Merge pull request #200 from simon987/dev

v2.11.3

README.md

@@ -50,7 +50,8 @@ sist2 (Simple incremental search tool)
         ```
 1. Download sist2 executable
     1. Download the [latest sist2 release](https://github.com/simon987/sist2/releases) *
-    1. *(or)* Download a [development snapshot](https://files.simon987.net/.gate/sist2/simon987_sist2/) *(Not recommended!)*
+    1. *(or)* Download a [development snapshot](https://files.simon987.net/.gate/sist2/simon987_sist2/) *(Not
+       recommended!)*
     1. *(or)* `docker pull simon987/sist2:2.11.2-x64-linux`
 
 1. See [Usage guide](docs/USAGE.md)
@@ -70,19 +71,20 @@ See [Usage guide](docs/USAGE.md) for more details
 File type | Library | Content | Thumbnail | Metadata
 :---|:---|:---|:---|:---
 pdf,xps,fb2,epub | MuPDF | text+ocr | yes | author, title |
-cbz,cbr | *(none)* | - | yes | - |
+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` | *(none)* | yes | no | - |
-html, xml | *(none)* | yes | no | - |
+`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 | *(none)* | yes | if embedded | creator, modified_by, title |
+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 | - | - |
 
 \* *See [Archive files](#archive-files)*
 
@@ -135,7 +137,7 @@ docker run --rm my-sist2-image cat /root/sist2 > sist2-x64-linux
    ```bash
    apt install gcc g++ python3 yasm ragel automake autotools-dev wget libtool libssl-dev curl zip unzip tar xorg-dev libglu1-mesa-dev libxcursor-dev libxml2-dev libxinerama-dev gettext nasm git
    ```
-   
+
 1. Apply vcpkg patches, as per [sist2-build](https://github.com/simon987/sist2-build) Dockerfile
 
 1. Install vcpkg dependencies

docs/USAGE.md

@@ -49,6 +49,7 @@ Scan options
     --mem-buffer=<int>            Maximum memory buffer size per thread in MB for files inside archives (see USAGE.md). DEFAULT: 2000
     --read-subtitles              Read subtitles from media files.
     --fast-epub                   Faster but less accurate EPUB parsing (no thumbnails, metadata)
+    --checksums                   Calculate file checksums when scanning.
 
 Index options
     -t, --threads=<int>           Number of threads. DEFAULT=1
@@ -129,6 +130,9 @@ Exec-script options
     To check if a media file can be parsed without *seek*, execute `cat file.mp4 | ffprobe -`
 * `--read-subtitles` When enabled, will attempt to read the subtitles stream from media files.
 * `--fast-epub` Much faster but less accurate EPUB parsing. When enabled, sist2 will use a simple HTML parser to read epub files instead of the MuPDF library. No thumbnails are generated and author/title metadata are not parsed.
+* `--checksums` Calculate file checksums (sha1) when scanning files. This option does not cause any additional read 
+  operations. Checksums are not calculated for all file types, unless the file is inside an archive. When enabled, duplicate
+  files are hidden in the web UI (this behaviour can be toggled in the Configuration page).
 
 ### Scan examples
 

schema/mappings.json

@@ -4,6 +4,10 @@
       "type": "keyword",
       "doc_values": true
     },
+    "checksum": {
+      "type": "keyword",
+      "index": false
+    },
     "_depth": {
       "type": "integer"
     },

schema/settings.json

@@ -2,7 +2,8 @@
   "index": {
     "refresh_interval": "30s",
     "codec": "best_compression",
-    "number_of_replicas": 0
+    "number_of_replicas": 0,
+    "highlight.max_analyzed_offset": 10000000
   },
   "analysis": {
     "tokenizer": {

scripts/mime.csv

@@ -22,6 +22,7 @@ application/java-archive, jar
 application/java, class
 application/javascript,
 application/json, json
+application/ndjson, jsonl|ndjson
 application/marc, mrc
 application/mbedlet, mbd
 application/mime, aps

sist2-vue/package-lock.json

@@ -23,7 +23,6 @@
         "vue-color": "^2.8.1",
         "vue-i18n": "^8.24.4",
         "vue-masonry-wall": "^0.3.2",
-        "vue-multiselect": "^2.1.6",
         "vue-router": "^3.2.0",
         "vue-simple-suggest": "^1.11.1",
         "vuex": "^3.4.0"
@@ -13604,15 +13603,6 @@
         "node": ">=10"
       }
     },
-    "node_modules/vue-multiselect": {
-      "version": "2.1.6",
-      "resolved": "https://registry.npmjs.org/vue-multiselect/-/vue-multiselect-2.1.6.tgz",
-      "integrity": "sha512-s7jmZPlm9FeueJg1RwJtnE9KNPtME/7C8uRWSfp9/yEN4M8XcS/d+bddoyVwVnvFyRh9msFo0HWeW0vTL8Qv+w==",
-      "engines": {
-        "node": ">= 4.0.0",
-        "npm": ">= 3.0.0"
-      }
-    },
     "node_modules/vue-observe-visibility": {
       "version": "0.4.6",
       "resolved": "https://registry.npmjs.org/vue-observe-visibility/-/vue-observe-visibility-0.4.6.tgz",
@@ -26376,11 +26366,6 @@
         "vue-observe-visibility": "^0.4.6"
       }
     },
-    "vue-multiselect": {
-      "version": "2.1.6",
-      "resolved": "https://registry.npmjs.org/vue-multiselect/-/vue-multiselect-2.1.6.tgz",
-      "integrity": "sha512-s7jmZPlm9FeueJg1RwJtnE9KNPtME/7C8uRWSfp9/yEN4M8XcS/d+bddoyVwVnvFyRh9msFo0HWeW0vTL8Qv+w=="
-    },
     "vue-observe-visibility": {
       "version": "0.4.6",
       "resolved": "https://registry.npmjs.org/vue-observe-visibility/-/vue-observe-visibility-0.4.6.tgz",

sist2-vue/package.json

@@ -22,7 +22,6 @@
     "vue-color": "^2.8.1",
     "vue-i18n": "^8.24.4",
     "vue-masonry-wall": "^0.3.2",
-    "vue-multiselect": "^2.1.6",
     "vue-router": "^3.2.0",
     "vue-simple-suggest": "^1.11.1",
     "vuex": "^3.4.0"

sist2-vue/src/Sist2Api.ts

@@ -50,6 +50,7 @@ export interface EsHit {
         height: number
         duration: number
         tag: string[]
+        checksum: string
     }
     _props: {
         isSubDocument: boolean

sist2-vue/src/Sist2Query.ts

@@ -187,7 +187,8 @@ class Sist2Query {
                     "name.nGram": {},
                     "content.nGram": {},
                     font_name: {},
-                }
+                },
+                max_analyzed_offset: 9_999_999
             };
             if (getters.optSearchInPath) {
                 q.highlight.fields["path.text"] = {};

sist2-vue/src/components/DocInfoModal.vue

@@ -4,7 +4,8 @@
     <template #modal-title>
       <h5 class="modal-title" :title="doc._source.name + ext(doc)">{{ doc._source.name + ext(doc) }}</h5>
     </template>
-    <img :src="`t/${doc._source.index}/${doc._id}`" alt="" class="fit card-img-top">
+
+    <img v-if="doc._props.hasThumbnail" :src="`t/${doc._source.index}/${doc._id}`" alt="" class="fit card-img-top">
 
     <InfoTable :doc="doc"></InfoTable>
 

sist2-vue/src/components/DocListItem.vue

@@ -1,5 +1,5 @@
 <template>
-  <b-list-group-item class="flex-column align-items-start mb-2">
+  <b-list-group-item class="flex-column align-items-start mb-2" :class="{'sub-document': doc._props.isSubDocument}">
 
     <!-- Info modal-->
     <DocInfoModal :show="showInfo" :doc="doc" @close="showInfo = false"></DocInfoModal>
@@ -40,9 +40,11 @@
         </div>
 
         <div v-if="doc._source.pages || doc._source.author" class="path-row text-muted">
-          <span v-if="doc._source.pages">{{ doc._source.pages }} {{ doc._source.pages > 1 ? $t("pages") : $t("page") }}</span>
+          <span v-if="doc._source.pages">{{ doc._source.pages }} {{
+              doc._source.pages > 1 ? $t("pages") : $t("page")
+            }}</span>
           <span v-if="doc._source.author && doc._source.pages" class="mx-1">-</span>
-          <span v-if="doc._source.author">{{doc._source.author}}</span>
+          <span v-if="doc._source.author">{{ doc._source.author }}</span>
         </div>
       </div>
     </div>
@@ -89,6 +91,14 @@ export default {
 </script>
 
 <style scoped>
+.sub-document {
+  background: #AB47BC1F !important;
+}
+
+.theme-black .sub-document {
+  background: #37474F !important;
+}
+
 .list-group {
   margin-top: 1em;
 }

sist2-vue/src/components/IndexPicker.vue

@@ -1,93 +1,95 @@
 <template>
-  <VueMultiselect
-      multiple
-      label="name"
-      :value="selectedIndices"
-      :options="indices"
-      :close-on-select="indices.length <= 1"
-      :placeholder="$t('indexPickerPlaceholder')"
-      @select="addItem"
-      @remove="removeItem">
-
-    <template slot="option" slot-scope="idx">
-      <b-row>
-        <b-col>
-          <span class="mr-1">{{ idx.option.name }}</span>
-          <SmallBadge pill :text="idx.option.version"></SmallBadge>
-        </b-col>
-      </b-row>
-      <b-row class="mt-1">
-        <b-col>
-          <span>{{ formatIdxDate(idx.option.timestamp) }}</span>
-        </b-col>
-      </b-row>
-    </template>
-
-  </VueMultiselect>
+  <div v-if="isMobile">
+    <b-form-select
+        :value="selectedIndicesIds"
+        @change="onSelect($event)"
+        :options="indices" multiple :select-size="6" text-field="name"
+        value-field="id"></b-form-select>
+  </div>
+  <div v-else>
+    <b-list-group id="index-picker-desktop">
+      <b-list-group-item
+          v-for="idx in indices"
+          @click="toggleIndex(idx)"
+          class="d-flex justify-content-between align-items-center list-group-item-action pointer">
+        <div class="d-flex">
+          <b-checkbox @change="toggleIndex(idx)" :checked="isSelected(idx)"></b-checkbox>
+          {{ idx.name }}
+          <span class="text-muted timestamp-text ml-2">{{ formatIdxDate(idx.timestamp) }}</span>
+        </div>
+        <b-badge class="version-badge">v{{ idx.version }}</b-badge>
+      </b-list-group-item>
+    </b-list-group>
+  </div>
 </template>
 
 <script lang="ts">
-import VueMultiselect from "vue-multiselect"
 import SmallBadge from "./SmallBadge.vue"
 import {mapActions, mapGetters} from "vuex";
-import {Index} from "@/Sist2Api";
 import Vue from "vue";
 import {format} from "date-fns";
 
 export default Vue.extend({
   components: {
-    VueMultiselect,
     SmallBadge
   },
   data() {
     return {
-      loading: true
+      loading: true,
     }
   },
   computed: {
     ...mapGetters([
       "indices", "selectedIndices"
     ]),
+    selectedIndicesIds() {
+      return this.selectedIndices.map(idx => idx.id)
+    },
+    isMobile() {
+      return window.innerWidth <= 650;
+    }
   },
   methods: {
     ...mapActions({
       setSelectedIndices: "setSelectedIndices"
     }),
-    removeItem(val: Index): void {
-      this.setSelectedIndices(this.selectedIndices.filter((item: Index) => item !== val))
-    },
-    addItem(val: Index): void {
-      this.setSelectedIndices([...this.selectedIndices, val])
+    onSelect(value) {
+      this.setSelectedIndices(this.indices.filter(idx => value.includes(idx.id)));
     },
     formatIdxDate(timestamp: number): string {
       return format(new Date(timestamp * 1000), "yyyy-MM-dd");
+    },
+    toggleIndex(index) {
+      if (this.isSelected(index)) {
+        this.setSelectedIndices(this.selectedIndices.filter(idx => idx.id != index.id));
+      } else {
+        this.setSelectedIndices([index, ...this.selectedIndices]);
+      }
+    },
+    isSelected(index) {
+      return this.selectedIndices.find(idx => idx.id == index.id) != null;
     }
   },
 })
 </script>
 
-<style src="vue-multiselect/dist/vue-multiselect.min.css"></style>
-
-<style>
-.multiselect__option {
-  padding: 5px 10px;
+<style scoped>
+.timestamp-text {
+  line-height: 24px;
+  font-size: 80%;
 }
 
-.multiselect__content-wrapper {
-  overflow: hidden;
+.version-badge {
+  color: #222 !important;
+  background: none;
 }
 
-.theme-black .multiselect__tags {
-  background: #37474F;
-  border: 1px solid #616161 !important
+.list-group-item {
+  padding: 0.2em 0.4em;
 }
 
-.theme-black .multiselect__input {
-  color: #dbdbdb;
-  background: #37474F;
-}
-
-.theme-black .multiselect__content-wrapper {
-  border: none
+#index-picker-desktop {
+  overflow-y: auto;
+  max-height: 132px;
 }
 </style>

sist2-vue/src/components/InfoTable.vue

@@ -3,7 +3,7 @@
 
     <template #cell(value)="data">
       <span v-if="'html' in data.item" v-html="data.item.html"></span>
-      <span v-else>{{data.value}}</span>
+      <span v-else>{{ data.value }}</span>
     </template>
   </b-table>
 </template>
@@ -57,7 +57,8 @@ export default {
         "bitrate", "artist", "album", "album_artist", "genre", "font_name", "author",
         "modified_by", "pages", "tag",
         "exif_make", "exif_software", "exif_exposure_time", "exif_fnumber", "exif_focal_length",
-          "exif_user_comment", "exif_iso_speed_ratings", "exif_model", "exif_datetime",
+        "exif_user_comment", "exif_iso_speed_ratings", "exif_model", "exif_datetime",
+        "checksum"
       ];
 
       fields.forEach(field => {
@@ -76,9 +77,9 @@ export default {
         items.push({
           key: "Exif GPS",
           html: makeGpsLink(
-                  dmsToDecimal(src["exif_gps_latitude_dms"], src["exif_gps_latitude_ref"]),
-                  dmsToDecimal(src["exif_gps_longitude_dms"], src["exif_gps_longitude_ref"]),
-              ),
+              dmsToDecimal(src["exif_gps_latitude_dms"], src["exif_gps_latitude_ref"]),
+              dmsToDecimal(src["exif_gps_longitude_dms"], src["exif_gps_longitude_ref"]),
+          ),
         });
       }
 

sist2-vue/src/i18n/messages.ts

@@ -62,7 +62,8 @@ export default {
             lightboxLoadOnlyCurrent: "Do not preload full-size images for adjacent slides in image viewer.",
             slideDuration: "Slide duration",
             resultSize: "Number of results per page",
-            tagOrOperator: "Use OR operator when specifying multiple tags."
+            tagOrOperator: "Use OR operator when specifying multiple tags.",
+            hideDuplicates: "Hide duplicate results based on checksum"
         },
         queryMode: {
             simple: "Simple",
@@ -209,7 +210,8 @@ export default {
             lightboxLoadOnlyCurrent: "Désactiver le chargement des diapositives adjacentes pour le visualiseur d'images",
             slideDuration: "Durée des diapositives",
             resultSize: "Nombre de résultats par page",
-            tagOrOperator: "Utiliser l'opérateur OU lors de la spécification de plusieurs tags"
+            tagOrOperator: "Utiliser l'opérateur OU lors de la spécification de plusieurs tags",
+            hideDuplicates: "Masquer les résultats en double"
         },
         queryMode: {
             simple: "Simple",

sist2-vue/src/store/index.ts

@@ -27,6 +27,7 @@ export default new Vuex.Store({
         size: 60,
 
         optLang: "en",
+        optHideDuplicates: true,
         optTheme: "light",
         optDisplay: "grid",
 
@@ -79,6 +80,7 @@ export default new Vuex.Store({
         setSizeMax: (state, val) => state.sizeMax = val,
         setSist2Info: (state, val) => state.sist2Info = val,
         setSeed: (state, val) => state.seed = val,
+        setOptHideDuplicates: (state, val) => state.optHideDuplicates = val,
         setOptLang: (state, val) => state.optLang = val,
         setSortMode: (state, val) => state.sortMode = val,
         setIndices: (state, val) => {
@@ -317,6 +319,7 @@ export default new Vuex.Store({
         uiLightboxKey: state => state.uiLightboxKey,
         uiLightboxSlide: state => state.uiLightboxSlide,
 
+        optHideDuplicates: state => state.optHideDuplicates,
         optLang: state => state.optLang,
         optTheme: state => state.optTheme,
         optDisplay: state => state.optDisplay,

sist2-vue/src/views/Configuration.vue

@@ -35,6 +35,11 @@
         <br/>
         <h4>{{ $t("searchOptions") }}</h4>
         <b-card>
+          <b-form-checkbox :checked="optHideDuplicates" @input="setOptHideDuplicates">{{
+              $t("opt.hideDuplicates")
+            }}
+          </b-form-checkbox>
+
           <b-form-checkbox :checked="optHighlight" @input="setOptHighlight">{{ $t("opt.highlight") }}</b-form-checkbox>
           <b-form-checkbox :checked="optTagOrOperator" @input="setOptTagOrOperator">{{
               $t("opt.tagOrOperator")
@@ -206,10 +211,10 @@ export default {
       "optTreemapSize",
       "optLightboxLoadOnlyCurrent",
       "optLightboxSlideDuration",
-      "optContainerWidth",
       "optResultSize",
       "optTagOrOperator",
-      "optLang"
+      "optLang",
+      "optHideDuplicates",
     ]),
     clientWidth() {
       return window.innerWidth;
@@ -248,7 +253,8 @@ export default {
       "setOptContainerWidth",
       "setOptResultSize",
       "setOptTagOrOperator",
-      "setOptLang"
+      "setOptLang",
+      "setOptHideDuplicates"
     ]),
     onResetClick() {
       localStorage.removeItem("sist2_configuration");

sist2-vue/src/views/SearchPage.vue

@@ -91,6 +91,7 @@ export default Vue.extend({
     search: undefined as any,
     docs: [] as EsHit[],
     docIds: new Set(),
+    docChecksums: new Set(),
     searchBusy: false,
     Sist2Query: Sist2Query,
     showHelp: false
@@ -193,6 +194,7 @@ export default Vue.extend({
     async clearResults() {
       this.docs = [];
       this.docIds.clear();
+      this.docChecksums.clear();
       await this.$store.dispatch("clearResults");
       this.$store.commit("setUiReachedScrollEnd", false);
     },
@@ -202,7 +204,19 @@ export default Vue.extend({
       }
 
       resp.hits.hits = resp.hits.hits.filter(hit => !this.docIds.has(hit._id));
-      resp.hits.hits.forEach(hit => this.docIds.add(hit._id));
+
+      if (this.$store.state.optHideDuplicates) {
+        resp.hits.hits = resp.hits.hits.filter(hit => {
+
+          if (!("checksum" in hit._source)) {
+              return true;
+          }
+
+          const isDupe = !this.docChecksums.has(hit._source.checksum);
+          this.docChecksums.add(hit._source.checksum);
+          return isDupe;
+        });
+      }
 
       for (const hit of resp.hits.hits) {
         if (hit._props.isPlayableImage || hit._props.isPlayableVideo) {

src/cli.h

@@ -28,6 +28,7 @@ typedef struct scan_args {
     int max_memory_buffer;
     int read_subtitles;
     int fast_epub;
+    int calculate_checksums;
 } scan_args_t;
 
 scan_args_t *scan_args_create();

src/ctx.h

@@ -15,6 +15,7 @@
 #include "libscan/raw/raw.h"
 #include "libscan/msdoc/msdoc.h"
 #include "libscan/wpd/wpd.h"
+#include "libscan/json/json.h"
 #include "src/io/store.h"
 
 #include <glib.h>
@@ -32,6 +33,7 @@ typedef struct {
 
     int threads;
     int depth;
+    int calculate_checksums;
 
     size_t stat_tn_size;
     size_t stat_index_size;
@@ -62,6 +64,7 @@ typedef struct {
     scan_raw_ctx_t raw_ctx;
     scan_msdoc_ctx_t msdoc_ctx;
     scan_wpd_ctx_t wpd_ctx;
+    scan_json_ctx_t json_ctx;
 } ScanCtx_t;
 
 typedef struct {

src/io/serialize.c

@@ -74,6 +74,8 @@ char *get_meta_key_text(enum metakey meta_key) {
             return "exif_gps_latitude_dms";
         case MetaExifGpsLatitudeDec:
             return "exif_gps_latitude_dec";
+        case MetaChecksum:
+            return "checksum";
         default:
         LOG_FATALF("serialize.c", "FIXME: Unknown meta key: %d", meta_key)
     }
@@ -165,6 +167,7 @@ char *build_json_string(document_t *doc) {
             case MetaExifGpsLatitudeDMS:
             case MetaExifGpsLatitudeDec:
             case MetaExifGpsLatitudeRef:
+            case MetaChecksum:
             case MetaTitle: {
                 cJSON_AddStringToObject(json, get_meta_key_text(meta->key), meta->str_val);
                 buffer_size_guess += (int) strlen(meta->str_val);

src/io/store.c

@@ -4,6 +4,7 @@
 store_t *store_create(const char *path, size_t chunk_size) {
     store_t *store = malloc(sizeof(struct store_t));
     mkdir(path, S_IWUSR | S_IRUSR | S_IXUSR);
+    strcpy(store->path, path);
 
 #if (SIST_FAKE_STORE != 1)
     store->chunk_size = chunk_size;
@@ -78,27 +79,57 @@ void store_write(store_t *store, char *key, size_t key_len, char *buf, size_t bu
     int put_ret = mdb_put(txn, store->dbi, &mdb_key, &mdb_value, 0);
     ScanCtx.stat_tn_size += buf_len;
 
+    int db_full = FALSE;
+    int should_abort_transaction = FALSE;
+
     if (put_ret == MDB_MAP_FULL) {
-        mdb_txn_abort(txn);
+        db_full = TRUE;
+        should_abort_transaction = TRUE;
+    } else {
+        int commit_ret = mdb_txn_commit(txn);
+
+        if (commit_ret == MDB_MAP_FULL) {
+            db_full = TRUE;
+        }
+    }
+
+    if (db_full) {
+        LOG_INFOF("store.c", "Updating mdb mapsize to %lu bytes", store->size)
+
+        if (should_abort_transaction) {
+            mdb_txn_abort(txn);
+        }
+
         pthread_rwlock_unlock(&store->lock);
 
         // Cannot resize when there is a opened transaction.
         //  Resize take effect on the next commit.
         pthread_rwlock_wrlock(&store->lock);
         store->size += store->chunk_size;
-        mdb_env_set_mapsize(store->env, store->size);
+        int resize_ret = mdb_env_set_mapsize(store->env, store->size);
+        if (resize_ret != 0) {
+            LOG_ERROR("store.c", mdb_strerror(put_ret))
+        }
         mdb_txn_begin(store->env, NULL, 0, &txn);
-        put_ret = mdb_put(txn, store->dbi, &mdb_key, &mdb_value, 0);
+        int put_ret_retry = mdb_put(txn, store->dbi, &mdb_key, &mdb_value, 0);
 
+        if (put_ret_retry != 0) {
+            LOG_ERROR("store.c", mdb_strerror(put_ret))
+        }
+
+        int ret = mdb_txn_commit(txn);
+        if (ret != 0) {
+            LOG_FATALF("store.c", "FIXME: Could not commit to store %s: %s (%d), %d, %d %d",
+                       store->path, mdb_strerror(ret), ret,
+                       put_ret, put_ret_retry);
+        }
         LOG_INFOF("store.c", "Updated mdb mapsize to %lu bytes", store->size)
-    }
-
-    mdb_txn_commit(txn);
-    pthread_rwlock_unlock(&store->lock);
-
-    if (put_ret != 0) {
+    } else if (put_ret != 0) {
         LOG_ERROR("store.c", mdb_strerror(put_ret))
     }
+
+    pthread_rwlock_unlock(&store->lock);
+
 #endif
 }
 

src/io/store.h

@@ -6,12 +6,12 @@
 
 #include <glib.h>
 
-#define STORE_SIZE_TN 1024 * 1024 * 5
-#define STORE_SIZE_TAG 1024 * 16
+#define STORE_SIZE_TN (1024 * 1024 * 5)
+#define STORE_SIZE_TAG (1024 * 1024)
 #define STORE_SIZE_META STORE_SIZE_TAG
 
 typedef struct store_t {
-    char *path;
+    char path[PATH_MAX];
     char *tmp_path;
     MDB_dbi dbi;
     MDB_env *env;

src/io/walk.c

@@ -24,16 +24,22 @@ parse_job_t *create_fs_parse_job(const char *filepath, const struct stat *info,
 
     job->vfile.filepath = job->filepath;
     job->vfile.read = fs_read;
+    // Filesystem reads are always rewindable
+    job->vfile.read_rewindable = fs_read;
     job->vfile.reset = fs_reset;
     job->vfile.close = fs_close;
     job->vfile.fd = -1;
     job->vfile.is_fs_file = TRUE;
+    job->vfile.has_checksum = FALSE;
+    job->vfile.rewind_buffer_size = 0;
+    job->vfile.rewind_buffer = NULL;
+    job->vfile.calculate_checksum = ScanCtx.calculate_checksums;
 
     return job;
 }
 
 int sub_strings[30];
-#define EXCLUDED(str) (pcre_exec(ScanCtx.exclude, ScanCtx.exclude_extra, filepath, strlen(filepath), 0, 0, sub_strings, sizeof(sub_strings)) >= 0)
+#define EXCLUDED(str) (pcre_exec(ScanCtx.exclude, ScanCtx.exclude_extra, str, strlen(str), 0, 0, sub_strings, sizeof(sub_strings)) >= 0)
 
 int handle_entry(const char *filepath, const struct stat *info, int typeflag, struct FTW *ftw) {
 

src/main.c

@@ -170,6 +170,8 @@ void initialize_scan_context(scan_args_t *args) {
     pthread_mutex_init(&ScanCtx.dbg_current_files_mu, NULL);
     pthread_mutex_init(&ScanCtx.dbg_file_counts_mu, NULL);
 
+    ScanCtx.calculate_checksums = args->calculate_checksums;
+
     // Archive
     ScanCtx.arc_ctx.mode = args->archive_mode;
     ScanCtx.arc_ctx.log = _log;
@@ -259,10 +261,18 @@ void initialize_scan_context(scan_args_t *args) {
     ScanCtx.raw_ctx.logf = _logf;
     ScanCtx.raw_ctx.store = _store;
 
+    // Wpd
     ScanCtx.wpd_ctx.content_size = args->content_size;
     ScanCtx.wpd_ctx.log = _log;
     ScanCtx.wpd_ctx.logf = _logf;
     ScanCtx.wpd_ctx.wpd_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/wordperfect");
+
+    // Json
+    ScanCtx.json_ctx.content_size = args->content_size;
+    ScanCtx.json_ctx.log = _log;
+    ScanCtx.json_ctx.logf = _logf;
+    ScanCtx.json_ctx.json_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/json");
+    ScanCtx.json_ctx.ndjson_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/ndjson");
 }
 
 
@@ -508,8 +518,8 @@ void sist2_web(web_args_t *args) {
 
 
 int main(int argc, const char *argv[]) {
-    sigsegv_handler = signal(SIGSEGV, sig_handler);
-    sigabrt_handler = signal(SIGABRT, sig_handler);
+//    sigsegv_handler = signal(SIGSEGV, sig_handler);
+//    sigabrt_handler = signal(SIGABRT, sig_handler);
 
     setlocale(LC_ALL, "");
 
@@ -566,6 +576,7 @@ int main(int argc, const char *argv[]) {
             OPT_BOOLEAN(0, "read-subtitles", &scan_args->read_subtitles, "Read subtitles from media files."),
             OPT_BOOLEAN(0, "fast-epub", &scan_args->fast_epub,
                         "Faster but less accurate EPUB parsing (no thumbnails, metadata)"),
+            OPT_BOOLEAN(0, "checksums", &scan_args->calculate_checksums, "Calculate file checksums when scanning."),
 
             OPT_GROUP("Index options"),
             OPT_INTEGER('t', "threads", &common_threads, "Number of threads. DEFAULT=1"),

src/parsing/mime_generated.c

@@ -35,425 +35,426 @@ enum mime {
     application_mime=655387,
     application_mspowerpoint=655388,
     application_msword=655389,
-    application_netmc=655390,
-    application_octet_stream=655391,
-    application_oda=655392,
-    application_ogg=655393,
-    application_pdf=655394 | 0x40000000,
-    application_pgp_keys=655395,
-    application_pgp_signature=655396,
-    application_pkcs7_signature=655397,
-    application_pkix_cert=655398,
-    application_postscript=655399,
-    application_pro_eng=655400,
-    application_ringing_tones=655401,
-    application_smil=655402,
-    application_solids=655403,
-    application_sounder=655404,
-    application_step=655405,
-    application_streamingmedia=655406,
-    application_vda=655407,
-    application_vnd_amazon_mobi8_ebook=655408 | 0x02000000,
-    application_vnd_coffeescript=655409,
-    application_vnd_fdf=655410,
-    application_vnd_font_fontforge_sfd=655411,
-    application_vnd_hp_hpgl=655412,
-    application_vnd_iccprofile=655413,
-    application_vnd_lotus_1_2_3=655414,
-    application_vnd_ms_cab_compressed=655415,
-    application_vnd_ms_excel=655416,
-    application_vnd_ms_fontobject=655417,
-    application_vnd_ms_opentype=655418 | 0x20000000,
-    application_vnd_ms_outlook=655419,
-    application_vnd_ms_pki_certstore=655420,
-    application_vnd_ms_pki_pko=655421,
-    application_vnd_ms_pki_seccat=655422,
-    application_vnd_ms_powerpoint=655423,
-    application_vnd_ms_project=655424,
-    application_vnd_oasis_opendocument_base=655425,
-    application_vnd_oasis_opendocument_formula=655426,
-    application_vnd_oasis_opendocument_graphics=655427,
-    application_vnd_oasis_opendocument_presentation=655428,
-    application_vnd_oasis_opendocument_spreadsheet=655429,
-    application_vnd_oasis_opendocument_text=655430,
-    application_vnd_openxmlformats_officedocument_presentationml_presentation=655431 | 0x04000000,
-    application_vnd_openxmlformats_officedocument_spreadsheetml_sheet=655432 | 0x04000000,
-    application_vnd_openxmlformats_officedocument_wordprocessingml_document=655433 | 0x04000000,
-    application_vnd_symbian_install=655434,
-    application_vnd_tcpdump_pcap=655435,
-    application_vnd_wap_wmlc=655436,
-    application_vnd_wap_wmlscriptc=655437,
-    application_vnd_xara=655438,
-    application_vocaltec_media_desc=655439,
-    application_vocaltec_media_file=655440,
-    application_warc=655441,
-    application_winhelp=655442,
-    application_wordperfect=655443,
-    application_x_123=655444,
-    application_x_7z_compressed=655445 | 0x10000000,
-    application_x_aim=655446,
-    application_x_apple_diskimage=655447,
-    application_x_arc=655448 | 0x10000000,
-    application_x_archive=655449,
-    application_x_atari_7800_rom=655450,
-    application_x_authorware_bin=655451,
-    application_x_authorware_map=655452,
-    application_x_authorware_seg=655453,
-    application_x_avira_qua=655454,
-    application_x_bcpio=655455,
-    application_x_bittorrent=655456,
-    application_x_bsh=655457,
-    application_x_bytecode_python=655458,
-    application_x_bzip=655459,
-    application_x_bzip2=655460 | 0x08000000,
-    application_x_cbr=655461,
-    application_x_cbz=655462,
-    application_x_cdlink=655463,
-    application_x_chat=655464,
-    application_x_chrome_extension=655465,
-    application_x_cocoa=655466,
-    application_x_conference=655467,
-    application_x_coredump=655468,
-    application_x_cpio=655469,
-    application_x_dbf=655470,
-    application_x_dbt=655471,
-    application_x_debian_package=655472,
-    application_x_deepv=655473,
-    application_x_director=655474,
-    application_x_dmp=655475,
-    application_x_dosdriver=655476,
-    application_x_dosexec=655477,
-    application_x_dvi=655478,
-    application_x_elc=655479,
+    application_ndjson=655390,
+    application_netmc=655391,
+    application_octet_stream=655392,
+    application_oda=655393,
+    application_ogg=655394,
+    application_pdf=655395 | 0x40000000,
+    application_pgp_keys=655396,
+    application_pgp_signature=655397,
+    application_pkcs7_signature=655398,
+    application_pkix_cert=655399,
+    application_postscript=655400,
+    application_pro_eng=655401,
+    application_ringing_tones=655402,
+    application_smil=655403,
+    application_solids=655404,
+    application_sounder=655405,
+    application_step=655406,
+    application_streamingmedia=655407,
+    application_vda=655408,
+    application_vnd_amazon_mobi8_ebook=655409 | 0x02000000,
+    application_vnd_coffeescript=655410,
+    application_vnd_fdf=655411,
+    application_vnd_font_fontforge_sfd=655412,
+    application_vnd_hp_hpgl=655413,
+    application_vnd_iccprofile=655414,
+    application_vnd_lotus_1_2_3=655415,
+    application_vnd_ms_cab_compressed=655416,
+    application_vnd_ms_excel=655417,
+    application_vnd_ms_fontobject=655418,
+    application_vnd_ms_opentype=655419 | 0x20000000,
+    application_vnd_ms_outlook=655420,
+    application_vnd_ms_pki_certstore=655421,
+    application_vnd_ms_pki_pko=655422,
+    application_vnd_ms_pki_seccat=655423,
+    application_vnd_ms_powerpoint=655424,
+    application_vnd_ms_project=655425,
+    application_vnd_oasis_opendocument_base=655426,
+    application_vnd_oasis_opendocument_formula=655427,
+    application_vnd_oasis_opendocument_graphics=655428,
+    application_vnd_oasis_opendocument_presentation=655429,
+    application_vnd_oasis_opendocument_spreadsheet=655430,
+    application_vnd_oasis_opendocument_text=655431,
+    application_vnd_openxmlformats_officedocument_presentationml_presentation=655432 | 0x04000000,
+    application_vnd_openxmlformats_officedocument_spreadsheetml_sheet=655433 | 0x04000000,
+    application_vnd_openxmlformats_officedocument_wordprocessingml_document=655434 | 0x04000000,
+    application_vnd_symbian_install=655435,
+    application_vnd_tcpdump_pcap=655436,
+    application_vnd_wap_wmlc=655437,
+    application_vnd_wap_wmlscriptc=655438,
+    application_vnd_xara=655439,
+    application_vocaltec_media_desc=655440,
+    application_vocaltec_media_file=655441,
+    application_warc=655442,
+    application_winhelp=655443,
+    application_wordperfect=655444,
+    application_x_123=655445,
+    application_x_7z_compressed=655446 | 0x10000000,
+    application_x_aim=655447,
+    application_x_apple_diskimage=655448,
+    application_x_arc=655449 | 0x10000000,
+    application_x_archive=655450,
+    application_x_atari_7800_rom=655451,
+    application_x_authorware_bin=655452,
+    application_x_authorware_map=655453,
+    application_x_authorware_seg=655454,
+    application_x_avira_qua=655455,
+    application_x_bcpio=655456,
+    application_x_bittorrent=655457,
+    application_x_bsh=655458,
+    application_x_bytecode_python=655459,
+    application_x_bzip=655460,
+    application_x_bzip2=655461 | 0x08000000,
+    application_x_cbr=655462,
+    application_x_cbz=655463,
+    application_x_cdlink=655464,
+    application_x_chat=655465,
+    application_x_chrome_extension=655466,
+    application_x_cocoa=655467,
+    application_x_conference=655468,
+    application_x_coredump=655469,
+    application_x_cpio=655470,
+    application_x_dbf=655471,
+    application_x_dbt=655472,
+    application_x_debian_package=655473,
+    application_x_deepv=655474,
+    application_x_director=655475,
+    application_x_dmp=655476,
+    application_x_dosdriver=655477,
+    application_x_dosexec=655478,
+    application_x_dvi=655479,
+    application_x_elc=655480,
     application_x_empty=1,
-    application_x_envoy=655480,
-    application_x_esrehber=655481,
-    application_x_excel=655482,
-    application_x_executable=655483,
-    application_x_font_gdos=655484,
-    application_x_font_pf2=655485,
-    application_x_font_pfm=655486,
-    application_x_font_sfn=655487,
-    application_x_font_ttf=655488 | 0x20000000,
-    application_x_fptapplication_x_dbt=655489,
-    application_x_freelance=655490,
-    application_x_gamecube_rom=655491,
-    application_x_gdbm=655492,
-    application_x_gettext_translation=655493,
-    application_x_git=655494,
-    application_x_gsp=655495,
-    application_x_gss=655496,
-    application_x_gtar=655497,
-    application_x_gzip=655498,
-    application_x_hdf=655499,
-    application_x_helpfile=655500,
-    application_x_httpd_imap=655501,
-    application_x_ima=655502,
-    application_x_innosetup=655503,
-    application_x_internett_signup=655504,
-    application_x_inventor=655505,
-    application_x_ip2=655506,
-    application_x_java_applet=655507,
-    application_x_java_commerce=655508,
-    application_x_java_image=655509,
-    application_x_java_jmod=655510,
-    application_x_java_keystore=655511,
-    application_x_kdelnk=655512,
-    application_x_koan=655513,
-    application_x_latex=655514,
-    application_x_livescreen=655515,
-    application_x_lotus=655516,
-    application_x_lz4=655517 | 0x08000000,
-    application_x_lz4_json=655518,
-    application_x_lzh=655519,
-    application_x_lzh_compressed=655520,
-    application_x_lzip=655521 | 0x08000000,
-    application_x_lzma=655522 | 0x08000000,
-    application_x_lzop=655523 | 0x08000000,
-    application_x_lzx=655524,
-    application_x_mach_binary=655525,
-    application_x_mach_executable=655526,
-    application_x_magic_cap_package_1_0=655527,
-    application_x_mathcad=655528,
-    application_x_maxis_dbpf=655529,
-    application_x_meme=655530,
-    application_x_midi=655531,
-    application_x_mif=655532,
-    application_x_mix_transfer=655533,
-    application_x_mobipocket_ebook=655534 | 0x02000000,
-    application_x_ms_compress_szdd=655535,
-    application_x_ms_pdb=655536,
-    application_x_ms_reader=655537,
-    application_x_msaccess=655538,
-    application_x_n64_rom=655539,
-    application_x_navi_animation=655540,
-    application_x_navidoc=655541,
-    application_x_navimap=655542,
-    application_x_navistyle=655543,
-    application_x_nes_rom=655544,
-    application_x_netcdf=655545,
-    application_x_newton_compatible_pkg=655546,
-    application_x_nintendo_ds_rom=655547,
-    application_x_object=655548,
-    application_x_omc=655549,
-    application_x_omcdatamaker=655550,
-    application_x_omcregerator=655551,
-    application_x_pagemaker=655552,
-    application_x_pcl=655553,
-    application_x_pgp_keyring=655554,
-    application_x_pixclscript=655555,
-    application_x_pkcs7_certreqresp=655556,
-    application_x_pkcs7_signature=655557,
-    application_x_project=655558,
-    application_x_qpro=655559,
-    application_x_rar=655560 | 0x10000000,
-    application_x_rpm=655561,
-    application_x_sdp=655562,
-    application_x_sea=655563,
-    application_x_seelogo=655564,
-    application_x_setupscript=655565,
-    application_x_shar=655566,
-    application_x_sharedlib=655567,
-    application_x_shockwave_flash=655568,
-    application_x_snappy_framed=655569,
-    application_x_sprite=655570,
-    application_x_sqlite3=655571,
-    application_x_stargallery_thm=655572,
-    application_x_stuffit=655573,
-    application_x_sv4cpio=655574,
-    application_x_sv4crc=655575,
-    application_x_tar=655576 | 0x10000000,
-    application_x_tbook=655577,
-    application_x_terminfo=655578,
-    application_x_terminfo2=655579,
-    application_x_tex_tfm=655580,
-    application_x_texinfo=655581,
-    application_x_ustar=655582,
-    application_x_visio=655583,
-    application_x_vnd_audioexplosion_mzz=655584,
-    application_x_vnd_ls_xpix=655585,
-    application_x_vrml=655586,
-    application_x_wais_source=655587,
-    application_x_wine_extension_ini=655588,
-    application_x_wintalk=655589,
-    application_x_world=655590,
-    application_x_wri=655591,
-    application_x_x509_ca_cert=655592,
-    application_x_xz=655593 | 0x08000000,
-    application_x_zip=655594,
-    application_x_zstd=655595 | 0x08000000,
-    application_x_zstd_dictionary=655596,
-    application_xml=655597,
-    application_zip=655598 | 0x10000000,
-    application_zlib=655599,
-    audio_basic=458992 | 0x80000000,
-    audio_it=458993,
-    audio_make=458994,
-    audio_mid=458995,
-    audio_midi=458996,
-    audio_mp4=458997,
-    audio_mpeg=458998,
-    audio_ogg=458999,
-    audio_s3m=459000,
-    audio_tsp_audio=459001,
-    audio_tsplayer=459002,
-    audio_vnd_qcelp=459003,
-    audio_voxware=459004,
-    audio_x_aiff=459005,
-    audio_x_flac=459006,
-    audio_x_gsm=459007,
-    audio_x_hx_aac_adts=459008,
-    audio_x_jam=459009,
-    audio_x_liveaudio=459010,
-    audio_x_m4a=459011,
-    audio_x_midi=459012,
-    audio_x_mod=459013,
-    audio_x_mp4a_latm=459014,
-    audio_x_mpeg_3=459015,
-    audio_x_mpequrl=459016,
-    audio_x_nspaudio=459017,
-    audio_x_pn_realaudio=459018,
-    audio_x_psid=459019,
-    audio_x_realaudio=459020,
-    audio_x_s3m=459021,
-    audio_x_twinvq=459022,
-    audio_x_twinvq_plugin=459023,
-    audio_x_voc=459024,
-    audio_x_wav=459025,
-    audio_x_xbox_executable=459026 | 0x80000000,
-    audio_x_xbox360_executable=459027 | 0x80000000,
-    audio_xm=459028,
-    font_otf=327957 | 0x20000000,
-    font_sfnt=327958 | 0x20000000,
-    font_woff=327959 | 0x20000000,
-    font_woff2=327960 | 0x20000000,
-    image_bmp=524569,
-    image_cmu_raster=524570,
-    image_fif=524571,
-    image_florian=524572,
-    image_g3fax=524573,
-    image_gif=524574,
-    image_heic=524575,
-    image_ief=524576,
-    image_jpeg=524577,
-    image_jutvision=524578,
-    image_naplps=524579,
-    image_pict=524580,
-    image_png=524581,
-    image_svg=524582 | 0x80000000,
-    image_svg_xml=524583 | 0x80000000,
-    image_tiff=524584,
-    image_vnd_adobe_photoshop=524585 | 0x80000000,
-    image_vnd_djvu=524586 | 0x80000000,
-    image_vnd_fpx=524587,
-    image_vnd_microsoft_icon=524588,
-    image_vnd_rn_realflash=524589,
-    image_vnd_rn_realpix=524590,
-    image_vnd_wap_wbmp=524591,
-    image_vnd_xiff=524592,
-    image_webp=524593,
-    image_wmf=524594,
-    image_x_3ds=524595,
-    image_x_adobe_dng=524596 | 0x00800000,
-    image_x_award_bioslogo=524597,
-    image_x_canon_cr2=524598 | 0x00800000,
-    image_x_canon_crw=524599 | 0x00800000,
-    image_x_cmu_raster=524600,
-    image_x_cur=524601,
-    image_x_dcraw=524602 | 0x00800000,
-    image_x_dwg=524603,
-    image_x_eps=524604,
-    image_x_epson_erf=524605 | 0x00800000,
-    image_x_exr=524606,
-    image_x_fuji_raf=524607 | 0x00800000,
-    image_x_gem=524608,
-    image_x_icns=524609,
-    image_x_icon=524610 | 0x80000000,
-    image_x_jg=524611,
-    image_x_jps=524612,
-    image_x_kodak_dcr=524613 | 0x00800000,
-    image_x_kodak_k25=524614 | 0x00800000,
-    image_x_kodak_kdc=524615 | 0x00800000,
-    image_x_minolta_mrw=524616 | 0x00800000,
-    image_x_ms_bmp=524617,
-    image_x_niff=524618,
-    image_x_nikon_nef=524619 | 0x00800000,
-    image_x_olympus_orf=524620 | 0x00800000,
-    image_x_panasonic_raw=524621 | 0x00800000,
-    image_x_pcx=524622,
-    image_x_pentax_pef=524623 | 0x00800000,
-    image_x_pict=524624,
-    image_x_portable_bitmap=524625,
-    image_x_portable_graymap=524626,
-    image_x_portable_pixmap=524627,
-    image_x_quicktime=524628,
-    image_x_rgb=524629,
-    image_x_sigma_x3f=524630 | 0x00800000,
-    image_x_sony_arw=524631 | 0x00800000,
-    image_x_sony_sr2=524632 | 0x00800000,
-    image_x_sony_srf=524633 | 0x00800000,
-    image_x_tga=524634,
-    image_x_tiff=524635,
-    image_x_win_bitmap=524636,
-    image_x_xcf=524637 | 0x80000000,
-    image_x_xpixmap=524638 | 0x80000000,
-    image_x_xwindowdump=524639,
-    message_news=196960,
-    message_rfc822=196961,
-    model_vnd_dwf=65890,
-    model_vnd_gdl=65891,
-    model_vnd_gs_gdl=65892,
-    model_vrml=65893,
-    model_x_pov=65894,
+    application_x_envoy=655481,
+    application_x_esrehber=655482,
+    application_x_excel=655483,
+    application_x_executable=655484,
+    application_x_font_gdos=655485,
+    application_x_font_pf2=655486,
+    application_x_font_pfm=655487,
+    application_x_font_sfn=655488,
+    application_x_font_ttf=655489 | 0x20000000,
+    application_x_fptapplication_x_dbt=655490,
+    application_x_freelance=655491,
+    application_x_gamecube_rom=655492,
+    application_x_gdbm=655493,
+    application_x_gettext_translation=655494,
+    application_x_git=655495,
+    application_x_gsp=655496,
+    application_x_gss=655497,
+    application_x_gtar=655498,
+    application_x_gzip=655499,
+    application_x_hdf=655500,
+    application_x_helpfile=655501,
+    application_x_httpd_imap=655502,
+    application_x_ima=655503,
+    application_x_innosetup=655504,
+    application_x_internett_signup=655505,
+    application_x_inventor=655506,
+    application_x_ip2=655507,
+    application_x_java_applet=655508,
+    application_x_java_commerce=655509,
+    application_x_java_image=655510,
+    application_x_java_jmod=655511,
+    application_x_java_keystore=655512,
+    application_x_kdelnk=655513,
+    application_x_koan=655514,
+    application_x_latex=655515,
+    application_x_livescreen=655516,
+    application_x_lotus=655517,
+    application_x_lz4=655518 | 0x08000000,
+    application_x_lz4_json=655519,
+    application_x_lzh=655520,
+    application_x_lzh_compressed=655521,
+    application_x_lzip=655522 | 0x08000000,
+    application_x_lzma=655523 | 0x08000000,
+    application_x_lzop=655524 | 0x08000000,
+    application_x_lzx=655525,
+    application_x_mach_binary=655526,
+    application_x_mach_executable=655527,
+    application_x_magic_cap_package_1_0=655528,
+    application_x_mathcad=655529,
+    application_x_maxis_dbpf=655530,
+    application_x_meme=655531,
+    application_x_midi=655532,
+    application_x_mif=655533,
+    application_x_mix_transfer=655534,
+    application_x_mobipocket_ebook=655535 | 0x02000000,
+    application_x_ms_compress_szdd=655536,
+    application_x_ms_pdb=655537,
+    application_x_ms_reader=655538,
+    application_x_msaccess=655539,
+    application_x_n64_rom=655540,
+    application_x_navi_animation=655541,
+    application_x_navidoc=655542,
+    application_x_navimap=655543,
+    application_x_navistyle=655544,
+    application_x_nes_rom=655545,
+    application_x_netcdf=655546,
+    application_x_newton_compatible_pkg=655547,
+    application_x_nintendo_ds_rom=655548,
+    application_x_object=655549,
+    application_x_omc=655550,
+    application_x_omcdatamaker=655551,
+    application_x_omcregerator=655552,
+    application_x_pagemaker=655553,
+    application_x_pcl=655554,
+    application_x_pgp_keyring=655555,
+    application_x_pixclscript=655556,
+    application_x_pkcs7_certreqresp=655557,
+    application_x_pkcs7_signature=655558,
+    application_x_project=655559,
+    application_x_qpro=655560,
+    application_x_rar=655561 | 0x10000000,
+    application_x_rpm=655562,
+    application_x_sdp=655563,
+    application_x_sea=655564,
+    application_x_seelogo=655565,
+    application_x_setupscript=655566,
+    application_x_shar=655567,
+    application_x_sharedlib=655568,
+    application_x_shockwave_flash=655569,
+    application_x_snappy_framed=655570,
+    application_x_sprite=655571,
+    application_x_sqlite3=655572,
+    application_x_stargallery_thm=655573,
+    application_x_stuffit=655574,
+    application_x_sv4cpio=655575,
+    application_x_sv4crc=655576,
+    application_x_tar=655577 | 0x10000000,
+    application_x_tbook=655578,
+    application_x_terminfo=655579,
+    application_x_terminfo2=655580,
+    application_x_tex_tfm=655581,
+    application_x_texinfo=655582,
+    application_x_ustar=655583,
+    application_x_visio=655584,
+    application_x_vnd_audioexplosion_mzz=655585,
+    application_x_vnd_ls_xpix=655586,
+    application_x_vrml=655587,
+    application_x_wais_source=655588,
+    application_x_wine_extension_ini=655589,
+    application_x_wintalk=655590,
+    application_x_world=655591,
+    application_x_wri=655592,
+    application_x_x509_ca_cert=655593,
+    application_x_xz=655594 | 0x08000000,
+    application_x_zip=655595,
+    application_x_zstd=655596 | 0x08000000,
+    application_x_zstd_dictionary=655597,
+    application_xml=655598,
+    application_zip=655599 | 0x10000000,
+    application_zlib=655600,
+    audio_basic=458993 | 0x80000000,
+    audio_it=458994,
+    audio_make=458995,
+    audio_mid=458996,
+    audio_midi=458997,
+    audio_mp4=458998,
+    audio_mpeg=458999,
+    audio_ogg=459000,
+    audio_s3m=459001,
+    audio_tsp_audio=459002,
+    audio_tsplayer=459003,
+    audio_vnd_qcelp=459004,
+    audio_voxware=459005,
+    audio_x_aiff=459006,
+    audio_x_flac=459007,
+    audio_x_gsm=459008,
+    audio_x_hx_aac_adts=459009,
+    audio_x_jam=459010,
+    audio_x_liveaudio=459011,
+    audio_x_m4a=459012,
+    audio_x_midi=459013,
+    audio_x_mod=459014,
+    audio_x_mp4a_latm=459015,
+    audio_x_mpeg_3=459016,
+    audio_x_mpequrl=459017,
+    audio_x_nspaudio=459018,
+    audio_x_pn_realaudio=459019,
+    audio_x_psid=459020,
+    audio_x_realaudio=459021,
+    audio_x_s3m=459022,
+    audio_x_twinvq=459023,
+    audio_x_twinvq_plugin=459024,
+    audio_x_voc=459025,
+    audio_x_wav=459026,
+    audio_x_xbox_executable=459027 | 0x80000000,
+    audio_x_xbox360_executable=459028 | 0x80000000,
+    audio_xm=459029,
+    font_otf=327958 | 0x20000000,
+    font_sfnt=327959 | 0x20000000,
+    font_woff=327960 | 0x20000000,
+    font_woff2=327961 | 0x20000000,
+    image_bmp=524570,
+    image_cmu_raster=524571,
+    image_fif=524572,
+    image_florian=524573,
+    image_g3fax=524574,
+    image_gif=524575,
+    image_heic=524576,
+    image_ief=524577,
+    image_jpeg=524578,
+    image_jutvision=524579,
+    image_naplps=524580,
+    image_pict=524581,
+    image_png=524582,
+    image_svg=524583 | 0x80000000,
+    image_svg_xml=524584 | 0x80000000,
+    image_tiff=524585,
+    image_vnd_adobe_photoshop=524586 | 0x80000000,
+    image_vnd_djvu=524587 | 0x80000000,
+    image_vnd_fpx=524588,
+    image_vnd_microsoft_icon=524589,
+    image_vnd_rn_realflash=524590,
+    image_vnd_rn_realpix=524591,
+    image_vnd_wap_wbmp=524592,
+    image_vnd_xiff=524593,
+    image_webp=524594,
+    image_wmf=524595,
+    image_x_3ds=524596,
+    image_x_adobe_dng=524597 | 0x00800000,
+    image_x_award_bioslogo=524598,
+    image_x_canon_cr2=524599 | 0x00800000,
+    image_x_canon_crw=524600 | 0x00800000,
+    image_x_cmu_raster=524601,
+    image_x_cur=524602,
+    image_x_dcraw=524603 | 0x00800000,
+    image_x_dwg=524604,
+    image_x_eps=524605,
+    image_x_epson_erf=524606 | 0x00800000,
+    image_x_exr=524607,
+    image_x_fuji_raf=524608 | 0x00800000,
+    image_x_gem=524609,
+    image_x_icns=524610,
+    image_x_icon=524611 | 0x80000000,
+    image_x_jg=524612,
+    image_x_jps=524613,
+    image_x_kodak_dcr=524614 | 0x00800000,
+    image_x_kodak_k25=524615 | 0x00800000,
+    image_x_kodak_kdc=524616 | 0x00800000,
+    image_x_minolta_mrw=524617 | 0x00800000,
+    image_x_ms_bmp=524618,
+    image_x_niff=524619,
+    image_x_nikon_nef=524620 | 0x00800000,
+    image_x_olympus_orf=524621 | 0x00800000,
+    image_x_panasonic_raw=524622 | 0x00800000,
+    image_x_pcx=524623,
+    image_x_pentax_pef=524624 | 0x00800000,
+    image_x_pict=524625,
+    image_x_portable_bitmap=524626,
+    image_x_portable_graymap=524627,
+    image_x_portable_pixmap=524628,
+    image_x_quicktime=524629,
+    image_x_rgb=524630,
+    image_x_sigma_x3f=524631 | 0x00800000,
+    image_x_sony_arw=524632 | 0x00800000,
+    image_x_sony_sr2=524633 | 0x00800000,
+    image_x_sony_srf=524634 | 0x00800000,
+    image_x_tga=524635,
+    image_x_tiff=524636,
+    image_x_win_bitmap=524637,
+    image_x_xcf=524638 | 0x80000000,
+    image_x_xpixmap=524639 | 0x80000000,
+    image_x_xwindowdump=524640,
+    message_news=196961,
+    message_rfc822=196962,
+    model_vnd_dwf=65891,
+    model_vnd_gdl=65892,
+    model_vnd_gs_gdl=65893,
+    model_vrml=65894,
+    model_x_pov=65895,
     sist2_sidecar=2,
-    text_PGP=590183,
-    text_asp=590184,
-    text_css=590185,
-    text_html=590186 | 0x01000000,
-    text_javascript=590187,
-    text_mcf=590188,
-    text_pascal=590189,
-    text_plain=590190,
-    text_richtext=590191,
-    text_rtf=590192,
-    text_scriplet=590193,
-    text_tab_separated_values=590194,
-    text_troff=590195,
-    text_uri_list=590196,
-    text_vnd_abc=590197,
-    text_vnd_fmi_flexstor=590198,
-    text_vnd_wap_wml=590199,
-    text_vnd_wap_wmlscript=590200,
-    text_webviewhtml=590201,
-    text_x_Algol68=590202,
-    text_x_asm=590203,
-    text_x_audiosoft_intra=590204,
-    text_x_awk=590205,
-    text_x_bcpl=590206,
-    text_x_c=590207,
-    text_x_c__=590208,
-    text_x_component=590209,
-    text_x_diff=590210,
-    text_x_fortran=590211,
-    text_x_java=590212,
-    text_x_la_asf=590213,
-    text_x_lisp=590214,
-    text_x_m=590215,
-    text_x_m4=590216,
-    text_x_makefile=590217,
-    text_x_ms_regedit=590218,
-    text_x_msdos_batch=590219,
-    text_x_objective_c=590220,
-    text_x_pascal=590221,
-    text_x_perl=590222,
-    text_x_php=590223,
-    text_x_po=590224,
-    text_x_python=590225,
-    text_x_ruby=590226,
-    text_x_sass=590227,
-    text_x_scss=590228,
-    text_x_server_parsed_html=590229,
-    text_x_setext=590230,
-    text_x_sgml=590231 | 0x01000000,
-    text_x_shellscript=590232,
-    text_x_speech=590233,
-    text_x_tcl=590234,
-    text_x_tex=590235,
-    text_x_uil=590236,
-    text_x_uuencode=590237,
-    text_x_vcalendar=590238,
-    text_x_vcard=590239,
-    text_xml=590240 | 0x01000000,
-    video_MP2T=393633,
-    video_animaflex=393634,
-    video_avi=393635,
-    video_avs_video=393636,
-    video_mp4=393637,
-    video_mpeg=393638,
-    video_quicktime=393639,
-    video_vdo=393640,
-    video_vivo=393641,
-    video_vnd_rn_realvideo=393642,
-    video_vosaic=393643,
-    video_webm=393644,
-    video_x_amt_demorun=393645,
-    video_x_amt_showrun=393646,
-    video_x_atomic3d_feature=393647,
-    video_x_dl=393648,
-    video_x_dv=393649,
-    video_x_fli=393650,
-    video_x_flv=393651,
-    video_x_isvideo=393652,
-    video_x_jng=393653 | 0x80000000,
-    video_x_m4v=393654,
-    video_x_matroska=393655,
-    video_x_mng=393656,
-    video_x_motion_jpeg=393657,
-    video_x_ms_asf=393658,
-    video_x_msvideo=393659,
-    video_x_qtc=393660,
-    video_x_sgi_movie=393661,
-    x_epoc_x_sisx_app=721342,
+    text_PGP=590184,
+    text_asp=590185,
+    text_css=590186,
+    text_html=590187 | 0x01000000,
+    text_javascript=590188,
+    text_mcf=590189,
+    text_pascal=590190,
+    text_plain=590191,
+    text_richtext=590192,
+    text_rtf=590193,
+    text_scriplet=590194,
+    text_tab_separated_values=590195,
+    text_troff=590196,
+    text_uri_list=590197,
+    text_vnd_abc=590198,
+    text_vnd_fmi_flexstor=590199,
+    text_vnd_wap_wml=590200,
+    text_vnd_wap_wmlscript=590201,
+    text_webviewhtml=590202,
+    text_x_Algol68=590203,
+    text_x_asm=590204,
+    text_x_audiosoft_intra=590205,
+    text_x_awk=590206,
+    text_x_bcpl=590207,
+    text_x_c=590208,
+    text_x_c__=590209,
+    text_x_component=590210,
+    text_x_diff=590211,
+    text_x_fortran=590212,
+    text_x_java=590213,
+    text_x_la_asf=590214,
+    text_x_lisp=590215,
+    text_x_m=590216,
+    text_x_m4=590217,
+    text_x_makefile=590218,
+    text_x_ms_regedit=590219,
+    text_x_msdos_batch=590220,
+    text_x_objective_c=590221,
+    text_x_pascal=590222,
+    text_x_perl=590223,
+    text_x_php=590224,
+    text_x_po=590225,
+    text_x_python=590226,
+    text_x_ruby=590227,
+    text_x_sass=590228,
+    text_x_scss=590229,
+    text_x_server_parsed_html=590230,
+    text_x_setext=590231,
+    text_x_sgml=590232 | 0x01000000,
+    text_x_shellscript=590233,
+    text_x_speech=590234,
+    text_x_tcl=590235,
+    text_x_tex=590236,
+    text_x_uil=590237,
+    text_x_uuencode=590238,
+    text_x_vcalendar=590239,
+    text_x_vcard=590240,
+    text_xml=590241 | 0x01000000,
+    video_MP2T=393634,
+    video_animaflex=393635,
+    video_avi=393636,
+    video_avs_video=393637,
+    video_mp4=393638,
+    video_mpeg=393639,
+    video_quicktime=393640,
+    video_vdo=393641,
+    video_vivo=393642,
+    video_vnd_rn_realvideo=393643,
+    video_vosaic=393644,
+    video_webm=393645,
+    video_x_amt_demorun=393646,
+    video_x_amt_showrun=393647,
+    video_x_atomic3d_feature=393648,
+    video_x_dl=393649,
+    video_x_dv=393650,
+    video_x_fli=393651,
+    video_x_flv=393652,
+    video_x_isvideo=393653,
+    video_x_jng=393654 | 0x80000000,
+    video_x_m4v=393655,
+    video_x_matroska=393656,
+    video_x_mng=393657,
+    video_x_motion_jpeg=393658,
+    video_x_ms_asf=393659,
+    video_x_msvideo=393660,
+    video_x_qtc=393661,
+    video_x_sgi_movie=393662,
+    x_epoc_x_sisx_app=721343,
 };
 char *mime_get_mime_text(unsigned int mime_id) {switch (mime_id) {
 case application_arj: return "application/arj";
@@ -480,6 +481,7 @@ case application_java_archive: return "application/java-archive";
 case application_java: return "application/java";
 case application_javascript: return "application/javascript";
 case application_json: return "application/json";
+case application_ndjson: return "application/ndjson";
 case application_marc: return "application/marc";
 case application_mbedlet: return "application/mbedlet";
 case application_mime: return "application/mime";
@@ -930,6 +932,8 @@ g_hash_table_insert(ext_table, "inf", (gpointer)application_inf);
 g_hash_table_insert(ext_table, "jar", (gpointer)application_java_archive);
 g_hash_table_insert(ext_table, "class", (gpointer)application_java);
 g_hash_table_insert(ext_table, "json", (gpointer)application_json);
+g_hash_table_insert(ext_table, "jsonl", (gpointer)application_ndjson);
+g_hash_table_insert(ext_table, "ndjson", (gpointer)application_ndjson);
 g_hash_table_insert(ext_table, "mrc", (gpointer)application_marc);
 g_hash_table_insert(ext_table, "mbd", (gpointer)application_mbedlet);
 g_hash_table_insert(ext_table, "aps", (gpointer)application_mime);
@@ -1474,6 +1478,7 @@ g_hash_table_insert(mime_table, "application/java-archive", (gpointer)applicatio
 g_hash_table_insert(mime_table, "application/java", (gpointer)application_java);
 g_hash_table_insert(mime_table, "application/javascript", (gpointer)application_javascript);
 g_hash_table_insert(mime_table, "application/json", (gpointer)application_json);
+g_hash_table_insert(mime_table, "application/ndjson", (gpointer)application_ndjson);
 g_hash_table_insert(mime_table, "application/marc", (gpointer)application_marc);
 g_hash_table_insert(mime_table, "application/mbedlet", (gpointer)application_mbedlet);
 g_hash_table_insert(mime_table, "application/mime", (gpointer)application_mime);

src/parsing/parse.c

@@ -10,25 +10,34 @@
 
 
 #define MIN_VIDEO_SIZE (1024 * 64)
-#define MIN_IMAGE_SIZE (1024 * 2)
+#define MIN_IMAGE_SIZE (512)
 
 int fs_read(struct vfile *f, void *buf, size_t size) {
 
     if (f->fd == -1) {
+        SHA1_Init(&f->sha1_ctx);
+
         f->fd = open(f->filepath, O_RDONLY);
         if (f->fd == -1) {
-            LOG_ERRORF(f->filepath, "open(): [%d] %s", errno, strerror(errno))
             return -1;
         }
     }
 
-    return read(f->fd, buf, size);
+    int ret = (int) read(f->fd, buf, size);
+
+    if (ret != 0 && f->calculate_checksum) {
+        f->has_checksum = TRUE;
+        safe_sha1_update(&f->sha1_ctx, (unsigned char *) buf, ret);
+    }
+
+    return ret;
 }
 
 #define CLOSE_FILE(f) if ((f).close != NULL) {(f).close(&(f));};
 
 void fs_close(struct vfile *f) {
     if (f->fd != -1) {
+        SHA1_Final(f->sha1_digest, &f->sha1_ctx);
         close(f->fd);
     }
 }
@@ -66,7 +75,7 @@ void parse(void *arg) {
     doc->meta_tail = NULL;
     doc->mime = 0;
     doc->size = job->vfile.info.st_size;
-    doc->mtime = job->vfile.info.st_mtim.tv_sec;
+    doc->mtime = (int) job->vfile.info.st_mtim.tv_sec;
 
     int inc_ts = incremental_get(ScanCtx.original_table, doc->path_md5);
     if (inc_ts != 0 && inc_ts == job->vfile.info.st_mtim.tv_sec) {
@@ -93,18 +102,17 @@ void parse(void *arg) {
         doc->mime = mime_get_mime_by_ext(ScanCtx.ext_table, job->filepath + job->ext);
     }
 
-    int bytes_read = 0;
 
     if (doc->mime == 0 && !ScanCtx.fast) {
 
         // Get mime type with libmagic
-        if (!job->vfile.is_fs_file) {
+        if (job->vfile.read_rewindable == NULL) {
             LOG_WARNING(job->filepath,
-                        "Guessing mime type with libmagic inside archive files is not currently supported");
+                        "File does not support rewindable reads, cannot guess Media type");
             goto abort;
         }
 
-        bytes_read = job->vfile.read(&job->vfile, buf, MAGIC_BUF_SIZE);
+        int bytes_read = job->vfile.read_rewindable(&job->vfile, buf, MAGIC_BUF_SIZE);
         if (bytes_read < 0) {
 
             if (job->vfile.is_fs_file) {
@@ -135,7 +143,9 @@ void parse(void *arg) {
             }
         }
 
-        job->vfile.reset(&job->vfile);
+        if (job->vfile.reset != NULL) {
+            job->vfile.reset(&job->vfile);
+        }
 
         magic_close(magic);
     }
@@ -149,7 +159,7 @@ void parse(void *arg) {
     } else if ((mmime == MimeVideo && doc->size >= MIN_VIDEO_SIZE) ||
                (mmime == MimeImage && doc->size >= MIN_IMAGE_SIZE) || mmime == MimeAudio) {
 
-        parse_media(&ScanCtx.media_ctx, &job->vfile, doc);
+        parse_media(&ScanCtx.media_ctx, &job->vfile, doc, mime_get_mime_text(doc->mime));
 
     } else if (IS_PDF(doc->mime)) {
         parse_ebook(&ScanCtx.ebook_ctx, &job->vfile, mime_get_mime_text(doc->mime), doc);
@@ -169,7 +179,7 @@ void parse(void *arg) {
                     IS_ARC(doc->mime) ||
                     (IS_ARC_FILTER(doc->mime) && should_parse_filtered_file(doc->filepath, doc->ext))
             )) {
-        parse_archive(&ScanCtx.arc_ctx, &job->vfile, doc);
+        parse_archive(&ScanCtx.arc_ctx, &job->vfile, doc, ScanCtx.exclude, ScanCtx.exclude_extra);
     } else if ((ScanCtx.ooxml_ctx.content_size > 0 || ScanCtx.media_ctx.tn_size > 0) && IS_DOC(doc->mime)) {
         parse_ooxml(&ScanCtx.ooxml_ctx, &job->vfile, doc);
     } else if (is_cbr(&ScanCtx.comic_ctx, doc->mime) || is_cbz(&ScanCtx.comic_ctx, doc->mime)) {
@@ -179,11 +189,15 @@ void parse(void *arg) {
     } else if (doc->mime == MIME_SIST2_SIDECAR) {
         parse_sidecar(&job->vfile, doc);
         CLOSE_FILE(job->vfile)
+        free(doc->filepath);
+        free(doc);
         return;
     } else if (is_msdoc(&ScanCtx.msdoc_ctx, doc->mime)) {
         parse_msdoc(&ScanCtx.msdoc_ctx, &job->vfile, doc);
-    } else if (is_wpd(&ScanCtx.wpd_ctx, doc->mime)) {
-        parse_wpd(&ScanCtx.wpd_ctx, &job->vfile, doc);
+    } else if (is_json(&ScanCtx.json_ctx, doc->mime)) {
+        parse_json(&ScanCtx.json_ctx, &job->vfile, doc);
+    } else if (is_ndjson(&ScanCtx.json_ctx, doc->mime)) {
+        parse_ndjson(&ScanCtx.json_ctx, &job->vfile, doc);
     }
 
     abort:
@@ -200,9 +214,15 @@ void parse(void *arg) {
         doc->has_parent = FALSE;
     }
 
-    write_document(doc);
-
     CLOSE_FILE(job->vfile)
+
+    if (job->vfile.has_checksum) {
+        char sha1_digest_str[SHA1_STR_LENGTH];
+        buf2hex((unsigned char *) job->vfile.sha1_digest, SHA1_DIGEST_LENGTH, (char *) sha1_digest_str);
+        APPEND_STR_META(doc, MetaChecksum, (const char *) sha1_digest_str);
+    }
+
+    write_document(doc);
 }
 
 void cleanup_parse() {

src/parsing/parse.h

@@ -3,7 +3,7 @@
 
 #include "../sist.h"
 
-#define MAGIC_BUF_SIZE 4096 * 6
+#define MAGIC_BUF_SIZE (4096 * 6)
 
 int fs_read(struct vfile *f, void *buf, size_t size);
 void fs_close(struct vfile *f);

src/parsing/sidecar.c

@@ -27,7 +27,10 @@ void parse_sidecar(vfile_t *vfile, document_t *doc) {
     MD5((unsigned char *) vfile->filepath + ScanCtx.index.desc.root_len, doc->ext - 1 - ScanCtx.index.desc.root_len,
         path_md5);
 
-    store_write(ScanCtx.index.meta_store, (char *) path_md5, sizeof(path_md5), json_str, strlen(json_str) + 1);
+    char path_md5_str[MD5_STR_LENGTH];
+    buf2hex(path_md5, MD5_DIGEST_LENGTH, path_md5_str);
+
+    store_write(ScanCtx.index.meta_store, path_md5_str, MD5_STR_LENGTH, json_str, strlen(json_str) + 1);
 
     cJSON_Delete(json);
     free(json_str);

src/sist.h

@@ -26,6 +26,8 @@
 #define UNUSED(x) __attribute__((__unused__))  x
 
 #define MD5_STR_LENGTH 33
+#define SHA1_STR_LENGTH 41
+#define SHA1_DIGEST_LENGTH 20
 
 #include "util.h"
 #include "log.h"
@@ -49,7 +51,7 @@
 #include <ctype.h>
 #include "git_hash.h"
 
-#define VERSION "2.11.2"
+#define VERSION "2.11.3"
 static const char *const Version = VERSION;
 
 #ifndef SIST_PLATFORM