tesseract ocr path fix

src/cli.c

@@ -15,6 +15,13 @@
 #define DEFAULT_BIND_ADDR "localhost"
 #define DEFAULT_PORT "4090"
 
+const char* TESS_DATAPATHS[] = {
+        "/usr/share/tessdata/",
+        "/usr/share/tesseract-ocr/tessdata/",
+        "./",
+        NULL
+};
+
 
 scan_args_t *scan_args_create() {
     scan_args_t *args = calloc(sizeof(scan_args_t), 1);
@@ -136,13 +143,23 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
 
     if (args->tesseract_lang != NULL) {
         TessBaseAPI *api = TessBaseAPICreate();
-        ret = TessBaseAPIInit3(api, TESS_DATAPATH, args->tesseract_lang);
+
+        char filename[128];
+        sprintf(filename, "%s.traineddata", args->tesseract_lang);
+        const char * path = find_file_in_paths(TESS_DATAPATHS, filename);
+        if (path == NULL) {
+            LOG_FATAL("cli.c", "Could not find tesseract language file!");
+        }
+
+        ret = TessBaseAPIInit3(api, path, args->tesseract_lang);
         if (ret != 0) {
             fprintf(stderr, "Could not initialize tesseract with lang '%s'\n", args->tesseract_lang);
             return 1;
         }
         TessBaseAPIEnd(api);
         TessBaseAPIDelete(api);
+
+        args->tesseract_path = path;
     }
 
     LOG_DEBUGF("cli.c", "arg quality=%f", args->quality)
@@ -156,7 +173,8 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
     LOG_DEBUGF("cli.c", "arg depth=%d", args->depth)
     LOG_DEBUGF("cli.c", "arg path=%s", args->path)
     LOG_DEBUGF("cli.c", "arg archive=%s", args->archive)
-    LOG_DEBUGF("cli.c", "arg ocr=%s", args->tesseract_lang)
+    LOG_DEBUGF("cli.c", "arg tesseract_lang=%s", args->tesseract_lang)
+    LOG_DEBUGF("cli.c", "arg tesseract_path=%s", args->tesseract_path)
 
     return 0;
 }

src/cli.h

@@ -17,6 +17,7 @@ typedef struct scan_args {
     char *archive;
     archive_mode_t archive_mode;
     char *tesseract_lang;
+    const char *tesseract_path;
 } scan_args_t;
 
 scan_args_t *scan_args_create();

src/ctx.h

@@ -28,6 +28,7 @@ struct {
 
     pthread_mutex_t mupdf_mu;
     char * tesseract_lang;
+    char * tesseract_path;
 } ScanCtx;
 
 struct {

src/main.c

@@ -51,6 +51,7 @@ void sist2_scan(scan_args_t *args) {
     strncpy(ScanCtx.index.desc.root, args->path, sizeof(ScanCtx.index.desc.root));
     ScanCtx.index.desc.root_len = (short) strlen(ScanCtx.index.desc.root);
     ScanCtx.tesseract_lang = args->tesseract_lang;
+    ScanCtx.tesseract_path = args->tesseract_path;
 
     init_dir(ScanCtx.index.path);
 

src/parsing/pdf.c

@@ -141,7 +141,7 @@ void fill_image(fz_context *ctx, UNUSED(fz_device *dev),
 
         if (pix->h > MIN_OCR_SIZE && img->h > MIN_OCR_SIZE && img->xres != 0) {
             TessBaseAPI *api = TessBaseAPICreate();
-            TessBaseAPIInit3(api, TESS_DATAPATH, ScanCtx.tesseract_lang);
+            TessBaseAPIInit3(api, ScanCtx.tesseract_path, ScanCtx.tesseract_lang);
 
             TessBaseAPISetImage(api, pix->samples, pix->w, pix->h, pix->n, pix->stride);
             TessBaseAPISetSourceResolution(api, pix->xres);
@@ -157,8 +157,8 @@ void fill_image(fz_context *ctx, UNUSED(fz_device *dev),
 
             TessBaseAPIEnd(api);
             TessBaseAPIDelete(api);
-            fz_drop_pixmap(ctx, pix);
         }
+        fz_drop_pixmap(ctx, pix);
     }
 }
 

src/sist.h

@@ -3,7 +3,6 @@
 
 #define UUID_STR_LEN 37
 #define UNUSED(x) __attribute__((__unused__))  x
-#define TESS_DATAPATH "/usr/share/tessdata/"
 
 #include <glib-2.0/glib.h>
 #include <unistd.h>

src/util.c

@@ -1,4 +1,5 @@
 #include "util.h"
+#include "src/ctx.h"
 
 dyn_buffer_t dyn_buffer_create() {
     dyn_buffer_t buf;
@@ -317,4 +318,29 @@ GHashTable *incremental_get_table() {
     return file_table;
 }
 
+const char *find_file_in_paths(const char *paths[], const char *filename) {
+
+    for (int i = 0; paths[i] != NULL; i++) {
+
+        char *apath = abspath(paths[i]);
+        if (apath == NULL) {
+            continue;
+        }
+
+        char path[PATH_MAX];
+        snprintf(path, sizeof(path), "%s%s", apath, filename);
+
+        LOG_DEBUGF("util.c", "Looking for '%s' in folder '%s'", filename, apath)
+        free(apath);
+
+        struct stat info;
+        int ret = stat(path, &info);
+        if (ret != -1) {
+            return paths[i];
+        }
+    }
+
+    return NULL;
+}
+
 

src/util.h

@@ -74,5 +74,6 @@ int incremental_get(GHashTable *table, unsigned long inode_no);
 
 int incremental_mark_file_for_copy(GHashTable *table, unsigned long inode_no);
 
+const char *find_file_in_paths(const char **paths, const char *filename);
 
 #endif