4 | SPDX-License-Identifier: LGPL-2.0-or-later |
|---|
| 5 | */ |
|---|
| 6 | |
|---|
| 7 | #ifndef TEXTBREAKS_H |
|---|
| 8 | #define TEXTBREAKS_H |
|---|
| 9 | |
|---|
| 10 | class QString; |
|---|
| 11 | |
|---|
| 12 | #include "sonnetcore_export.h" |
|---|
| 13 | |
|---|
| 14 | #include <memory> |
|---|
| 15 | |
|---|
| 16 | namespace Sonnet |
|---|
| 17 | { |
|---|
| 18 | class TextBreaksPrivate; |
|---|
| 19 | |
|---|
| 20 | /** |
|---|
| 21 | * @short TextBreaks determines the barriers between linguistic structures in any given text. |
|---|
| 22 | * |
|---|
| 23 | * TextBreaks is a class that determines the boundaries between graphemes |
|---|
| 24 | * (characters as per the unicode definition,) words and sentences. The |
|---|
| 25 | * default implementation conforms to Unicode Standard Annex #29 https://unicode.org/reports/tr29/. |
|---|
| 26 | * You can subclass TextBreaks to create the correct behaviour for languages that require it. |
|---|
| 27 | * |
|---|
| 28 | * @author Jacob Rideout <kde@jacobrideout.net> |
|---|
| 29 | * @since 4.3 |
|---|
| 30 | */ |
|---|
| 31 | class SONNETCORE_EXPORT TextBreaks |
|---|
| 32 | { |
|---|
| 33 | public: |
|---|
| 34 | struct Position { |
|---|
| 35 | int start, length; |
|---|
| 36 | }; |
|---|
| 37 | |
|---|
| 38 | /** |
|---|
| 39 | * This structure abstracts the positions of breaks in the test. As per the |
|---|
| 40 | * unicode annex, both the start and end of the text are returned. |
|---|
| 41 | */ |
|---|
| 42 | typedef QList<Position> Positions; |
|---|
| 43 | |
|---|
| 44 | /** Constructor |
|---|
| 45 | * Creates a new TextBreaks instance. If @p text is specified, |
|---|
| 46 | * it sets the text to be checked. |
|---|
| 47 | * @param text the text that is to be checked |
|---|
| 48 | */ |
|---|
| 49 | explicit TextBreaks(const QString &text = QString()); |
|---|
| 50 | |
|---|
| 51 | /** Virtual Destructor |
|---|
| 52 | */ |
|---|
| 53 | virtual ~TextBreaks(); |
|---|
| 54 | |
|---|
| 55 | /** |
|---|
| 56 | * Returns the text to be checked |
|---|
| 57 | * @return text |
|---|
| 58 | */ |
|---|
| 59 | QString text() const; |
|---|
| 60 | |
|---|
| 61 | /** |
|---|
| 62 | * Sets the text to @p text |
|---|
| 63 | * @param text to be set |
|---|
| 64 | * @return true if the word is misspelled. false otherwise |
|---|
| 65 | */ |
|---|
| 66 | void setText(const QString &text); |
|---|
| 67 | |
|---|
| 68 | /** |
|---|
| 69 | * Return the Positions of each word for the given @p text. |
|---|
| 70 | * @param text to be checked |
|---|
| 71 | * @return positions of breaks |
|---|
| 72 | */ |
|---|
| 73 | static Positions wordBreaks(const QString &text); |
|---|
| 74 | |
|---|
| 75 | /** |
|---|
| 76 | * Return the Positions of each sentence for the given @p text. |
|---|
| 77 | * @param text to be checked |
|---|
| 78 | * @return positions of breaks |
|---|
| 79 | */ |
|---|
| 80 | static Positions sentenceBreaks(const QString &text); |
|---|
| 81 | |
|---|
| 82 | /** |
|---|
| 83 | * Return the Positions of each word for the text previously set. |
|---|
| 84 | * @return positions of breaks |
|---|
| 85 | */ |
|---|
| 86 | virtual Positions wordBreaks() const; |
|---|
| 87 | |
|---|
| 88 | /** |
|---|
| 89 | * Return the Positions of each sentence for the text previously set. |
|---|
| 90 | * @return positions of breaks |
|---|
| 91 | */ |
|---|
| 92 | virtual Positions sentenceBreaks() const; |
|---|
| 93 | |
|---|
| 94 | private: |
|---|
| 95 | std::unique_ptr<TextBreaksPrivate> const d; |
|---|
| 96 | }; |
|---|
| 97 | } |
|---|
| 98 | |
|---|
| 99 | Q_DECLARE_TYPEINFO(Sonnet::TextBreaks::Position, Q_PRIMITIVE_TYPE); |
|---|
| 100 | |
|---|
| 101 | #endif |
|---|
| 102 | |
|---|
