Add text dump to pdfium_test.

samples/pdfium_test.cc

Index: samples/pdfium_test.cc
diff --git a/samples/pdfium_test.cc b/samples/pdfium_test.cc
index 5ee14642e4b7993f2d5de807f67f7427ca65ddfb..dc858db250221ce87a693129820a3b852dce549a 100644
--- a/samples/pdfium_test.cc
+++ b/samples/pdfium_test.cc
@@ -42,6 +42,7 @@
 
 enum OutputFormat {
   OUTPUT_NONE,
+  OUTPUT_TEXT,
   OUTPUT_PPM,
   OUTPUT_PNG,
 #ifdef _WIN32
@@ -112,6 +113,36 @@ static void WritePpm(const char* pdf_name, int num, const void* buffer_void,
   fclose(fp);
 }
 
+void WriteText(FPDF_PAGE page, const char* pdf_name, int num) {
+  char filename[256];
+  int chars_formatted =
+      snprintf(filename, sizeof(filename), "%s.%d.txt", pdf_name, num);
+  if (chars_formatted < 0 ||
+      static_cast<size_t>(chars_formatted) >= sizeof(filename)) {
+    fprintf(stderr, "Filename %s is too long\n", filename);
+    return;
+  }
+
+  FILE* fp = fopen(filename, "w");
+  if (!fp) {
+    fprintf(stderr, "Failed to open %s for output\n", filename);
+    return;
+  }
+
+  unsigned char bom[] = {0xFF, 0xFE, 0x00, 0x00};

[hal.canary, 2016/06/14 19:27:57]

What is the output format?  UTF-32-LE?

[dsinclair, 2016/06/14 20:25:27]

Yup, UTF32-LE.

+  fwrite(bom, sizeof(unsigned char), sizeof(bom), fp);
+
+  FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
+  for (int i = 0; i < FPDFText_CountChars(textpage); i++) {
+    wchar_t c = FPDFText_GetUnicode(textpage, i);
+    fwrite(&c, sizeof(wchar_t), 1, fp);

[hal.canary, 2016/06/14 19:23:12]

sizeof(wchar_t) is "compiler-dependent and therefore not very portable"

[dsinclair, 2016/06/14 20:25:27]

Done.

+  }
+
+  FPDFText_ClosePage(textpage);
+
+  (void)fclose(fp);
+}
+
 static void WritePng(const char* pdf_name, int num, const void* buffer_void,
                      int stride, int width, int height) {
   if (!CheckDimensions(stride, width, height))
@@ -354,6 +385,12 @@ bool ParseCommandLine(const std::vector<std::string>& args,
         return false;
       }
       options->output_format = OUTPUT_PNG;
+    } else if (cur_arg == "--text") {
+      if (options->output_format != OUTPUT_NONE) {
+        fprintf(stderr, "Duplicate or conflicting --text argument\n");
+        return false;
+      }
+      options->output_format = OUTPUT_TEXT;
 #ifdef PDF_ENABLE_SKIA
     } else if (cur_arg == "--skp") {
       if (options->output_format != OUTPUT_NONE) {
@@ -528,6 +565,10 @@ bool RenderPage(const std::string& name,
         WriteEmf(page, name.c_str(), page_index);
         break;
 #endif
+      case OUTPUT_TEXT:
+        WriteText(page, name.c_str(), page_index);
+        break;
+
       case OUTPUT_PNG:
         WritePng(name.c_str(), page_index, buffer, stride, width, height);
         break;