michael@0: /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
michael@0:  * This Source Code Form is subject to the terms of the Mozilla Public
michael@0:  * License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0:  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
michael@0: 
michael@0: #include "nsTextFrameUtils.h"
michael@0: 
michael@0: #include "nsUnicharUtils.h"
michael@0: #include "nsBidiUtils.h"
michael@0: #include "nsIContent.h"
michael@0: #include "nsStyleStruct.h"
michael@0: #include "nsTextFragment.h"
michael@0: #include <algorithm>
michael@0: 
michael@0: static bool IsDiscardable(char16_t ch, uint32_t* aFlags)
michael@0: {
michael@0:   // Unlike IS_DISCARDABLE, we don't discard \r. \r will be ignored by gfxTextRun
michael@0:   // and discarding it would force us to copy text in many cases of preformatted
michael@0:   // text containing \r\n.
michael@0:   if (ch == CH_SHY) {
michael@0:     *aFlags |= nsTextFrameUtils::TEXT_HAS_SHY;
michael@0:     return true;
michael@0:   }
michael@0:   return IsBidiControl(ch);
michael@0: }
michael@0: 
michael@0: static bool IsDiscardable(uint8_t ch, uint32_t* aFlags)
michael@0: {
michael@0:   if (ch == CH_SHY) {
michael@0:     *aFlags |= nsTextFrameUtils::TEXT_HAS_SHY;
michael@0:     return true;
michael@0:   }
michael@0:   return false;
michael@0: }
michael@0: 
michael@0: char16_t*
michael@0: nsTextFrameUtils::TransformText(const char16_t* aText, uint32_t aLength,
michael@0:                                 char16_t* aOutput,
michael@0:                                 CompressionMode aCompression,
michael@0:                                 uint8_t* aIncomingFlags,
michael@0:                                 gfxSkipChars* aSkipChars,
michael@0:                                 uint32_t* aAnalysisFlags)
michael@0: {
michael@0:   uint32_t flags = 0;
michael@0:   char16_t* outputStart = aOutput;
michael@0: 
michael@0:   bool lastCharArabic = false;
michael@0: 
michael@0:   if (aCompression == COMPRESS_NONE ||
michael@0:       aCompression == DISCARD_NEWLINE) {
michael@0:     // Skip discardables.
michael@0:     uint32_t i;
michael@0:     for (i = 0; i < aLength; ++i) {
michael@0:       char16_t ch = *aText++;
michael@0:       if (IsDiscardable(ch, &flags) ||
michael@0:           (ch == '\n' && aCompression == DISCARD_NEWLINE)) {
michael@0:         aSkipChars->SkipChar();
michael@0:       } else {
michael@0:         aSkipChars->KeepChar();
michael@0:         if (ch > ' ') {
michael@0:           lastCharArabic = IS_ARABIC_CHAR(ch);
michael@0:         } else if (ch == '\t') {
michael@0:           flags |= TEXT_HAS_TAB;
michael@0:         }
michael@0:         *aOutput++ = ch;
michael@0:       }
michael@0:     }
michael@0:     if (lastCharArabic) {
michael@0:       *aIncomingFlags |= INCOMING_ARABICCHAR;
michael@0:     } else {
michael@0:       *aIncomingFlags &= ~INCOMING_ARABICCHAR;
michael@0:     }
michael@0:     *aIncomingFlags &= ~INCOMING_WHITESPACE;
michael@0:   } else {
michael@0:     bool inWhitespace = (*aIncomingFlags & INCOMING_WHITESPACE) != 0;
michael@0:     uint32_t i;
michael@0:     for (i = 0; i < aLength; ++i) {
michael@0:       char16_t ch = *aText++;
michael@0:       bool nowInWhitespace;
michael@0:       if (ch == ' ' &&
michael@0:           (i + 1 >= aLength ||
michael@0:            !IsSpaceCombiningSequenceTail(aText, aLength - (i + 1)))) {
michael@0:         nowInWhitespace = true;
michael@0:       } else if (ch == '\n' && aCompression == COMPRESS_WHITESPACE_NEWLINE) {
michael@0:         if (i > 0 && IS_CJ_CHAR(aText[-1]) &&
michael@0:             i + 1 < aLength && IS_CJ_CHAR(aText[1])) {
michael@0:           // Discard newlines between CJK chars.
michael@0:           // XXX this really requires more context to get right!
michael@0:           aSkipChars->SkipChar();
michael@0:           continue;
michael@0:         }
michael@0:         nowInWhitespace = true;
michael@0:       } else {
michael@0:         nowInWhitespace = ch == '\t';
michael@0:       }
michael@0: 
michael@0:       if (!nowInWhitespace) {
michael@0:         if (IsDiscardable(ch, &flags)) {
michael@0:           aSkipChars->SkipChar();
michael@0:           nowInWhitespace = inWhitespace;
michael@0:         } else {
michael@0:           *aOutput++ = ch;
michael@0:           aSkipChars->KeepChar();
michael@0:           lastCharArabic = IS_ARABIC_CHAR(ch);
michael@0:         }
michael@0:       } else {
michael@0:         if (inWhitespace) {
michael@0:           aSkipChars->SkipChar();
michael@0:         } else {
michael@0:           if (ch != ' ') {
michael@0:             flags |= TEXT_WAS_TRANSFORMED;
michael@0:           }
michael@0:           *aOutput++ = ' ';
michael@0:           aSkipChars->KeepChar();
michael@0:         }
michael@0:       }
michael@0:       inWhitespace = nowInWhitespace;
michael@0:     }
michael@0:     if (lastCharArabic) {
michael@0:       *aIncomingFlags |= INCOMING_ARABICCHAR;
michael@0:     } else {
michael@0:       *aIncomingFlags &= ~INCOMING_ARABICCHAR;
michael@0:     }
michael@0:     if (inWhitespace) {
michael@0:       *aIncomingFlags |= INCOMING_WHITESPACE;
michael@0:     } else {
michael@0:       *aIncomingFlags &= ~INCOMING_WHITESPACE;
michael@0:     }
michael@0:   }
michael@0: 
michael@0:   if (outputStart + aLength != aOutput) {
michael@0:     flags |= TEXT_WAS_TRANSFORMED;
michael@0:   }
michael@0:   *aAnalysisFlags = flags;
michael@0:   return aOutput;
michael@0: }
michael@0: 
michael@0: uint8_t*
michael@0: nsTextFrameUtils::TransformText(const uint8_t* aText, uint32_t aLength,
michael@0:                                 uint8_t* aOutput,
michael@0:                                 CompressionMode aCompression,
michael@0:                                 uint8_t* aIncomingFlags,
michael@0:                                 gfxSkipChars* aSkipChars,
michael@0:                                 uint32_t* aAnalysisFlags)
michael@0: {
michael@0:   uint32_t flags = 0;
michael@0:   uint8_t* outputStart = aOutput;
michael@0: 
michael@0:   if (aCompression == COMPRESS_NONE ||
michael@0:       aCompression == DISCARD_NEWLINE) {
michael@0:     // Skip discardables.
michael@0:     uint32_t i;
michael@0:     for (i = 0; i < aLength; ++i) {
michael@0:       uint8_t ch = *aText++;
michael@0:       if (IsDiscardable(ch, &flags) ||
michael@0:           (ch == '\n' && aCompression == DISCARD_NEWLINE)) {
michael@0:         aSkipChars->SkipChar();
michael@0:       } else {
michael@0:         aSkipChars->KeepChar();
michael@0:         if (ch == '\t') {
michael@0:           flags |= TEXT_HAS_TAB;
michael@0:         }
michael@0:         *aOutput++ = ch;
michael@0:       }
michael@0:     }
michael@0:     *aIncomingFlags &= ~(INCOMING_ARABICCHAR | INCOMING_WHITESPACE);
michael@0:   } else {
michael@0:     bool inWhitespace = (*aIncomingFlags & INCOMING_WHITESPACE) != 0;
michael@0:     uint32_t i;
michael@0:     for (i = 0; i < aLength; ++i) {
michael@0:       uint8_t ch = *aText++;
michael@0:       bool nowInWhitespace = ch == ' ' || ch == '\t' ||
michael@0:         (ch == '\n' && aCompression == COMPRESS_WHITESPACE_NEWLINE);
michael@0:       if (!nowInWhitespace) {
michael@0:         if (IsDiscardable(ch, &flags)) {
michael@0:           aSkipChars->SkipChar();
michael@0:           nowInWhitespace = inWhitespace;
michael@0:         } else {
michael@0:           *aOutput++ = ch;
michael@0:           aSkipChars->KeepChar();
michael@0:         }
michael@0:       } else {
michael@0:         if (inWhitespace) {
michael@0:           aSkipChars->SkipChar();
michael@0:         } else {
michael@0:           if (ch != ' ') {
michael@0:             flags |= TEXT_WAS_TRANSFORMED;
michael@0:           }
michael@0:           *aOutput++ = ' ';
michael@0:           aSkipChars->KeepChar();
michael@0:         }
michael@0:       }
michael@0:       inWhitespace = nowInWhitespace;
michael@0:     }
michael@0:     *aIncomingFlags &= ~INCOMING_ARABICCHAR;
michael@0:     if (inWhitespace) {
michael@0:       *aIncomingFlags |= INCOMING_WHITESPACE;
michael@0:     } else {
michael@0:       *aIncomingFlags &= ~INCOMING_WHITESPACE;
michael@0:     }
michael@0:   }
michael@0: 
michael@0:   if (outputStart + aLength != aOutput) {
michael@0:     flags |= TEXT_WAS_TRANSFORMED;
michael@0:   }
michael@0:   *aAnalysisFlags = flags;
michael@0:   return aOutput;
michael@0: }
michael@0: 
michael@0: uint32_t
michael@0: nsTextFrameUtils::ComputeApproximateLengthWithWhitespaceCompression(
michael@0:                     nsIContent *aContent, const nsStyleText *aStyleText)
michael@0: {
michael@0:   const nsTextFragment *frag = aContent->GetText();
michael@0:   // This is an approximation so we don't really need anything
michael@0:   // too fancy here.
michael@0:   uint32_t len;
michael@0:   if (aStyleText->WhiteSpaceIsSignificant()) {
michael@0:     len = frag->GetLength();
michael@0:   } else {
michael@0:     bool is2b = frag->Is2b();
michael@0:     union {
michael@0:       const char *s1b;
michael@0:       const char16_t *s2b;
michael@0:     } u;
michael@0:     if (is2b) {
michael@0:       u.s2b = frag->Get2b();
michael@0:     } else {
michael@0:       u.s1b = frag->Get1b();
michael@0:     }
michael@0:     bool prevWS = true; // more important to ignore blocks with
michael@0:                         // only whitespace than get inline boundaries
michael@0:                         // exactly right
michael@0:     len = 0;
michael@0:     for (uint32_t i = 0, i_end = frag->GetLength(); i < i_end; ++i) {
michael@0:       char16_t c = is2b ? u.s2b[i] : u.s1b[i];
michael@0:       if (c == ' ' || c == '\n' || c == '\t' || c == '\r') {
michael@0:         if (!prevWS) {
michael@0:           ++len;
michael@0:         }
michael@0:         prevWS = true;
michael@0:       } else {
michael@0:         ++len;
michael@0:         prevWS = false;
michael@0:       }
michael@0:     }
michael@0:   }
michael@0:   return len;
michael@0: }
michael@0: 
michael@0: bool nsSkipCharsRunIterator::NextRun() {
michael@0:   do {
michael@0:     if (mRunLength) {
michael@0:       mIterator.AdvanceOriginal(mRunLength);
michael@0:       NS_ASSERTION(mRunLength > 0, "No characters in run (initial length too large?)");
michael@0:       if (!mSkipped || mLengthIncludesSkipped) {
michael@0:         mRemainingLength -= mRunLength;
michael@0:       }
michael@0:     }
michael@0:     if (!mRemainingLength)
michael@0:       return false;
michael@0:     int32_t length;
michael@0:     mSkipped = mIterator.IsOriginalCharSkipped(&length);
michael@0:     mRunLength = std::min(length, mRemainingLength);
michael@0:   } while (!mVisitSkipped && mSkipped);
michael@0: 
michael@0:   return true;
michael@0: }