From ce5ed9e6498a832f070400fb281e46cd04e39ea1 Mon Sep 17 00:00:00 2001 From: Bruno de Oliveira Abinader Date: Mon, 12 Dec 2016 12:46:24 +0200 Subject: [build] ICU is now configured per platform --- platform/default/bidi.cpp | 125 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 125 insertions(+) create mode 100644 platform/default/bidi.cpp (limited to 'platform/default') diff --git a/platform/default/bidi.cpp b/platform/default/bidi.cpp new file mode 100644 index 0000000000..7d5f6313bc --- /dev/null +++ b/platform/default/bidi.cpp @@ -0,0 +1,125 @@ +#include + +#include +#include +#include + +namespace mbgl { + +// Takes UTF16 input in logical order and applies Arabic shaping to the input while maintaining +// logical order +// Output won't be intelligible until the bidirectional algorithm is applied +std::u16string applyArabicShaping(const std::u16string& input) { + UErrorCode errorCode = U_ZERO_ERROR; + + int32_t outputLength = + u_shapeArabic(input.c_str(), static_cast(input.size()), NULL, 0, + (U_SHAPE_LETTERS_SHAPE & U_SHAPE_LETTERS_MASK) | + (U_SHAPE_TEXT_DIRECTION_LOGICAL & U_SHAPE_TEXT_DIRECTION_MASK), + &errorCode); + + // Pre-flighting will always set U_BUFFER_OVERFLOW_ERROR + errorCode = U_ZERO_ERROR; + + std::unique_ptr outputText = std::make_unique(outputLength); + u_shapeArabic(input.c_str(), static_cast(input.size()), outputText.get(), outputLength, + (U_SHAPE_LETTERS_SHAPE & U_SHAPE_LETTERS_MASK) | + (U_SHAPE_TEXT_DIRECTION_LOGICAL & U_SHAPE_TEXT_DIRECTION_MASK), + &errorCode); + + // If the algorithm fails for any reason, fall back to non-transformed text + if (U_FAILURE(errorCode)) + return input; + + return std::u16string(outputText.get(), outputLength); +} + +ProcessedBiDiText::ProcessedBiDiText(BiDi& p_bidi) : bidi(p_bidi) { +} + +void ProcessedBiDiText::mergeParagraphLineBreaks(std::set& lineBreakPoints) { + int32_t paragraphCount = ubidi_countParagraphs(bidi.bidiText); + for (int32_t i = 0; i < paragraphCount; i++) { + UErrorCode errorCode = U_ZERO_ERROR; + int32_t paragraphEndIndex; + ubidi_getParagraphByIndex(bidi.bidiText, i, NULL, ¶graphEndIndex, NULL, &errorCode); + + if (U_FAILURE(errorCode)) + throw std::runtime_error(std::string("ProcessedBiDiText::mergeParagraphLineBreaks: ") + + u_errorName(errorCode)); + + lineBreakPoints.insert(paragraphEndIndex); + } +} + +std::vector +ProcessedBiDiText::applyLineBreaking(std::set lineBreakPoints) { + // BiDi::getLine will error if called across a paragraph boundary, so we need to ensure that all + // paragraph + // boundaries are included in the set of line break points. The calling code might not include + // the line break because it + // didn't need to wrap at that point, or because the text was separated with a more exotic code + // point such as (U+001C) + mergeParagraphLineBreaks(lineBreakPoints); + + std::vector transformedLines; + int32_t start = 0; + for (int32_t lineBreakPoint : lineBreakPoints) { + transformedLines.push_back(bidi.getLine(start, lineBreakPoint)); + start = lineBreakPoint; + } + + return transformedLines; +} + +BiDi::BiDi() { + bidiText = ubidi_open(); + bidiLine = ubidi_open(); +} + +BiDi::~BiDi() { + if (bidiText) + ubidi_close(bidiText); + + if (bidiLine) + ubidi_close(bidiLine); +} + +ProcessedBiDiText BiDi::processText(const std::u16string& input) { + UErrorCode errorCode = U_ZERO_ERROR; + + ubidi_setPara(bidiText, input.c_str(), static_cast(input.size()), UBIDI_DEFAULT_LTR, + NULL, &errorCode); + + if (U_FAILURE(errorCode)) + throw std::runtime_error(std::string("BiDi::processText: ") + u_errorName(errorCode)); + + return ProcessedBiDiText(*this); +} + +std::u16string BiDi::getLine(int32_t start, int32_t end) { + UErrorCode errorCode = U_ZERO_ERROR; + ubidi_setLine(bidiText, start, end, bidiLine, &errorCode); + + if (U_FAILURE(errorCode)) + throw std::runtime_error(std::string("BiDi::getLine (setLine): ") + u_errorName(errorCode)); + + // Because we set UBIDI_REMOVE_BIDI_CONTROLS, the output may be smaller than what we reserve + // Setting UBIDI_INSERT_LRM_FOR_NUMERIC would require + // ubidi_getLength(pBiDi)+2*ubidi_countRuns(pBiDi) + int32_t outputLength = ubidi_getProcessedLength(bidiLine); + std::unique_ptr outputText = std::make_unique(outputLength); + + // UBIDI_DO_MIRRORING: Apply unicode mirroring of characters like parentheses + // UBIDI_REMOVE_BIDI_CONTROLS: Now that all the lines are set, remove control characters so that + // they don't show up on screen (some fonts have glyphs representing them) + ubidi_writeReordered(bidiLine, outputText.get(), outputLength, + UBIDI_DO_MIRRORING | UBIDI_REMOVE_BIDI_CONTROLS, &errorCode); + + if (U_FAILURE(errorCode)) + throw std::runtime_error(std::string("BiDi::getLine (writeReordered): ") + u_errorName(errorCode)); + + return std::u16string(outputText.get(), outputLength); +} + +} // end namespace mbgl -- cgit v1.2.1