add support for the ratings file to helium and tesseract\'s simple test app

Signed-off-by: Iliyan Malchev <malchev@google.com>

6 files changed, 49 insertions, 15 deletions

diff --git a/ccmain/test.cpp b/ccmain/test.cpp
index 15531c1..55d0a7b 100644
--- a/ccmain/test.cpp
+++ b/ccmain/test.cpp
@@ -9,6 +9,8 @@
 #include <sys/stat.h>
 
 #include "baseapi.h"
+#include "varable.h"
+#include "tessvars.h"
 
 #define FAILIF(cond, msg...) do {                        	 \
         if (cond) { 	                                	 \
@@ -18,21 +20,25 @@
         }                                                	 \
 } while(0)
 
+// This is to make the ratings file parser happy.
+BOOL_VAR (tessedit_write_images, FALSE,
+                 "Capture the image from the IPE");
 
 int main(int argc, char **argv) {
-	const char *infile, *outfile, *lang;
+	const char *infile, *outfile, *lang, *ratings;
 	void *buffer;
 	struct stat s;
 	int x, y, ifd;
 
-	FAILIF(argc != 6, 
-		"tesstest infile xres yres outfile lang\n"); 
+	FAILIF(argc < 6 || argc > 7, 
+		"tesstest infile xres yres outfile lang [ratings]\n"); 
 
 	infile = argv[1];
 	FAILIF(sscanf(argv[2], "%d", &x) != 1, "could not parse x!\n");
 	FAILIF(sscanf(argv[3], "%d", &y) != 1, "could not parse y!\n");
 	outfile = argv[4];
 	lang = argv[5];
+	ratings = argv[6];
 
 	printf("input file %s\n", infile);
 	ifd = open(infile, O_RDONLY);
@@ -47,15 +53,19 @@ int main(int argc, char **argv) {
 
 	printf("lang %s\n", lang);
 	FAILIF(api.Init("/sdcard/", lang), "could not initialize tesseract\n");
-#if 0
-	FAILIF(false == api.ReadConfigFile("/sdcard/tessdata/ratings"),
-		"could not read config file\n");
-#endif
+	if (ratings) {
+		printf("ratings %s\n", ratings);
+		FAILIF(false == api.ReadConfigFile("/sdcard/tessdata/ratings"),
+			"could not read config file\n");
+	}
 
 	printf("set image x=%d, y=%d\n", x, y);
 	api.SetImage((const unsigned char *)buffer, x, y, 1, x); 
 	printf("recognize\n");
 	char * text = api.GetUTF8Text();
+	if (tessedit_write_images) {
+		page_image.write("tessinput.tif");
+	}
 	FAILIF(text == NULL, "didn't recognize\n");
 
 	printf("write to output %s\n", outfile);
diff --git a/helium/tesseract.cpp b/helium/tesseract.cpp
index 8b00760..861f7ce 100644
--- a/helium/tesseract.cpp
+++ b/helium/tesseract.cpp
@@ -16,6 +16,10 @@
 #include "ccmain/control.h"
 #endif
 
+// This is to make the ratings file parser happy.
+BOOL_VAR (tessedit_write_images, FALSE,
+                 "Capture the image from the IPE");
+
 #undef LOG
 
 // Local includes
@@ -29,8 +33,13 @@ using namespace helium;
 
 const char* kArguments[3] = { "tesseract", "out", "batch" };
 
-int Tesseract::Init(const char* datapath, const char *lang) {
-  return api_.Init(datapath, lang);
+int Tesseract::Init(const char* datapath,
+                    const char *lang,
+                    const char *configfile) {
+  int res = api_.Init(datapath, lang);
+  if (!res && configfile)
+      api_.ReadConfigFile(configfile);
+  return res;
 }
 
 void Tesseract::ReadMask(const Mask& mask, bool flipped) {
@@ -70,8 +79,14 @@ char* Tesseract::RecognizeText(const Mask& mask) {
   MaskToBuffer(mask, buf);
   api_.SetImage(buf, mask.width(), mask.height(), 1, mask.width());
   api_.Recognize(NULL);
+
   delete[] buf;
-  return api_.GetUTF8Text();
+  char *text = api_.GetUTF8Text();
+
+  if (tessedit_write_images) 
+      page_image.write("tessinput.tif");
+
+  return text;
 }
 
 void Tesseract::End() {
diff --git a/helium/tesseract.h b/helium/tesseract.h
index f92d21d..dc88ebe 100644
--- a/helium/tesseract.h
+++ b/helium/tesseract.h
@@ -38,7 +38,9 @@ class Tesseract {
     // Call this method exactly once to initialize the Tesseract engine with
     // the data files at the specified path (This should be the path, that 
     // contains the 'tessdata' folder).
-    static int Init(const char* datapath, const char *lang = NULL);
+    static int Init(const char* datapath,
+                    const char *lang = NULL,
+                    const char *configfile = NULL);
                 
     // Find the baseline, specified by the offset and slope, for the given
     // Mask. If flipped is true, this method will flip the image vertically
diff --git a/helium/test.cpp b/helium/test.cpp
index 4c4891d..1d384aa 100644
--- a/helium/test.cpp
+++ b/helium/test.cpp
@@ -82,7 +82,10 @@ int main(int argc, char** argv) {
     // Run OCR
     printf("OCRing (language %s)...\n", lang);
     TextAreas text;
-    TextRecognition::Init("/sdcard/", lang);
+    TextRecognition::Init("/sdcard/", 
+                          lang,
+                          "/sdcard/tessdata/ratings");
+
     TextRecognition::RecognizeUsingBinarizer(&binarizer, text);
     
     // Output Text
diff --git a/helium/textrecognition.cpp b/helium/textrecognition.cpp
index 5768449..f2fdf90 100644
--- a/helium/textrecognition.cpp
+++ b/helium/textrecognition.cpp
@@ -16,9 +16,11 @@ using namespace helium;
 
 bool TextRecognition::recognizer_initialized_ = false;
 
-void TextRecognition::Init(const char* data_path, const char *lang) {
+void TextRecognition::Init(const char* data_path,
+                           const char *lang,
+                           const char *configfile) {
   // Allow reinitialization of Tesseract to get around its adaptation.
-  Tesseract::Init(data_path, lang);
+  Tesseract::Init(data_path, lang, configfile);
   recognizer_initialized_ = true;
 }
 
diff --git a/helium/textrecognition.h b/helium/textrecognition.h
index 53f2233..fa4ecf5 100644
--- a/helium/textrecognition.h
+++ b/helium/textrecognition.h
@@ -23,7 +23,9 @@ class TextRecognition {
     // tessdata directory. 
     // This must be called before using RecognizeUsingBinarizer(...), but it
     // can be callled multiple times to clear OCR's internal adaptation.
-    static void Init(const char* data_path, const char *lang = NULL);
+    static void Init(const char* data_path,
+                     const char *lang = NULL,
+                     const char *configfile = NULL);
     
     // This method passes all the binarized masks, that were extracted by the
     // specified Binarizer, through perspective correction, and