pdfium: replace FPDFTextObj_CountChars/GetTextProcessed patches with backport

Change-Id: I9c198ff0cd31de064fcc4a3bb0ed92066f38e4c5
Reviewed-on: https://gerrit.libreoffice.org/60007
Tested-by: Jenkins
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>

7 files changed, 169 insertions, 248 deletions

diff --git a/external/pdfium/0001-Add-FPDFTextObj_GetText-API.patch.1 b/external/pdfium/0001-Add-FPDFTextObj_GetText-API.patch.1
new file mode 100644
index 000000000000..22926462cdac
--- /dev/null
+++ b/external/pdfium/0001-Add-FPDFTextObj_GetText-API.patch.1
@@ -0,0 +1,164 @@
+From 3bee9c60f013b8b7e99c39ee35699d132b330334 Mon Sep 17 00:00:00 2001
+Date: Tue, 7 Aug 2018 21:45:34 +0000
+Subject: [PATCH] Add FPDFTextObj_GetText() API
+
+Generalize CPDF_TextPage::GetTextByRect(), so that it's possible to get
+the text from a text page using a predicate, that way we can easily
+get the text that belongs to single text object as well.
+
+Change-Id: Ia457af0f41184694dc1481709be72b35685bce7f
+Reviewed-on: https://pdfium-review.googlesource.com/39530
+Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>
+Reviewed-by: Lei Zhang <thestig@chromium.org>
+Commit-Queue: Lei Zhang <thestig@chromium.org>
+---
+ core/fpdftext/cpdf_textpage.cpp    | 18 +++++++++++++--
+ core/fpdftext/cpdf_textpage.h      |  4 ++++
+ fpdfsdk/fpdf_edittext.cpp          | 18 +++++++++++++++
+ fpdfsdk/fpdf_text_embeddertest.cpp | 45 ++++++++++++++++++++++++++++++++++++++
+ fpdfsdk/fpdf_view_c_api_test.c     |  1 +
+ public/fpdf_edit.h                 | 20 +++++++++++++++++
+ 6 files changed, 104 insertions(+), 2 deletions(-)
+
+diff --git a/core/fpdftext/cpdf_textpage.cpp b/core/fpdftext/cpdf_textpage.cpp
+index 289416043..ed7f36fb6 100644
+--- a/core/fpdftext/cpdf_textpage.cpp
++++ b/core/fpdftext/cpdf_textpage.cpp
+@@ -426,7 +426,8 @@ int CPDF_TextPage::GetIndexAtPos(const CFX_PointF& point,
+   return pos < nCount ? pos : NearPos;
+ }
+ 
+-WideString CPDF_TextPage::GetTextByRect(const CFX_FloatRect& rect) const {
++WideString CPDF_TextPage::GetTextByPredicate(
++    const std::function<bool(const PAGECHAR_INFO&)>& predicate) const {
+   if (!m_bIsParsed)
+     return WideString();
+ 
+@@ -435,7 +436,7 @@ WideString CPDF_TextPage::GetTextByRect(const CFX_FloatRect& rect) const {
+   bool IsAddLineFeed = false;
+   WideString strText;
+   for (const auto& charinfo : m_CharList) {
+-    if (IsRectIntersect(rect, charinfo.m_CharBox)) {
++    if (predicate(charinfo)) {
+       if (fabs(posy - charinfo.m_Origin.y) > 0 && !IsContainPreChar &&
+           IsAddLineFeed) {
+         posy = charinfo.m_Origin.y;
+@@ -460,6 +461,19 @@ WideString CPDF_TextPage::GetTextByRect(const CFX_FloatRect& rect) const {
+   return strText;
+ }
+ 
++WideString CPDF_TextPage::GetTextByRect(const CFX_FloatRect& rect) const {
++  return GetTextByPredicate([&rect](const PAGECHAR_INFO& charinfo) {
++    return IsRectIntersect(rect, charinfo.m_CharBox);
++  });
++}
++
++WideString CPDF_TextPage::GetTextByObject(
++    const CPDF_TextObject* pTextObj) const {
++  return GetTextByPredicate([pTextObj](const PAGECHAR_INFO& charinfo) {
++    return charinfo.m_pTextObj == pTextObj;
++  });
++}
++
+ void CPDF_TextPage::GetCharInfo(int index, FPDF_CHAR_INFO* info) const {
+   if (!m_bIsParsed || !pdfium::IndexInBounds(m_CharList, index))
+     return;
+diff --git a/core/fpdftext/cpdf_textpage.h b/core/fpdftext/cpdf_textpage.h
+index 36d01854f..90b45bd96 100644
+--- a/core/fpdftext/cpdf_textpage.h
++++ b/core/fpdftext/cpdf_textpage.h
+@@ -8,6 +8,7 @@
+ #define CORE_FPDFTEXT_CPDF_TEXTPAGE_H_
+ 
+ #include <deque>
++#include <functional>
+ #include <vector>
+ 
+ #include "core/fpdfapi/page/cpdf_pageobjectlist.h"
+@@ -97,6 +98,7 @@ class CPDF_TextPage {
+   std::vector<CFX_FloatRect> GetRectArray(int start, int nCount) const;
+   int GetIndexAtPos(const CFX_PointF& point, const CFX_SizeF& tolerance) const;
+   WideString GetTextByRect(const CFX_FloatRect& rect) const;
++  WideString GetTextByObject(const CPDF_TextObject* pTextObj) const;
+ 
+   // Returns string with the text from |m_TextBuf| that are covered by the input
+   // range. |start| and |count| are in terms of the |m_CharIndex|, so the range
+@@ -151,6 +153,8 @@ class CPDF_TextPage {
+   TextOrientation FindTextlineFlowOrientation() const;
+   void AppendGeneratedCharacter(wchar_t unicode, const CFX_Matrix& formMatrix);
+   void SwapTempTextBuf(int32_t iCharListStartAppend, int32_t iBufStartAppend);
++  WideString GetTextByPredicate(
++      const std::function<bool(const PAGECHAR_INFO&)>& predicate) const;
+ 
+   UnownedPtr<const CPDF_Page> const m_pPage;
+   std::vector<uint16_t> m_CharIndex;
+diff --git a/fpdfsdk/fpdf_edittext.cpp b/fpdfsdk/fpdf_edittext.cpp
+index 6aa44b3b2..2773763b9 100644
+--- a/fpdfsdk/fpdf_edittext.cpp
++++ b/fpdfsdk/fpdf_edittext.cpp
+@@ -22,6 +22,7 @@
+ #include "core/fpdfapi/parser/cpdf_number.h"
+ #include "core/fpdfapi/parser/cpdf_reference.h"
+ #include "core/fpdfapi/parser/cpdf_stream.h"
++#include "core/fpdftext/cpdf_textpage.h"
+ #include "core/fxcrt/fx_extension.h"
+ #include "core/fxge/cfx_fontmgr.h"
+ #include "core/fxge/fx_font.h"
+@@ -564,6 +565,23 @@ FPDFTextObj_GetFontName(FPDF_PAGEOBJECT text,
+   return dwStringLen;
+ }
+ 
++FPDF_EXPORT unsigned long FPDF_CALLCONV
++FPDFTextObj_GetText(FPDF_PAGEOBJECT text_object,
++                    FPDF_TEXTPAGE text_page,
++                    void* buffer,
++                    unsigned long length) {
++  CPDF_TextObject* pTextObj = CPDFTextObjectFromFPDFPageObject(text_object);
++  if (!pTextObj)
++    return 0;
++
++  CPDF_TextPage* pTextPage = CPDFTextPageFromFPDFTextPage(text_page);
++  if (!pTextPage)
++    return 0;
++
++  WideString text = pTextPage->GetTextByObject(pTextObj);
++  return Utf16EncodeMaybeCopyAndReturnLength(text, buffer, length);
++}
++
+ FPDF_EXPORT void FPDF_CALLCONV FPDFFont_Close(FPDF_FONT font) {
+   CPDF_Font* pFont = CPDFFontFromFPDFFont(font);
+   if (!pFont)
+diff --git a/public/fpdf_edit.h b/public/fpdf_edit.h
+index 4d5aa9c48..83fedba90 100644
+--- a/public/fpdf_edit.h
++++ b/public/fpdf_edit.h
+@@ -1274,6 +1274,26 @@ FPDFTextObj_GetFontName(FPDF_PAGEOBJECT text,
+                         void* buffer,
+                         unsigned long length);
+ 
++// Experimental API.
++// Get the text of a text object.
++//
++// text_object      - the handle to the text object.
++// text_page        - the handle to the text page.
++// buffer           - the address of a buffer that receives the text.
++// length           - the size, in bytes, of |buffer|.
++//
++// Returns the number of bytes in the text (including the trailing NUL
++// character) on success, 0 on error.
++//
++// Regardless of the platform, the |buffer| is always in UTF16-LE encoding.
++// If |length| is less than the returned length, or |buffer| is NULL, |buffer|
++// will not be modified.
++FPDF_EXPORT unsigned long FPDF_CALLCONV
++FPDFTextObj_GetText(FPDF_PAGEOBJECT text_object,
++                    FPDF_TEXTPAGE text_page,
++                    void* buffer,
++                    unsigned long length);
++
+ // Experimental API.
+ // Get number of page objects inside |form_object|.
+ //
+-- 
+2.16.4
+
diff --git a/external/pdfium/0002-svx-more-accurate-PDF-text-importing.patch.2 b/external/pdfium/0002-svx-more-accurate-PDF-text-importing.patch.2
deleted file mode 100644
index 6288dccba6b4..000000000000
--- a/external/pdfium/0002-svx-more-accurate-PDF-text-importing.patch.2
+++ /dev/null
@@ -1,69 +0,0 @@
-From 5f83d0a3fac4f8ccef457c03b74433ffd7b12e2a Mon Sep 17 00:00:00 2001
-From: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
-Date: Tue, 5 Jun 2018 11:28:30 +0200
-Subject: [PATCH 02/14] svx: more accurate PDF text importing
-
----
- pdfium/fpdfsdk/fpdf_editpage.cpp | 84 ++++++++++++++++++++++++++++++++++++++++
- pdfium/public/fpdf_edit.h        | 36 +++++++++++++++++
- 2 files changed, 120 insertions(+)
-
-diff --git a/pdfium/fpdfsdk/fpdf_editpage.cpp b/pdfium/fpdfsdk/fpdf_editpage.cpp
-index 912df63..3244943 100644
---- a/pdfium/fpdfsdk/fpdf_editpage.cpp
-+++ b/pdfium/fpdfsdk/fpdf_editpage.cpp
-@@ -13,6 +13,7 @@
- 
- #include "constants/page_object.h"
- #include "core/fpdfapi/edit/cpdf_pagecontentgenerator.h"
-+#include "core/fpdfapi/font/cpdf_font.h"
- #include "core/fpdfapi/page/cpdf_form.h"
- #include "core/fpdfapi/page/cpdf_formobject.h"
- #include "core/fpdfapi/page/cpdf_imageobject.h"
-@@ -26,6 +27,7 @@
- #include "core/fpdfapi/parser/cpdf_string.h"
- #include "core/fpdfdoc/cpdf_annot.h"
- #include "core/fpdfdoc/cpdf_annotlist.h"
-+#include "core/fpdfapi/page/cpdf_textobject.h"
- #include "fpdfsdk/cpdfsdk_helpers.h"
- #include "public/fpdf_formfill.h"
- #include "third_party/base/logging.h"
-@@ -457,6 +459,16 @@ FPDFPageObj_Transform(FPDF_PAGEOBJECT page_object,
-   pPageObj->Transform(matrix);
- }
- 
-+FPDF_EXPORT int FPDF_CALLCONV
-+FPDFTextObj_CountChars(FPDF_PAGEOBJECT text_object)
-+{
-+  if (!text_object)
-+    return 0;
-+
-+  CPDF_TextObject* pTxtObj = static_cast<CPDF_TextObject*>(text_object);
-+  return pTxtObj->CountChars();
-+}
-+
- FPDF_EXPORT void FPDF_CALLCONV
- FPDFPageObj_SetBlendMode(FPDF_PAGEOBJECT page_object,
-                          FPDF_BYTESTRING blend_mode) {
-diff --git a/pdfium/public/fpdf_edit.h b/pdfium/public/fpdf_edit.h
---- a/pdfium/public/fpdf_edit.h
-+++ b/pdfium/public/fpdf_edit.h
-@@ -1152,6 +1152,15 @@ FPDFFormObj_CountObjects(FPDF_PAGEOBJECT form_object);
- FPDF_EXPORT FPDF_PAGEOBJECT FPDF_CALLCONV
- FPDFFormObj_GetObject(FPDF_PAGEOBJECT form_object, unsigned long index);
- 
-+// Get the number of characters from a text object.
-+//
-+// text_object - Handle of text object returned by FPDFPageObj_NewTextObj
-+//               or FPDFPageObj_NewTextObjEx.
-+// Return Value:
-+// A character count in the text object.
-+FPDF_EXPORT int FPDF_CALLCONV
-+FPDFTextObj_CountChars(FPDF_PAGEOBJECT text_object);
-+
- #ifdef __cplusplus
- }  // extern "C"
- #endif  // __cplusplus
--- 
-2.16.3
-
diff --git a/external/pdfium/0004-svx-support-PDF-text-color.patch.2 b/external/pdfium/0004-svx-support-PDF-text-color.patch.2
index 5f492007da3b..cffb8fb80530 100644
--- a/external/pdfium/0004-svx-support-PDF-text-color.patch.2
+++ b/external/pdfium/0004-svx-support-PDF-text-color.patch.2
@@ -25,19 +25,6 @@ index d93ecfc..13362cf 100644
  ByteString CFXByteStringFromFPDFWideString(FPDF_WIDESTRING wide_string);
  
  #ifdef PDF_ENABLE_XFA
-diff --git a/pdfium/fpdfsdk/fpdf_editpage.cpp b/pdfium/fpdfsdk/fpdf_editpage.cpp
-index 3244943..f8e2418 100644
---- a/pdfium/fpdfsdk/fpdf_editpage.cpp
-+++ b/pdfium/fpdfsdk/fpdf_editpage.cpp
-@@ -443,7 +443,7 @@ FPDFTextObj_CountChars(FPDF_PAGEOBJECT text_object)
-   if (!text_object)
-     return 0;
- 
--  CPDF_TextObject* pTxtObj = static_cast<CPDF_TextObject*>(text_object);
-+  CPDF_TextObject* pTxtObj = CPDFTextObjectFromFPDFPageObject(text_object);
-   return pTxtObj->CountChars();
- }
- 
 diff --git a/pdfium/fpdfsdk/fpdf_edittext.cpp b/pdfium/fpdfsdk/fpdf_edittext.cpp
 index c38873faa..aa3287ef4 100644
 --- a/pdfium/fpdfsdk/fpdf_edittext.cpp
diff --git a/external/pdfium/0011-svx-correctly-possition-form-objects-from-PDF.patch.2 b/external/pdfium/0011-svx-correctly-possition-form-objects-from-PDF.patch.2
index 3849de8b7c24..db6057899bc7 100644
--- a/external/pdfium/0011-svx-correctly-possition-form-objects-from-PDF.patch.2
+++ b/external/pdfium/0011-svx-correctly-possition-form-objects-from-PDF.patch.2
@@ -61,9 +61,9 @@ diff --git a/pdfium/public/fpdf_edit.h b/pdfium/public/fpdf_edit.h
 index ca76954..f249e64 100644
 --- a/pdfium/public/fpdf_edit.h
 +++ b/pdfium/public/fpdf_edit.h
-@@ -1161,6 +1161,24 @@ FPDFFormObj_GetObject(FPDF_PAGEOBJECT form_object, unsigned long index);
- FPDF_EXPORT int FPDF_CALLCONV
- FPDFTextObj_CountChars(FPDF_PAGEOBJECT text_object);
+@@ -1190,6 +1190,24 @@ FPDFFormObj_CountObjects(FPDF_PAGEOBJECT form_object);
+ FPDF_EXPORT FPDF_PAGEOBJECT FPDF_CALLCONV
+ FPDFFormObj_GetObject(FPDF_PAGEOBJECT form_object, unsigned long index);
  
 +// Get the matrix of a particular form object.
 +//
diff --git a/external/pdfium/0012-svx-import-processed-PDF-text.patch.2 b/external/pdfium/0012-svx-import-processed-PDF-text.patch.2
deleted file mode 100644
index 23629184603f..000000000000
--- a/external/pdfium/0012-svx-import-processed-PDF-text.patch.2
+++ /dev/null
@@ -1,148 +0,0 @@
-From 7e8ecec81f102993e3fe73256415dcf049c09e29 Mon Sep 17 00:00:00 2001
-From: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
-Date: Tue, 5 Jun 2018 11:35:39 +0200
-Subject: [PATCH 12/14] svx: import processed PDF text
-
----
- pdfium/core/fpdftext/cpdf_textpage.cpp | 29 ++++++++++++++++++++++++
- pdfium/core/fpdftext/cpdf_textpage.h   |  2 ++
- pdfium/fpdfsdk/fpdf_editpage.cpp       | 41 ++++++++++++++++++++++++++++++++++
- pdfium/public/fpdf_edit.h              | 13 +++++++++++
- 4 files changed, 85 insertions(+)
-
-diff --git a/pdfium/core/fpdftext/cpdf_textpage.cpp b/pdfium/core/fpdftext/cpdf_textpage.cpp
-index 5690698..4d7c48a 100644
---- a/pdfium/core/fpdftext/cpdf_textpage.cpp
-+++ b/pdfium/core/fpdftext/cpdf_textpage.cpp
-@@ -1464,3 +1464,32 @@ Optional<PAGECHAR_INFO> CPDF_TextPage::GenerateCharInfo(wchar_t unicode) {
-                                  info.m_Origin.x, info.m_Origin.y);
-   return info;
- }
-+
-+WideString CPDF_TextPage::GetTextObjectText(CPDF_TextObject* pTextObj)
-+{
-+  if (!m_bIsParsed)
-+    return WideString();
-+
-+  float posy = 0;
-+  bool IsContainPreChar = false;
-+  bool IsAddLineFeed = false;
-+  WideString strText;
-+  for (const auto& charinfo : m_CharList) {
-+    if (charinfo.m_pTextObj == pTextObj) {
-+      IsContainPreChar = true;
-+      IsAddLineFeed = false;
-+      if (charinfo.m_Unicode)
-+        strText += charinfo.m_Unicode;
-+    } else if (charinfo.m_Unicode == 32) {
-+      if (IsContainPreChar && charinfo.m_Unicode) {
-+        strText += charinfo.m_Unicode;
-+        IsContainPreChar = false;
-+        IsAddLineFeed = false;
-+      }
-+    } else {
-+      IsContainPreChar = false;
-+      IsAddLineFeed = true;
-+    }
-+  }
-+  return strText;
-+}
-diff --git a/pdfium/core/fpdftext/cpdf_textpage.h b/pdfium/core/fpdftext/cpdf_textpage.h
-index 43a0312..7d5d5ec 100644
---- a/pdfium/core/fpdftext/cpdf_textpage.h
-+++ b/pdfium/core/fpdftext/cpdf_textpage.h
-@@ -105,6 +105,8 @@ class CPDF_TextPage {
-   WideString GetPageText(int start, int count) const;
-   WideString GetAllPageText() const { return GetPageText(0, CountChars()); }
- 
-+  WideString GetTextObjectText(CPDF_TextObject* pTextObj);
-+
-   int CountRects(int start, int nCount);
-   bool GetRect(int rectIndex, CFX_FloatRect* pRect) const;
- 
-diff --git a/pdfium/fpdfsdk/fpdf_editpage.cpp b/pdfium/fpdfsdk/fpdf_editpage.cpp
-index f4a1688..f34d3b5 100644
---- a/pdfium/fpdfsdk/fpdf_editpage.cpp
-+++ b/pdfium/fpdfsdk/fpdf_editpage.cpp
-@@ -28,6 +28,7 @@
- #include "core/fpdfdoc/cpdf_annot.h"
- #include "core/fpdfdoc/cpdf_annotlist.h"
- #include "core/fpdfapi/page/cpdf_textobject.h"
-+#include "core/fpdftext/cpdf_textpage.h"
- #include "fpdfsdk/cpdfsdk_helpers.h"
- #include "public/fpdf_formfill.h"
- #include "third_party/base/logging.h"
-@@ -668,6 +669,46 @@ FPDFPageObj_SetLineCap(FPDF_PAGEOBJECT page_object, int line_cap) {
-   return true;
- }
- 
-+FPDF_EXPORT int FPDF_CALLCONV
-+FPDFTextObj_GetTextProcessed(FPDF_PAGEOBJECT text_object,
-+                             FPDF_TEXTPAGE page,
-+                             int char_start,
-+                             int char_count,
-+                             unsigned short* result)
-+{
-+  if (!page || !text_object || char_start < 0 || char_count < 0 || !result)
-+    return 0;
-+
-+  CPDF_TextObject* pTxtObj = CPDFTextObjectFromFPDFPageObject(text_object);
-+  CPDF_TextPage* textpage = CPDFTextPageFromFPDFTextPage(page);
-+  int char_available = textpage->CountChars() - char_start;
-+  if (char_available <= 0)
-+    return 0;
-+
-+  char_count = std::min(char_count, char_available);
-+  if (char_count == 0) {
-+    // Writing out "", which has a character count of 1 due to the NUL.
-+    *result = '\0';
-+    return 1;
-+  }
-+
-+  WideString str = textpage->GetTextObjectText(pTxtObj);
-+
-+  if (str.GetLength() > static_cast<size_t>(char_count))
-+    str = str.Left(static_cast<size_t>(char_count));
-+
-+  // UFT16LE_Encode doesn't handle surrogate pairs properly, so it is expected
-+  // the number of items to stay the same.
-+  ByteString byte_str = str.UTF16LE_Encode();
-+  size_t byte_str_len = byte_str.GetLength();
-+  constexpr size_t kBytesPerCharacter = sizeof(unsigned short);
-+  int ret_count = byte_str_len / kBytesPerCharacter;
-+
-+  ASSERT(ret_count <= char_count + 1);  // +1 to account for the NUL terminator.
-+  memcpy(result, byte_str.GetBuffer(byte_str_len), byte_str_len);
-+  return ret_count;
-+}
-+
- FPDF_EXPORT int FPDF_CALLCONV
- FPDFFormObj_CountObjects(FPDF_PAGEOBJECT page_object) {
-   const CPDF_PageObjectList* pObjectList =
-diff --git a/pdfium/public/fpdf_edit.h b/pdfium/public/fpdf_edit.h
-index f249e64..e14b2a5 100644
---- a/pdfium/public/fpdf_edit.h
-+++ b/pdfium/public/fpdf_edit.h
-@@ -1152,6 +1152,19 @@ FPDFFormObj_CountObjects(FPDF_PAGEOBJECT form_object);
- FPDF_EXPORT FPDF_PAGEOBJECT FPDF_CALLCONV
- FPDFFormObj_GetObject(FPDF_PAGEOBJECT form_object, unsigned long index);
- 
-+// Get the processed text of a text object.
-+//
-+// text_object - Handle of text object returned by FPDFPageObj_NewTextObj
-+//               or FPDFPageObj_NewTextObjEx.
-+// Return Value:
-+// The number of characters (not bytes) written in result.
-+FPDF_EXPORT int FPDF_CALLCONV
-+FPDFTextObj_GetTextProcessed(FPDF_PAGEOBJECT text_object,
-+                             FPDF_TEXTPAGE page,
-+                             int char_start,
-+                             int char_count,
-+                             unsigned short* result);
-+
- // Get the number of characters from a text object.
- //
- // text_object - Handle of text object returned by FPDFPageObj_NewTextObj
--- 
-2.16.3
-
diff --git a/external/pdfium/0014-svx-update-PDFium-patch-and-code.patch.2 b/external/pdfium/0014-svx-update-PDFium-patch-and-code.patch.2
index f664c80232a5..3c53a2772e05 100644
--- a/external/pdfium/0014-svx-update-PDFium-patch-and-code.patch.2
+++ b/external/pdfium/0014-svx-update-PDFium-patch-and-code.patch.2
@@ -36,19 +36,6 @@ index 3f400c7..968b84a 100644
  }
  
  FPDF_EXPORT unsigned long FPDF_CALLCONV
-diff --git a/pdfium/fpdfsdk/fpdf_editpage.cpp b/pdfium/fpdfsdk/fpdf_editpage.cpp
-index 29c8b01..a52e1a9 100644
---- a/pdfium/fpdfsdk/fpdf_editpage.cpp
-+++ b/pdfium/fpdfsdk/fpdf_editpage.cpp
-@@ -688,7 +688,7 @@ FPDFTextObj_GetTextProcessed(FPDF_PAGEOBJECT text_object,
-   int ret_count = byte_str_len / kBytesPerCharacter;
- 
-   ASSERT(ret_count <= char_count + 1);  // +1 to account for the NUL terminator.
--  memcpy(result, byte_str.GetBuffer(byte_str_len), byte_str_len);
-+  memcpy(result, byte_str.GetBuffer(byte_str_len).data(), byte_str_len);
-   return ret_count;
- }
- 
 -- 
 2.16.3
 
diff --git a/external/pdfium/UnpackedTarball_pdfium.mk b/external/pdfium/UnpackedTarball_pdfium.mk
index d5295106db4b..0357fdfd4ec1 100644
--- a/external/pdfium/UnpackedTarball_pdfium.mk
+++ b/external/pdfium/UnpackedTarball_pdfium.mk
@@ -23,11 +23,11 @@ pdfium_patches += 0001-Add-FPDFFormObj_CountObjects-API.patch.1
 pdfium_patches += 0001-Add-FPDFFormObj_GetObject-API.patch.1
 # Backport of <https://pdfium-review.googlesource.com/38870>.
 pdfium_patches += 0001-Add-FPDFText_GetFontName-API.patch.1
-pdfium_patches += 0002-svx-more-accurate-PDF-text-importing.patch.2
+# Backport of <https://pdfium-review.googlesource.com/39530>.
+pdfium_patches += 0001-Add-FPDFTextObj_GetText-API.patch.1
 pdfium_patches += 0003-svx-import-PDF-images-as-BGRA.patch.2
 pdfium_patches += 0004-svx-support-PDF-text-color.patch.2
 pdfium_patches += 0011-svx-correctly-possition-form-objects-from-PDF.patch.2
-pdfium_patches += 0012-svx-import-processed-PDF-text.patch.2
 pdfium_patches += 0014-svx-update-PDFium-patch-and-code.patch.2
 
 $(eval $(call gb_UnpackedTarball_UnpackedTarball,pdfium))