Update libmupdf

.github/ISSUE_TEMPLATE/issue-template.md

@@ -9,7 +9,7 @@ assignees: ''
 
 sist2 version:
 
-Platform (Linux or Docker):
+Platform (Linux or Docker, x86-64 or arm64):
 
 Elasticsearch version:
 

CMakeLists.txt

@@ -9,8 +9,11 @@ add_subdirectory(third-party/libscan)
 set(ARGPARSE_SHARED off)
 add_subdirectory(third-party/argparse)
 
-add_executable(
-        sist2
+add_executable(sist2
+
+        # argparse
+        third-party/argparse/argparse.h third-party/argparse/argparse.c
+
         src/main.c
         src/sist.h
         src/io/walk.h src/io/walk.c
@@ -25,12 +28,9 @@ add_executable(
         src/util.c src/util.h
         src/ctx.h src/types.h
         src/log.c src/log.h
-
-        # argparse
-        third-party/argparse/argparse.h third-party/argparse/argparse.c
-
         src/cli.c src/cli.h
-        src/stats.c src/stats.h src/ctx.c)
+        src/stats.c src/stats.h src/ctx.c
+        src/parsing/sidecar.c src/parsing/sidecar.h)
 
 target_link_directories(sist2 PRIVATE BEFORE ${_VCPKG_INSTALLED_DIR}/${VCPKG_TARGET_TRIPLET}/lib/)
 set(CMAKE_FIND_LIBRARY_SUFFIXES .a .lib)
@@ -73,7 +73,6 @@ if (SIST_DEBUG)
             sist2
             PRIVATE
             -fsanitize=address
-#            -static
     )
     set_target_properties(
             sist2
@@ -81,7 +80,7 @@ if (SIST_DEBUG)
             OUTPUT_NAME sist2_debug
     )
 else ()
-#    set(VCPKG_BUILD_TYPE release)
+    #    set(VCPKG_BUILD_TYPE release)
     target_compile_options(
             sist2
             PRIVATE
@@ -106,7 +105,6 @@ target_link_libraries(
         argparse
         unofficial::glib::glib
         unofficial::mongoose::mongoose
-#        OpenSSL::SSL OpenSSL::Crypto
         CURL::libcurl
 
         ${UUID_LIB}

DockerArm64/Dockerfile

@@ -0,0 +1,22 @@
+FROM ubuntu:19.10
+MAINTAINER simon987 <me@simon987.net>
+
+RUN apt update
+RUN apt install -y libglib2.0-0 libcurl4 libmagic1 libharfbuzz-bin libopenjp2-7 libarchive13 liblzma5 libzstd1 liblz4-1 \
+ curl libtiff5 libpng16-16 libpcre3
+
+RUN mkdir -p /usr/share/tessdata && \
+    cd /usr/share/tessdata/ && \
+    curl -o /usr/share/tessdata/hin.traineddata https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/hin.traineddata &&\
+    curl -o /usr/share/tessdata/jpn.traineddata https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/jpn.traineddata &&\
+    curl -o /usr/share/tessdata/eng.traineddata https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.traineddata &&\
+    curl -o /usr/share/tessdata/fra.traineddata https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/fra.traineddata &&\
+    curl -o /usr/share/tessdata/rus.traineddata https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/rus.traineddata &&\
+    curl -o /usr/share/tessdata/spa.traineddata https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/spa.traineddata && ls -lh
+
+ADD sist2_arm64 /root/sist2
+
+ENV LANG C.UTF-8
+ENV LC_ALL C.UTF-8
+
+ENTRYPOINT ["/root/sist2"]

DockerArm64/build.sh

@@ -0,0 +1,13 @@
+rm ./sist2_arm64
+cp ../sist2_arm64.gz .
+gzip -d sist2_arm64.gz
+
+version=$(./sist2_arm64 --version)
+
+echo "Version ${version}"
+docker build . -t simon987/sist2-arm64:"${version}" -t simon987/sist2-arm64:latest
+
+docker push simon987/sist2-arm64:"${version}"
+docker push simon987/sist2-arm64:latest
+
+docker run --rm simon987/sist2-arm64 -v

README.md

@@ -126,7 +126,7 @@ binaries (GCC 7+ required).
 1. Install compile-time dependencies
 
    ```bash
-   vcpkg install lmdb cjson glib libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 ffmpeg zstd gtest mongoose libuuid libmagic libraw curl[core,ssl] jbig2dec
+   vcpkg install lmdb cjson glib libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 ffmpeg zstd gtest mongoose libuuid libmagic libraw curl[core,ssl] jbig2dec brotli libmupdf
    ```
 
 2. Build

ci/build_arm64.sh

@@ -0,0 +1,12 @@
+#!/usr/bin/env bash
+
+VCPKG_ROOT="/vcpkg"
+
+rm *.gz
+
+rm -rf CMakeFiles CMakeCache.txt
+cmake -DSIST_DEBUG=off -DCMAKE_TOOLCHAIN_FILE="${VCPKG_ROOT}/scripts/buildsystems/vcpkg.cmake" .
+make -j 4
+strip sist2
+mv sist2 sist2_arm64
+gzip -9 sist2_arm64

docs/USAGE.md

@@ -16,6 +16,7 @@
     * [link to specific indices](#link-to-specific-indices)
 * [exec-script](#exec-script)
 * [tagging](#tagging)
+* [sidecar files](#sidecar-files)
 
 ```
 Usage: sist2 scan [OPTION]... PATH
@@ -52,6 +53,8 @@ Index options
     --es-index=<str>              Elasticsearch index name. DEFAULT=sist2
     -p, --print                   Just print JSON documents to stdout.
     --script-file=<str>           Path to user script.
+    --mappings-file=<str>         Path to Elasticsearch mappings.
+    --settings-file=<str>         Path to Elasticsearch settings.
     --async-script                Execute user script asynchronously.
     --batch-size=<int>            Index batch size. DEFAULT: 100
     -f, --force-reset             Reset Elasticsearch mappings and settings. (You must use this option the first time you use the index command)
@@ -153,10 +156,13 @@ documents.idx/
 ├── agg_mime.csv
 ├── agg_date.csv
 ├── add_size.csv
-├── thumbs
+├── thumbs/
 |   ├── data.mdb
 |   └── lock.mdb
-└── tags
+├── tags/
+|   ├── data.mdb
+|   └── lock.mdb
+└── meta/
     ├── data.mdb
     └── lock.mdb
 ```
@@ -183,9 +189,11 @@ by a third party application. The 'external' index must have the following forma
 my_index/
 ├── descriptor.json
 ├── _index_0
-└── thumbs
-    ├── data.mdb
-    └── lock.mdb
+└── thumbs/
+|   ├── data.mdb
+|   └── lock.mdb
+└── meta/
+    └── <empty>
 ```
 
 *descriptor.json*:
@@ -251,6 +259,10 @@ it is currently unsupported and has no guaranties of back/forward compatibility.
     Print index in JSON format to stdout.
  * `--script-file` 
     Path to user script. See [Scripting](scripting.md).
+ * `--mappings-file`
+    Path to custom Elasticsearch mappings. If none is specified, [the bundled mappings](https://github.com/simon987/sist2/tree/master/schema) will be used.
+ * `--settings-file`
+    Path to custom Elasticsearch settings. *(See above)*
  * `--async-script` 
     Use `wait_for_completion=false` elasticsearch option while executing user script.
      (See [Elasticsearch documentation](https://www.elastic.co/guide/en/elasticsearch/reference/current/tasks.html))
@@ -260,7 +272,6 @@ it is currently unsupported and has no guaranties of back/forward compatibility.
     down the process.
  * `-f, --force-reset` 
     Reset Elasticsearch mappings and settings.
-    **(You must use this option the first time you use the index command)**.
     
 ### Index examples
 
@@ -343,9 +354,48 @@ See [Automatic tagging](#automatic-tagging) for information about tag
  hierarchies and tag colors.
 
 \* *It can take a few seconds to take effect in new search queries, and the page needs 
-    to be reloaded for the tag tab to update*
+    to be reloaded for the tags tab to update*
 
 
 ### Automatic tagging
 
 See [scripting](scripting.md) documentation.
+
+# Sidecar files
+
+When scanning, sist2 will read metadata from `.s2meta` JSON files and overwrite the 
+original document's metadata. Sidecar metadata files will also work inside archives.
+Sidecar files themselves are not saved in the index.
+
+This feature is useful to leverage third-party applications such as speech-to-text or
+OCR to add additional metadata to a file.
+
+**Example**
+
+```
+~/Documents/
+├── Video.mp4
+└── Video.mp4.s2meta
+```
+
+The sidecar file must have exactly the same file path and the `.s2meta` suffix.
+
+`Video.mp4.s2meta`:
+```json
+{
+  "content": "This sidecar file will overwrite some metadata fields of Video.mp4",
+  "author": "Some author",
+  "duration": 12345,
+  "bitrate": 67890,
+  "some_arbitrary_field": [1,2,3]
+}
+```
+
+```
+sist2 scan ~/Documents -o ./docs.idx
+sist2 index ./docs.idx
+```
+
+*NOTE*: It is technically possible to overwrite the `tag` value using sidecar files, however,
+it is not currently possible to restore both manual tags and sidecar tags without user scripts
+while reindexing.

scripts/mime.csv

@@ -449,3 +449,4 @@ image/x-sony-arw, arw
 image/x-sony-sr2, sr2
 image/x-sony-srf, srf
 image/x-epson-erf, erf
+sist2/sidecar, s2meta

scripts/mime.py

@@ -3,6 +3,7 @@ noparse = set()
 ext_in_hash = set()
 
 major_mime = {
+    "sist2": 0,
     "model": 1,
     "example": 2,
     "message": 3,
@@ -122,7 +123,11 @@ def mime_id(mime):
     elif mime in raw:
         mime_id += " | 0x00800000"
     elif mime == "application/x-empty":
+        cnt -= 1
         return "1"
+    elif mime == "sist2/sidecar":
+        cnt -= 1
+        return "2"
     return mime_id
 
 

src/cli.c

@@ -56,6 +56,12 @@ void scan_args_destroy(scan_args_t *args) {
 
 void index_args_destroy(index_args_t *args) {
     //todo
+    if (args->es_mappings_path) {
+        free(args->es_mappings);
+    }
+    if (args->es_settings_path) {
+        free(args->es_settings);
+    }
     free(args);
 }
 
@@ -231,25 +237,25 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv) {
     return 0;
 }
 
-int load_script(const char *script_path, char **dst) {
+int load_external_file(const char *file_path, char **dst) {
     struct stat info;
-    int res = stat(script_path, &info);
+    int res = stat(file_path, &info);
 
     if (res == -1) {
-        fprintf(stderr, "Error opening script file '%s': %s\n", script_path, strerror(errno));
+        LOG_ERRORF("cli.c", "Error opening file '%s': %s\n", file_path, strerror(errno))
         return 1;
     }
 
-    int fd = open(script_path, O_RDONLY);
+    int fd = open(file_path, O_RDONLY);
     if (fd == -1) {
-        fprintf(stderr, "Error opening script file '%s': %s\n", script_path, strerror(errno));
+        LOG_ERRORF("cli.c", "Error opening file '%s': %s\n", file_path, strerror(errno))
         return 1;
     }
 
     *dst = malloc(info.st_size + 1);
     res = read(fd, *dst, info.st_size);
     if (res < 0) {
-        fprintf(stderr, "Error reading script file '%s': %s\n", script_path, strerror(errno));
+        LOG_ERRORF("cli.c", "Error reading file '%s': %s\n", file_path, strerror(errno))
         return 1;
     }
 
@@ -293,7 +299,19 @@ int index_args_validate(index_args_t *args, int argc, const char **argv) {
     }
 
     if (args->script_path != NULL) {
-        if (load_script(args->script_path, &args->script) != 0) {
+        if (load_external_file(args->script_path, &args->script) != 0) {
+            return 1;
+        }
+    }
+
+    if (args->es_settings_path != NULL) {
+        if (load_external_file(args->es_settings_path, &args->es_settings) != 0) {
+            return 1;
+        }
+    }
+
+    if (args->es_mappings_path != NULL) {
+        if (load_external_file(args->es_mappings_path, &args->es_mappings) != 0) {
             return 1;
         }
     }
@@ -309,6 +327,10 @@ int index_args_validate(index_args_t *args, int argc, const char **argv) {
     LOG_DEBUGF("cli.c", "arg async_script=%s", args->async_script)
     LOG_DEBUGF("cli.c", "arg script=%s", args->script)
     LOG_DEBUGF("cli.c", "arg print=%d", args->print)
+    LOG_DEBUGF("cli.c", "arg es_mappings_path=%s", args->es_mappings_path)
+    LOG_DEBUGF("cli.c", "arg es_mappings=%s", args->es_mappings)
+    LOG_DEBUGF("cli.c", "arg es_settings_path=%s", args->es_settings_path)
+    LOG_DEBUGF("cli.c", "arg es_settings=%s", args->es_settings)
     LOG_DEBUGF("cli.c", "arg batch_size=%d", args->batch_size)
     LOG_DEBUGF("cli.c", "arg force_reset=%d", args->force_reset)
 
@@ -445,7 +467,7 @@ int exec_args_validate(exec_args_t *args, int argc, const char **argv) {
         LOG_FATAL("cli.c", "--script-file argument is required");
     }
 
-    if (load_script(args->script_path, &args->script) != 0) {
+    if (load_external_file(args->script_path, &args->script) != 0) {
         return 1;
     }
 

src/cli.h

@@ -39,6 +39,10 @@ typedef struct index_args {
     const char *index_path;
     const char *script_path;
     char *script;
+    const char *es_settings_path;
+    char *es_settings;
+    const char *es_mappings_path;
+    char *es_mappings;
     int print;
     int batch_size;
     int async_script;

src/ctx.h

@@ -63,6 +63,8 @@ typedef struct {
     tpool_t *pool;
     store_t *tag_store;
     GHashTable *tags;
+    store_t *meta_store;
+    GHashTable *meta;
 } IndexCtx_t;
 
 typedef struct {

src/index/elastic.c

@@ -356,7 +356,7 @@ void finish_indexer(char *script, int async_script, char *index_id) {
     free_response(r);
 }
 
-void elastic_init(int force_reset) {
+void elastic_init(int force_reset, const char* user_mappings, const char* user_settings) {
 
     // Check if index exists
     char url[4096];
@@ -392,13 +392,13 @@ void elastic_init(int force_reset) {
         free_response(r);
 
         snprintf(url, sizeof(url), "%s/%s/_settings", IndexCtx.es_url, IndexCtx.es_index);
-        r = web_put(url, settings_json);
-        LOG_INFOF("elastic.c", "Update settings <%d>", r->status_code);
+        r = web_put(url, user_settings ? user_settings : settings_json);
+        LOG_INFOF("elastic.c", "Update user_settings <%d>", r->status_code);
         free_response(r);
 
         snprintf(url, sizeof(url), "%s/%s/_mappings/_doc?include_type_name=true", IndexCtx.es_url, IndexCtx.es_index);
-        r = web_put(url, mappings_json);
-        LOG_INFOF("elastic.c", "Update mappings <%d>", r->status_code);
+        r = web_put(url, user_mappings ? user_mappings : mappings_json);
+        LOG_INFOF("elastic.c", "Update user_mappings <%d>", r->status_code);
         free_response(r);
 
         snprintf(url, sizeof(url), "%s/%s/_open", IndexCtx.es_url, IndexCtx.es_index);

src/index/elastic.h

@@ -25,7 +25,7 @@ es_indexer_t *create_indexer(const char *url, const char *index);
 void elastic_cleanup();
 void finish_indexer(char *script, int async_script, char *index_id);
 
-void elastic_init(int force_reset);
+void elastic_init(int force_reset, const char* user_mappings, const char* user_settings);
 
 cJSON *elastic_get_document(const char *uuid_str);
 

src/index/web.c

@@ -1,5 +1,6 @@
 #include "web.h"
 #include "src/sist.h"
+#include "src/ctx.h"
 
 #include <mongoose.h>
 #include <pthread.h>
@@ -21,95 +22,82 @@ void free_response(response_t *resp) {
     free(resp);
 }
 
-#define SIST2_HEADERS "User-Agent: sist2\r\nContent-Type: application/json\r\n"
+void web_post_async_poll(subreq_ctx_t* req) {
+    fd_set fdread;
+    fd_set fdwrite;
+    fd_set fdexcep;
+    int maxfd = -1;
 
+    FD_ZERO(&fdread);
+    FD_ZERO(&fdwrite);
+    FD_ZERO(&fdexcep);
 
-void http_req_ev(struct mg_connection *nc, int ev, void *ptr) {
+    CURLMcode mc = curl_multi_fdset(req->multi, &fdread, &fdwrite, &fdexcep, &maxfd);
 
-    http_ev_data_t *ev_data = (http_ev_data_t *) nc->user_data;
+    if(mc != CURLM_OK) {
+        req->done = TRUE;
+        return;
+    }
 
-    switch (ev) {
-        case MG_EV_CONNECT: {
-            int connect_status = *(int *) ptr;
-            if (connect_status != 0) {
-                ev_data->done = TRUE;
-                ev_data->resp->status_code = 0;
-            }
+    if (maxfd == -1) {
+        // no fds ready yet
+        return;
+    }
+
+    struct timeval timeout = {1, 0};
+    int rc = select(maxfd + 1, &fdread, &fdwrite, &fdexcep, &timeout);
+
+    switch(rc) {
+        case -1:
+            req->done = TRUE;
             break;
-        }
-        case MG_EV_HTTP_REPLY: {
-            struct http_message *hm = (struct http_message *) ptr;
-
-            //TODO: Check errors?
-
-            ev_data->resp->size = hm->body.len;
-            ev_data->resp->status_code = hm->resp_code;
-            ev_data->resp->body = malloc(hm->body.len + 1);
-            memcpy(ev_data->resp->body, hm->body.p, hm->body.len);
-            *(ev_data->resp->body + hm->body.len) = '\0';
-
-            ev_data->done = TRUE;
+        case 0:
             break;
-        }
-        case MG_EV_CLOSE: {
-            ev_data->done = TRUE;
-            break;
-        }
         default:
+            curl_multi_perform(req->multi, &req->running_handles);
             break;
     }
+
+    if (req->running_handles == 0) {
+        req->done = TRUE;
+        req->response->body = req->response_buf.buf;
+        req->response->size = req->response_buf.cur;
+        curl_easy_getinfo(req->handle, CURLINFO_RESPONSE_CODE, &req->response->status_code);
+
+        curl_multi_cleanup(req->multi);
+        curl_easy_cleanup(req->handle);
+        curl_slist_free_all(req->headers);
+        return;
+    }
 }
 
-subreq_ctx_t *http_req(const char *url, const char *extra_headers, const char *post_data, const char *method) {
+subreq_ctx_t *web_post_async(const char *url, char *data) {
+    subreq_ctx_t *req = calloc(1, sizeof(subreq_ctx_t));
+    req->response = calloc(1, sizeof(response_t));
+    req->data = data;
+    req->response_buf = dyn_buffer_create();
 
-    struct mg_str scheme;
-    struct mg_str user_info;
-    struct mg_str host;
-    unsigned int port;
-    struct mg_str path;
-    struct mg_str query;
-    struct mg_str fragment;
+    req->handle = curl_easy_init();
+    CURL *curl = req->handle;
+    curl_easy_setopt(curl, CURLOPT_URL, url);
+    curl_easy_setopt(curl, CURLOPT_WRITEDATA, (void *) (&req->response_buf));
+    curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_cb);
+    curl_easy_setopt(curl, CURLOPT_POST, 1);
+    curl_easy_setopt(curl, CURLOPT_USERAGENT, "sist2");
 
-    if (post_data == NULL) post_data = "";
-    if (extra_headers == NULL) extra_headers = "";
-    if (path.len == 0) path = mg_mk_str("/");
-    if (host.len == 0) host = mg_mk_str("");
+    struct curl_slist *headers = NULL;
+    headers = curl_slist_append(headers, "Content-Type: application/json");
+    curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 
-    // [scheme://[user_info@]]host[:port][/path][?query][#fragment]
-    mg_parse_uri(mg_mk_str(url), &scheme, &user_info, &host, &port, &path, &query, &fragment);
+    curl_easy_setopt(curl, CURLOPT_POSTFIELDS, data);
 
-    if (query.len > 0) path.len += query.len + 1;
+    req->multi = curl_multi_init();
+    curl_multi_add_handle(req->multi, curl);
+    curl_multi_perform(req->multi, &req->running_handles);
 
-    subreq_ctx_t *ctx = malloc(sizeof(subreq_ctx_t));
-    mg_mgr_init(&ctx->mgr, NULL);
+    LOG_DEBUGF("web.c", "async request POST %s", url)
 
-    char address[8192];
-    snprintf(address, sizeof(address), "tcp://%.*s:%u", (int) host.len, host.p, port);
-    struct mg_connection *nc = mg_connect(&ctx->mgr, address, http_req_ev);
-    nc->user_data = &ctx->ev_data;
-    mg_set_protocol_http_websocket(nc);
-
-    ctx->ev_data.resp = calloc(1, sizeof(response_t));
-    ctx->ev_data.done = FALSE;
-
-    mg_printf(
-            nc, "%s %.*s HTTP/1.1\r\n"
-                  "Host: %.*s\r\n"
-                  "Content-Length: %zu\r\n"
-                  "%s\r\n"
-                  "%s",
-            method, (int) path.len, path.p,
-            (int) (path.p - host.p), host.p,
-            strlen(post_data),
-            extra_headers,
-            post_data
-    );
-
-    return ctx;
-}
-
-subreq_ctx_t *web_post_async(const char *url, const char *data) {
-    return http_req(url, SIST2_HEADERS, data, "POST");
+    return req;
 }
 
 response_t *web_get(const char *url, int timeout) {

src/index/web.h

@@ -3,6 +3,7 @@
 
 #include "src/sist.h"
 #include <mongoose.h>
+#include <curl/curl.h>
 
 typedef struct response {
     char *body;
@@ -16,13 +17,20 @@ typedef struct {
 } http_ev_data_t;
 
 typedef struct {
-    http_ev_data_t ev_data;
-    struct mg_mgr mgr;
+    char* data;
+    dyn_buffer_t response_buf;
+    struct curl_slist *headers;
+    CURL *handle;
+    CURLM *multi;
+    response_t *response;
+    int running_handles;
+    int done;
 } subreq_ctx_t;
 
 response_t *web_get(const char *url, int timeout);
 response_t *web_post(const char * url, const char * data);
-subreq_ctx_t *web_post_async(const char *url, const char *data);
+void web_post_async_poll(subreq_ctx_t* req);
+subreq_ctx_t *web_post_async(const char *url, char *data);
 response_t *web_put(const char *url, const char *data);
 response_t *web_delete(const char *url);
 

src/io/serialize.c

@@ -247,13 +247,8 @@ void read_index_bin(const char *path, const char *index_id, index_func func) {
         }
         dyn_buffer_write_char(&buf, '\0');
 
-        if (IndexCtx.tags != NULL) {
-            const char *tags_string = g_hash_table_lookup(IndexCtx.tags, buf.buf);
-            if (tags_string != NULL) {
-                cJSON *tags_arr = cJSON_Parse(tags_string);
-                cJSON_AddItemToObject(document, "tag", tags_arr);
-            }
-        }
+        char full_filename[PATH_MAX];
+        strcpy(full_filename, buf.buf);
 
         cJSON_AddStringToObject(document, "extension", buf.buf + line.ext);
         if (*(buf.buf + line.ext - 1) == '.') {
@@ -334,8 +329,36 @@ void read_index_bin(const char *path, const char *index_id, index_func func) {
             key = getc(file);
         }
 
+        cJSON *meta_obj = NULL;
+        if (IndexCtx.meta != NULL) {
+            const char *meta_string = g_hash_table_lookup(IndexCtx.meta, full_filename);
+            if (meta_string != NULL) {
+                meta_obj = cJSON_Parse(meta_string);
+
+                cJSON *child;
+                for (child = meta_obj->child; child != NULL; child = child->next) {
+                    char meta_key[4096];
+                    strcpy(meta_key, child->string);
+                    cJSON_DeleteItemFromObject(document, meta_key);
+                    cJSON_AddItemReferenceToObject(document, meta_key, child);
+                }
+            }
+        }
+
+        if (IndexCtx.tags != NULL) {
+            const char *tags_string = g_hash_table_lookup(IndexCtx.tags, full_filename);
+            if (tags_string != NULL) {
+                cJSON *tags_arr = cJSON_Parse(tags_string);
+                cJSON_DeleteItemFromObject(document, "tag");
+                cJSON_AddItemToObject(document, "tag", tags_arr);
+            }
+        }
+
         func(document, uuid_str);
         cJSON_Delete(document);
+        if (meta_obj) {
+            cJSON_Delete(meta_obj);
+        }
     }
     dyn_buffer_destroy(&buf);
     fclose(file);

src/io/store.c

@@ -43,9 +43,13 @@ void store_destroy(store_t *store) {
 void store_write(store_t *store, char *key, size_t key_len, char *buf, size_t buf_len) {
 
     if (LogCtx.very_verbose) {
-        char uuid_str[UUID_STR_LEN];
-        uuid_unparse((unsigned char *) key, uuid_str);
-        LOG_DEBUGF("store.c", "Store write {%s} %lu bytes", uuid_str, buf_len)
+        if (key_len == 16) {
+            char uuid_str[UUID_STR_LEN] = {0, };
+            uuid_unparse((unsigned char *) key, uuid_str);
+            LOG_DEBUGF("store.c", "Store write {%s} %lu bytes", uuid_str, buf_len)
+        } else {
+            LOG_DEBUGF("store.c", "Store write {%s} %lu bytes", key, buf_len)
+        }
     }
 
     MDB_val mdb_key;
@@ -136,7 +140,9 @@ GHashTable *store_read_all(store_t *store) {
         count += 1;
     }
 
-    LOG_DEBUGF("store.c", "Read tags for %d documents", count);
+    const char *path;
+    mdb_env_get_path(store->env, &path);
+    LOG_DEBUGF("store.c", "Read %d entries from %s", count, path);
 
     mdb_cursor_close(cur);
     mdb_txn_abort(txn);

src/io/store.h

@@ -8,6 +8,7 @@
 
 #define STORE_SIZE_TN 1024 * 1024 * 5
 #define STORE_SIZE_TAG 1024 * 16
+#define STORE_SIZE_META STORE_SIZE_TAG
 
 typedef struct store_t {
     MDB_dbi dbi;

src/main.c

@@ -21,7 +21,7 @@
 #define EPILOG "Made by simon987 <me@simon987.net>. Released under GPL-3.0"
 
 
-static const char *const Version = "2.8.1";
+static const char *const Version = "2.8.5";
 static const char *const usage[] = {
         "sist2 scan [OPTION]... PATH",
         "sist2 index [OPTION]... INDEX",
@@ -182,6 +182,10 @@ void sist2_scan(scan_args_t *args) {
     mkdir(store_path, S_IWUSR | S_IRUSR | S_IXUSR);
     ScanCtx.index.store = store_create(store_path, STORE_SIZE_TN);
 
+    snprintf(store_path, PATH_MAX, "%smeta", ScanCtx.index.path);
+    mkdir(store_path, S_IWUSR | S_IRUSR | S_IXUSR);
+    ScanCtx.index.meta_store = store_create(store_path, STORE_SIZE_META);
+
     scan_print_header();
 
     if (args->incremental != NULL) {
@@ -263,7 +267,7 @@ void sist2_index(index_args_t *args) {
     IndexCtx.batch_size = args->batch_size;
 
     if (!args->print) {
-        elastic_init(args->force_reset);
+        elastic_init(args->force_reset, args->es_mappings, args->es_settings);
     }
 
     char descriptor_path[PATH_MAX];
@@ -289,6 +293,10 @@ void sist2_index(index_args_t *args) {
     IndexCtx.tag_store = store_create(path_tmp, STORE_SIZE_TAG);
     IndexCtx.tags = store_read_all(IndexCtx.tag_store);
 
+    snprintf(path_tmp, sizeof(path_tmp), "%s/meta", args->index_path);
+    IndexCtx.meta_store = store_create(path_tmp, STORE_SIZE_META);
+    IndexCtx.meta = store_read_all(IndexCtx.meta_store);
+
     index_func f;
     if (args->print) {
         f = print_json;
@@ -438,6 +446,8 @@ int main(int argc, const char *argv[]) {
             OPT_STRING(0, "es-index", &common_es_index, "Elasticsearch index name. DEFAULT=sist2"),
             OPT_BOOLEAN('p', "print", &index_args->print, "Just print JSON documents to stdout."),
             OPT_STRING(0, "script-file", &common_script_path, "Path to user script."),
+            OPT_STRING(0, "mappings-file", &index_args->es_mappings_path, "Path to Elasticsearch mappings."),
+            OPT_STRING(0, "settings-file", &index_args->es_settings_path, "Path to Elasticsearch settings."),
             OPT_BOOLEAN(0, "async-script", &common_async_script, "Execute user script asynchronously."),
             OPT_INTEGER(0, "batch-size", &index_args->batch_size, "Index batch size. DEFAULT: 100"),
             OPT_BOOLEAN('f', "force-reset", &index_args->force_reset, "Reset Elasticsearch mappings and settings. "

src/parsing/mime.h

@@ -6,6 +6,7 @@
 #define MAJOR_MIME(mime_id) (mime_id & 0x000F0000) >> 16
 
 #define MIME_EMPTY 1
+#define MIME_SIST2_SIDECAR 2
 
 #define DONT_PARSE 0x80000000
 #define SHOULD_PARSE(mime_id) (ScanCtx.fast == 0 && (mime_id & DONT_PARSE) != DONT_PARSE && mime_id != 0)

src/parsing/mime_generated.c

@@ -128,333 +128,334 @@ enum mime {
     application_x_dvi=655480,
     application_x_elc=655481,
     application_x_empty=1,
-    application_x_envoy=655483,
-    application_x_esrehber=655484,
-    application_x_excel=655485,
-    application_x_executable=655486,
-    application_x_font_gdos=655487,
-    application_x_font_pf2=655488,
-    application_x_font_pfm=655489,
-    application_x_font_sfn=655490,
-    application_x_font_ttf=655491 | 0x20000000,
-    application_x_fptapplication_x_dbt=655492,
-    application_x_freelance=655493,
-    application_x_gamecube_rom=655494,
-    application_x_gdbm=655495,
-    application_x_gettext_translation=655496,
-    application_x_git=655497,
-    application_x_gsp=655498,
-    application_x_gss=655499,
-    application_x_gtar=655500,
-    application_x_gzip=655501,
-    application_x_hdf=655502,
-    application_x_helpfile=655503,
-    application_x_httpd_imap=655504,
-    application_x_ima=655505,
-    application_x_innosetup=655506,
-    application_x_internett_signup=655507,
-    application_x_inventor=655508,
-    application_x_ip2=655509,
-    application_x_java_applet=655510,
-    application_x_java_commerce=655511,
-    application_x_java_image=655512,
-    application_x_java_jmod=655513,
-    application_x_java_keystore=655514,
-    application_x_kdelnk=655515,
-    application_x_koan=655516,
-    application_x_latex=655517,
-    application_x_livescreen=655518,
-    application_x_lotus=655519,
-    application_x_lz4=655520 | 0x08000000,
-    application_x_lz4_json=655521,
-    application_x_lzh=655522,
-    application_x_lzh_compressed=655523,
-    application_x_lzip=655524 | 0x08000000,
-    application_x_lzma=655525 | 0x08000000,
-    application_x_lzop=655526 | 0x08000000,
-    application_x_lzx=655527,
-    application_x_mach_binary=655528,
-    application_x_mach_executable=655529,
-    application_x_magic_cap_package_1_0=655530,
-    application_x_mathcad=655531,
-    application_x_maxis_dbpf=655532,
-    application_x_meme=655533,
-    application_x_midi=655534,
-    application_x_mif=655535,
-    application_x_mix_transfer=655536,
-    application_x_mobipocket_ebook=655537 | 0x02000000,
-    application_x_ms_compress_szdd=655538,
-    application_x_ms_pdb=655539,
-    application_x_ms_reader=655540,
-    application_x_msaccess=655541,
-    application_x_n64_rom=655542,
-    application_x_navi_animation=655543,
-    application_x_navidoc=655544,
-    application_x_navimap=655545,
-    application_x_navistyle=655546,
-    application_x_nes_rom=655547,
-    application_x_netcdf=655548,
-    application_x_newton_compatible_pkg=655549,
-    application_x_nintendo_ds_rom=655550,
-    application_x_object=655551,
-    application_x_omc=655552,
-    application_x_omcdatamaker=655553,
-    application_x_omcregerator=655554,
-    application_x_pagemaker=655555,
-    application_x_pcl=655556,
-    application_x_pgp_keyring=655557,
-    application_x_pixclscript=655558,
-    application_x_pkcs7_certreqresp=655559,
-    application_x_pkcs7_signature=655560,
-    application_x_project=655561,
-    application_x_qpro=655562,
-    application_x_rar=655563 | 0x10000000,
-    application_x_rpm=655564,
-    application_x_sdp=655565,
-    application_x_sea=655566,
-    application_x_seelogo=655567,
-    application_x_setupscript=655568,
-    application_x_shar=655569,
-    application_x_sharedlib=655570,
-    application_x_shockwave_flash=655571,
-    application_x_snappy_framed=655572,
-    application_x_sprite=655573,
-    application_x_sqlite3=655574,
-    application_x_stargallery_thm=655575,
-    application_x_stuffit=655576,
-    application_x_sv4cpio=655577,
-    application_x_sv4crc=655578,
-    application_x_tar=655579 | 0x10000000,
-    application_x_tbook=655580,
-    application_x_terminfo=655581,
-    application_x_terminfo2=655582,
-    application_x_tex_tfm=655583,
-    application_x_texinfo=655584,
-    application_x_ustar=655585,
-    application_x_visio=655586,
-    application_x_vnd_audioexplosion_mzz=655587,
-    application_x_vnd_ls_xpix=655588,
-    application_x_vrml=655589,
-    application_x_wais_source=655590,
-    application_x_wine_extension_ini=655591,
-    application_x_wintalk=655592,
-    application_x_world=655593,
-    application_x_wri=655594,
-    application_x_x509_ca_cert=655595,
-    application_x_xz=655596 | 0x08000000,
-    application_x_zip=655597,
-    application_x_zstd=655598 | 0x08000000,
-    application_x_zstd_dictionary=655599,
-    application_xml=655600,
-    application_zip=655601 | 0x10000000,
-    application_zlib=655602,
-    audio_basic=458995 | 0x80000000,
-    audio_it=458996,
-    audio_make=458997,
-    audio_mid=458998,
-    audio_midi=458999,
-    audio_mp4=459000,
-    audio_mpeg=459001,
-    audio_ogg=459002,
-    audio_s3m=459003,
-    audio_tsp_audio=459004,
-    audio_tsplayer=459005,
-    audio_vnd_qcelp=459006,
-    audio_voxware=459007,
-    audio_x_aiff=459008,
-    audio_x_flac=459009,
-    audio_x_gsm=459010,
-    audio_x_hx_aac_adts=459011,
-    audio_x_jam=459012,
-    audio_x_liveaudio=459013,
-    audio_x_m4a=459014,
-    audio_x_midi=459015,
-    audio_x_mod=459016,
-    audio_x_mp4a_latm=459017,
-    audio_x_mpeg_3=459018,
-    audio_x_mpequrl=459019,
-    audio_x_nspaudio=459020,
-    audio_x_pn_realaudio=459021,
-    audio_x_psid=459022,
-    audio_x_realaudio=459023,
-    audio_x_s3m=459024,
-    audio_x_twinvq=459025,
-    audio_x_twinvq_plugin=459026,
-    audio_x_voc=459027,
-    audio_x_wav=459028,
-    audio_x_xbox_executable=459029 | 0x80000000,
-    audio_x_xbox360_executable=459030 | 0x80000000,
-    audio_xm=459031,
-    font_otf=327960 | 0x20000000,
-    font_sfnt=327961 | 0x20000000,
-    font_woff=327962 | 0x20000000,
-    font_woff2=327963 | 0x20000000,
-    image_bmp=524572,
-    image_cmu_raster=524573,
-    image_fif=524574,
-    image_florian=524575,
-    image_g3fax=524576,
-    image_gif=524577,
-    image_heic=524578,
-    image_ief=524579,
-    image_jpeg=524580,
-    image_jutvision=524581,
-    image_naplps=524582,
-    image_pict=524583,
-    image_png=524584,
-    image_svg=524585 | 0x80000000,
-    image_svg_xml=524586 | 0x80000000,
-    image_tiff=524587,
-    image_vnd_adobe_photoshop=524588 | 0x80000000,
-    image_vnd_djvu=524589 | 0x80000000,
-    image_vnd_fpx=524590,
-    image_vnd_microsoft_icon=524591,
-    image_vnd_rn_realflash=524592,
-    image_vnd_rn_realpix=524593,
-    image_vnd_wap_wbmp=524594,
-    image_vnd_xiff=524595,
-    image_webp=524596,
-    image_wmf=524597,
-    image_x_3ds=524598,
-    image_x_adobe_dng=524599 | 0x00800000,
-    image_x_award_bioslogo=524600,
-    image_x_canon_cr2=524601 | 0x00800000,
-    image_x_canon_crw=524602 | 0x00800000,
-    image_x_cmu_raster=524603,
-    image_x_cur=524604,
-    image_x_dcraw=524605 | 0x00800000,
-    image_x_dwg=524606,
-    image_x_eps=524607,
-    image_x_epson_erf=524608 | 0x00800000,
-    image_x_exr=524609,
-    image_x_fuji_raf=524610 | 0x00800000,
-    image_x_gem=524611,
-    image_x_icns=524612,
-    image_x_icon=524613 | 0x80000000,
-    image_x_jg=524614,
-    image_x_jps=524615,
-    image_x_kodak_dcr=524616 | 0x00800000,
-    image_x_kodak_k25=524617 | 0x00800000,
-    image_x_kodak_kdc=524618 | 0x00800000,
-    image_x_minolta_mrw=524619 | 0x00800000,
-    image_x_ms_bmp=524620,
-    image_x_niff=524621,
-    image_x_nikon_nef=524622 | 0x00800000,
-    image_x_olympus_orf=524623 | 0x00800000,
-    image_x_panasonic_raw=524624 | 0x00800000,
-    image_x_pcx=524625,
-    image_x_pentax_pef=524626 | 0x00800000,
-    image_x_pict=524627,
-    image_x_portable_bitmap=524628,
-    image_x_portable_graymap=524629,
-    image_x_portable_pixmap=524630,
-    image_x_quicktime=524631,
-    image_x_rgb=524632,
-    image_x_sigma_x3f=524633 | 0x00800000,
-    image_x_sony_arw=524634 | 0x00800000,
-    image_x_sony_sr2=524635 | 0x00800000,
-    image_x_sony_srf=524636 | 0x00800000,
-    image_x_tga=524637,
-    image_x_tiff=524638,
-    image_x_win_bitmap=524639,
-    image_x_xcf=524640 | 0x80000000,
-    image_x_xpixmap=524641 | 0x80000000,
-    image_x_xwindowdump=524642,
-    message_news=196963,
-    message_rfc822=196964,
-    model_vnd_dwf=65893,
-    model_vnd_gdl=65894,
-    model_vnd_gs_gdl=65895,
-    model_vrml=65896,
-    model_x_pov=65897,
-    text_PGP=590186,
-    text_asp=590187,
-    text_css=590188,
-    text_html=590189 | 0x01000000,
-    text_javascript=590190,
-    text_mcf=590191,
-    text_pascal=590192,
-    text_plain=590193,
-    text_richtext=590194,
-    text_rtf=590195,
-    text_scriplet=590196,
-    text_tab_separated_values=590197,
-    text_troff=590198,
-    text_uri_list=590199,
-    text_vnd_abc=590200,
-    text_vnd_fmi_flexstor=590201,
-    text_vnd_wap_wml=590202,
-    text_vnd_wap_wmlscript=590203,
-    text_webviewhtml=590204,
-    text_x_Algol68=590205,
-    text_x_asm=590206,
-    text_x_audiosoft_intra=590207,
-    text_x_awk=590208,
-    text_x_bcpl=590209,
-    text_x_c=590210,
-    text_x_c__=590211,
-    text_x_component=590212,
-    text_x_diff=590213,
-    text_x_fortran=590214,
-    text_x_java=590215,
-    text_x_la_asf=590216,
-    text_x_lisp=590217,
-    text_x_m=590218,
-    text_x_m4=590219,
-    text_x_makefile=590220,
-    text_x_ms_regedit=590221,
-    text_x_msdos_batch=590222,
-    text_x_objective_c=590223,
-    text_x_pascal=590224,
-    text_x_perl=590225,
-    text_x_php=590226,
-    text_x_po=590227,
-    text_x_python=590228,
-    text_x_ruby=590229,
-    text_x_sass=590230,
-    text_x_scss=590231,
-    text_x_server_parsed_html=590232,
-    text_x_setext=590233,
-    text_x_sgml=590234 | 0x01000000,
-    text_x_shellscript=590235,
-    text_x_speech=590236,
-    text_x_tcl=590237,
-    text_x_tex=590238,
-    text_x_uil=590239,
-    text_x_uuencode=590240,
-    text_x_vcalendar=590241,
-    text_x_vcard=590242,
-    text_xml=590243 | 0x01000000,
-    video_MP2T=393636,
-    video_animaflex=393637,
-    video_avi=393638,
-    video_avs_video=393639,
-    video_mp4=393640,
-    video_mpeg=393641,
-    video_quicktime=393642,
-    video_vdo=393643,
-    video_vivo=393644,
-    video_vnd_rn_realvideo=393645,
-    video_vosaic=393646,
-    video_webm=393647,
-    video_x_amt_demorun=393648,
-    video_x_amt_showrun=393649,
-    video_x_atomic3d_feature=393650,
-    video_x_dl=393651,
-    video_x_dv=393652,
-    video_x_fli=393653,
-    video_x_flv=393654,
-    video_x_isvideo=393655,
-    video_x_jng=393656 | 0x80000000,
-    video_x_m4v=393657,
-    video_x_matroska=393658,
-    video_x_mng=393659,
-    video_x_motion_jpeg=393660,
-    video_x_ms_asf=393661,
-    video_x_msvideo=393662,
-    video_x_qtc=393663,
-    video_x_sgi_movie=393664,
-    x_epoc_x_sisx_app=721345,
+    application_x_envoy=655482,
+    application_x_esrehber=655483,
+    application_x_excel=655484,
+    application_x_executable=655485,
+    application_x_font_gdos=655486,
+    application_x_font_pf2=655487,
+    application_x_font_pfm=655488,
+    application_x_font_sfn=655489,
+    application_x_font_ttf=655490 | 0x20000000,
+    application_x_fptapplication_x_dbt=655491,
+    application_x_freelance=655492,
+    application_x_gamecube_rom=655493,
+    application_x_gdbm=655494,
+    application_x_gettext_translation=655495,
+    application_x_git=655496,
+    application_x_gsp=655497,
+    application_x_gss=655498,
+    application_x_gtar=655499,
+    application_x_gzip=655500,
+    application_x_hdf=655501,
+    application_x_helpfile=655502,
+    application_x_httpd_imap=655503,
+    application_x_ima=655504,
+    application_x_innosetup=655505,
+    application_x_internett_signup=655506,
+    application_x_inventor=655507,
+    application_x_ip2=655508,
+    application_x_java_applet=655509,
+    application_x_java_commerce=655510,
+    application_x_java_image=655511,
+    application_x_java_jmod=655512,
+    application_x_java_keystore=655513,
+    application_x_kdelnk=655514,
+    application_x_koan=655515,
+    application_x_latex=655516,
+    application_x_livescreen=655517,
+    application_x_lotus=655518,
+    application_x_lz4=655519 | 0x08000000,
+    application_x_lz4_json=655520,
+    application_x_lzh=655521,
+    application_x_lzh_compressed=655522,
+    application_x_lzip=655523 | 0x08000000,
+    application_x_lzma=655524 | 0x08000000,
+    application_x_lzop=655525 | 0x08000000,
+    application_x_lzx=655526,
+    application_x_mach_binary=655527,
+    application_x_mach_executable=655528,
+    application_x_magic_cap_package_1_0=655529,
+    application_x_mathcad=655530,
+    application_x_maxis_dbpf=655531,
+    application_x_meme=655532,
+    application_x_midi=655533,
+    application_x_mif=655534,
+    application_x_mix_transfer=655535,
+    application_x_mobipocket_ebook=655536 | 0x02000000,
+    application_x_ms_compress_szdd=655537,
+    application_x_ms_pdb=655538,
+    application_x_ms_reader=655539,
+    application_x_msaccess=655540,
+    application_x_n64_rom=655541,
+    application_x_navi_animation=655542,
+    application_x_navidoc=655543,
+    application_x_navimap=655544,
+    application_x_navistyle=655545,
+    application_x_nes_rom=655546,
+    application_x_netcdf=655547,
+    application_x_newton_compatible_pkg=655548,
+    application_x_nintendo_ds_rom=655549,
+    application_x_object=655550,
+    application_x_omc=655551,
+    application_x_omcdatamaker=655552,
+    application_x_omcregerator=655553,
+    application_x_pagemaker=655554,
+    application_x_pcl=655555,
+    application_x_pgp_keyring=655556,
+    application_x_pixclscript=655557,
+    application_x_pkcs7_certreqresp=655558,
+    application_x_pkcs7_signature=655559,
+    application_x_project=655560,
+    application_x_qpro=655561,
+    application_x_rar=655562 | 0x10000000,
+    application_x_rpm=655563,
+    application_x_sdp=655564,
+    application_x_sea=655565,
+    application_x_seelogo=655566,
+    application_x_setupscript=655567,
+    application_x_shar=655568,
+    application_x_sharedlib=655569,
+    application_x_shockwave_flash=655570,
+    application_x_snappy_framed=655571,
+    application_x_sprite=655572,
+    application_x_sqlite3=655573,
+    application_x_stargallery_thm=655574,
+    application_x_stuffit=655575,
+    application_x_sv4cpio=655576,
+    application_x_sv4crc=655577,
+    application_x_tar=655578 | 0x10000000,
+    application_x_tbook=655579,
+    application_x_terminfo=655580,
+    application_x_terminfo2=655581,
+    application_x_tex_tfm=655582,
+    application_x_texinfo=655583,
+    application_x_ustar=655584,
+    application_x_visio=655585,
+    application_x_vnd_audioexplosion_mzz=655586,
+    application_x_vnd_ls_xpix=655587,
+    application_x_vrml=655588,
+    application_x_wais_source=655589,
+    application_x_wine_extension_ini=655590,
+    application_x_wintalk=655591,
+    application_x_world=655592,
+    application_x_wri=655593,
+    application_x_x509_ca_cert=655594,
+    application_x_xz=655595 | 0x08000000,
+    application_x_zip=655596,
+    application_x_zstd=655597 | 0x08000000,
+    application_x_zstd_dictionary=655598,
+    application_xml=655599,
+    application_zip=655600 | 0x10000000,
+    application_zlib=655601,
+    audio_basic=458994 | 0x80000000,
+    audio_it=458995,
+    audio_make=458996,
+    audio_mid=458997,
+    audio_midi=458998,
+    audio_mp4=458999,
+    audio_mpeg=459000,
+    audio_ogg=459001,
+    audio_s3m=459002,
+    audio_tsp_audio=459003,
+    audio_tsplayer=459004,
+    audio_vnd_qcelp=459005,
+    audio_voxware=459006,
+    audio_x_aiff=459007,
+    audio_x_flac=459008,
+    audio_x_gsm=459009,
+    audio_x_hx_aac_adts=459010,
+    audio_x_jam=459011,
+    audio_x_liveaudio=459012,
+    audio_x_m4a=459013,
+    audio_x_midi=459014,
+    audio_x_mod=459015,
+    audio_x_mp4a_latm=459016,
+    audio_x_mpeg_3=459017,
+    audio_x_mpequrl=459018,
+    audio_x_nspaudio=459019,
+    audio_x_pn_realaudio=459020,
+    audio_x_psid=459021,
+    audio_x_realaudio=459022,
+    audio_x_s3m=459023,
+    audio_x_twinvq=459024,
+    audio_x_twinvq_plugin=459025,
+    audio_x_voc=459026,
+    audio_x_wav=459027,
+    audio_x_xbox_executable=459028 | 0x80000000,
+    audio_x_xbox360_executable=459029 | 0x80000000,
+    audio_xm=459030,
+    font_otf=327959 | 0x20000000,
+    font_sfnt=327960 | 0x20000000,
+    font_woff=327961 | 0x20000000,
+    font_woff2=327962 | 0x20000000,
+    image_bmp=524571,
+    image_cmu_raster=524572,
+    image_fif=524573,
+    image_florian=524574,
+    image_g3fax=524575,
+    image_gif=524576,
+    image_heic=524577,
+    image_ief=524578,
+    image_jpeg=524579,
+    image_jutvision=524580,
+    image_naplps=524581,
+    image_pict=524582,
+    image_png=524583,
+    image_svg=524584 | 0x80000000,
+    image_svg_xml=524585 | 0x80000000,
+    image_tiff=524586,
+    image_vnd_adobe_photoshop=524587 | 0x80000000,
+    image_vnd_djvu=524588 | 0x80000000,
+    image_vnd_fpx=524589,
+    image_vnd_microsoft_icon=524590,
+    image_vnd_rn_realflash=524591,
+    image_vnd_rn_realpix=524592,
+    image_vnd_wap_wbmp=524593,
+    image_vnd_xiff=524594,
+    image_webp=524595,
+    image_wmf=524596,
+    image_x_3ds=524597,
+    image_x_adobe_dng=524598 | 0x00800000,
+    image_x_award_bioslogo=524599,
+    image_x_canon_cr2=524600 | 0x00800000,
+    image_x_canon_crw=524601 | 0x00800000,
+    image_x_cmu_raster=524602,
+    image_x_cur=524603,
+    image_x_dcraw=524604 | 0x00800000,
+    image_x_dwg=524605,
+    image_x_eps=524606,
+    image_x_epson_erf=524607 | 0x00800000,
+    image_x_exr=524608,
+    image_x_fuji_raf=524609 | 0x00800000,
+    image_x_gem=524610,
+    image_x_icns=524611,
+    image_x_icon=524612 | 0x80000000,
+    image_x_jg=524613,
+    image_x_jps=524614,
+    image_x_kodak_dcr=524615 | 0x00800000,
+    image_x_kodak_k25=524616 | 0x00800000,
+    image_x_kodak_kdc=524617 | 0x00800000,
+    image_x_minolta_mrw=524618 | 0x00800000,
+    image_x_ms_bmp=524619,
+    image_x_niff=524620,
+    image_x_nikon_nef=524621 | 0x00800000,
+    image_x_olympus_orf=524622 | 0x00800000,
+    image_x_panasonic_raw=524623 | 0x00800000,
+    image_x_pcx=524624,
+    image_x_pentax_pef=524625 | 0x00800000,
+    image_x_pict=524626,
+    image_x_portable_bitmap=524627,
+    image_x_portable_graymap=524628,
+    image_x_portable_pixmap=524629,
+    image_x_quicktime=524630,
+    image_x_rgb=524631,
+    image_x_sigma_x3f=524632 | 0x00800000,
+    image_x_sony_arw=524633 | 0x00800000,
+    image_x_sony_sr2=524634 | 0x00800000,
+    image_x_sony_srf=524635 | 0x00800000,
+    image_x_tga=524636,
+    image_x_tiff=524637,
+    image_x_win_bitmap=524638,
+    image_x_xcf=524639 | 0x80000000,
+    image_x_xpixmap=524640 | 0x80000000,
+    image_x_xwindowdump=524641,
+    message_news=196962,
+    message_rfc822=196963,
+    model_vnd_dwf=65892,
+    model_vnd_gdl=65893,
+    model_vnd_gs_gdl=65894,
+    model_vrml=65895,
+    model_x_pov=65896,
+    sist2_sidecar=2,
+    text_PGP=590185,
+    text_asp=590186,
+    text_css=590187,
+    text_html=590188 | 0x01000000,
+    text_javascript=590189,
+    text_mcf=590190,
+    text_pascal=590191,
+    text_plain=590192,
+    text_richtext=590193,
+    text_rtf=590194,
+    text_scriplet=590195,
+    text_tab_separated_values=590196,
+    text_troff=590197,
+    text_uri_list=590198,
+    text_vnd_abc=590199,
+    text_vnd_fmi_flexstor=590200,
+    text_vnd_wap_wml=590201,
+    text_vnd_wap_wmlscript=590202,
+    text_webviewhtml=590203,
+    text_x_Algol68=590204,
+    text_x_asm=590205,
+    text_x_audiosoft_intra=590206,
+    text_x_awk=590207,
+    text_x_bcpl=590208,
+    text_x_c=590209,
+    text_x_c__=590210,
+    text_x_component=590211,
+    text_x_diff=590212,
+    text_x_fortran=590213,
+    text_x_java=590214,
+    text_x_la_asf=590215,
+    text_x_lisp=590216,
+    text_x_m=590217,
+    text_x_m4=590218,
+    text_x_makefile=590219,
+    text_x_ms_regedit=590220,
+    text_x_msdos_batch=590221,
+    text_x_objective_c=590222,
+    text_x_pascal=590223,
+    text_x_perl=590224,
+    text_x_php=590225,
+    text_x_po=590226,
+    text_x_python=590227,
+    text_x_ruby=590228,
+    text_x_sass=590229,
+    text_x_scss=590230,
+    text_x_server_parsed_html=590231,
+    text_x_setext=590232,
+    text_x_sgml=590233 | 0x01000000,
+    text_x_shellscript=590234,
+    text_x_speech=590235,
+    text_x_tcl=590236,
+    text_x_tex=590237,
+    text_x_uil=590238,
+    text_x_uuencode=590239,
+    text_x_vcalendar=590240,
+    text_x_vcard=590241,
+    text_xml=590242 | 0x01000000,
+    video_MP2T=393635,
+    video_animaflex=393636,
+    video_avi=393637,
+    video_avs_video=393638,
+    video_mp4=393639,
+    video_mpeg=393640,
+    video_quicktime=393641,
+    video_vdo=393642,
+    video_vivo=393643,
+    video_vnd_rn_realvideo=393644,
+    video_vosaic=393645,
+    video_webm=393646,
+    video_x_amt_demorun=393647,
+    video_x_amt_showrun=393648,
+    video_x_atomic3d_feature=393649,
+    video_x_dl=393650,
+    video_x_dv=393651,
+    video_x_fli=393652,
+    video_x_flv=393653,
+    video_x_isvideo=393654,
+    video_x_jng=393655 | 0x80000000,
+    video_x_m4v=393656,
+    video_x_matroska=393657,
+    video_x_mng=393658,
+    video_x_motion_jpeg=393659,
+    video_x_ms_asf=393660,
+    video_x_msvideo=393661,
+    video_x_qtc=393662,
+    video_x_sgi_movie=393663,
+    x_epoc_x_sisx_app=721344,
 };
 char *mime_get_mime_text(unsigned int mime_id) {switch (mime_id) {
 case application_arj: return "application/arj";
@@ -906,6 +907,7 @@ case image_x_sony_arw: return "image/x-sony-arw";
 case image_x_sony_sr2: return "image/x-sony-sr2";
 case image_x_sony_srf: return "image/x-sony-srf";
 case image_x_epson_erf: return "image/x-epson-erf";
+case sist2_sidecar: return "sist2/sidecar";
 default: return NULL;}}
 GHashTable *mime_get_ext_table() {GHashTable *ext_table = g_hash_table_new(g_str_hash, g_str_equal);
 g_hash_table_insert(ext_table, "arj", (gpointer)application_arj);
@@ -1449,6 +1451,7 @@ g_hash_table_insert(ext_table, "arw", (gpointer)image_x_sony_arw);
 g_hash_table_insert(ext_table, "sr2", (gpointer)image_x_sony_sr2);
 g_hash_table_insert(ext_table, "srf", (gpointer)image_x_sony_srf);
 g_hash_table_insert(ext_table, "erf", (gpointer)image_x_epson_erf);
+g_hash_table_insert(ext_table, "s2meta", (gpointer)sist2_sidecar);
 return ext_table;}
 GHashTable *mime_get_mime_table() {GHashTable *mime_table = g_hash_table_new(g_str_hash, g_str_equal);
 g_hash_table_insert(mime_table, "application/arj", (gpointer)application_arj);
@@ -1900,5 +1903,6 @@ g_hash_table_insert(mime_table, "image/x-sony-arw", (gpointer)image_x_sony_arw);
 g_hash_table_insert(mime_table, "image/x-sony-sr2", (gpointer)image_x_sony_sr2);
 g_hash_table_insert(mime_table, "image/x-sony-srf", (gpointer)image_x_sony_srf);
 g_hash_table_insert(mime_table, "image/x-epson-erf", (gpointer)image_x_epson_erf);
+g_hash_table_insert(mime_table, "sist2/sidecar", (gpointer)sist2_sidecar);
 return mime_table;}
 #endif

src/parsing/parse.c

@@ -4,6 +4,7 @@
 #include "src/ctx.h"
 #include "mime.h"
 #include "src/io/serialize.h"
+#include "src/parsing/sidecar.h"
 
 #include <magic.h>
 
@@ -157,6 +158,10 @@ void parse(void *arg) {
         parse_comic(&ScanCtx.comic_ctx, &job->vfile, &doc);
     } else if (IS_MOBI(doc.mime)) {
         parse_mobi(&ScanCtx.mobi_ctx, &job->vfile, &doc);
+    } else if (doc.mime == MIME_SIST2_SIDECAR) {
+        parse_sidecar(&job->vfile, &doc);
+        CLOSE_FILE(job->vfile)
+        return;
     }
 
     abort:

src/parsing/sidecar.c

@@ -0,0 +1,35 @@
+#include "sidecar.h"
+
+#include "src/ctx.h"
+
+void parse_sidecar(vfile_t *vfile, document_t *doc) {
+
+    LOG_DEBUGF("sidecar.c", "Parsing sidecar file %s", vfile->filepath)
+
+    size_t size;
+    char* buf = read_all(vfile, &size);
+    if (buf == NULL) {
+        LOG_ERRORF("sidecar.c", "Read error for %s", vfile->filepath)
+        return;
+    }
+
+    buf = realloc(buf, size + 1);
+    *(buf + size) = '\0';
+
+    cJSON *json = cJSON_Parse(buf);
+    if (json == NULL) {
+        LOG_ERRORF("sidecar.c", "Could not parse JSON sidecar %s", vfile->filepath)
+        return;
+    }
+    char *json_str = cJSON_PrintUnformatted(json);
+
+    char filepath[PATH_MAX];
+    memcpy(filepath, vfile->filepath + ScanCtx.index.desc.root_len, doc->ext - 1 - ScanCtx.index.desc.root_len);
+    *(filepath + doc->ext - 1) = '\0';
+
+    store_write(ScanCtx.index.meta_store, filepath, doc->ext, json_str, strlen(json_str) + 1);
+
+    cJSON_Delete(json);
+    free(json_str);
+    free(buf);
+}

src/parsing/sidecar.h

@@ -0,0 +1,8 @@
+#ifndef SIST2_SIDECAR_H
+#define SIST2_SIDECAR_H
+
+#include "src/sist.h"
+
+void parse_sidecar(vfile_t *vfile, document_t *doc);
+
+#endif

src/static/css/dark.css

@@ -229,6 +229,7 @@ a:hover, .btn:hover {
     max-width: 100%;
     max-height: 400px;
     margin: 0 auto 0;
+    width: auto;
     height: auto;
 }
 
@@ -537,3 +538,7 @@ svg {
 .stat > .card-body {
     padding: 0.7em 1.25em;
 }
+
+#modal-body > .img-wrapper {
+    margin-bottom: 1em;
+}

src/static/css/light.css

@@ -416,4 +416,8 @@ mark {
 
 .stat > .card-body {
     padding: 0.7em 1.25em;
+}
+
+#modal-body > .img-wrapper {
+    margin-bottom: 1em;
 }

src/static/js/dom.js

@@ -195,10 +195,21 @@ function makeUserTag(tag, hit) {
 function infoButtonCb(hit) {
     return () => {
         getDocumentInfo(hit["_id"]).then(doc => {
+            $("#modal-body").empty()
+
             $("#modal-title").text(doc["name"] + ext(hit));
 
+            if (doc["mime"]) {
+                const mimeCategory = doc["mime"].split("/")[0];
+                const imgWrapper = document.createElement("div");
+                imgWrapper.setAttribute("style", "position: relative");
+                imgWrapper.setAttribute("class", "img-wrapper");
+                makeThumbnail(mimeCategory, hit, imgWrapper, false);
+                $("#modal-body").append(imgWrapper);
+            }
+
             const tbody = $("<tbody>");
-            $("#modal-body").empty()
+            $("#modal-body")
                 .append($("<table class='table table-sm'>")
                     .append($("<thead>")
                         .append($("<tr>")
@@ -212,11 +223,16 @@ function infoButtonCb(hit) {
             tbody.append($("<tr>")
                 .append($("<td>").text("index"))
                 .append($("<td>").text(`[${indexMap[doc["index"]]}]`))
+            ).append($("<tr>")
+                .append($("<td>").text("mtime"))
+                .append($("<td>")
+                    .text(new Date(doc["mtime"] * 1000).toISOString().split(".")[0].replace("T", " "))
+                    .attr("title", doc["mtime"]))
             );
             const displayFields = new Set([
-                "mime", "size", "mtime", "path", "title", "width", "height", "duration", "audioc", "videoc",
+                "mime", "size", "path", "title", "width", "height", "duration", "audioc", "videoc",
                 "bitrate", "artist", "album", "album_artist", "genre", "title", "font_name", "tag", "author",
-                "modified_by"
+                "modified_by", "pages"
             ]);
             Object.keys(doc)
                 .filter(key => key.startsWith("_keyword.") || key.startsWith("_text.") || displayFields.has(key) || key.startsWith("exif_"))

src/static/js/search.js

@@ -168,7 +168,7 @@ window.onload = () => {
 };
 
 function saveTag(tag, hit) {
-    const relPath = hit["_source"]["path"] + "/" + hit["_source"]["name"] + ext(hit);
+    const relPath = hit["_source"]["path"] + (hit["_source"]["path"] ? "/" : "") + hit["_source"]["name"] + ext(hit);
 
     return $.jsonPost("/tag/" + hit["_source"]["index"], {
         delete: false,

src/static/search.html

@@ -12,7 +12,7 @@
 
 <nav class="navbar navbar-expand-lg">
     <a class="navbar-brand" href="/">sist2</a>
-    <span class="badge badge-pill version">2.8.1</span>
+    <span class="badge badge-pill version">2.8.5</span>
     <span class="tagline">Lightning-fast file system indexer and search tool </span>
     <a class="btn ml-auto" href="/stats">Stats</a>
     <button class="btn" type="button" data-toggle="modal" data-target="#settings" onclick="loadSettings()">Settings

src/static/stats.html

@@ -10,7 +10,7 @@
 
 <nav class="navbar navbar-expand-lg">
     <a class="navbar-brand" href="/">sist2</a>
-    <span class="badge badge-pill version">2.8.1</span>
+    <span class="badge badge-pill version">2.8.5</span>
     <span class="tagline">Lightning-fast file system indexer and search tool </span>
     <a style="margin-left: auto" class="btn" href="/">Back</a>
     <button class="btn" type="button" data-toggle="modal" data-target="#settings"

src/types.h

@@ -20,6 +20,7 @@ typedef struct index_t {
     struct index_descriptor desc;
     struct store_t *store;
     struct store_t *tag_store;
+    struct store_t *meta_store;
     char path[PATH_MAX];
 } index_t;
 

src/web/serve.c

@@ -10,8 +10,6 @@
 
 #include <mongoose.h>
 
-#define CHUNK_SIZE 1024 * 1024 * 10
-
 
 static int has_prefix(const struct mg_str *str, const struct mg_str *prefix) {
     return str->len > prefix->len && memcmp(str->p, prefix->p, prefix->len) == 0;
@@ -240,7 +238,6 @@ void search(struct mg_connection *nc, struct http_message *hm) {
     snprintf(url, 4096, "%s/%s/_search", WebCtx.es_url, WebCtx.es_index);
 
     nc->user_data = web_post_async(url, body);
-    free(body);
 }
 
 void serve_file_from_url(cJSON *json, index_t *idx, struct mg_connection *nc) {
@@ -538,8 +535,8 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
             }
         }
 
-        char buf[8192];
-        snprintf(buf, sizeof(buf),
+        char *buf = malloc(sizeof(char) * 8192);
+        snprintf(buf, 8192,
                  "{"
                  "    \"script\" : {"
                  "        \"source\": \"if (ctx._source.tag.contains(params.tag)) { ctx._source.tag.remove(ctx._source.tag.indexOf(params.tag)) }\","
@@ -558,8 +555,8 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
     } else {
         cJSON_AddItemToArray(arr, cJSON_CreateString(arg_req->name));
 
-        char buf[8192];
-        snprintf(buf, sizeof(buf),
+        char *buf = malloc(sizeof(char) * 8192);
+        snprintf(buf, 8192,
                  "{"
                  "    \"script\" : {"
                  "        \"source\": \"if(ctx._source.tag == null) {ctx._source.tag = new ArrayList()} ctx._source.tag.add(params.tag)\","
@@ -668,11 +665,11 @@ static void ev_router(struct mg_connection *nc, int ev, void *p) {
         if (nc->user_data != NULL) {
             //Waiting for ES reply
             subreq_ctx_t *ctx = (subreq_ctx_t *) nc->user_data;
-            mg_mgr_poll(&ctx->mgr, 0);
+            web_post_async_poll(ctx);
 
-            if (ctx->ev_data.done == TRUE) {
+            if (ctx->done == TRUE) {
 
-                response_t *r = ctx->ev_data.resp;
+                response_t *r = ctx->response;
 
                 if (r->status_code == 200) {
                     send_response_line(nc, 200, r->size, "Content-Type: application/json");
@@ -695,6 +692,8 @@ static void ev_router(struct mg_connection *nc, int ev, void *p) {
                 }
 
                 free_response(r);
+                free(ctx->data);
+                free(ctx);
                 nc->flags |= MG_F_SEND_AND_CLOSE;
                 nc->user_data = NULL;
             }