svx: import processed PDF text

Some PDFs don't include spaces in the text.
Instead, they rely on the explicit positioning
of each character to render visually separated words.
Latex seems to be prone to this approach, though not
exclusively.

Luckily, PDFium does process text and inserts
"generated" spaces where necessary, which is what
we retrieve and use as the text string while importing.

Change-Id: Ic21fe6c8416ecaba66f06b6260f1d6b040ff12af

1 files changed, 140 insertions, 11 deletions

diff --git a/external/pdfium/edit.patch.1 b/external/pdfium/edit.patch.1
index 270dceb871b6..a110313017da 100644
--- a/external/pdfium/edit.patch.1
+++ b/external/pdfium/edit.patch.1
@@ -147,6 +147,56 @@ index 0a01ae0..6947e3a 100644
    if (bPattern) {
      DrawTextPathWithPattern(textobj, pObj2Device, pFont, font_size,
                              &text_matrix, bFill, bStroke);
+diff --git a/core/fpdftext/cpdf_textpage.cpp b/core/fpdftext/cpdf_textpage.cpp
+index e712549..a7973f7 100644
+--- a/core/fpdftext/cpdf_textpage.cpp
++++ b/core/fpdftext/cpdf_textpage.cpp
+@@ -1490,3 +1490,32 @@ bool CPDF_TextPage::IsRectIntersect(const CFX_FloatRect& rect1,
+   rect.Intersect(rect2);
+   return !rect.IsEmpty();
+ }
++
++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/core/fpdftext/cpdf_textpage.h b/core/fpdftext/cpdf_textpage.h
+index c87ab00..e5a1ba8 100644
+--- a/core/fpdftext/cpdf_textpage.h
++++ b/core/fpdftext/cpdf_textpage.h
+@@ -110,6 +110,8 @@ 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/core/fxge/cfx_pathdata.cpp b/core/fxge/cfx_pathdata.cpp
 index 4ac5cf6..4286de4 100644
 --- a/core/fxge/cfx_pathdata.cpp
@@ -199,7 +249,7 @@ index 0d7ba56..37bdf99 100644
  FPDFImageObj_GetImageDataDecoded(FPDF_PAGEOBJECT image_object,
                                   void* buffer,
 diff --git a/fpdfsdk/fpdfeditpage.cpp b/fpdfsdk/fpdfeditpage.cpp
-index ca2cf3f..832a9ae 100644
+index ca2cf3f..2162625 100644
 --- a/fpdfsdk/fpdfeditpage.cpp
 +++ b/fpdfsdk/fpdfeditpage.cpp
 @@ -11,12 +11,14 @@
@@ -217,7 +267,15 @@ index ca2cf3f..832a9ae 100644
  #include "core/fpdfapi/page/cpdf_shadingobject.h"
  #include "core/fpdfapi/parser/cpdf_array.h"
  #include "core/fpdfapi/parser/cpdf_document.h"
-@@ -363,3 +365,212 @@ FPDFPageObj_GetBounds(FPDF_PAGEOBJECT pageObject,
+@@ -24,6 +26,7 @@
+ #include "core/fpdfapi/parser/cpdf_string.h"
+ #include "core/fpdfdoc/cpdf_annot.h"
+ #include "core/fpdfdoc/cpdf_annotlist.h"
++#include "core/fpdftext/cpdf_textpage.h"
+ #include "fpdfsdk/fsdk_define.h"
+ #include "public/fpdf_formfill.h"
+ #include "third_party/base/logging.h"
+@@ -363,3 +366,252 @@ FPDFPageObj_GetBounds(FPDF_PAGEOBJECT pageObject,
    *top = bbox.top;
    return true;
  }
@@ -327,6 +385,46 @@ index ca2cf3f..832a9ae 100644
 +  return ret_count;
 +}
 +
++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 FPDF_BOOL FPDF_CALLCONV
 +FPDFTextObj_GetColor(FPDF_PAGEOBJECT text_object,
 +                     unsigned int* R,
@@ -509,10 +607,21 @@ index a291987..0202284 100644
    auto* pPathPoint = FXPathPointFromFPDFPathSegment(segment);
    if (!pPathPoint || !x || !y)
 diff --git a/fpdfsdk/fpdftext.cpp b/fpdfsdk/fpdftext.cpp
-index 68bf4f8..e073b20 100644
+index 68bf4f8..1cac9c7 100644
 --- a/fpdfsdk/fpdftext.cpp
 +++ b/fpdfsdk/fpdftext.cpp
-@@ -105,6 +105,28 @@ FPDF_EXPORT double FPDF_CALLCONV FPDFText_GetFontSize(FPDF_TEXTPAGE text_page,
+@@ -31,10 +31,6 @@ namespace {
+ 
+ constexpr size_t kBytesPerCharacter = sizeof(unsigned short);
+ 
+-CPDF_TextPage* CPDFTextPageFromFPDFTextPage(FPDF_TEXTPAGE text_page) {
+-  return static_cast<CPDF_TextPage*>(text_page);
+-}
+-
+ CPDF_TextPageFind* CPDFTextPageFindFromFPDFSchHandle(FPDF_SCHHANDLE handle) {
+   return static_cast<CPDF_TextPageFind*>(handle);
+ }
+@@ -105,6 +101,28 @@ FPDF_EXPORT double FPDF_CALLCONV FPDFText_GetFontSize(FPDF_TEXTPAGE text_page,
    return charinfo.m_FontSize;
  }
  
@@ -542,13 +651,17 @@ index 68bf4f8..e073b20 100644
                                                          int index,
                                                          double* left,
 diff --git a/fpdfsdk/fpdfview.cpp b/fpdfsdk/fpdfview.cpp
-index e890aa0..b62283f 100644
+index e890aa0..09d2345 100644
 --- a/fpdfsdk/fpdfview.cpp
 +++ b/fpdfsdk/fpdfview.cpp
-@@ -336,6 +336,16 @@ CPDF_Page* CPDFPageFromFPDFPage(FPDF_PAGE page) {
+@@ -336,6 +336,20 @@ CPDF_Page* CPDFPageFromFPDFPage(FPDF_PAGE page) {
  #endif  // PDF_ENABLE_XFA
  }
  
++CPDF_TextPage* CPDFTextPageFromFPDFTextPage(FPDF_TEXTPAGE text_page) {
++  return static_cast<CPDF_TextPage*>(text_page);
++}
++
 +CPDF_TextObject* CPDFTextObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object) {
 +  auto* obj = CPDFPageObjectFromFPDFPageObject(page_object);
 +  return obj ? obj->AsText() : nullptr;
@@ -563,22 +676,25 @@ index e890aa0..b62283f 100644
    auto* obj = CPDFPageObjectFromFPDFPageObject(page_object);
    return obj ? obj->AsPath() : nullptr;
 diff --git a/fpdfsdk/fsdk_define.h b/fpdfsdk/fsdk_define.h
-index 77c2315..b61f447 100644
+index 77c2315..e9a309a 100644
 --- a/fpdfsdk/fsdk_define.h
 +++ b/fpdfsdk/fsdk_define.h
-@@ -25,6 +25,8 @@ class CPDF_Annot;
+@@ -25,6 +25,9 @@ class CPDF_Annot;
  class CPDF_Page;
  class CPDF_PageObject;
  class CPDF_PageRenderContext;
 +class CPDF_TextObject;
++class CPDF_TextPage;
 +class CPDF_FormObject;
  class CPDF_PathObject;
  class CPDF_Stream;
  class IFSDK_PAUSE_Adapter;
-@@ -65,6 +67,10 @@ FPDF_DOCUMENT FPDFDocumentFromCPDFDocument(CPDF_Document* doc);
+@@ -65,6 +68,12 @@ FPDF_DOCUMENT FPDFDocumentFromCPDFDocument(CPDF_Document* doc);
  
  CPDF_Page* CPDFPageFromFPDFPage(FPDF_PAGE page);
  
++CPDF_TextPage* CPDFTextPageFromFPDFTextPage(FPDF_TEXTPAGE text_page);
++
 +CPDF_TextObject* CPDFTextObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object);
 +
 +CPDF_FormObject* CPDFFormObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object);
@@ -587,7 +703,7 @@ index 77c2315..b61f447 100644
  
  CPDF_PageObject* CPDFPageObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object);
 diff --git a/public/fpdf_edit.h b/public/fpdf_edit.h
-index 54735a3..2e7e2e7 100644
+index 54735a3..4d81aac 100644
 --- a/public/fpdf_edit.h
 +++ b/public/fpdf_edit.h
 @@ -520,6 +520,15 @@ FPDFPath_GetStrokeColor(FPDF_PAGEOBJECT path,
@@ -643,7 +759,7 @@ index 54735a3..2e7e2e7 100644
  // Create a new text object using one of the standard PDF fonts.
  //
  // document   - handle to the document.
-@@ -761,6 +800,112 @@ FPDFPageObj_CreateTextObj(FPDF_DOCUMENT document,
+@@ -761,6 +800,125 @@ FPDFPageObj_CreateTextObj(FPDF_DOCUMENT document,
                            FPDF_FONT font,
                            float font_size);
  
@@ -702,6 +818,19 @@ index 54735a3..2e7e2e7 100644
 +                    int char_count,
 +                    unsigned short* result);
 +
++// 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 stroke RGBA of a text. Range of values: 0 - 255.
 +//
 +// path   - the handle to the path object.