Update compose file to avoid confusion. Fixes #490

README.md

@@ -57,7 +57,7 @@ services:
     restart: unless-stopped
     volumes:
       - /data/sist2-admin-data/:/sist2-admin/
-      - /:/host
+      - /<path to index>/:/host
     ports:
       - 4090:4090
       # NOTE: Don't expose this port publicly!

docker-compose.yml

@@ -18,7 +18,7 @@ services:
     container_name: sist2-admin
     volumes:
       - /data/sist2-admin-data/:/sist2-admin/
-      - /:/host
+      - /<path to index>/:/host
     ports:
       - 4090:4090
       # NOTE: Don't export this port publicly!

scripts/magic_static.py

@@ -1,5 +1,16 @@
 
-with open("/usr/lib/file/magic.mgc", "rb") as f:
-    data = f.read()
+MAGIC_PATHS = [
+    "/vcpkg/installed/x64-linux/share/libmagic/misc/magic.mgc",
+    "/work/vcpkg/installed/x64-linux/share/libmagic/misc/magic.mgc",
+    "/usr/lib/file/magic.mgc"
+]
+
+for path in MAGIC_PATHS:
+    try:
+        with open(path, "rb") as f:
+            data = f.read()
+        break
+    except:
+        continue
 
 print("char magic_database_buffer[%d] = {%s};" % (len(data), ",".join(str(int(b)) for b in data)))

src/cli.c

@@ -25,6 +25,7 @@ const char *TESS_DATAPATHS[] = {
         "/usr/share/tessdata/",
         "/usr/share/tesseract-ocr/tessdata/",
         "/usr/share/tesseract-ocr/4.00/tessdata/",
+        "/usr/share/tesseract-ocr/5/tessdata/",
         "./",
         NULL
 };

src/sist.h

@@ -55,7 +55,7 @@
 static const char *const Version = VERSION;
 static const int VersionMajor = 3;
 static const int VersionMinor = 4;
-static const int VersionPatch = 2;
+static const int VersionPatch = 3;
 
 #ifndef SIST_PLATFORM
 #define SIST_PLATFORM unknown

third-party/libscan/libscan/ebook/ebook.c

@@ -175,9 +175,19 @@ int render_cover(scan_ebook_ctx_t *ctx, fz_context *fzctx, document_t *doc, fz_d
     return TRUE;
 }
 
+#define IS_IGNORED_MESSAGE(message)           \
+    (                                          \
+        strstr(message, "invalid glyph index") \
+        || strstr(message, "... repeated")     \
+    )                                          \
+
 void fz_err_callback(void *user, const char *message) {
     document_t *doc = (document_t *) user;
 
+    if (IS_IGNORED_MESSAGE(message)) {
+        return;
+    }
+
     const scan_ebook_ctx_t *ctx = &thread_ctx;
     CTX_LOG_WARNINGF(doc->filepath, "FZ: %s", message);
 }
@@ -185,6 +195,10 @@ void fz_err_callback(void *user, const char *message) {
 void fz_warn_callback(void *user, const char *message) {
     document_t *doc = (document_t *) user;
 
+    if (IS_IGNORED_MESSAGE(message)) {
+        return;
+    }
+
     const scan_ebook_ctx_t *ctx = &thread_ctx;
     CTX_LOG_DEBUGF(doc->filepath, "FZ: %s", message);
 }

third-party/libscan/libscan/media/media.c

@@ -223,14 +223,10 @@ read_frame(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx, AVCodecContext *d
 
 void append_tag_meta_if_not_exists(scan_media_ctx_t *ctx, document_t *doc, AVDictionaryEntry *tag, enum metakey key) {
 
-    meta_line_t *meta = doc->meta_head;
-    while (meta != NULL) {
-        if (meta->key == key) {
-            CTX_LOG_DEBUGF(doc->filepath, "Ignoring duplicate tag: '%02x=%s' and '%02x=%s'",
-                           key, meta->str_val, key, tag->value);
-            return;
-        }
-        meta = meta->next;
+    if (meta_contains_key(doc->meta_head, key)) {
+        CTX_LOG_DEBUGF(doc->filepath, "Ignoring duplicate tag: '%02x=%s'",
+                       key, tag->value);
+        return;
     }
 
     text_buffer_t tex = text_buffer_create(-1);
@@ -445,7 +441,7 @@ int decode_frame_and_save_thumbnail(scan_media_ctx_t *ctx, AVFormatContext *pFor
         return SAVE_THUMBNAIL_FAILED;
     }
 
-    if (ctx->tesseract_lang != NULL && thumbnail_index == 0) {
+    if (ctx->tesseract_lang != NULL && thumbnail_index == 0 && !meta_contains_key(doc->meta_head, MetaContent)) {
         ocr_image(ctx, doc, decoder, frame_and_packet->frame);
     }
 

third-party/libscan/libscan/util.h

@@ -392,4 +392,18 @@ static parse_job_t *create_parse_job(const char *filepath, int mtime, size_t st_
     return job;
 }
 
+
+static int meta_contains_key (meta_line_t *meta_head, enum metakey key) {
+
+    meta_line_t *meta = meta_head;
+    while (meta != NULL) {
+        if (meta->key == key) {
+            return TRUE;
+        }
+        meta = meta->next;
+    }
+
+    return FALSE;
+}
+
 #endif