add tessdata-path cmdline option to tesstest

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

1 files changed, 13 insertions, 8 deletions

diff --git a/ccmain/test.cpp b/ccmain/test.cpp
index afe252d..10e8e72 100644
--- a/ccmain/test.cpp
+++ b/ccmain/test.cpp
@@ -25,13 +25,13 @@ BOOL_VAR (tessedit_write_images, FALSE,
                  "Capture the image from the IPE");
 
 int main(int argc, char **argv) {
-	const char *infile, *outfile, *lang, *ratings;
+	const char *tessdata, *infile, *outfile, *lang, *ratings;
 	void *buffer;
 	struct stat s;
 	int x, y, bpp, ifd;
 
-	FAILIF(argc < 7 || argc > 8, 
-		"tesstest infile xres yres bpp outfile lang [ratings]\n"); 
+	FAILIF(argc < 8 || argc > 9,
+		"tesstest infile xres yres bpp outfile lang tessdata [ratings]\n"); 
 
 	infile = argv[1];
 	FAILIF(sscanf(argv[2], "%d", &x) != 1, "could not parse x!\n");
@@ -39,7 +39,8 @@ int main(int argc, char **argv) {
 	FAILIF(sscanf(argv[4], "%d", &bpp) != 1, "could not parse bpp!\n");
 	outfile = argv[5];
 	lang = argv[6];
-	ratings = argv[7];
+	tessdata = argv[7];
+	ratings = argv[8];
 
 	printf("input file %s\n", infile);
 	ifd = open(infile, O_RDONLY);
@@ -49,23 +50,27 @@ int main(int argc, char **argv) {
 	buffer = mmap(NULL, s.st_size, PROT_READ, MAP_PRIVATE, ifd, 0);
 	FAILIF(buffer == MAP_FAILED, "mmap(): %s\n", strerror(errno));
 	printf("infile mmapped at %p\n", buffer);
+	FAILIF(tessdata, "You must specify a path for tessdata.\n");
 
 	tesseract::TessBaseAPI  api;
 
+	printf("tessdata %s\n", tessdata);
 	printf("lang %s\n", lang);
-	FAILIF(api.Init("/sdcard/", lang), "could not initialize tesseract\n");
+	FAILIF(api.Init(tessdata, lang), "could not initialize tesseract\n");
 	if (ratings) {
 		printf("ratings %s\n", ratings);
-		FAILIF(false == api.ReadConfigFile("/sdcard/tessdata/ratings"),
+		FAILIF(false == api.ReadConfigFile(ratings),
 			"could not read config file\n");
 	}
 
 	printf("set image x=%d, y=%d bpp=%d\n", x, y, bpp);
 
-    FAILIF(!bpp || bpp == 2 || bpp > 4, 
-           "Invalid value %d of bpp\n", bpp);
+	FAILIF(!bpp || bpp == 2 || bpp > 4, 
+		"Invalid value %d of bpp\n", bpp);
 
 	api.SetImage((const unsigned char *)buffer, x, y, bpp, bpp*x); 
+	printf("set page seg mode to single word\n");
+	api.SetPageSegMode(tesseract::PSM_SINGLE_WORD);
 	printf("recognize\n");
 	char * text = api.GetUTF8Text();
 	if (tessedit_write_images) {