7#include "core/fpdfapi/page/cpdf_textobject.h"
11#include "core/fpdfapi/font/cpdf_cidfont.h"
12#include "core/fpdfapi/font/cpdf_font.h"
13#include "core/fxcrt/check.h"
14#include "core/fxcrt/fx_coordinates.h"
15#include "core/fxcrt/span.h"
16#include "core/fxcrt/span_util.h"
18#define ISLATINWORD(u) (u != 0x20
&& u <= 0x28FF
)
22bool IsVertWritingCIDFont(
const CPDF_CIDFont* font) {
28CPDF_TextObject::
Item::
Item() =
default;
32CPDF_TextObject::
Item::~
Item() =
default;
42 return m_CharCodes.size();
46 DCHECK(index < m_CharCodes.size());
49 info.m_CharCode = m_CharCodes[index];
50 info.m_Origin = CFX_PointF(index > 0 ? m_CharPos[index - 1] : 0, 0);
55 const CPDF_CIDFont* pCIDFont = pFont->AsCIDFont();
56 if (!IsVertWritingCIDFont(pCIDFont))
60 info.m_Origin =
CFX_PointF(0, info.m_Origin.x);
62 CFX_Point16 vertical_origin = pCIDFont->GetVertOrigin(cid);
64 info.m_Origin.x -= fontsize * vertical_origin.x / 1000;
65 info.m_Origin.y -= fontsize * vertical_origin.y / 1000;
71 for (uint32_t charcode : m_CharCodes) {
72 if (charcode != CPDF_Font::kInvalidCharCode)
80 for (uint32_t code : m_CharCodes) {
81 if (code == CPDF_Font::kInvalidCharCode)
92 for (size_t i = 0; i < m_CharCodes.size(); ++i) {
93 uint32_t charcode = m_CharCodes[i];
97 return GetItemInfo(i);
104 bool bInLatinWord =
false;
106 for (size_t i = 0, sz = CountChars(); i < sz; ++i) {
107 uint32_t charcode = GetCharCode(i);
109 WideString swUnicode = pFont->UnicodeFromCharCode(charcode);
110 uint16_t unicode = 0;
111 if (swUnicode.GetLength() > 0)
112 unicode = swUnicode[0];
115 if (bIsLatin && bInLatinWord)
118 bInLatinWord = bIsLatin;
130 bool bInLatinWord =
false;
131 for (size_t i = 0, sz = CountChars(); i < sz; ++i) {
132 uint32_t charcode = GetCharCode(i);
134 WideString swUnicode = pFont->UnicodeFromCharCode(charcode);
135 uint16_t unicode = 0;
136 if (swUnicode.GetLength() > 0)
137 unicode = swUnicode[0];
140 if (!bIsLatin || !bInLatinWord) {
141 bInLatinWord = bIsLatin;
145 if (nWords - 1 == nWordIndex)
152 auto obj =
std::make_unique<CPDF_TextObject>();
154 obj->m_CharCodes = m_CharCodes;
155 obj->m_CharPos = m_CharPos;
173CPDF_TextObject* CPDF_TextObject::
AsText() {
177const CPDF_TextObject* CPDF_TextObject::
AsText()
const {
182 pdfium::span<
const float> pTextMatrix =
text_state().GetMatrix();
183 return CFX_Matrix(pTextMatrix[0], pTextMatrix[2], pTextMatrix[1],
184 pTextMatrix[3], m_Pos.x, m_Pos.y);
189 pTextMatrix[0] = matrix
.a;
190 pTextMatrix[1] = matrix
.c;
191 pTextMatrix[2] = matrix
.b;
192 pTextMatrix[3] = matrix
.d;
193 m_Pos = CFX_PointF(matrix.e, matrix.f);
194 CalcPositionDataInternal(
GetFont());
197void CPDF_TextObject::
SetSegments(pdfium::span<
const ByteString> strings,
198 pdfium::span<
const float> kernings) {
199 size_t nSegs = strings.size();
204 size_t nChars = nSegs - 1;
205 for (
const auto& str : strings) {
206 nChars += pFont->CountChar(str.AsStringView());
209 m_CharCodes.resize(nChars);
210 m_CharPos.resize(nChars - 1);
212 for (size_t i = 0; i < nSegs; ++i) {
215 while (offset < segment.GetLength()) {
216 DCHECK(index < m_CharCodes.size());
217 m_CharCodes[index++] = pFont->GetNextChar(segment, &offset);
219 if (i != nSegs - 1) {
220 m_CharPos[index - 1] = kernings[i];
221 m_CharCodes[index++] = CPDF_Font::kInvalidCharCode;
227 SetSegments(pdfium::span_from_ref(str), pdfium::span<
float>());
228 CalcPositionDataInternal(
GetFont());
235 const CPDF_CIDFont* pCIDFont = pFont->AsCIDFont();
236 if (!IsVertWritingCIDFont(pCIDFont))
237 return pFont->GetCharWidthF(charcode) * fontsize;
262 const float curpos = CalcPositionDataInternal(pFont);
263 if (IsVertWritingCIDFont(pFont->AsCIDFont()))
265 return {curpos * horz_scale, 0};
268float CPDF_TextObject::CalcPositionDataInternal(
271 float min_x = 10000.0f;
272 float max_x = -10000.0f;
273 float min_y = 10000.0f;
274 float max_y = -10000.0f;
275 const CPDF_CIDFont* pCIDFont = pFont->AsCIDFont();
276 const bool bVertWriting = IsVertWritingCIDFont(pCIDFont);
279 for (size_t i = 0; i < m_CharCodes.size(); ++i) {
280 const uint32_t charcode = m_CharCodes[i];
283 curpos -= (m_CharPos[i - 1] * fontsize) / 1000;
286 m_CharPos[i - 1] = curpos;
289 FX_RECT char_rect = pFont->GetCharBBox(charcode);
293 CFX_Point16 vertical_origin = pCIDFont->GetVertOrigin(cid);
294 char_rect
.Offset(-vertical_origin.x
, -vertical_origin.y
);
295 min_x =
std::min({min_x,
static_cast<
float>(char_rect
.left),
296 static_cast<
float>(char_rect
.right)});
297 max_x =
std::max({max_x,
static_cast<
float>(char_rect
.left),
298 static_cast<
float>(char_rect
.right)});
299 const float char_top = curpos + char_rect
.top * fontsize / 1000;
300 const float char_bottom = curpos + char_rect
.bottom * fontsize / 1000;
301 min_y =
std::min({min_y, char_top, char_bottom});
302 max_y =
std::max({max_y, char_top, char_bottom});
305 min_y =
std::min({min_y,
static_cast<
float>(char_rect
.top),
306 static_cast<
float>(char_rect
.bottom)});
307 max_y =
std::max({max_y,
static_cast<
float>(char_rect
.top),
308 static_cast<
float>(char_rect
.bottom)});
309 const float char_left = curpos + char_rect
.left * fontsize / 1000;
310 const float char_right = curpos + char_rect
.right * fontsize / 1000;
311 min_x =
std::min({min_x, char_left, char_right});
312 max_x =
std::max({max_x, char_left, char_right});
313 charwidth = pFont->GetCharWidthF(charcode) * fontsize / 1000;
323 min_x = min_x * fontsize / 1000;
324 max_x = max_x * fontsize / 1000;
326 min_y = min_y * fontsize / 1000;
327 max_y = max_y * fontsize / 1000;
332 if (TextRenderingModeIsStrokeMode(
text_state().GetTextMode())) {
335 const float half_width =
graph_state().GetLineWidth() / 2;
fxcrt::ByteString ByteString
constexpr CFX_FloatRect(float l, float b, float r, float t)
void Inflate(float x, float y)
CFX_Matrix operator*(const CFX_Matrix &right) const
uint16_t CIDFromCharCode(uint32_t charcode) const
int16_t GetVertWidth(uint16_t cid) const
bool IsVertWriting() const override
int GetCharSize(uint32_t charcode) const
static constexpr uint32_t kInvalidCharCode
void SetRect(const CFX_FloatRect &rect)
const CFX_FloatRect & GetOriginalRect() const
void SetDirty(bool value)
CPDF_PageObject(int32_t content_stream)
void SetOriginalRect(const CFX_FloatRect &rect)
static constexpr int32_t kNoContentStream
CPDF_TextState & mutable_text_state()
const CPDF_TextState & text_state() const
const CFX_GraphState & graph_state() const
RetainPtr< CPDF_Font > GetFont() const
void SetSegments(pdfium::span< const ByteString > strings, pdfium::span< const float > kernings)
void SetText(const ByteString &str)
CPDF_TextObject * AsText() override
Item GetItemInfo(size_t index) const
void SetTextRenderMode(TextRenderingMode mode)
CPDF_TextObject(int32_t content_stream)
CFX_Matrix GetTextMatrix() const
WideString GetWordString(int nWordIndex) const
std::unique_ptr< CPDF_TextObject > Clone() const
Type GetType() const override
bool IsText() const override
uint32_t GetCharCode(size_t index) const
Item GetCharInfo(size_t index) const
float GetCharWidth(uint32_t charcode) const
float GetFontSize() const
TextRenderingMode GetTextRenderMode() const
~CPDF_TextObject() override
size_t CountChars() const
const CPDF_TextObject * AsText() const override
size_t CountItems() const
void Transform(const CFX_Matrix &matrix) override
CFX_PointF CalcPositionData(float horz_scale)
void SetTextMatrix(const CFX_Matrix &matrix)
CFX_PTemplate< int16_t > CFX_Point16
CFX_PTemplate< float > CFX_PointF
fxcrt::ByteStringView ByteStringView
void Offset(int dx, int dy)
fxcrt::WideString WideString