Add ProcessText supporting standard fonts

BUG=pdfium:667

Change-Id: I701719144127881ebdb5be01a51e833d1e576477
Reviewed-on: https://pdfium-review.googlesource.com/2691
Commit-Queue: Nicolás Peña <npm@chromium.org>
Reviewed-by: dsinclair <dsinclair@chromium.org>

3 files changed, 95 insertions, 6 deletions

diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp b/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
index 9de97ee718..4de89a47de 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
@@ -9,12 +9,14 @@
 #include <tuple>
 #include <utility>
 
+#include "core/fpdfapi/font/cpdf_font.h"
 #include "core/fpdfapi/page/cpdf_docpagedata.h"
 #include "core/fpdfapi/page/cpdf_image.h"
 #include "core/fpdfapi/page/cpdf_imageobject.h"
 #include "core/fpdfapi/page/cpdf_page.h"
 #include "core/fpdfapi/page/cpdf_path.h"
 #include "core/fpdfapi/page/cpdf_pathobject.h"
+#include "core/fpdfapi/page/cpdf_textobject.h"
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
 #include "core/fpdfapi/parser/cpdf_document.h"
 #include "core/fpdfapi/parser/cpdf_name.h"
@@ -63,6 +65,8 @@ void CPDF_PageContentGenerator::GenerateContent() {
       ProcessImage(&buf, pImageObject);
     else if (CPDF_PathObject* pPathObj = pPageObj->AsPath())
       ProcessPath(&buf, pPathObj);
+    else if (CPDF_TextObject* pTextObj = pPageObj->AsText())
+      ProcessText(&buf, pTextObj);
   }
   CPDF_Dictionary* pPageDict = m_pPage->m_pFormDict;
   CPDF_Object* pContent =
@@ -240,3 +244,46 @@ bool CPDF_PageContentGenerator::GraphicsData::operator<(
     return fillAlpha < other.fillAlpha;
   return strokeAlpha < other.strokeAlpha;
 }
+
+bool CPDF_PageContentGenerator::FontData::operator<(
+    const FontData& other) const {
+  return baseFont < other.baseFont;
+}
+
+// This method adds text to the buffer, BT begins the text object, ET ends it.
+// Tm sets the text matrix (allows positioning and transforming text).
+// Tf sets the font name (from Font in Resources) and font size.
+// Tj sets the actual text, <####...> is used when specifying charcodes.
+void CPDF_PageContentGenerator::ProcessText(CFX_ByteTextBuf* buf,
+                                            CPDF_TextObject* pTextObj) {
+  // TODO(npm): Add support for something other than standard type1 fonts.
+  *buf << "BT " << pTextObj->GetTextMatrix() << " Tm ";
+  CPDF_Font* pFont = pTextObj->GetFont();
+  if (!pFont)
+    pFont = CPDF_Font::GetStockFont(m_pDocument, "Helvetica");
+  FontData fontD;
+  fontD.baseFont = pFont->GetBaseFont();
+  auto it = m_FontsMap.find(fontD);
+  CFX_ByteString dictName;
+  if (it != m_FontsMap.end()) {
+    dictName = it->second;
+  } else {
+    auto fontDict = pdfium::MakeUnique<CPDF_Dictionary>();
+    fontDict->SetNewFor<CPDF_Name>("Type", "Font");
+    fontDict->SetNewFor<CPDF_Name>("Subtype", "Type1");
+    fontDict->SetNewFor<CPDF_Name>("BaseFont", fontD.baseFont);
+    CPDF_Object* pDict = m_pDocument->AddIndirectObject(std::move(fontDict));
+    uint32_t dwObjNum = pDict->GetObjNum();
+    dictName = RealizeResource(dwObjNum, "Font");
+    m_FontsMap[fontD] = dictName;
+  }
+  *buf << "/" << PDF_NameEncode(dictName) << " " << pTextObj->GetFontSize()
+       << " Tf ";
+  CFX_ByteString text;
+  for (uint32_t charcode : pTextObj->m_CharCodes) {
+    if (charcode == CPDF_Font::kInvalidCharCode)
+      continue;
+    pFont->AppendChar(text, charcode);
+  }
+  *buf << PDF_EncodeString(text, true) << " Tj ET\n";
+}
diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator.h b/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
index e48ea4a7c9..fd80bd8f44 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
@@ -18,6 +18,7 @@ class CPDF_ImageObject;
 class CPDF_Page;
 class CPDF_PageObject;
 class CPDF_PathObject;
+class CPDF_TextObject;
 
 class CPDF_PageContentGenerator {
  public:
@@ -32,6 +33,7 @@ class CPDF_PageContentGenerator {
   void ProcessPath(CFX_ByteTextBuf* buf, CPDF_PathObject* pPathObj);
   void ProcessImage(CFX_ByteTextBuf* buf, CPDF_ImageObject* pImageObj);
   void ProcessGraphics(CFX_ByteTextBuf* buf, CPDF_PageObject* pPageObj);
+  void ProcessText(CFX_ByteTextBuf* buf, CPDF_TextObject* pTextObj);
   CFX_ByteString RealizeResource(uint32_t dwResourceObjNum,
                                  const CFX_ByteString& bsType);
 
@@ -41,7 +43,13 @@ class CPDF_PageContentGenerator {
     bool operator<(const GraphicsData& other) const;
   };
 
+  struct FontData {
+    CFX_ByteString baseFont;
+    bool operator<(const FontData& other) const;
+  };
+
   std::map<GraphicsData, CFX_ByteString> m_GraphicsMap;
+  std::map<FontData, CFX_ByteString> m_FontsMap;
   CPDF_Page* const m_pPage;
   CPDF_Document* const m_pDocument;
   std::vector<CPDF_PageObject*> m_pageObjects;
diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp b/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp
index 4846b1bd3c..0a636a1888 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp
@@ -5,8 +5,10 @@
 #include "core/fpdfapi/edit/cpdf_pagecontentgenerator.h"
 
 #include "core/fpdfapi/cpdf_modulemgr.h"
+#include "core/fpdfapi/font/cpdf_font.h"
 #include "core/fpdfapi/page/cpdf_page.h"
 #include "core/fpdfapi/page/cpdf_pathobject.h"
+#include "core/fpdfapi/page/cpdf_textobject.h"
 #include "core/fpdfapi/parser/cpdf_document.h"
 #include "core/fpdfapi/parser/cpdf_parser.h"
 #include "testing/gtest/include/gtest/gtest.h"
@@ -24,10 +26,16 @@ class CPDF_PageContentGeneratorTest : public testing::Test {
     pGen->ProcessPath(buf, pPathObj);
   }
 
-  CPDF_Dictionary* TestGetGS(CPDF_PageContentGenerator* pGen,
-                             const CFX_ByteString& name) {
-    return pGen->m_pPage->m_pResources->GetDictFor("ExtGState")
-        ->GetDictFor(name);
+  CPDF_Dictionary* TestGetResource(CPDF_PageContentGenerator* pGen,
+                                   const CFX_ByteString& type,
+                                   const CFX_ByteString& name) {
+    return pGen->m_pPage->m_pResources->GetDictFor(type)->GetDictFor(name);
+  }
+
+  void TestProcessText(CPDF_PageContentGenerator* pGen,
+                       CFX_ByteTextBuf* buf,
+                       CPDF_TextObject* pTextObj) {
+    pGen->ProcessText(buf, pTextObj);
   }
 };
 
@@ -162,8 +170,8 @@ TEST_F(CPDF_PageContentGeneratorTest, ProcessGraphics) {
             pathString.Left(48));
   EXPECT_EQ(" gs 1 2 m 3 4 l 5 6 l h B Q\n", pathString.Right(28));
   ASSERT_TRUE(pathString.GetLength() > 76);
-  CPDF_Dictionary* externalGS =
-      TestGetGS(&generator, pathString.Mid(48, pathString.GetLength() - 76));
+  CPDF_Dictionary* externalGS = TestGetResource(
+      &generator, "ExtGState", pathString.Mid(48, pathString.GetLength() - 76));
   ASSERT_TRUE(externalGS);
   EXPECT_EQ(0.5f, externalGS->GetNumberFor("ca"));
   EXPECT_EQ(0.8f, externalGS->GetNumberFor("CA"));
@@ -181,3 +189,29 @@ TEST_F(CPDF_PageContentGeneratorTest, ProcessGraphics) {
   EXPECT_EQ(pathString.Mid(48, pathString.GetLength() - 76),
             pathString2.Mid(55, pathString2.GetLength() - 83));
 }
+
+TEST_F(CPDF_PageContentGeneratorTest, ProcessText) {
+  auto pDoc = pdfium::MakeUnique<CPDF_Document>(nullptr);
+  pDoc->CreateNewDoc();
+  CPDF_Dictionary* pPageDict = pDoc->CreateNewPage(0);
+  auto pTestPage = pdfium::MakeUnique<CPDF_Page>(pDoc.get(), pPageDict, false);
+  CPDF_PageContentGenerator generator(pTestPage.get());
+  auto pTextObj = pdfium::MakeUnique<CPDF_TextObject>();
+  CPDF_Font* pFont = CPDF_Font::GetStockFont(pDoc.get(), "Times-Roman");
+  pTextObj->m_TextState.SetFont(pFont);
+  pTextObj->m_TextState.SetFontSize(10.0f);
+  pTextObj->Transform(CFX_Matrix(1, 0, 0, 1, 100, 100));
+  pTextObj->SetText("Hello World");
+  CFX_ByteTextBuf buf;
+  TestProcessText(&generator, &buf, pTextObj.get());
+  CFX_ByteString textString = buf.MakeString();
+  EXPECT_LT(61, textString.GetLength());
+  EXPECT_EQ("BT 1 0 0 1 100 100 Tm /", textString.Left(23));
+  EXPECT_EQ(" 10 Tf <48656C6C6F20576F726C64> Tj ET\n", textString.Right(38));
+  CPDF_Dictionary* fontDict = TestGetResource(
+      &generator, "Font", textString.Mid(23, textString.GetLength() - 61));
+  ASSERT_TRUE(fontDict);
+  EXPECT_EQ("Font", fontDict->GetStringFor("Type"));
+  EXPECT_EQ("Type1", fontDict->GetStringFor("Subtype"));
+  EXPECT_EQ("Times-Roman", fontDict->GetStringFor("BaseFont"));
+}