script_run_iterator.cc | Explore in Territory

// Copyright 2015 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#ifdef UNSAFE_BUFFERS_BUILD
// TODO(crbug.com/351564777): Remove this and convert code to safer constructs.
#pragma allow_unsafe_buffers
#endif

#include "third_party/blink/renderer/platform/fonts/script_run_iterator.h"

#include <algorithm>

#include "base/containers/contains.h"
#include "base/logging.h"
#include "base/notreached.h"
#include "base/ranges/algorithm.h"
#include "third_party/blink/renderer/platform/text/icu_error.h"
#include "third_party/blink/renderer/platform/wtf/text/character_names.h"
#include "third_party/blink/renderer/platform/wtf/threading.h"

namespace blink {

namespace {

// UScriptCode and OpenType script are not 1:1; specifically, both Hiragana and
// Katakana map to 'kana' in OpenType. They will be mapped correctly in
// HarfBuzz, but normalizing earlier helps to reduce splitting runs between
// these scripts.
// https://docs.microsoft.com/en-us/typography/opentype/spec/scripttags
inline UScriptCode GetScriptForOpenType(UChar32 ch, UErrorCode* status) { … }

inline bool IsHanScript(UScriptCode script) { … }

inline UScriptCode FirstHanScript(
    const ScriptRunIterator::UScriptCodeList& list) { … }

ScriptRunIterator::UScriptCodeList GetHanScriptExtensions() { … }

// This function updates the script list to the Han ideographic-based scripts if
// the East Asian Width property[1] indicates it is an East Asian character.
//
// Most East Asian punctuation characters have East Asian scripts in the script
// extensions. However, not all of them are so. For example, when they are
// halfwidth/fullwidth forms, they must have the same properties as their
// canonical equivalent[2] code points that are not East Asian. Such code points
// can split runs in the middle of consecutive CJK punctuation characters when
// they are preceded by non-CJK characters, and prevent applying font features
// to consecutive CJK punctuation characters.
//
// TODO(crbug.com/1273998): This function is not needed if Unicode changes the
// script extension for these code points.
//
// [1]: https://www.unicode.org/reports/tr11/
// [2]: https://unicode.org/reports/tr15/#Canon_Compat_Equivalence
void FixScriptsByEastAsianWidth(UChar32 ch,
                                ScriptRunIterator::UScriptCodeList* set) { … }

}  // namespace

PairedBracketType;

constexpr int ScriptRunIterator::kMaxScriptCount;
constexpr int ScriptData::kMaxScriptCount;

ScriptData::~ScriptData() = default;

void ICUScriptData::GetScripts(UChar32 ch, UScriptCodeList& dst) const { … }

UChar32 ICUScriptData::GetPairedBracket(UChar32 ch) const { … }

PairedBracketType ICUScriptData::GetPairedBracketType(UChar32 ch) const { … }

const ICUScriptData* ICUScriptData::Instance() { … }

ScriptRunIterator::ScriptRunIterator(const UChar* text,
                                     wtf_size_t length,
                                     const ScriptData* data)
    : … { … }

ScriptRunIterator::ScriptRunIterator(const UChar* text, wtf_size_t length)
    : … { … }

bool ScriptRunIterator::Consume(unsigned* limit, UScriptCode* script) { … }

void ScriptRunIterator::OpenBracket(UChar32 ch) { … }

void ScriptRunIterator::CloseBracket(UChar32 ch) { … }

// Keep items in current_set_ that are in next_set_.
//
// If the sets are disjoint, return false and leave current_set_ unchanged. Else
// return true and make current set the intersection. Make sure to maintain
// current priority script as priority if it remains, else retain next priority
// script if it remains.
//
// Also maintain a common preferred script.  If current and next are both
// common, and there is no common preferred script and next has a preferred
// script, set the common preferred script to that of next.
bool ScriptRunIterator::MergeSets() { … }

// When we hit the end of the run, and resolve the script, we now know the
// resolved script of any open bracket that was pushed on the stack since
// the start of the run. Fixup depth records how many of these there
// were. We've maintained this count during pushes, and taken care to
// adjust it if the stack got overfull and open brackets were pushed off
// the bottom. This sets the script of the fixup_depth topmost entries of the
// stack to the resolved script.
void ScriptRunIterator::FixupStack(UScriptCode resolved_script,
                                   bool exclude_last) { … }

bool ScriptRunIterator::Fetch(wtf_size_t* pos, UChar32* ch) { … }

UScriptCode ScriptRunIterator::ResolveCurrentScript() const { … }

}  // namespace blink
chromium/third_party/blink/renderer/platform/fonts/script_run_iterator.cc
