Merge 4ec6a14252a5bf93a8e0eaeaa83e9317a1211990 into 7a03a2202ec52db2f4ba35413259799fcc6cdcbc

Added the version badge-pill to the sist2-admin page so you can check the version without going to the main search frontend.

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:
+MAGIC_PATHS = [
-    data = f.read()
+    "/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)))

sist2-admin/frontend/src/components/NavBar.vue

@@ -3,7 +3,11 @@
     <b-navbar-brand to="/">
       <Sist2Icon></Sist2Icon>
     </b-navbar-brand>
-
+    
+    <span class="badge badge-pill version" v-if="$store && $store.state.sist2Info">
+      v{{ sist2Version() }}
+    </span>
+      
     <b-button class="ml-auto" to="/task" variant="link">{{ $t("tasks") }}</b-button>
   </b-navbar>
 </template>
@@ -66,4 +70,4 @@ export default {
 .btn-link {
   color: #222;
 }
-</style>
+</style>

sist2-vue/src/Sist2Api.js

@@ -309,7 +309,7 @@ class Sist2Api {
     }
 
     getTagsSqlite() {
-        return axios.get(`${this.baseUrl}/fts/tags`)
+        return axios.get(`${this.baseUrl}fts/tags`)
             .then(resp => {
                 return resp.data.map(tag => this._createEsTag(tag.tag, tag.count))
             });
@@ -566,7 +566,7 @@ class Sist2Api {
     }
 
     getDocumentSqlite(sid) {
-        return axios.get(`${this.baseUrl}/fts/d/${sid}`)
+        return axios.get(`${this.baseUrl}fts/d/${sid}`)
             .then(resp => ({
                 _source: resp.data
             }));
@@ -589,7 +589,7 @@ class Sist2Api {
     }
 
     getTagSuggestionsSqlite(prefix) {
-        return axios.post(`${this.baseUrl}/fts/suggestTags`, prefix)
+        return axios.post(`${this.baseUrl}fts/suggestTags`, prefix)
             .then(resp => (resp.data));
     }
 
@@ -620,7 +620,7 @@ class Sist2Api {
     }
 
     getEmbeddings(sid, modelId) {
-        return axios.post(`${this.baseUrl}/e/${sid}/${modelId.toString().padStart(3, '0')}`)
+        return axios.post(`${this.baseUrl}e/${sid}/${modelId.toString().padStart(3, '0')}`)
             .then(resp => (resp.data));
     }
 }

sist2-vue/src/Sist2ElasticsearchQuery.js

@@ -117,11 +117,11 @@ class Sist2ElasticsearchQuery {
             }
 
             if (dateMin && dateMax) {
-                filters.push({range: {mtime: {gte: dateMin, lte: dateMax}}})
+                filters.push({range: {mtime: {gte: dateMin, lte: dateMax, format: "epoch_second"}}})
             } else if (dateMin) {
-                filters.push({range: {mtime: {gte: dateMin}}})
+                filters.push({range: {mtime: {gte: dateMin, format: "epoch_second"}}})
             } else if (dateMax) {
-                filters.push({range: {mtime: {lte: dateMax}}})
+                filters.push({range: {mtime: {lte: dateMax, format: "epoch_second"}}})
             }
 
             const path = pathText.replace(/\/$/, "").toLowerCase(); //remove trailing slashes

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;
+    if (meta_contains_key(doc->meta_head, key)) {
-    while (meta != NULL) {
+        CTX_LOG_DEBUGF(doc->filepath, "Ignoring duplicate tag: '%02x=%s'",
-        if (meta->key == key) {
+                       key, tag->value);
-            CTX_LOG_DEBUGF(doc->filepath, "Ignoring duplicate tag: '%02x=%s' and '%02x=%s'",
+        return;
-                           key, meta->str_val, key, tag->value);
-            return;
-        }
-        meta = meta->next;
     }
 
     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