Validate that all the tesseract data files are in the same folder

refactor: split ocr_extract_text from ebook

.gitignore

@@ -23,4 +23,6 @@ git_hash.h
 Testing/
 test_i
 test_i_inc
-node_modules/
+node_modules/
+.cmake/
+i_inc/

README.md

@@ -67,23 +67,23 @@ See [Usage guide](docs/USAGE.md) for more details
 
 ## Format support
 
-File type | Library | Content | Thumbnail | Metadata
-:---|:---|:---|:---|:---
-pdf,xps,fb2,epub | MuPDF | text+ocr | yes | author, title |
-cbz,cbr | [libscan](https://github.com/simon987/libscan) | - | yes | - |
-`audio/*` | ffmpeg | - | yes | ID3 tags |
-`video/*` | ffmpeg | - | yes | title, comment, artist |
-`image/*` | ffmpeg | - | yes | [Common EXIF tags](https://github.com/simon987/sist2/blob/efdde2734eca9b14a54f84568863b7ffd59bdba3/src/parsing/media.c#L190), GPS tags |
-raw, rw2, dng, cr2, crw, dcr, k25, kdc, mrw, pef, xf3, arw, sr2, srf, erf  | LibRaw | - | yes | Common EXIF tags, GPS tags |
-ttf,ttc,cff,woff,fnt,otf | Freetype2 | - | yes, `bmp` | Name & style |
-`text/plain` | [libscan](https://github.com/simon987/libscan) | yes | no | - |
-html, xml | [libscan](https://github.com/simon987/libscan) | yes | no | - |
-tar, zip, rar, 7z, ar ...  | Libarchive | yes\* | - | no |
-docx, xlsx, pptx | [libscan](https://github.com/simon987/libscan) | yes | if embedded | creator, modified_by, title |
-doc (MS Word 97-2003) | antiword | yes | yes | author, title |
-mobi, azw, azw3 | libmobi | yes | no | author, title |
-wpd (WordPerfect) | libwpd | yes | no | *planned* |
-json, jsonl, ndjson | [libscan](https://github.com/simon987/libscan) | yes | - | - |
+| File type                                                                 | Library                                                                      | Content  | Thumbnail   | Metadata                                                                                                                               |
+|:--------------------------------------------------------------------------|:-----------------------------------------------------------------------------|:---------|:------------|:---------------------------------------------------------------------------------------------------------------------------------------|
+| pdf,xps,fb2,epub                                                          | MuPDF                                                                        | text+ocr | yes         | author, title                                                                                                                          |
+| cbz,cbr                                                                   | [libscan](https://github.com/simon987/sist2/tree/master/third-party/libscan) | -        | yes         | -                                                                                                                                      |
+| `audio/*`                                                                 | ffmpeg                                                                       | -        | yes         | ID3 tags                                                                                                                               |
+| `video/*`                                                                 | ffmpeg                                                                       | -        | yes         | title, comment, artist                                                                                                                 |
+| `image/*`                                                                 | ffmpeg                                                                       | -        | yes         | [Common EXIF tags](https://github.com/simon987/sist2/blob/efdde2734eca9b14a54f84568863b7ffd59bdba3/src/parsing/media.c#L190), GPS tags |
+| raw, rw2, dng, cr2, crw, dcr, k25, kdc, mrw, pef, xf3, arw, sr2, srf, erf | LibRaw                                                                       | -        | yes         | Common EXIF tags, GPS tags                                                                                                             |
+| ttf,ttc,cff,woff,fnt,otf                                                  | Freetype2                                                                    | -        | yes, `bmp`  | Name & style                                                                                                                           |
+| `text/plain`                                                              | [libscan](https://github.com/simon987/sist2/tree/master/third-party/libscan) | yes      | no          | -                                                                                                                                      |
+| html, xml                                                                 | [libscan](https://github.com/simon987/sist2/tree/master/third-party/libscan) | yes      | no          | -                                                                                                                                      |
+| tar, zip, rar, 7z, ar ...                                                 | Libarchive                                                                   | yes\*    | -           | no                                                                                                                                     |
+| docx, xlsx, pptx                                                          | [libscan](https://github.com/simon987/sist2/tree/master/third-party/libscan) | yes      | if embedded | creator, modified_by, title                                                                                                            |
+| doc (MS Word 97-2003)                                                     | antiword                                                                     | yes      | yes         | author, title                                                                                                                          |
+| mobi, azw, azw3                                                           | libmobi                                                                      | yes      | no          | author, title                                                                                                                          |
+| wpd (WordPerfect)                                                         | libwpd                                                                       | yes      | no          | *planned*                                                                                                                              |
+| json, jsonl, ndjson                                                       | [libscan](https://github.com/simon987/sist2/tree/master/third-party/libscan) | yes      | -           | -                                                                                                                                      |
 
 \* *See [Archive files](#archive-files)*
 
@@ -102,18 +102,24 @@ scan is also supported.
 
 ### OCR
 
-You can enable OCR support for pdf,xps,fb2,epub file types with the
-`--ocr <lang>` option. Download the language data files with your package manager (`apt install tesseract-ocr-eng`) or
+You can enable OCR support for ebook (pdf,xps,fb2,epub) or image file types with the
+`--ocr-lang <lang>` option in combination with `--ocr-images` and/or `--ocr-ebooks`.
+Download the language data files with your package manager (`apt install tesseract-ocr-eng`) or
 directly [from Github](https://github.com/tesseract-ocr/tesseract/wiki/Data-Files).
 
 The `simon987/sist2` image comes with common languages
 (hin, jpn, eng, fra, rus, spa) pre-installed.
 
-Examples
+You can use the `+` separator to specify multiple languages. The language
+name must be identical to the `*.traineddata` file installed on your system 
+(use `chi_sim` rather than `chi-sim`).
+
+Examples:
 
 ```bash
-sist2 scan --ocr jpn ~/Books/Manga/
-sist2 scan --ocr eng ~/Books/Textbooks/
+sist2 scan --ocr-ebooks --ocr-lang jpn ~/Books/Manga/
+sist2 scan --ocr-images --ocr-lang eng ~/Images/Screenshots/
+sist2 scan --ocr-ebooks --ocr-images --ocr-lang eng+chi_sim ~/Chinese-Bilingual/
 ```
 
 ## Build from source
@@ -126,7 +132,7 @@ You can compile **sist2** by yourself if you don't want to use the pre-compiled
 git clone --recursive https://github.com/simon987/sist2/
 cd sist2
 docker build . -f ./Dockerfile -t my-sist2-image
-docker run --rm my-sist2-image cat /root/sist2 > sist2-x64-linux
+docker run --rm --entrypoint cat my-sist2-image /root/sist2 > sist2-x64-linux
 ```
 
 ### On a linux computer

docs/USAGE.md

@@ -43,7 +43,7 @@ Scan options
     --depth=<int>                 Scan up to DEPTH subdirectories deep. Use 0 to only scan files in PATH. DEFAULT: -1
     --archive=<str>               Archive file mode (skip|list|shallow|recurse). skip: Don't parse, list: only get file names as text, shallow: Don't parse archives inside archives. DEFAULT: recurse
     --archive-passphrase=<str>    Passphrase for encrypted archive files
-    --ocr=<str>                   Tesseract language (use tesseract --list-langs to see which are installed on your machine)
+    # TODO: add new --ocr-* options here
     -e, --exclude=<str>           Files that match this regex will not be scanned
     --fast                        Only index file names & mime type
     --treemap-threshold=<str>     Relative size threshold for treemap (see USAGE.md). DEFAULT: 0.0005

scripts/before_build.sh

@@ -6,5 +6,4 @@ python3 scripts/mime.py > src/parsing/mime_generated.c
 python3 scripts/serve_static.py > src/web/static_generated.c
 python3 scripts/index_static.py > src/index/static_generated.c
 
-printf "static const char *const Sist2CommitHash = \"%s\";\n" $(git rev-parse HEAD) > src/git_hash.h
-printf "static const char *const LibScanCommitHash = \"%s\";\n" $(cd third-party/libscan/ && git rev-parse HEAD) >> src/git_hash.h
+printf "static const char *const Sist2CommitHash = \"%s\";\n" $(git rev-parse HEAD) > src/git_hash.h

sist2-vue/src/components/DebugInfo.vue

@@ -15,7 +15,7 @@
 
 <script>
 import IndexDebugInfo from "@/components/IndexDebugInfo";
-import DebugIcon from "@/components/DebugIcon";
+import DebugIcon from "@/components/icons/DebugIcon";
 
 export default {
   name: "DebugInfo.vue",
@@ -27,7 +27,6 @@ export default {
         {key: "platform", value: this.$store.state.sist2Info.platform},
         {key: "debugBinary", value: this.$store.state.sist2Info.debug},
         {key: "sist2CommitHash", value: this.$store.state.sist2Info.sist2Hash},
-        {key: "libscanCommitHash", value: this.$store.state.sist2Info.libscanHash},
         {key: "esIndex", value: this.$store.state.sist2Info.esIndex},
         {key: "tagline", value: this.$store.state.sist2Info.tagline},
         {key: "dev", value: this.$store.state.sist2Info.dev},

sist2-vue/src/components/DocCard.vue

@@ -34,9 +34,11 @@
           </svg>
         </div>
 
-        <img v-if="doc._props.isPlayableImage || doc._props.isPlayableVideo"
+        <img ref="tn"
+             v-if="doc._props.isPlayableImage || doc._props.isPlayableVideo"
              :src="(doc._props.isGif && hover) ? `f/${doc._id}` : `t/${doc._source.index}/${doc._id}`"
              alt=""
+             :style="{height: (doc._props.isGif && hover) ? `${tnHeight()}px` : undefined}"
              class="pointer fit card-img-top" @click="onThumbnailClick()">
         <img v-else :src="`t/${doc._source.index}/${doc._id}`" alt=""
              class="fit card-img-top">
@@ -122,6 +124,9 @@ export default {
     },
     onTnLeave() {
       this.hover = false;
+    },
+    tnHeight() {
+      return this.$refs.tn.height;
     }
   },
 }

sist2-vue/src/components/DocListItem.vue

@@ -1,5 +1,6 @@
 <template>
-  <b-list-group-item class="flex-column align-items-start mb-2" :class="{'sub-document': doc._props.isSubDocument}">
+  <b-list-group-item class="flex-column align-items-start mb-2" :class="{'sub-document': doc._props.isSubDocument}"
+                     @mouseenter="onTnEnter()" @mouseleave="onTnLeave()" >
 
     <!-- Info modal-->
     <DocInfoModal :show="showInfo" :doc="doc" @close="showInfo = false"></DocInfoModal>
@@ -56,7 +57,7 @@ import TagContainer from "@/components/TagContainer";
 import DocFileTitle from "@/components/DocFileTitle";
 import DocInfoModal from "@/components/DocInfoModal";
 import ContentDiv from "@/components/ContentDiv";
-import FileIcon from "@/components/FileIcon";
+import FileIcon from "@/components/icons/FileIcon";
 
 export default {
   name: "DocListItem",
@@ -85,7 +86,13 @@ export default {
         return this.doc.highlight["path.nGram"] + "/"
       }
       return this.doc._source.path + "/"
-    }
+    },
+    onTnEnter() {
+      this.hover = true;
+    },
+    onTnLeave() {
+      this.hover = false;
+    },
   }
 }
 </script>

sist2-vue/src/components/Lightbox.vue

@@ -1,6 +1,5 @@
 <template>
   <div>
-    <!-- TODO: Set slideshowTime as a configurable option-->
     <FsLightbox
         :key="lightboxKey"
         :toggler="showLightbox"
@@ -10,7 +9,7 @@
         :types="lightboxTypes"
         :source-index="lightboxSlide"
         :custom-toolbar-buttons="customButtons"
-        :slideshow-time="1000 * 10"
+        :slideshow-time="$store.getters.optLightboxSlideDuration * 1000"
         :zoom-increment="0.5"
         :load-only-current-source="$store.getters.optLightboxLoadOnlyCurrent"
         :on-close="onClose"

sist2-vue/src/components/NavBar.vue

@@ -20,7 +20,7 @@
 </template>
 
 <script>
-import Sist2Icon from "@/components/Sist2Icon";
+import Sist2Icon from "@/components/icons/Sist2Icon";
 
 export default {
   name: "NavBar",

sist2-vue/src/components/TagContainer.vue

@@ -51,7 +51,7 @@
         >{{ tag.text.split(".").pop() }}</span>
 
         <b-popover :target="hit._id+tag.rawText" triggers="focus blur" placement="top">
-          <b-button variant="danger" @click="onTagDeleteClick(tag, $event)">Delete</b-button>
+          <b-button variant="danger" @click="onTagDeleteClick(tag, $event)">{{$t("deleteTag")}}</b-button>
         </b-popover>
       </div>
 
@@ -63,7 +63,7 @@
     </template>
 
     <!-- Add button -->
-    <small v-if="showAddButton" class="badge add-tag-button" @click="tagAdd()">Add</small>
+    <small v-if="showAddButton" class="badge add-tag-button" @click="tagAdd()">{{$t("addTag")}}</small>
 
     <!-- Size tag-->
     <small v-else class="text-muted badge-size">{{

sist2-vue/src/components/icons/LanguageIcon.vue

@@ -0,0 +1,21 @@
+<template>
+  <svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24">
+    <path
+        fill="currentColor"
+        d="M12 0c-6.627 0-12 5.373-12 12s5.373 12 12 12 12-5.373 12-12-5.373-12-12-12zm1 16.057v-3.057h2.994c-.059 1.143-.212 2.24-.456 3.279-.823-.12-1.674-.188-2.538-.222zm1.957 2.162c-.499 1.33-1.159 2.497-1.957 3.456v-3.62c.666.028 1.319.081 1.957.164zm-1.957-7.219v-3.015c.868-.034 1.721-.103 2.548-.224.238 1.027.389 2.111.446 3.239h-2.994zm0-5.014v-3.661c.806.969 1.471 2.15 1.971 3.496-.642.084-1.3.137-1.971.165zm2.703-3.267c1.237.496 2.354 1.228 3.29 2.146-.642.234-1.311.442-2.019.607-.344-.992-.775-1.91-1.271-2.753zm-7.241 13.56c-.244-1.039-.398-2.136-.456-3.279h2.994v3.057c-.865.034-1.714.102-2.538.222zm2.538 1.776v3.62c-.798-.959-1.458-2.126-1.957-3.456.638-.083 1.291-.136 1.957-.164zm-2.994-7.055c.057-1.128.207-2.212.446-3.239.827.121 1.68.19 2.548.224v3.015h-2.994zm1.024-5.179c.5-1.346 1.165-2.527 1.97-3.496v3.661c-.671-.028-1.329-.081-1.97-.165zm-2.005-.35c-.708-.165-1.377-.373-2.018-.607.937-.918 2.053-1.65 3.29-2.146-.496.844-.927 1.762-1.272 2.753zm-.549 1.918c-.264 1.151-.434 2.36-.492 3.611h-3.933c.165-1.658.739-3.197 1.617-4.518.88.361 1.816.67 2.808.907zm.009 9.262c-.988.236-1.92.542-2.797.9-.89-1.328-1.471-2.879-1.637-4.551h3.934c.058 1.265.231 2.488.5 3.651zm.553 1.917c.342.976.768 1.881 1.257 2.712-1.223-.49-2.326-1.211-3.256-2.115.636-.229 1.299-.435 1.999-.597zm9.924 0c.7.163 1.362.367 1.999.597-.931.903-2.034 1.625-3.257 2.116.489-.832.915-1.737 1.258-2.713zm.553-1.917c.27-1.163.442-2.386.501-3.651h3.934c-.167 1.672-.748 3.223-1.638 4.551-.877-.358-1.81-.664-2.797-.9zm.501-5.651c-.058-1.251-.229-2.46-.492-3.611.992-.237 1.929-.546 2.809-.907.877 1.321 1.451 2.86 1.616 4.518h-3.933z"/>
+  </svg>
+</template>
+
+<script>
+export default {
+  name: "LanguageIcon"
+}
+</script>
+
+<style scoped>
+svg {
+  display: inline-block;
+  width: 20px;
+  height: 20px;
+}
+</style>

sist2-vue/src/i18n/messages.ts

@@ -5,6 +5,8 @@ export default {
             advanced: "Advanced search",
             fuzzy: "Fuzzy"
         },
+        addTag: "Add",
+        deleteTag: "Delete",
         download: "Download",
         and: "and",
         page: "page",
@@ -132,6 +134,7 @@ export default {
         saveTagModalTitle: "Add tag",
         saveTagPlaceholder: "Tag name",
         confirm: "Confirm",
+        indexPickerPlaceholder: "Select an index",
         sort: {
             relevance: "Relevance",
             dateAsc: "Date (Older first)",
@@ -161,6 +164,8 @@ export default {
             advanced: "Recherche avancée",
             fuzzy: "Approximatif"
         },
+        addTag: "Ajouter",
+        deleteTag: "Supprimer",
         download: "Télécharger",
         and: "et",
         page: "page",
@@ -320,6 +325,8 @@ export default {
             advanced: "高级搜索",
             fuzzy: "模糊搜索"
         },
+        addTag: "添加",
+        deleteTag: "删除",
         download: "下载",
         and: "与",
         page: "页",
@@ -447,6 +454,7 @@ export default {
         saveTagModalTitle: "增加标签",
         saveTagPlaceholder: "标签名",
         confirm: "确认",
+        indexPickerPlaceholder: "选择一个索引",
         sort: {
             relevance: "相关度",
             dateAsc: "日期（由旧到新）",

sist2-vue/src/store/index.ts

@@ -27,6 +27,7 @@ export default new Vuex.Store({
         size: 60,
 
         optLang: "en",
+        optLangIsDefault: true,
         optHideDuplicates: true,
         optTheme: "light",
         optDisplay: "grid",
@@ -82,7 +83,10 @@ export default new Vuex.Store({
         setSist2Info: (state, val) => state.sist2Info = val,
         setSeed: (state, val) => state.seed = val,
         setOptHideDuplicates: (state, val) => state.optHideDuplicates = val,
-        setOptLang: (state, val) => state.optLang = val,
+        setOptLang: (state, val) => {
+            state.optLang = val;
+            state.optLangIsDefault = false;
+        },
         setSortMode: (state, val) => state.sortMode = val,
         setIndices: (state, val) => {
             state.indices = val;
@@ -148,6 +152,7 @@ export default new Vuex.Store({
         setOptHideLegacy: (state, val) => state.optHideLegacy = val,
 
         setOptLightboxLoadOnlyCurrent: (state, val) => state.optLightboxLoadOnlyCurrent = val,
+        setOptLightboxSlideDuration: (state, val) => state.optLightboxSlideDuration = val,
 
         setUiMimeMap: (state, val) => state.uiMimeMap = val,
 
@@ -159,6 +164,13 @@ export default new Vuex.Store({
         },
     },
     actions: {
+        setSist2Info: (store, val) => {
+            store.commit("setSist2Info", val);
+
+            if (store.state.optLangIsDefault) {
+                store.commit("setOptLang", val.lang);
+            }
+        },
         loadFromArgs({commit}, route: Route) {
 
             if (route.query.q) {

sist2-vue/src/views/Configuration.vue

@@ -15,15 +15,8 @@
         <h4>{{ $t("displayOptions") }}</h4>
 
         <b-card>
-          <b-form-checkbox :checked="optLightboxLoadOnlyCurrent" @input="setOptLightboxLoadOnlyCurrent">
-            {{ $t("opt.lightboxLoadOnlyCurrent") }}
-          </b-form-checkbox>
 
-          <b-form-checkbox :checked="optHideLegacy" @input="setOptHideLegacy">
-            {{ $t("opt.hideLegacy") }}
-          </b-form-checkbox>
-
-          <label>{{ $t("opt.lang") }}</label>
+          <label><LanguageIcon/><span style="vertical-align: middle">&nbsp;{{ $t("opt.lang") }}</span></label>
           <b-form-select :options="langOptions" :value="optLang" @input="setOptLang"></b-form-select>
 
           <label>{{ $t("opt.theme") }}</label>
@@ -34,6 +27,16 @@
 
           <label>{{ $t("opt.columns") }}</label>
           <b-form-select :options="columnsOptions" :value="optColumns" @input="setOptColumns"></b-form-select>
+
+          <div style="height: 10px"></div>
+
+          <b-form-checkbox :checked="optLightboxLoadOnlyCurrent" @input="setOptLightboxLoadOnlyCurrent">
+            {{ $t("opt.lightboxLoadOnlyCurrent") }}
+          </b-form-checkbox>
+
+          <b-form-checkbox :checked="optHideLegacy" @input="setOptHideLegacy">
+            {{ $t("opt.hideLegacy") }}
+          </b-form-checkbox>
         </b-card>
 
         <br/>
@@ -117,15 +120,15 @@
 </template>
 
 <script>
-import Vue from "vue";
-import {mapGetters, mapMutations} from "vuex";
+import {mapActions, mapGetters, mapMutations} from "vuex";
 import DebugInfo from "@/components/DebugInfo.vue";
 import Preloader from "@/components/Preloader.vue";
 import sist2 from "@/Sist2Api";
-import GearIcon from "@/components/GearIcon.vue";
+import GearIcon from "@/components/icons/GearIcon.vue";
+import LanguageIcon from "@/components/icons/LanguageIcon";
 
 export default {
-  components: {GearIcon, DebugInfo, Preloader},
+  components: {LanguageIcon, GearIcon, DebugInfo, Preloader},
   data() {
     return {
       loading: true,
@@ -228,7 +231,7 @@ export default {
   },
   mounted() {
     sist2.getSist2Info().then(data => {
-      this.$store.commit("setSist2Info", data)
+      this.setSist2Info(data);
       this.loading = false;
     });
 
@@ -239,6 +242,9 @@ export default {
     });
   },
   methods: {
+    ...mapActions({
+      setSist2Info: "setSist2Info",
+    }),
     ...mapMutations([
       "setOptTheme",
       "setOptDisplay",
@@ -256,7 +262,6 @@ export default {
       "setOptTreemapSize",
       "setOptLightboxLoadOnlyCurrent",
       "setOptLightboxSlideDuration",
-      "setOptContainerWidth",
       "setOptResultSize",
       "setOptTagOrOperator",
       "setOptLang",

sist2-vue/src/views/SearchPage.vue

@@ -60,7 +60,7 @@
 
 <script lang="ts">
 import Preloader from "@/components/Preloader.vue";
-import {mapGetters, mapMutations} from "vuex";
+import {mapActions, mapGetters, mapMutations} from "vuex";
 import sist2 from "../Sist2Api";
 import Sist2Api, {EsHit, EsResult} from "../Sist2Api";
 import SearchBar from "@/components/SearchBar.vue";
@@ -151,8 +151,10 @@ export default Vue.extend({
     });
   },
   methods: {
-    ...mapMutations({
+    ...mapActions({
       setSist2Info: "setSist2Info",
+    }),
+    ...mapMutations({
       setIndices: "setIndices",
       setDateBoundsMin: "setDateBoundsMin",
       setDateBoundsMax: "setDateBoundsMax",

src/cli.c

@@ -146,7 +146,7 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
     if (args->name == NULL) {
         args->name = g_path_get_basename(args->output);
     } else {
-        char* tmp = malloc(strlen(args->name) + 1);
+        char *tmp = malloc(strlen(args->name) + 1);
         strcpy(tmp, args->name);
         args->name = tmp;
     }
@@ -168,17 +168,50 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
         return 1;
     }
 
-    if (args->tesseract_lang != NULL) {
-        TessBaseAPI *api = TessBaseAPICreate();
+    if (args->ocr_images && args->tesseract_lang == NULL) {
+        fprintf(stderr, "You must specify --ocr-lang <LANG> to use --ocr-images");
+        return 1;
+    }
 
-        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!");
+    if (args->ocr_ebooks && args->tesseract_lang == NULL) {
+        fprintf(stderr, "You must specify --ocr-lang <LANG> to use --ocr-ebooks");
+        return 1;
+    }
+
+    if (args->tesseract_lang != NULL) {
+
+        if (!args->ocr_ebooks && !args->ocr_images) {
+            fprintf(stderr, "You must specify at least one of --ocr-ebooks, --ocr-images");
+            return 1;
         }
 
-        ret = TessBaseAPIInit3(api, path, args->tesseract_lang);
+        TessBaseAPI *api = TessBaseAPICreate();
+
+        const char *trained_data_path = NULL;
+        char *lang = malloc(strlen(args->tesseract_lang) + 1);
+        strcpy(lang, args->tesseract_lang);
+
+        lang = strtok(lang, "+");
+
+        while (lang != NULL) {
+            char filename[128];
+            sprintf(filename, "%s.traineddata", lang);
+
+            const char *path = find_file_in_paths(TESS_DATAPATHS, filename);
+            if (path == NULL) {
+                LOG_FATALF("cli.c", "Could not find tesseract language file: %s!", filename);
+            }
+            if (trained_data_path != NULL && path != trained_data_path) {
+                LOG_FATAL("cli.c", "When specifying more than one tesseract language, all the traineddata "
+                                   "files must be in the same folder")
+            }
+            trained_data_path = path;
+
+            lang = strtok(NULL, "+");
+        }
+        free(lang);
+
+        ret = TessBaseAPIInit3(api, trained_data_path, args->tesseract_lang);
         if (ret != 0) {
             fprintf(stderr, "Could not initialize tesseract with lang '%s'\n", args->tesseract_lang);
             return 1;
@@ -186,7 +219,7 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
         TessBaseAPIEnd(api);
         TessBaseAPIDelete(api);
 
-        args->tesseract_path = path;
+        args->tesseract_path = trained_data_path;
     }
 
     if (args->exclude_regex != NULL) {
@@ -220,7 +253,7 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
     }
 
     if (args->list_path != NULL) {
-        if(strcmp(args->list_path, "-") == 0) {
+        if (strcmp(args->list_path, "-") == 0) {
             args->list_file = stdin;
             LOG_DEBUG("cli.c", "Using stdin as list file")
         } else {
@@ -377,15 +410,15 @@ int web_args_validate(web_args_t *args, int argc, const char **argv) {
         args->es_index = DEFAULT_ES_INDEX;
     }
 
-    if (args->lang == NULL) {
-        args->lang = DEFAULT_LANG;
-    }
-
     if (args->tagline == NULL) {
         args->tagline = DEFAULT_TAGLINE;
     }
 
-    if (strlen(args->lang) != 2) {
+    if (args->lang == NULL) {
+        args->lang = DEFAULT_LANG;
+    }
+
+    if (strlen(args->lang) != 2 && strlen(args->lang) != 5) {
         fprintf(stderr, "Invalid --lang value, see usage\n");
         return 1;
     }

src/cli.h

@@ -21,6 +21,8 @@ typedef struct scan_args {
     char *archive_passphrase;
     char *tesseract_lang;
     const char *tesseract_path;
+    int ocr_images;
+    int ocr_ebooks;
     char *exclude_regex;
     int fast;
     const char* treemap_threshold_str;

src/ctx.h

@@ -98,7 +98,7 @@ typedef struct {
     int tag_auth_enabled;
     char *tagline;
     struct index_t indices[256];
-    char lang[3];
+    char lang[10];
     int dev;
 } WebCtx_t;
 

src/io/serialize.c

@@ -38,6 +38,8 @@ char *get_meta_key_text(enum metakey meta_key) {
             return "parent";
         case MetaExifMake:
             return "exif_make";
+        case MetaExifDescription:
+            return "exif_description";
         case MetaExifSoftware:
             return "exif_software";
         case MetaExifExposureTime:
@@ -150,6 +152,7 @@ char *build_json_string(document_t *doc) {
             case MetaFontName:
             case MetaParent:
             case MetaExifMake:
+            case MetaExifDescription:
             case MetaExifSoftware:
             case MetaExifExposureTime:
             case MetaExifFNumber:

src/main.c

@@ -220,6 +220,11 @@ void initialize_scan_context(scan_args_t *args) {
     ScanCtx.media_ctx.store = _store;
     ScanCtx.media_ctx.max_media_buffer = (long) args->max_memory_buffer * 1024 * 1024;
     ScanCtx.media_ctx.read_subtitles = args->read_subtitles;
+
+    if (args->ocr_images) {
+        ScanCtx.media_ctx.tesseract_lang = args->tesseract_lang;
+        ScanCtx.media_ctx.tesseract_path = args->tesseract_path;
+    }
     init_media();
 
     // OOXML
@@ -501,7 +506,7 @@ void sist2_web(web_args_t *args) {
     WebCtx.tag_auth_enabled = args->tag_auth_enabled;
     WebCtx.tagline = args->tagline;
     WebCtx.dev = args->dev;
-    strcpy(WebCtx.lang, "en");
+    strcpy(WebCtx.lang, args->lang);
 
     for (int i = 0; i < args->index_count; i++) {
         char *abs_path = abspath(args->indices[i]);
@@ -576,8 +581,11 @@ int main(int argc, const char *argv[]) {
             OPT_STRING(0, "archive-passphrase", &scan_args->archive_passphrase,
                        "Passphrase for encrypted archive files"),
 
-            OPT_STRING(0, "ocr", &scan_args->tesseract_lang, "Tesseract language (use tesseract --list-langs to see "
-                                                             "which are installed on your machine)"),
+            OPT_STRING(0, "ocr-lang", &scan_args->tesseract_lang,
+                       "Tesseract language (use 'tesseract --list-langs' to see "
+                       "which are installed on your machine)"),
+            OPT_BOOLEAN(0, "ocr-images", &scan_args->ocr_images, "Enable OCR'ing of image files."),
+            OPT_BOOLEAN(0, "ocr-ebooks", &scan_args->ocr_ebooks, "Enable OCR'ing of ebook files."),
             OPT_STRING('e', "exclude", &scan_args->exclude_regex, "Files that match this regex will not be scanned"),
             OPT_BOOLEAN(0, "fast", &scan_args->fast, "Only index file names & mime type"),
             OPT_STRING(0, "treemap-threshold", &scan_args->treemap_threshold_str, "Relative size threshold for treemap "
@@ -614,6 +622,7 @@ int main(int argc, const char *argv[]) {
             OPT_STRING(0, "tag-auth", &web_args->tag_credentials, "Basic auth in user:password format for tagging"),
             OPT_STRING(0, "tagline", &web_args->tagline, "Tagline in navbar"),
             OPT_BOOLEAN(0, "dev", &web_args->dev, "Serve html & js files from disk (for development)"),
+            OPT_STRING(0, "lang", &web_args->lang, "Default UI language. Can be changed by the user"),
 
             OPT_GROUP("Exec-script options"),
             OPT_STRING(0, "es-url", &common_es_url, "Elasticsearch url. DEFAULT=http://localhost:9200"),

src/web/serve.c

@@ -280,7 +280,6 @@ void index_info(struct mg_connection *nc) {
     cJSON_AddBoolToObject(json, "esVersionLegacy", USE_LEGACY_ES_SETTINGS(WebCtx.es_version));
     cJSON_AddStringToObject(json, "platform", QUOTE(SIST_PLATFORM));
     cJSON_AddStringToObject(json, "sist2Hash", Sist2CommitHash);
-    cJSON_AddStringToObject(json, "libscanHash", LibScanCommitHash);
     cJSON_AddStringToObject(json, "lang", WebCtx.lang);
     cJSON_AddBoolToObject(json, "dev", WebCtx.dev);
 #ifdef SIST_DEBUG

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

@@ -5,9 +5,7 @@
 
 #include "../media/media.h"
 #include "../arc/arc.h"
-
-#define MIN_OCR_SIZE 350
-#define MIN_OCR_LEN 10
+#include "../ocr/ocr.h"
 
 /* fill_image callback doesn't let us pass opaque pointers unless I create my own device */
 __thread text_buffer_t thread_buffer;
@@ -225,7 +223,9 @@ static int read_stext_block(fz_stext_block *block, text_buffer_t *tex) {
     return 0;
 }
 
-#define IS_VALID_BPP(d) ((d)==1 || (d)==2 || (d)==4 || (d)==8 || (d)==16 || (d)==24 || (d)==32)
+static void fill_image_ocr_cb(const char* text, size_t len) {
+  text_buffer_append_string(&thread_buffer, text, len - 1);
+}
 
 void fill_image(fz_context *fzctx, UNUSED(fz_device *dev),
                 fz_image *img, UNUSED(fz_matrix ctm), UNUSED(float alpha),
@@ -233,26 +233,9 @@ void fill_image(fz_context *fzctx, UNUSED(fz_device *dev),
 
     int l2factor = 0;
 
-    if (img->w > MIN_OCR_SIZE && img->h > MIN_OCR_SIZE && IS_VALID_BPP(img->n)) {
-
+    if (img->w >= MIN_OCR_WIDTH && img->h >= MIN_OCR_HEIGHT && OCR_IS_VALID_BPP(img->n)) {
         fz_pixmap *pix = img->get_pixmap(fzctx, img, NULL, img->w, img->h, &l2factor);
-
-        if (pix->h > MIN_OCR_SIZE && img->h > MIN_OCR_SIZE && img->xres != 0) {
-            TessBaseAPI *api = TessBaseAPICreate();
-            TessBaseAPIInit3(api, thread_ctx.tesseract_path, thread_ctx.tesseract_lang);
-
-            TessBaseAPISetImage(api, pix->samples, pix->w, pix->h, pix->n, pix->stride);
-            TessBaseAPISetSourceResolution(api, pix->xres);
-
-            char *text = TessBaseAPIGetUTF8Text(api);
-            size_t len = strlen(text);
-            if (len >= MIN_OCR_LEN) {
-                text_buffer_append_string(&thread_buffer, text, len - 1);
-            }
-
-            TessBaseAPIEnd(api);
-            TessBaseAPIDelete(api);
-        }
+        ocr_extract_text(thread_ctx.tesseract_path, thread_ctx.tesseract_lang, pix->samples, pix->w, pix->h, pix->n, pix->stride, pix->xres, fill_image_ocr_cb);
         fz_drop_pixmap(fzctx, pix);
     }
 }

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

@@ -1,12 +1,18 @@
 #include "media.h"
+#include "../ocr/ocr.h"
 #include <ctype.h>
 
 #define MIN_SIZE 32
 #define AVIO_BUF_SIZE 8192
-#define IS_VIDEO(fmt) (fmt->iformat->name && strcmp(fmt->iformat->name, "image2") != 0)
+#define IS_VIDEO(fmt) ((fmt)->iformat->name && strcmp((fmt)->iformat->name, "image2") != 0)
+
+#define STREAM_IS_IMAGE (stream->nb_frames <= 1)
 
 #define STORE_AS_IS ((void*)-1)
 
+// Pointer to document being processed
+__thread document_t *thread_doc;
+
 const char *get_filepath_with_ext(document_t *doc, const char *filepath, const char *mime_str) {
 
     int has_extension = doc->ext > doc->base;
@@ -311,7 +317,7 @@ append_video_meta(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx, AVFrame *f
             if (strcmp(key, "artist") == 0) {
                 append_tag_meta_if_not_exists(ctx, doc, tag, MetaArtist);
             } else if (strcmp(key, "imagedescription") == 0) {
-                APPEND_TAG_META(MetaContent)
+                append_tag_meta_if_not_exists(ctx, doc, tag, MetaContent);
             } else if (strcmp(key, "make") == 0) {
                 APPEND_TAG_META(MetaExifMake)
             } else if (strcmp(key, "model") == 0) {
@@ -343,6 +349,55 @@ append_video_meta(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx, AVFrame *f
     }
 }
 
+static void ocr_image_cb(const char *text, size_t len) {
+    APPEND_STR_META(thread_doc, MetaContent, text);
+}
+
+#define OCR_PIXEL_FORMAT AV_PIX_FMT_RGB32
+#define OCR_BYTES_PER_PIXEL 4
+#define OCR_PIXELS_PER_INCH 70
+
+void ocr_image(scan_media_ctx_t *ctx, document_t *doc, const AVCodecContext *decoder, AVFrame *frame) {
+
+    // Convert to RGB32
+    AVFrame *rgb_frame = av_frame_alloc();
+
+    struct SwsContext *sws_ctx = sws_getContext(
+            frame->width, frame->height, decoder->pix_fmt,
+            frame->width, frame->height, OCR_PIXEL_FORMAT,
+            SWS_LANCZOS, 0, 0, 0
+    );
+
+    int dst_buf_len = av_image_get_buffer_size(OCR_PIXEL_FORMAT, frame->width, frame->height, 1);
+    uint8_t *dst_buf = (uint8_t *) av_malloc(dst_buf_len * 2);
+
+    av_image_fill_arrays(rgb_frame->data, rgb_frame->linesize, dst_buf, OCR_PIXEL_FORMAT, frame->width, frame->height,
+                         1);
+
+    sws_scale(sws_ctx,
+              (const uint8_t *const *) frame->data, frame->linesize,
+              0, frame->height,
+              rgb_frame->data, rgb_frame->linesize
+    );
+
+    thread_doc = doc;
+    ocr_extract_text(
+            ctx->tesseract_path,
+            ctx->tesseract_lang,
+            rgb_frame->data[0],
+            frame->width,
+            frame->height,
+            OCR_BYTES_PER_PIXEL,
+            rgb_frame->linesize[0],
+            OCR_PIXELS_PER_INCH,
+            ocr_image_cb
+    );
+
+    sws_freeContext(sws_ctx);
+    av_free(*rgb_frame->data);
+    av_frame_free(&rgb_frame);
+}
+
 void parse_media_format_ctx(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx, document_t *doc) {
 
     int video_stream = -1;
@@ -419,11 +474,11 @@ void parse_media_format_ctx(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx,
         avcodec_open2(decoder, video_codec, NULL);
 
         //Seek
-        if (stream->nb_frames > 1 && stream->codecpar->codec_id != AV_CODEC_ID_GIF) {
+        if (!STREAM_IS_IMAGE && stream->codecpar->codec_id != AV_CODEC_ID_GIF) {
             int seek_ret;
             for (int i = 20; i >= 0; i--) {
                 seek_ret = av_seek_frame(pFormatCtx, video_stream,
-                                         stream->duration * 0.10, 0);
+                                         (long) ((double) stream->duration * 0.10), 0);
                 if (seek_ret == 0) {
                     break;
                 }
@@ -438,6 +493,11 @@ void parse_media_format_ctx(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx,
             return;
         }
 
+        if (ctx->tesseract_lang != NULL && STREAM_IS_IMAGE) {
+            ocr_image(ctx, doc, decoder, frame_and_packet->frame);
+        }
+
+        // NOTE: OCR'd content takes precedence over exif image description
         append_video_meta(ctx, pFormatCtx, frame_and_packet->frame, doc, IS_VIDEO(pFormatCtx));
 
         // Scale frame
@@ -534,7 +594,7 @@ long memfile_seek(void *ptr, long offset, int whence) {
     memfile_t *mem = ptr;
 
     if (whence == 0x10000) {
-        return mem->size;
+        return (long) mem->size;
     }
 
     int ret = fseek(mem->file, offset, whence);

third-party/libscan/libscan/media/media.h

@@ -19,6 +19,9 @@ typedef struct {
     float tn_qscale;
     long max_media_buffer;
     int read_subtitles;
+
+    const char *tesseract_lang;
+    const char *tesseract_path;
 } scan_media_ctx_t;
 
 __always_inline

third-party/libscan/libscan/ocr/ocr.h

@@ -0,0 +1,47 @@
+#ifndef OCR_H
+#define OCR_H
+
+#include "../scan.h"
+#include <tesseract/capi.h>
+
+#define MIN_OCR_WIDTH 350
+#define MIN_OCR_HEIGHT 100
+#define MIN_OCR_LEN 10
+
+#define OCR_IS_VALID_BPP(d)                                                    \
+  ((d) == 1 || (d) == 2 || (d) == 4 || (d) == 8 || (d) == 16 || (d) == 24 ||   \
+   (d) == 32)
+
+typedef void (*ocr_extract_callback_t)(const char *, size_t);
+
+__always_inline static void
+ocr_extract_text(const char *tesseract_path, const char *tesseract_lang,
+                 const unsigned char *img_buf, const int img_w, const int img_h,
+                 const int img_bpp, const int img_stride, const int img_xres,
+                 const ocr_extract_callback_t cb) {
+
+    if (img_w < MIN_OCR_WIDTH || img_h < MIN_OCR_HEIGHT || img_xres <= 0 ||
+        !OCR_IS_VALID_BPP(img_bpp)) {
+        return;
+    }
+
+    TessBaseAPI *api = TessBaseAPICreate();
+    TessBaseAPIInit3(api, tesseract_path, tesseract_lang);
+
+    TessBaseAPISetImage(api, img_buf, img_w, img_h, img_bpp, img_stride);
+    TessBaseAPISetSourceResolution(api, img_xres);
+
+    char *text = TessBaseAPIGetUTF8Text(api);
+    if (text != NULL) {
+        size_t len = strlen(text);
+        if (len >= MIN_OCR_LEN) {
+            cb(text, len);
+        }
+        TessDeleteText(text);
+    }
+
+    TessBaseAPIEnd(api);
+    TessBaseAPIDelete(api);
+}
+
+#endif

third-party/libscan/libscan/scan.h

@@ -61,6 +61,7 @@ enum metakey {
     MetaFontName,
     MetaParent,
     MetaExifMake,
+    MetaExifDescription,
     MetaExifSoftware,
     MetaExifExposureTime,
     MetaExifFNumber,