Add md5 client-side lib

Fix relative stats URL

.gitignore

@@ -1,6 +1,5 @@
 .idea
 thumbs
-test
 *.cbp
 CMakeCache.txt
 CMakeFiles

CMakeLists.txt

@@ -5,12 +5,16 @@ project(sist2 C)
 
 option(SIST_DEBUG "Build a debug executable" on)
 
+set(BUILD_TESTS on)
 add_subdirectory(third-party/libscan)
 set(ARGPARSE_SHARED off)
 add_subdirectory(third-party/argparse)
 
-add_executable(
+add_executable(sist2
-        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 +29,9 @@ 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)
@@ -39,7 +40,6 @@ find_package(lmdb CONFIG REQUIRED)
 find_package(cJSON CONFIG REQUIRED)
 find_package(unofficial-glib CONFIG REQUIRED)
 find_package(unofficial-mongoose CONFIG REQUIRED)
-find_library(UUID_LIB NAMES uuid)
 find_package(CURL CONFIG REQUIRED)
 
 #find_package(OpenSSL REQUIRED)
@@ -67,13 +67,13 @@ if (SIST_DEBUG)
             -fstack-protector
             -fno-omit-frame-pointer
             -fsanitize=address
-            -O2
+            -fno-inline
+#            -O2
     )
     target_link_options(
             sist2
             PRIVATE
             -fsanitize=address
-#            -static
     )
     set_target_properties(
             sist2
@@ -81,7 +81,6 @@ if (SIST_DEBUG)
             OUTPUT_NAME sist2_debug
     )
 else ()
-#    set(VCPKG_BUILD_TYPE release)
     target_compile_options(
             sist2
             PRIVATE
@@ -106,13 +105,13 @@ target_link_libraries(
         argparse
         unofficial::glib::glib
         unofficial::mongoose::mongoose
-#        OpenSSL::SSL OpenSSL::Crypto
         CURL::libcurl
 
-        ${UUID_LIB}
         pthread
         magic
 
+        c
+
         scan
 )
 

README.md

@@ -2,7 +2,7 @@
 [![CodeFactor](https://www.codefactor.io/repository/github/simon987/sist2/badge?s=05daa325188aac4eae32c786f3d9cf4e0593f822)](https://www.codefactor.io/repository/github/simon987/sist2)
 [![Development snapshots](https://ci.simon987.net/app/rest/builds/buildType(Sist2_Build)/statusIcon)](https://files.simon987.net/artifacts/Sist2/Build/)
 
-**Demo**: [sist2.simon987.net](https://sist2.simon987.net/)
+**Demo**: [sist2.simon987.net](https://sist2.simon987.net/?i=Demo%20files)
 
 # sist2
 
@@ -74,7 +74,7 @@ See [Usage guide](docs/USAGE.md) for more details
 
 File type | Library | Content | Thumbnail | Metadata
 :---|:---|:---|:---|:---
-pdf,xps,fb2,epub | MuPDF | text+ocr | yes | title |
+pdf,xps,fb2,epub | MuPDF | text+ocr | yes | author, title |
 cbz,cbr | *(none)* | - | yes | - |
 `audio/*` | ffmpeg | - | yes | ID3 tags |
 `video/*` | ffmpeg | - | yes | title, comment, artist |
@@ -85,6 +85,7 @@ ttf,ttc,cff,woff,fnt,otf | Freetype2 | - | yes, `bmp` | Name & style |
 html, xml | *(none)* | yes | no | - |
 tar, zip, rar, 7z, ar ...  | Libarchive | yes\* | - | no |
 docx, xlsx, pptx | *(none)* | 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 |
 
 \* *See [Archive files](#archive-files)*
@@ -126,12 +127,12 @@ 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 brotli libmupdf
+   vcpkg install lmdb cjson glib libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 ffmpeg zstd gtest mongoose libmagic libraw curl[core,ssl] jbig2dec brotli libmupdf
    ```
 
 2. Build
     ```bash
     git clone --recursive https://github.com/simon987/sist2/
-    cmake -DCMAKE_TOOLCHAIN_FILE=<VCPKG_ROOT>/scripts/buildsystems/vcpkg.cmake .
+    cmake -DSIST_DEBUG=off -DCMAKE_TOOLCHAIN_FILE=<VCPKG_ROOT>/scripts/buildsystems/vcpkg.cmake .
     make
     ```

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
+└── thumbs/
-    ├── data.mdb
+|   ├── data.mdb
-    └── lock.mdb
+|   └── lock.mdb
+└── meta/
+    └── <empty>
 ```
 
 *descriptor.json*:
@@ -233,9 +241,11 @@ The `_text.*` items will be indexed and searchable as **text** fields (fuzzy sea
 
 *thumbs/*:
 
-LMDB key-value store. Keys are **binary** 128-bit UUID4s (`_id` field)
+LMDB key-value store. Keys are **binary** 16-byte md5 hash* (`_id` field)
 and values are raw image bytes.
 
+*\* Hash is calculated from the full path of the file, including the extension, relative to the index root*
+
 Importing an external `binary` type index is technically possible but
 it is currently unsupported and has no guaranties of back/forward compatibility.
 
@@ -251,6 +261,10 @@ 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 +274,6 @@ 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 +356,48 @@ 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.

schema/mappings.json

@@ -30,6 +30,10 @@
     "mime": {
       "type": "keyword"
     },
+    "parent": {
+      "type": "keyword",
+      "index": false
+    },
     "thumbnail": {
       "type": "keyword",
       "index": false

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

@@ -13,6 +13,7 @@
 #include "libscan/text/text.h"
 #include "libscan/mobi/scan_mobi.h"
 #include "libscan/raw/raw.h"
+#include "libscan/msdoc/msdoc.h"
 #include "src/io/store.h"
 
 #include <glib.h>
@@ -48,6 +49,7 @@ typedef struct {
     scan_text_ctx_t text_ctx;
     scan_mobi_ctx_t mobi_ctx;
     scan_raw_ctx_t raw_ctx;
+    scan_msdoc_ctx_t msdoc_ctx;
 } ScanCtx_t;
 
 typedef struct {
@@ -63,6 +65,8 @@ 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

@@ -30,11 +30,11 @@ void elastic_cleanup() {
     }
 }
 
-void print_json(cJSON *document, const char uuid_str[UUID_STR_LEN]) {
+void print_json(cJSON *document, const char id_str[MD5_STR_LENGTH]) {
 
     cJSON *line = cJSON_CreateObject();
 
-    cJSON_AddStringToObject(line, "_id", uuid_str);
+    cJSON_AddStringToObject(line, "_id", id_str);
     cJSON_AddStringToObject(line, "_index", IndexCtx.es_index);
     cJSON_AddStringToObject(line, "_type", "_doc");
     cJSON_AddItemReferenceToObject(line, "_source", document);
@@ -52,13 +52,13 @@ void index_json_func(void *arg) {
     elastic_index_line(line);
 }
 
-void index_json(cJSON *document, const char uuid_str[UUID_STR_LEN]) {
+void index_json(cJSON *document, const char index_id_str[MD5_STR_LENGTH]) {
     char *json = cJSON_PrintUnformatted(document);
 
     size_t json_len = strlen(json);
     es_bulk_line_t *bulk_line = malloc(sizeof(es_bulk_line_t) + json_len + 2);
     memcpy(bulk_line->line, json, json_len);
-    memcpy(bulk_line->uuid_str, uuid_str, UUID_STR_LEN);
+    memcpy(bulk_line->path_md5_str, index_id_str, MD5_STR_LENGTH);
     *(bulk_line->line + json_len) = '\n';
     *(bulk_line->line + json_len + 1) = '\0';
     bulk_line->next = NULL;
@@ -67,7 +67,7 @@ void index_json(cJSON *document, const char uuid_str[UUID_STR_LEN]) {
     tpool_add_work(IndexCtx.pool, index_json_func, bulk_line);
 }
 
-void execute_update_script(const char *script, int async, const char index_id[UUID_STR_LEN]) {
+void execute_update_script(const char *script, int async, const char index_id[MD5_STR_LENGTH]) {
 
     if (Indexer == NULL) {
         Indexer = create_indexer(IndexCtx.es_url, IndexCtx.es_index);
@@ -129,9 +129,9 @@ void *create_bulk_buffer(int max, int *count, size_t *buf_len) {
     while (line != NULL && *count < max) {
         char action_str[256];
         snprintf(
-                action_str, 256,
+                action_str, sizeof(action_str),
                 "{\"index\":{\"_id\":\"%s\",\"_type\":\"_doc\",\"_index\":\"%s\"}}\n",
-                line->uuid_str, Indexer->es_index
+                line->path_md5_str, Indexer->es_index
         );
 
         size_t action_str_len = strlen(action_str);
@@ -220,7 +220,7 @@ void _elastic_flush(int max) {
     if (r->status_code == 413) {
 
         if (max <= 1) {
-            LOG_ERRORF("elastic.c", "Single document too large, giving up: {%s}", Indexer->line_head->uuid_str)
+            LOG_ERRORF("elastic.c", "Single document too large, giving up: {%s}", Indexer->line_head->path_md5_str)
             free_response(r);
             free(buf);
             delete_queue(1);
@@ -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);
+        r = web_put(url, user_settings ? user_settings : settings_json);
-        LOG_INFOF("elastic.c", "Update settings <%d>", r->status_code);
+        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);
+        r = web_put(url, user_mappings ? user_mappings : mappings_json);
-        LOG_INFOF("elastic.c", "Update mappings <%d>", r->status_code);
+        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);
@@ -408,9 +408,9 @@ void elastic_init(int force_reset) {
     }
 }
 
-cJSON *elastic_get_document(const char *uuid_str) {
+cJSON *elastic_get_document(const char *id_str) {
     char url[4096];
-    snprintf(url, sizeof(url), "%s/%s/_doc/%s", WebCtx.es_url, WebCtx.es_index, uuid_str);
+    snprintf(url, sizeof(url), "%s/%s/_doc/%s", WebCtx.es_url, WebCtx.es_index, id_str);
 
     response_t *r = web_get(url, 3);
     cJSON *json = NULL;

src/index/elastic.h

@@ -5,7 +5,7 @@
 
 typedef struct es_bulk_line {
     struct es_bulk_line *next;
-    char uuid_str[UUID_STR_LEN];
+    char path_md5_str[MD5_STR_LENGTH];
     char line[0];
 } es_bulk_line_t;
 
@@ -16,21 +16,21 @@ typedef struct es_indexer es_indexer_t;
 
 void elastic_index_line(es_bulk_line_t *line);
 
-void print_json(cJSON *document, const char uuid_str[UUID_STR_LEN]);
+void print_json(cJSON *document, const char index_id_str[MD5_STR_LENGTH]);
 
-void index_json(cJSON *document, const char uuid_str[UUID_STR_LEN]);
+void index_json(cJSON *document, const char index_id_str[MD5_STR_LENGTH]);
 
 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);
+cJSON *elastic_get_document(const char *id_str);
 
 char *elastic_get_status();
 
-void execute_update_script(const char *script, int async, const char index_id[UUID_STR_LEN]);
+void execute_update_script(const char *script, int async, const char index_id[MD5_STR_LENGTH]);
 
 #endif

src/io/serialize.c

@@ -6,13 +6,13 @@
 static __thread int index_fd = -1;
 
 typedef struct {
-    unsigned char uuid[16];
+    unsigned char path_md5[MD5_DIGEST_LENGTH];
-    unsigned long ino;
     unsigned long size;
     unsigned int mime;
     int mtime;
     short base;
     short ext;
+    char has_parent;
 } line_t;
 
 void skip_meta(FILE *file) {
@@ -32,7 +32,7 @@ void skip_meta(FILE *file) {
 
 void write_index_descriptor(char *path, index_descriptor_t *desc) {
     cJSON *json = cJSON_CreateObject();
-    cJSON_AddStringToObject(json, "uuid", desc->uuid);
+    cJSON_AddStringToObject(json, "id", desc->id);
     cJSON_AddStringToObject(json, "version", desc->version);
     cJSON_AddStringToObject(json, "root", desc->root);
     cJSON_AddStringToObject(json, "name", desc->name);
@@ -82,7 +82,7 @@ index_descriptor_t read_index_descriptor(char *path) {
     strcpy(descriptor.rewrite_url, cJSON_GetObjectItem(json, "rewrite_url")->valuestring);
     descriptor.root_len = (short) strlen(descriptor.root);
     strcpy(descriptor.version, cJSON_GetObjectItem(json, "version")->valuestring);
-    strcpy(descriptor.uuid, cJSON_GetObjectItem(json, "uuid")->valuestring);
+    strcpy(descriptor.id, cJSON_GetObjectItem(json, "id")->valuestring);
     if (cJSON_GetObjectItem(json, "type") == NULL) {
         strcpy(descriptor.type, INDEX_TYPE_BIN);
     } else {
@@ -219,7 +219,7 @@ void read_index_bin(const char *path, const char *index_id, index_func func) {
     dyn_buffer_t buf = dyn_buffer_create();
 
     FILE *file = fopen(path, "rb");
-    while (1) {
+    while (TRUE) {
         buf.cur = 0;
         size_t _ = fread((void *) &line, 1, sizeof(line_t), file);
         if (feof(file)) {
@@ -229,8 +229,8 @@ void read_index_bin(const char *path, const char *index_id, index_func func) {
         cJSON *document = cJSON_CreateObject();
         cJSON_AddStringToObject(document, "index", index_id);
 
-        char uuid_str[UUID_STR_LEN];
+        char path_md5_str[MD5_STR_LENGTH];
-        uuid_unparse(line.uuid, uuid_str);
+        buf2hex(line.path_md5, sizeof(line.path_md5), path_md5_str);
 
         const char *mime_text = mime_get_mime_text(line.mime);
         if (mime_text == NULL) {
@@ -247,14 +247,6 @@ 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);
-            }
-        }
-
         cJSON_AddStringToObject(document, "extension", buf.buf + line.ext);
         if (*(buf.buf + line.ext - 1) == '.') {
             *(buf.buf + line.ext - 1) = '\0';
@@ -334,8 +326,36 @@ void read_index_bin(const char *path, const char *index_id, index_func func) {
             key = getc(file);
         }
 
-        func(document, uuid_str);
+        cJSON *meta_obj = NULL;
+        if (IndexCtx.meta != NULL) {
+            const char *meta_string = g_hash_table_lookup(IndexCtx.meta, path_md5_str);
+            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, path_md5_str);
+            if (tags_string != NULL) {
+                cJSON *tags_arr = cJSON_Parse(tags_string);
+                cJSON_DeleteItemFromObject(document, "tag");
+                cJSON_AddItemToObject(document, "tag", tags_arr);
+            }
+        }
+
+        func(document, path_md5_str);
         cJSON_Delete(document);
+        if (meta_obj) {
+            cJSON_Delete(meta_obj);
+        }
     }
     dyn_buffer_destroy(&buf);
     fclose(file);
@@ -359,7 +379,7 @@ const char *json_type_array_fields[] = {
 void read_index_json(const char *path, UNUSED(const char *index_id), index_func func) {
 
     FILE *file = fopen(path, "r");
-    while (1) {
+    while (TRUE) {
         char *line = NULL;
         size_t len;
         size_t read = getline(&line, &len, file);
@@ -379,7 +399,7 @@ void read_index_json(const char *path, UNUSED(const char *index_id), index_func
         }
 
         cJSON *document = cJSON_CreateObject();
-        const char *uuid_str = cJSON_GetObjectItem(input, "_id")->valuestring;
+        const char *id_str = cJSON_GetObjectItem(input, "_id")->valuestring;
 
         for (int i = 0; i < (sizeof(json_type_copy_fields) / sizeof(json_type_copy_fields[0])); i++) {
             cJSON *value = cJSON_GetObjectItem(input, json_type_copy_fields[i]);
@@ -407,7 +427,7 @@ void read_index_json(const char *path, UNUSED(const char *index_id), index_func
             }
         }
 
-        func(document, uuid_str);
+        func(document, id_str);
         cJSON_Delete(document);
         cJSON_Delete(input);
 
@@ -415,7 +435,7 @@ void read_index_json(const char *path, UNUSED(const char *index_id), index_func
     fclose(file);
 }
 
-void read_index(const char *path, const char index_id[UUID_STR_LEN], const char *type, index_func func) {
+void read_index(const char *path, const char index_id[MD5_STR_LENGTH], const char *type, index_func func) {
 
     if (strcmp(type, INDEX_TYPE_BIN) == 0) {
         read_index_bin(path, index_id, func);
@@ -428,13 +448,15 @@ void incremental_read(GHashTable *table, const char *filepath) {
     FILE *file = fopen(filepath, "rb");
     line_t line;
 
+    LOG_DEBUGF("serialize.c", "Incremental read %s", filepath)
+
     while (1) {
-        size_t ret = fread((void *) &line, 1, sizeof(line_t), file);
+        size_t ret = fread((void *) &line, sizeof(line_t), 1, file);
         if (ret != 1 || feof(file)) {
             break;
         }
 
-        incremental_put(table, line.ino, line.mtime);
+        incremental_put(table, line.path_md5, line.mtime);
 
         while ((getc(file))) {}
         skip_meta(file);
@@ -452,33 +474,47 @@ void incremental_copy(store_t *store, store_t *dst_store, const char *filepath,
     FILE *dst_file = fopen(dst_filepath, "ab");
     line_t line;
 
-    while (1) {
+    LOG_DEBUGF("serialize.c", "Incremental copy %s", filepath)
-        size_t ret = fread((void *) &line, 1, sizeof(line_t), file);
+
+    while (TRUE) {
+        size_t ret = fread((void *) &line, sizeof(line_t), 1, file);
         if (ret != 1 || feof(file)) {
             break;
         }
 
-        if (incremental_get(copy_table, line.ino)) {
+        // Assume that files with parents still exist.
+        //  One way to "fix" this would be to check if the parent is marked for copy but it would consistently
+        //  delete files with grandparents, which is a side-effect worse than having orphaned files
+        if (line.has_parent || incremental_get(copy_table, line.path_md5)) {
             fwrite(&line, sizeof(line), 1, dst_file);
 
-            size_t buf_len;
+            // Copy filepath
-            char *buf = store_read(store, (char *) line.uuid, 16, &buf_len);
+            char filepath_buf[PATH_MAX];
-            store_write(dst_store, (char *) line.uuid, 16, buf, buf_len);
-            free(buf);
-
             char c;
+            char *ptr = filepath_buf;
             while ((c = (char) getc(file))) {
-                fwrite(&c, sizeof(c), 1, dst_file);
+                *ptr++ = c;
+            }
+            *ptr = '\0';
+            fwrite(filepath_buf, (ptr - filepath_buf) + 1, 1, dst_file);
+
+            // Copy tn store contents
+            size_t buf_len;
+            char path_md5[MD5_DIGEST_LENGTH];
+            MD5((unsigned char *) filepath_buf, (ptr - filepath_buf), (unsigned char *) path_md5);
+            char *buf = store_read(store, path_md5, sizeof(path_md5), &buf_len);
+            if (buf_len != 0) {
+                store_write(dst_store, path_md5, sizeof(path_md5), buf, buf_len);
+                free(buf);
             }
-            fwrite("\0", sizeof(c), 1, dst_file);
 
             enum metakey key;
             while (1) {
                 key = getc(file);
+                fwrite(&key, sizeof(char), 1, dst_file);
                 if (key == '\n') {
                     break;
                 }
-                fwrite(&key, sizeof(char), 1, dst_file);
 
                 if (IS_META_INT(key)) {
                     int val;
@@ -494,14 +530,12 @@ void incremental_copy(store_t *store, store_t *dst_store, const char *filepath,
                     }
                     fwrite("\0", sizeof(c), 1, dst_file);
                 }
-
-                if (ret != 1) {
-                    break;
-                }
             }
         } else {
+            while ((getc(file))) {}
             skip_meta(file);
         }
     }
     fclose(file);
+    fclose(dst_file);
 }

src/io/serialize.h

@@ -7,14 +7,14 @@
 #include <sys/syscall.h>
 #include <glib.h>
 
-typedef void(*index_func)(cJSON *, const char[UUID_STR_LEN]);
+typedef void(*index_func)(cJSON *, const char[MD5_STR_LENGTH]);
 
 void incremental_copy(store_t *store, store_t *dst_store, const char *filepath,
                       const char *dst_filepath, GHashTable *copy_table);
 
 void write_document(document_t *doc);
 
-void read_index(const char *path, const char[UUID_STR_LEN], const char *type, index_func);
+void read_index(const char *path, const char[MD5_STR_LENGTH], const char *type, index_func);
 
 void incremental_read(GHashTable *table, const char *filepath);
 

src/io/store.c

@@ -40,12 +40,20 @@ void store_destroy(store_t *store) {
     free(store);
 }
 
+void store_flush(store_t *store) {
+    mdb_env_sync(store->env, TRUE);
+}
+
 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];
+        if (key_len == MD5_DIGEST_LENGTH) {
-        uuid_unparse((unsigned char *) key, uuid_str);
+            char path_md5_str[MD5_STR_LENGTH];
-        LOG_DEBUGF("store.c", "Store write {%s} %lu bytes", uuid_str, buf_len)
+            buf2hex((unsigned char *) key, MD5_DIGEST_LENGTH, path_md5_str);
+            LOG_DEBUGF("store.c", "Store write {%s} %lu bytes", path_md5_str, buf_len)
+        } else {
+            LOG_DEBUGF("store.c", "Store write {%s} %lu bytes", key, buf_len)
+        }
     }
 
     MDB_val mdb_key;
@@ -136,7 +144,9 @@ 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;
@@ -23,6 +24,8 @@ void store_destroy(store_t *store);
 
 void store_write(store_t *store, char *key, size_t key_len, char *buf, size_t buf_len);
 
+void store_flush(store_t *store);
+
 char *store_read(store_t *store, char *key, size_t key_len, size_t *ret_vallen);
 
 GHashTable *store_read_all(store_t *store);

src/io/walk.c

@@ -20,7 +20,7 @@ parse_job_t *create_fs_parse_job(const char *filepath, const struct stat *info,
 
     job->vfile.info = *info;
 
-    memset(job->parent, 0, 16);
+    memset(job->parent, 0, MD5_DIGEST_LENGTH);
 
     job->vfile.filepath = job->filepath;
     job->vfile.read = fs_read;

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 = "2.8.4";
+static const char *const Version = "2.9.0";
 static const char *const usage[] = {
         "sist2 scan [OPTION]... PATH",
         "sist2 index [OPTION]... INDEX",
@@ -34,9 +34,10 @@ void init_dir(const char *dirpath) {
     char path[PATH_MAX];
     snprintf(path, PATH_MAX, "%sdescriptor.json", dirpath);
 
-    uuid_t uuid;
+    unsigned char index_md5[MD5_DIGEST_LENGTH];
-    uuid_generate(uuid);
+    MD5((unsigned char *) ScanCtx.index.desc.name, strlen(ScanCtx.index.desc.name), index_md5);
-    uuid_unparse(uuid, ScanCtx.index.desc.uuid);
+    buf2hex(index_md5, MD5_DIGEST_LENGTH, ScanCtx.index.desc.id);
+
     time(&ScanCtx.index.desc.timestamp);
     strcpy(ScanCtx.index.desc.version, Version);
     strcpy(ScanCtx.index.desc.type, INDEX_TYPE_BIN);
@@ -149,6 +150,14 @@ void initialize_scan_context(scan_args_t *args) {
     ScanCtx.text_ctx.log = _log;
     ScanCtx.text_ctx.logf = _logf;
 
+    // MSDOC
+    ScanCtx.msdoc_ctx.tn_size = args->size;
+    ScanCtx.msdoc_ctx.content_size = args->content_size;
+    ScanCtx.msdoc_ctx.log = _log;
+    ScanCtx.msdoc_ctx.logf = _logf;
+    ScanCtx.msdoc_ctx.store = _store;
+    ScanCtx.msdoc_ctx.msdoc_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/msword");
+
     ScanCtx.threads = args->threads;
     ScanCtx.depth = args->depth;
 
@@ -182,6 +191,10 @@ 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) {
@@ -206,7 +219,7 @@ void sist2_scan(scan_args_t *args) {
         while ((de = readdir(dir)) != NULL) {
             if (strncmp(de->d_name, "_index_", sizeof("_index_") - 1) == 0) {
                 char file_path[PATH_MAX];
-                snprintf(file_path, PATH_MAX, "%s/%s", args->incremental, de->d_name);
+                snprintf(file_path, PATH_MAX, "%s%s", args->incremental, de->d_name);
                 incremental_read(ScanCtx.original_table, file_path);
             }
         }
@@ -221,8 +234,6 @@ void sist2_scan(scan_args_t *args) {
     tpool_wait(ScanCtx.pool);
     tpool_destroy(ScanCtx.pool);
 
-    generate_stats(&ScanCtx.index, args->treemap_threshold, ScanCtx.index.path);
-
     if (args->incremental != NULL) {
         char dst_path[PATH_MAX];
         snprintf(store_path, PATH_MAX, "%sthumbs", args->incremental);
@@ -238,7 +249,7 @@ void sist2_scan(scan_args_t *args) {
         while ((de = readdir(dir)) != NULL) {
             if (strncmp(de->d_name, "_index_", sizeof("_index_") - 1) == 0) {
                 char file_path[PATH_MAX];
-                snprintf(file_path, PATH_MAX, "%s/%s", args->incremental, de->d_name);
+                snprintf(file_path, PATH_MAX, "%s%s", args->incremental, de->d_name);
                 incremental_copy(source, ScanCtx.index.store, file_path, dst_path, ScanCtx.copy_table);
             }
         }
@@ -253,6 +264,8 @@ void sist2_scan(scan_args_t *args) {
         store_destroy(source_tags);
     }
 
+    generate_stats(&ScanCtx.index, args->treemap_threshold, ScanCtx.index.path);
+
     store_destroy(ScanCtx.index.store);
 }
 
@@ -263,7 +276,7 @@ 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 +302,10 @@ 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;
@@ -311,7 +328,7 @@ void sist2_index(index_args_t *args) {
         if (strncmp(de->d_name, "_index_", sizeof("_index_") - 1) == 0) {
             char file_path[PATH_MAX];
             snprintf(file_path, PATH_MAX, "%s/%s", args->index_path, de->d_name);
-            read_index(file_path, desc.uuid, desc.type, f);
+            read_index(file_path, desc.id, desc.type, f);
         }
     }
     closedir(dir);
@@ -321,7 +338,7 @@ void sist2_index(index_args_t *args) {
     tpool_destroy(IndexCtx.pool);
 
     if (!args->print) {
-        finish_indexer(args->script, args->async_script, desc.uuid);
+        finish_indexer(args->script, args->async_script, desc.id);
     }
 
     store_destroy(IndexCtx.tag_store);
@@ -341,7 +358,7 @@ void sist2_exec_script(exec_args_t *args) {
 
     LOG_DEBUGF("main.c", "descriptor version %s (%s)", desc.version, desc.type)
 
-    execute_update_script(args->script, args->async_script, desc.uuid);
+    execute_update_script(args->script, args->async_script, desc.id);
     free(args->script);
 }
 
@@ -438,6 +455,8 @@ 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_envoy=655482,
-    application_x_esrehber=655484,
+    application_x_esrehber=655483,
-    application_x_excel=655485,
+    application_x_excel=655484,
-    application_x_executable=655486,
+    application_x_executable=655485,
-    application_x_font_gdos=655487,
+    application_x_font_gdos=655486,
-    application_x_font_pf2=655488,
+    application_x_font_pf2=655487,
-    application_x_font_pfm=655489,
+    application_x_font_pfm=655488,
-    application_x_font_sfn=655490,
+    application_x_font_sfn=655489,
-    application_x_font_ttf=655491 | 0x20000000,
+    application_x_font_ttf=655490 | 0x20000000,
-    application_x_fptapplication_x_dbt=655492,
+    application_x_fptapplication_x_dbt=655491,
-    application_x_freelance=655493,
+    application_x_freelance=655492,
-    application_x_gamecube_rom=655494,
+    application_x_gamecube_rom=655493,
-    application_x_gdbm=655495,
+    application_x_gdbm=655494,
-    application_x_gettext_translation=655496,
+    application_x_gettext_translation=655495,
-    application_x_git=655497,
+    application_x_git=655496,
-    application_x_gsp=655498,
+    application_x_gsp=655497,
-    application_x_gss=655499,
+    application_x_gss=655498,
-    application_x_gtar=655500,
+    application_x_gtar=655499,
-    application_x_gzip=655501,
+    application_x_gzip=655500,
-    application_x_hdf=655502,
+    application_x_hdf=655501,
-    application_x_helpfile=655503,
+    application_x_helpfile=655502,
-    application_x_httpd_imap=655504,
+    application_x_httpd_imap=655503,
-    application_x_ima=655505,
+    application_x_ima=655504,
-    application_x_innosetup=655506,
+    application_x_innosetup=655505,
-    application_x_internett_signup=655507,
+    application_x_internett_signup=655506,
-    application_x_inventor=655508,
+    application_x_inventor=655507,
-    application_x_ip2=655509,
+    application_x_ip2=655508,
-    application_x_java_applet=655510,
+    application_x_java_applet=655509,
-    application_x_java_commerce=655511,
+    application_x_java_commerce=655510,
-    application_x_java_image=655512,
+    application_x_java_image=655511,
-    application_x_java_jmod=655513,
+    application_x_java_jmod=655512,
-    application_x_java_keystore=655514,
+    application_x_java_keystore=655513,
-    application_x_kdelnk=655515,
+    application_x_kdelnk=655514,
-    application_x_koan=655516,
+    application_x_koan=655515,
-    application_x_latex=655517,
+    application_x_latex=655516,
-    application_x_livescreen=655518,
+    application_x_livescreen=655517,
-    application_x_lotus=655519,
+    application_x_lotus=655518,
-    application_x_lz4=655520 | 0x08000000,
+    application_x_lz4=655519 | 0x08000000,
-    application_x_lz4_json=655521,
+    application_x_lz4_json=655520,
-    application_x_lzh=655522,
+    application_x_lzh=655521,
-    application_x_lzh_compressed=655523,
+    application_x_lzh_compressed=655522,
-    application_x_lzip=655524 | 0x08000000,
+    application_x_lzip=655523 | 0x08000000,
-    application_x_lzma=655525 | 0x08000000,
+    application_x_lzma=655524 | 0x08000000,
-    application_x_lzop=655526 | 0x08000000,
+    application_x_lzop=655525 | 0x08000000,
-    application_x_lzx=655527,
+    application_x_lzx=655526,
-    application_x_mach_binary=655528,
+    application_x_mach_binary=655527,
-    application_x_mach_executable=655529,
+    application_x_mach_executable=655528,
-    application_x_magic_cap_package_1_0=655530,
+    application_x_magic_cap_package_1_0=655529,
-    application_x_mathcad=655531,
+    application_x_mathcad=655530,
-    application_x_maxis_dbpf=655532,
+    application_x_maxis_dbpf=655531,
-    application_x_meme=655533,
+    application_x_meme=655532,
-    application_x_midi=655534,
+    application_x_midi=655533,
-    application_x_mif=655535,
+    application_x_mif=655534,
-    application_x_mix_transfer=655536,
+    application_x_mix_transfer=655535,
-    application_x_mobipocket_ebook=655537 | 0x02000000,
+    application_x_mobipocket_ebook=655536 | 0x02000000,
-    application_x_ms_compress_szdd=655538,
+    application_x_ms_compress_szdd=655537,
-    application_x_ms_pdb=655539,
+    application_x_ms_pdb=655538,
-    application_x_ms_reader=655540,
+    application_x_ms_reader=655539,
-    application_x_msaccess=655541,
+    application_x_msaccess=655540,
-    application_x_n64_rom=655542,
+    application_x_n64_rom=655541,
-    application_x_navi_animation=655543,
+    application_x_navi_animation=655542,
-    application_x_navidoc=655544,
+    application_x_navidoc=655543,
-    application_x_navimap=655545,
+    application_x_navimap=655544,
-    application_x_navistyle=655546,
+    application_x_navistyle=655545,
-    application_x_nes_rom=655547,
+    application_x_nes_rom=655546,
-    application_x_netcdf=655548,
+    application_x_netcdf=655547,
-    application_x_newton_compatible_pkg=655549,
+    application_x_newton_compatible_pkg=655548,
-    application_x_nintendo_ds_rom=655550,
+    application_x_nintendo_ds_rom=655549,
-    application_x_object=655551,
+    application_x_object=655550,
-    application_x_omc=655552,
+    application_x_omc=655551,
-    application_x_omcdatamaker=655553,
+    application_x_omcdatamaker=655552,
-    application_x_omcregerator=655554,
+    application_x_omcregerator=655553,
-    application_x_pagemaker=655555,
+    application_x_pagemaker=655554,
-    application_x_pcl=655556,
+    application_x_pcl=655555,
-    application_x_pgp_keyring=655557,
+    application_x_pgp_keyring=655556,
-    application_x_pixclscript=655558,
+    application_x_pixclscript=655557,
-    application_x_pkcs7_certreqresp=655559,
+    application_x_pkcs7_certreqresp=655558,
-    application_x_pkcs7_signature=655560,
+    application_x_pkcs7_signature=655559,
-    application_x_project=655561,
+    application_x_project=655560,
-    application_x_qpro=655562,
+    application_x_qpro=655561,
-    application_x_rar=655563 | 0x10000000,
+    application_x_rar=655562 | 0x10000000,
-    application_x_rpm=655564,
+    application_x_rpm=655563,
-    application_x_sdp=655565,
+    application_x_sdp=655564,
-    application_x_sea=655566,
+    application_x_sea=655565,
-    application_x_seelogo=655567,
+    application_x_seelogo=655566,
-    application_x_setupscript=655568,
+    application_x_setupscript=655567,
-    application_x_shar=655569,
+    application_x_shar=655568,
-    application_x_sharedlib=655570,
+    application_x_sharedlib=655569,
-    application_x_shockwave_flash=655571,
+    application_x_shockwave_flash=655570,
-    application_x_snappy_framed=655572,
+    application_x_snappy_framed=655571,
-    application_x_sprite=655573,
+    application_x_sprite=655572,
-    application_x_sqlite3=655574,
+    application_x_sqlite3=655573,
-    application_x_stargallery_thm=655575,
+    application_x_stargallery_thm=655574,
-    application_x_stuffit=655576,
+    application_x_stuffit=655575,
-    application_x_sv4cpio=655577,
+    application_x_sv4cpio=655576,
-    application_x_sv4crc=655578,
+    application_x_sv4crc=655577,
-    application_x_tar=655579 | 0x10000000,
+    application_x_tar=655578 | 0x10000000,
-    application_x_tbook=655580,
+    application_x_tbook=655579,
-    application_x_terminfo=655581,
+    application_x_terminfo=655580,
-    application_x_terminfo2=655582,
+    application_x_terminfo2=655581,
-    application_x_tex_tfm=655583,
+    application_x_tex_tfm=655582,
-    application_x_texinfo=655584,
+    application_x_texinfo=655583,
-    application_x_ustar=655585,
+    application_x_ustar=655584,
-    application_x_visio=655586,
+    application_x_visio=655585,
-    application_x_vnd_audioexplosion_mzz=655587,
+    application_x_vnd_audioexplosion_mzz=655586,
-    application_x_vnd_ls_xpix=655588,
+    application_x_vnd_ls_xpix=655587,
-    application_x_vrml=655589,
+    application_x_vrml=655588,
-    application_x_wais_source=655590,
+    application_x_wais_source=655589,
-    application_x_wine_extension_ini=655591,
+    application_x_wine_extension_ini=655590,
-    application_x_wintalk=655592,
+    application_x_wintalk=655591,
-    application_x_world=655593,
+    application_x_world=655592,
-    application_x_wri=655594,
+    application_x_wri=655593,
-    application_x_x509_ca_cert=655595,
+    application_x_x509_ca_cert=655594,
-    application_x_xz=655596 | 0x08000000,
+    application_x_xz=655595 | 0x08000000,
-    application_x_zip=655597,
+    application_x_zip=655596,
-    application_x_zstd=655598 | 0x08000000,
+    application_x_zstd=655597 | 0x08000000,
-    application_x_zstd_dictionary=655599,
+    application_x_zstd_dictionary=655598,
-    application_xml=655600,
+    application_xml=655599,
-    application_zip=655601 | 0x10000000,
+    application_zip=655600 | 0x10000000,
-    application_zlib=655602,
+    application_zlib=655601,
-    audio_basic=458995 | 0x80000000,
+    audio_basic=458994 | 0x80000000,
-    audio_it=458996,
+    audio_it=458995,
-    audio_make=458997,
+    audio_make=458996,
-    audio_mid=458998,
+    audio_mid=458997,
-    audio_midi=458999,
+    audio_midi=458998,
-    audio_mp4=459000,
+    audio_mp4=458999,
-    audio_mpeg=459001,
+    audio_mpeg=459000,
-    audio_ogg=459002,
+    audio_ogg=459001,
-    audio_s3m=459003,
+    audio_s3m=459002,
-    audio_tsp_audio=459004,
+    audio_tsp_audio=459003,
-    audio_tsplayer=459005,
+    audio_tsplayer=459004,
-    audio_vnd_qcelp=459006,
+    audio_vnd_qcelp=459005,
-    audio_voxware=459007,
+    audio_voxware=459006,
-    audio_x_aiff=459008,
+    audio_x_aiff=459007,
-    audio_x_flac=459009,
+    audio_x_flac=459008,
-    audio_x_gsm=459010,
+    audio_x_gsm=459009,
-    audio_x_hx_aac_adts=459011,
+    audio_x_hx_aac_adts=459010,
-    audio_x_jam=459012,
+    audio_x_jam=459011,
-    audio_x_liveaudio=459013,
+    audio_x_liveaudio=459012,
-    audio_x_m4a=459014,
+    audio_x_m4a=459013,
-    audio_x_midi=459015,
+    audio_x_midi=459014,
-    audio_x_mod=459016,
+    audio_x_mod=459015,
-    audio_x_mp4a_latm=459017,
+    audio_x_mp4a_latm=459016,
-    audio_x_mpeg_3=459018,
+    audio_x_mpeg_3=459017,
-    audio_x_mpequrl=459019,
+    audio_x_mpequrl=459018,
-    audio_x_nspaudio=459020,
+    audio_x_nspaudio=459019,
-    audio_x_pn_realaudio=459021,
+    audio_x_pn_realaudio=459020,
-    audio_x_psid=459022,
+    audio_x_psid=459021,
-    audio_x_realaudio=459023,
+    audio_x_realaudio=459022,
-    audio_x_s3m=459024,
+    audio_x_s3m=459023,
-    audio_x_twinvq=459025,
+    audio_x_twinvq=459024,
-    audio_x_twinvq_plugin=459026,
+    audio_x_twinvq_plugin=459025,
-    audio_x_voc=459027,
+    audio_x_voc=459026,
-    audio_x_wav=459028,
+    audio_x_wav=459027,
-    audio_x_xbox_executable=459029 | 0x80000000,
+    audio_x_xbox_executable=459028 | 0x80000000,
-    audio_x_xbox360_executable=459030 | 0x80000000,
+    audio_x_xbox360_executable=459029 | 0x80000000,
-    audio_xm=459031,
+    audio_xm=459030,
-    font_otf=327960 | 0x20000000,
+    font_otf=327959 | 0x20000000,
-    font_sfnt=327961 | 0x20000000,
+    font_sfnt=327960 | 0x20000000,
-    font_woff=327962 | 0x20000000,
+    font_woff=327961 | 0x20000000,
-    font_woff2=327963 | 0x20000000,
+    font_woff2=327962 | 0x20000000,
-    image_bmp=524572,
+    image_bmp=524571,
-    image_cmu_raster=524573,
+    image_cmu_raster=524572,
-    image_fif=524574,
+    image_fif=524573,
-    image_florian=524575,
+    image_florian=524574,
-    image_g3fax=524576,
+    image_g3fax=524575,
-    image_gif=524577,
+    image_gif=524576,
-    image_heic=524578,
+    image_heic=524577,
-    image_ief=524579,
+    image_ief=524578,
-    image_jpeg=524580,
+    image_jpeg=524579,
-    image_jutvision=524581,
+    image_jutvision=524580,
-    image_naplps=524582,
+    image_naplps=524581,
-    image_pict=524583,
+    image_pict=524582,
-    image_png=524584,
+    image_png=524583,
-    image_svg=524585 | 0x80000000,
+    image_svg=524584 | 0x80000000,
-    image_svg_xml=524586 | 0x80000000,
+    image_svg_xml=524585 | 0x80000000,
-    image_tiff=524587,
+    image_tiff=524586,
-    image_vnd_adobe_photoshop=524588 | 0x80000000,
+    image_vnd_adobe_photoshop=524587 | 0x80000000,
-    image_vnd_djvu=524589 | 0x80000000,
+    image_vnd_djvu=524588 | 0x80000000,
-    image_vnd_fpx=524590,
+    image_vnd_fpx=524589,
-    image_vnd_microsoft_icon=524591,
+    image_vnd_microsoft_icon=524590,
-    image_vnd_rn_realflash=524592,
+    image_vnd_rn_realflash=524591,
-    image_vnd_rn_realpix=524593,
+    image_vnd_rn_realpix=524592,
-    image_vnd_wap_wbmp=524594,
+    image_vnd_wap_wbmp=524593,
-    image_vnd_xiff=524595,
+    image_vnd_xiff=524594,
-    image_webp=524596,
+    image_webp=524595,
-    image_wmf=524597,
+    image_wmf=524596,
-    image_x_3ds=524598,
+    image_x_3ds=524597,
-    image_x_adobe_dng=524599 | 0x00800000,
+    image_x_adobe_dng=524598 | 0x00800000,
-    image_x_award_bioslogo=524600,
+    image_x_award_bioslogo=524599,
-    image_x_canon_cr2=524601 | 0x00800000,
+    image_x_canon_cr2=524600 | 0x00800000,
-    image_x_canon_crw=524602 | 0x00800000,
+    image_x_canon_crw=524601 | 0x00800000,
-    image_x_cmu_raster=524603,
+    image_x_cmu_raster=524602,
-    image_x_cur=524604,
+    image_x_cur=524603,
-    image_x_dcraw=524605 | 0x00800000,
+    image_x_dcraw=524604 | 0x00800000,
-    image_x_dwg=524606,
+    image_x_dwg=524605,
-    image_x_eps=524607,
+    image_x_eps=524606,
-    image_x_epson_erf=524608 | 0x00800000,
+    image_x_epson_erf=524607 | 0x00800000,
-    image_x_exr=524609,
+    image_x_exr=524608,
-    image_x_fuji_raf=524610 | 0x00800000,
+    image_x_fuji_raf=524609 | 0x00800000,
-    image_x_gem=524611,
+    image_x_gem=524610,
-    image_x_icns=524612,
+    image_x_icns=524611,
-    image_x_icon=524613 | 0x80000000,
+    image_x_icon=524612 | 0x80000000,
-    image_x_jg=524614,
+    image_x_jg=524613,
-    image_x_jps=524615,
+    image_x_jps=524614,
-    image_x_kodak_dcr=524616 | 0x00800000,
+    image_x_kodak_dcr=524615 | 0x00800000,
-    image_x_kodak_k25=524617 | 0x00800000,
+    image_x_kodak_k25=524616 | 0x00800000,
-    image_x_kodak_kdc=524618 | 0x00800000,
+    image_x_kodak_kdc=524617 | 0x00800000,
-    image_x_minolta_mrw=524619 | 0x00800000,
+    image_x_minolta_mrw=524618 | 0x00800000,
-    image_x_ms_bmp=524620,
+    image_x_ms_bmp=524619,
-    image_x_niff=524621,
+    image_x_niff=524620,
-    image_x_nikon_nef=524622 | 0x00800000,
+    image_x_nikon_nef=524621 | 0x00800000,
-    image_x_olympus_orf=524623 | 0x00800000,
+    image_x_olympus_orf=524622 | 0x00800000,
-    image_x_panasonic_raw=524624 | 0x00800000,
+    image_x_panasonic_raw=524623 | 0x00800000,
-    image_x_pcx=524625,
+    image_x_pcx=524624,
-    image_x_pentax_pef=524626 | 0x00800000,
+    image_x_pentax_pef=524625 | 0x00800000,
-    image_x_pict=524627,
+    image_x_pict=524626,
-    image_x_portable_bitmap=524628,
+    image_x_portable_bitmap=524627,
-    image_x_portable_graymap=524629,
+    image_x_portable_graymap=524628,
-    image_x_portable_pixmap=524630,
+    image_x_portable_pixmap=524629,
-    image_x_quicktime=524631,
+    image_x_quicktime=524630,
-    image_x_rgb=524632,
+    image_x_rgb=524631,
-    image_x_sigma_x3f=524633 | 0x00800000,
+    image_x_sigma_x3f=524632 | 0x00800000,
-    image_x_sony_arw=524634 | 0x00800000,
+    image_x_sony_arw=524633 | 0x00800000,
-    image_x_sony_sr2=524635 | 0x00800000,
+    image_x_sony_sr2=524634 | 0x00800000,
-    image_x_sony_srf=524636 | 0x00800000,
+    image_x_sony_srf=524635 | 0x00800000,
-    image_x_tga=524637,
+    image_x_tga=524636,
-    image_x_tiff=524638,
+    image_x_tiff=524637,
-    image_x_win_bitmap=524639,
+    image_x_win_bitmap=524638,
-    image_x_xcf=524640 | 0x80000000,
+    image_x_xcf=524639 | 0x80000000,
-    image_x_xpixmap=524641 | 0x80000000,
+    image_x_xpixmap=524640 | 0x80000000,
-    image_x_xwindowdump=524642,
+    image_x_xwindowdump=524641,
-    message_news=196963,
+    message_news=196962,
-    message_rfc822=196964,
+    message_rfc822=196963,
-    model_vnd_dwf=65893,
+    model_vnd_dwf=65892,
-    model_vnd_gdl=65894,
+    model_vnd_gdl=65893,
-    model_vnd_gs_gdl=65895,
+    model_vnd_gs_gdl=65894,
-    model_vrml=65896,
+    model_vrml=65895,
-    model_x_pov=65897,
+    model_x_pov=65896,
-    text_PGP=590186,
+    sist2_sidecar=2,
-    text_asp=590187,
+    text_PGP=590185,
-    text_css=590188,
+    text_asp=590186,
-    text_html=590189 | 0x01000000,
+    text_css=590187,
-    text_javascript=590190,
+    text_html=590188 | 0x01000000,
-    text_mcf=590191,
+    text_javascript=590189,
-    text_pascal=590192,
+    text_mcf=590190,
-    text_plain=590193,
+    text_pascal=590191,
-    text_richtext=590194,
+    text_plain=590192,
-    text_rtf=590195,
+    text_richtext=590193,
-    text_scriplet=590196,
+    text_rtf=590194,
-    text_tab_separated_values=590197,
+    text_scriplet=590195,
-    text_troff=590198,
+    text_tab_separated_values=590196,
-    text_uri_list=590199,
+    text_troff=590197,
-    text_vnd_abc=590200,
+    text_uri_list=590198,
-    text_vnd_fmi_flexstor=590201,
+    text_vnd_abc=590199,
-    text_vnd_wap_wml=590202,
+    text_vnd_fmi_flexstor=590200,
-    text_vnd_wap_wmlscript=590203,
+    text_vnd_wap_wml=590201,
-    text_webviewhtml=590204,
+    text_vnd_wap_wmlscript=590202,
-    text_x_Algol68=590205,
+    text_webviewhtml=590203,
-    text_x_asm=590206,
+    text_x_Algol68=590204,
-    text_x_audiosoft_intra=590207,
+    text_x_asm=590205,
-    text_x_awk=590208,
+    text_x_audiosoft_intra=590206,
-    text_x_bcpl=590209,
+    text_x_awk=590207,
-    text_x_c=590210,
+    text_x_bcpl=590208,
-    text_x_c__=590211,
+    text_x_c=590209,
-    text_x_component=590212,
+    text_x_c__=590210,
-    text_x_diff=590213,
+    text_x_component=590211,
-    text_x_fortran=590214,
+    text_x_diff=590212,
-    text_x_java=590215,
+    text_x_fortran=590213,
-    text_x_la_asf=590216,
+    text_x_java=590214,
-    text_x_lisp=590217,
+    text_x_la_asf=590215,
-    text_x_m=590218,
+    text_x_lisp=590216,
-    text_x_m4=590219,
+    text_x_m=590217,
-    text_x_makefile=590220,
+    text_x_m4=590218,
-    text_x_ms_regedit=590221,
+    text_x_makefile=590219,
-    text_x_msdos_batch=590222,
+    text_x_ms_regedit=590220,
-    text_x_objective_c=590223,
+    text_x_msdos_batch=590221,
-    text_x_pascal=590224,
+    text_x_objective_c=590222,
-    text_x_perl=590225,
+    text_x_pascal=590223,
-    text_x_php=590226,
+    text_x_perl=590224,
-    text_x_po=590227,
+    text_x_php=590225,
-    text_x_python=590228,
+    text_x_po=590226,
-    text_x_ruby=590229,
+    text_x_python=590227,
-    text_x_sass=590230,
+    text_x_ruby=590228,
-    text_x_scss=590231,
+    text_x_sass=590229,
-    text_x_server_parsed_html=590232,
+    text_x_scss=590230,
-    text_x_setext=590233,
+    text_x_server_parsed_html=590231,
-    text_x_sgml=590234 | 0x01000000,
+    text_x_setext=590232,
-    text_x_shellscript=590235,
+    text_x_sgml=590233 | 0x01000000,
-    text_x_speech=590236,
+    text_x_shellscript=590234,
-    text_x_tcl=590237,
+    text_x_speech=590235,
-    text_x_tex=590238,
+    text_x_tcl=590236,
-    text_x_uil=590239,
+    text_x_tex=590237,
-    text_x_uuencode=590240,
+    text_x_uil=590238,
-    text_x_vcalendar=590241,
+    text_x_uuencode=590239,
-    text_x_vcard=590242,
+    text_x_vcalendar=590240,
-    text_xml=590243 | 0x01000000,
+    text_x_vcard=590241,
-    video_MP2T=393636,
+    text_xml=590242 | 0x01000000,
-    video_animaflex=393637,
+    video_MP2T=393635,
-    video_avi=393638,
+    video_animaflex=393636,
-    video_avs_video=393639,
+    video_avi=393637,
-    video_mp4=393640,
+    video_avs_video=393638,
-    video_mpeg=393641,
+    video_mp4=393639,
-    video_quicktime=393642,
+    video_mpeg=393640,
-    video_vdo=393643,
+    video_quicktime=393641,
-    video_vivo=393644,
+    video_vdo=393642,
-    video_vnd_rn_realvideo=393645,
+    video_vivo=393643,
-    video_vosaic=393646,
+    video_vnd_rn_realvideo=393644,
-    video_webm=393647,
+    video_vosaic=393645,
-    video_x_amt_demorun=393648,
+    video_webm=393646,
-    video_x_amt_showrun=393649,
+    video_x_amt_demorun=393647,
-    video_x_atomic3d_feature=393650,
+    video_x_amt_showrun=393648,
-    video_x_dl=393651,
+    video_x_atomic3d_feature=393649,
-    video_x_dv=393652,
+    video_x_dl=393650,
-    video_x_fli=393653,
+    video_x_dv=393651,
-    video_x_flv=393654,
+    video_x_fli=393652,
-    video_x_isvideo=393655,
+    video_x_flv=393653,
-    video_x_jng=393656 | 0x80000000,
+    video_x_isvideo=393654,
-    video_x_m4v=393657,
+    video_x_jng=393655 | 0x80000000,
-    video_x_matroska=393658,
+    video_x_m4v=393656,
-    video_x_mng=393659,
+    video_x_matroska=393657,
-    video_x_motion_jpeg=393660,
+    video_x_mng=393658,
-    video_x_ms_asf=393661,
+    video_x_motion_jpeg=393659,
-    video_x_msvideo=393662,
+    video_x_ms_asf=393660,
-    video_x_qtc=393663,
+    video_x_msvideo=393661,
-    video_x_sgi_movie=393664,
+    video_x_qtc=393662,
-    x_epoc_x_sisx_app=721345,
+    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>
 
@@ -45,29 +46,31 @@ void parse(void *arg) {
     parse_job_t *job = arg;
     document_t doc;
 
-    int inc_ts = incremental_get(ScanCtx.original_table, job->vfile.info.st_ino);
-    if (inc_ts != 0 && inc_ts == job->vfile.info.st_mtim.tv_sec) {
-        incremental_mark_file_for_copy(ScanCtx.copy_table, job->vfile.info.st_ino);
-        return;
-    }
-
     doc.filepath = job->filepath;
     doc.ext = (short) job->ext;
     doc.base = (short) job->base;
+
+    char *rel_path = doc.filepath + ScanCtx.index.desc.root_len;
+    MD5((unsigned char *) rel_path, strlen(rel_path), doc.path_md5);
+
     doc.meta_head = NULL;
     doc.meta_tail = NULL;
     doc.mime = 0;
     doc.size = job->vfile.info.st_size;
-    doc.ino = job->vfile.info.st_ino;
     doc.mtime = job->vfile.info.st_mtim.tv_sec;
 
-    uuid_generate(doc.uuid);
+    int inc_ts = incremental_get(ScanCtx.original_table, doc.path_md5);
+    if (inc_ts != 0 && inc_ts == job->vfile.info.st_mtim.tv_sec) {
+        incremental_mark_file_for_copy(ScanCtx.copy_table, doc.path_md5);
+        return;
+    }
+
     char *buf[MAGIC_BUF_SIZE];
 
     if (LogCtx.very_verbose) {
-        char uuid_str[UUID_STR_LEN];
+        char path_md5_str[MD5_STR_LENGTH];
-        uuid_unparse(doc.uuid, uuid_str);
+        buf2hex(doc.path_md5, MD5_DIGEST_LENGTH, path_md5_str);
-        LOG_DEBUGF(job->filepath, "Starting parse job {%s}", uuid_str)
+        LOG_DEBUGF(job->filepath, "Starting parse job {%s}", path_md5_str)
     }
 
     if (job->vfile.info.st_size == 0) {
@@ -85,7 +88,8 @@ void parse(void *arg) {
 
         // Get mime type with libmagic
         if (!job->vfile.is_fs_file) {
-            LOG_WARNING(job->filepath, "Guessing mime type with libmagic inside archive files is not currently supported");
+            LOG_WARNING(job->filepath,
+                        "Guessing mime type with libmagic inside archive files is not currently supported");
             goto abort;
         }
 
@@ -157,16 +161,26 @@ 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;
+    } else if (is_msdoc(&ScanCtx.msdoc_ctx, doc.mime)) {
+        parse_msdoc(&ScanCtx.msdoc_ctx, &job->vfile, &doc);
     }
 
     abort:
 
     //Parent meta
-    if (!uuid_is_null(job->parent)) {
+    if (!md5_digest_is_null(job->parent)) {
-        meta_line_t *meta_parent = malloc(sizeof(meta_line_t) + UUID_STR_LEN + 1);
+        meta_line_t *meta_parent = malloc(sizeof(meta_line_t) + MD5_STR_LENGTH);
         meta_parent->key = MetaParent;
-        uuid_unparse(job->parent, meta_parent->str_val);
+        buf2hex(job->parent, MD5_DIGEST_LENGTH, meta_parent->str_val);
         APPEND_META((&doc), meta_parent)
+
+        doc.has_parent = TRUE;
+    } else {
+        doc.has_parent = FALSE;
     }
 
     write_document(&doc);

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);
+
+    unsigned char path_md5[MD5_DIGEST_LENGTH];
+    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);
+
+    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/sist.h

@@ -23,9 +23,10 @@
 #undef ABS
 #define ABS(a)	   (((a) < 0) ? -(a) : (a))
 
-#define UUID_STR_LEN 37
 #define UNUSED(x) __attribute__((__unused__))  x
 
+#define MD5_STR_LENGTH 33
+
 #include "util.h"
 #include "log.h"
 #include "types.h"
@@ -47,5 +48,4 @@
 #include <errno.h>
 #include <ctype.h>
 
-
 #endif

src/static/css/ion.rangeSlider.skinFlat.min.css

@@ -1 +1 @@
-.irs-bar,.irs-bar-edge,.irs-line-left,.irs-line-mid,.irs-line-right,.irs-slider{background:url("../img/sprite-skin-flat.png") repeat-x}.irs{height:40px}.irs-with-grid{height:60px}.irs-line{height:12px;top:25px}.irs-line-left{height:12px;background-position:0 -30px}.irs-line-mid{height:12px;background-position:0 0}.irs-line-right{height:12px;background-position:100% -30px}.irs-bar{height:12px;top:25px;background-position:0 -60px}.irs-bar-edge{top:25px;height:12px;width:9px;background-position:0 -90px}.irs-shadow{height:3px;top:34px;background:#000;opacity:0.25}.lt-ie9 .irs-shadow{filter: alpha(opacity=25)}.irs-slider{width:16px;height:18px;top:22px;background-position:0 -120px}.irs-slider.state_hover,.irs-slider:hover{background-position:0 -150px}.irs-max,.irs-min{color:#999;font-size:10px;line-height:1.333;text-shadow:none;top:0;padding:1px 3px;background:#e1e4e9;-moz-border-radius:4px;border-radius:4px}.irs-from,.irs-single,.irs-to{color:#fff;font-size:10px;line-height:1.333;text-shadow:none;padding:1px 5px;background:#2196F3;-moz-border-radius:4px;border-radius:4px}.irs-from:after,.irs-single:after,.irs-to:after{position:absolute;display:block;content:"";bottom:-6px;left:50%;width:0;height:0;margin-left:-3px;overflow:hidden;border:3px solid transparent;border-top-color:#2196F3}.irs-grid-pol{background:#e1e4e9}.irs-grid-text{color:#999}.irs-disabled{}
+.irs-bar,.irs-bar-edge,.irs-line-left,.irs-line-mid,.irs-line-right,.irs-slider{background:url("./img/sprite-skin-flat.png") repeat-x}.irs{height:40px}.irs-with-grid{height:60px}.irs-line{height:12px;top:25px}.irs-line-left{height:12px;background-position:0 -30px}.irs-line-mid{height:12px;background-position:0 0}.irs-line-right{height:12px;background-position:100% -30px}.irs-bar{height:12px;top:25px;background-position:0 -60px}.irs-bar-edge{top:25px;height:12px;width:9px;background-position:0 -90px}.irs-shadow{height:3px;top:34px;background:#000;opacity:0.25}.lt-ie9 .irs-shadow{filter: alpha(opacity=25)}.irs-slider{width:16px;height:18px;top:22px;background-position:0 -120px}.irs-slider.state_hover,.irs-slider:hover{background-position:0 -150px}.irs-max,.irs-min{color:#999;font-size:10px;line-height:1.333;text-shadow:none;top:0;padding:1px 3px;background:#e1e4e9;-moz-border-radius:4px;border-radius:4px}.irs-from,.irs-single,.irs-to{color:#fff;font-size:10px;line-height:1.333;text-shadow:none;padding:1px 5px;background:#2196F3;-moz-border-radius:4px;border-radius:4px}.irs-from:after,.irs-single:after,.irs-to:after{position:absolute;display:block;content:"";bottom:-6px;left:50%;width:0;height:0;margin-left:-3px;overflow:hidden;border:3px solid transparent;border-top-color:#2196F3}.irs-grid-pol{background:#e1e4e9}.irs-grid-text{color:#999}.irs-disabled{}

src/static/js/8_md5.min.js

@@ -0,0 +1 @@
+!function(n){"use strict";function d(n,t){var r=(65535&n)+(65535&t);return(n>>16)+(t>>16)+(r>>16)<<16|65535&r}function f(n,t,r,e,o,u){return d((c=d(d(t,n),d(e,u)))<<(f=o)|c>>>32-f,r);var c,f}function l(n,t,r,e,o,u,c){return f(t&r|~t&e,n,t,o,u,c)}function v(n,t,r,e,o,u,c){return f(t&e|r&~e,n,t,o,u,c)}function g(n,t,r,e,o,u,c){return f(t^r^e,n,t,o,u,c)}function m(n,t,r,e,o,u,c){return f(r^(t|~e),n,t,o,u,c)}function i(n,t){var r,e,o,u;n[t>>5]|=128<<t%32,n[14+(t+64>>>9<<4)]=t;for(var c=1732584193,f=-271733879,i=-1732584194,a=271733878,h=0;h<n.length;h+=16)c=l(r=c,e=f,o=i,u=a,n[h],7,-680876936),a=l(a,c,f,i,n[h+1],12,-389564586),i=l(i,a,c,f,n[h+2],17,606105819),f=l(f,i,a,c,n[h+3],22,-1044525330),c=l(c,f,i,a,n[h+4],7,-176418897),a=l(a,c,f,i,n[h+5],12,1200080426),i=l(i,a,c,f,n[h+6],17,-1473231341),f=l(f,i,a,c,n[h+7],22,-45705983),c=l(c,f,i,a,n[h+8],7,1770035416),a=l(a,c,f,i,n[h+9],12,-1958414417),i=l(i,a,c,f,n[h+10],17,-42063),f=l(f,i,a,c,n[h+11],22,-1990404162),c=l(c,f,i,a,n[h+12],7,1804603682),a=l(a,c,f,i,n[h+13],12,-40341101),i=l(i,a,c,f,n[h+14],17,-1502002290),c=v(c,f=l(f,i,a,c,n[h+15],22,1236535329),i,a,n[h+1],5,-165796510),a=v(a,c,f,i,n[h+6],9,-1069501632),i=v(i,a,c,f,n[h+11],14,643717713),f=v(f,i,a,c,n[h],20,-373897302),c=v(c,f,i,a,n[h+5],5,-701558691),a=v(a,c,f,i,n[h+10],9,38016083),i=v(i,a,c,f,n[h+15],14,-660478335),f=v(f,i,a,c,n[h+4],20,-405537848),c=v(c,f,i,a,n[h+9],5,568446438),a=v(a,c,f,i,n[h+14],9,-1019803690),i=v(i,a,c,f,n[h+3],14,-187363961),f=v(f,i,a,c,n[h+8],20,1163531501),c=v(c,f,i,a,n[h+13],5,-1444681467),a=v(a,c,f,i,n[h+2],9,-51403784),i=v(i,a,c,f,n[h+7],14,1735328473),c=g(c,f=v(f,i,a,c,n[h+12],20,-1926607734),i,a,n[h+5],4,-378558),a=g(a,c,f,i,n[h+8],11,-2022574463),i=g(i,a,c,f,n[h+11],16,1839030562),f=g(f,i,a,c,n[h+14],23,-35309556),c=g(c,f,i,a,n[h+1],4,-1530992060),a=g(a,c,f,i,n[h+4],11,1272893353),i=g(i,a,c,f,n[h+7],16,-155497632),f=g(f,i,a,c,n[h+10],23,-1094730640),c=g(c,f,i,a,n[h+13],4,681279174),a=g(a,c,f,i,n[h],11,-358537222),i=g(i,a,c,f,n[h+3],16,-722521979),f=g(f,i,a,c,n[h+6],23,76029189),c=g(c,f,i,a,n[h+9],4,-640364487),a=g(a,c,f,i,n[h+12],11,-421815835),i=g(i,a,c,f,n[h+15],16,530742520),c=m(c,f=g(f,i,a,c,n[h+2],23,-995338651),i,a,n[h],6,-198630844),a=m(a,c,f,i,n[h+7],10,1126891415),i=m(i,a,c,f,n[h+14],15,-1416354905),f=m(f,i,a,c,n[h+5],21,-57434055),c=m(c,f,i,a,n[h+12],6,1700485571),a=m(a,c,f,i,n[h+3],10,-1894986606),i=m(i,a,c,f,n[h+10],15,-1051523),f=m(f,i,a,c,n[h+1],21,-2054922799),c=m(c,f,i,a,n[h+8],6,1873313359),a=m(a,c,f,i,n[h+15],10,-30611744),i=m(i,a,c,f,n[h+6],15,-1560198380),f=m(f,i,a,c,n[h+13],21,1309151649),c=m(c,f,i,a,n[h+4],6,-145523070),a=m(a,c,f,i,n[h+11],10,-1120210379),i=m(i,a,c,f,n[h+2],15,718787259),f=m(f,i,a,c,n[h+9],21,-343485551),c=d(c,r),f=d(f,e),i=d(i,o),a=d(a,u);return[c,f,i,a]}function a(n){for(var t="",r=32*n.length,e=0;e<r;e+=8)t+=String.fromCharCode(n[e>>5]>>>e%32&255);return t}function h(n){var t=[];for(t[(n.length>>2)-1]=void 0,e=0;e<t.length;e+=1)t[e]=0;for(var r=8*n.length,e=0;e<r;e+=8)t[e>>5]|=(255&n.charCodeAt(e/8))<<e%32;return t}function e(n){for(var t,r="0123456789abcdef",e="",o=0;o<n.length;o+=1)t=n.charCodeAt(o),e+=r.charAt(t>>>4&15)+r.charAt(15&t);return e}function r(n){return unescape(encodeURIComponent(n))}function o(n){return a(i(h(t=r(n)),8*t.length));var t}function u(n,t){return function(n,t){var r,e,o=h(n),u=[],c=[];for(u[15]=c[15]=void 0,16<o.length&&(o=i(o,8*n.length)),r=0;r<16;r+=1)u[r]=909522486^o[r],c[r]=1549556828^o[r];return e=i(u.concat(h(t)),512+8*t.length),a(i(c.concat(e),640))}(r(n),r(t))}function t(n,t,r){return t?r?u(t,n):e(u(t,n)):r?o(n):e(o(n))}"function"==typeof define&&define.amd?define(function(){return t}):"object"==typeof module&&module.exports?module.exports=t:n.md5=t}(this);

src/static/js/dom.js

@@ -22,7 +22,7 @@ function gifOver(thumbnail, hit) {
     thumbnail.addEventListener("mouseout", function () {
         //Reset timer
         thumbnail.mouseStayedOver = false;
-        thumbnail.setAttribute("src", `t/${hit["_source"]["index"]}/${hit["_id"]}`);
+        thumbnail.setAttribute("src", `t/${hit["_source"]["index"]}/${hit["_path_md5"]}`);
     })
 }
 
@@ -419,7 +419,7 @@ function makeThumbnail(mimeCategory, hit, imgWrapper, small) {
             thumbnail.setAttribute("class", "card-img-top fit");
         }
     }
-    thumbnail.setAttribute("src", `t/${hit["_source"]["index"]}/${hit["_id"]}`);
+    thumbnail.setAttribute("src", `t/${hit["_source"]["index"]}/${hit["_path_md5"]}`);
 
     if (shouldDisplayRawImage(hit)) {
         thumbnail.addEventListener("click", () => {

src/static/js/search.js

@@ -168,13 +168,13 @@ 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,
         name: tag,
         doc_id: hit["_id"],
-        relpath: relPath
+        path_md5: md5(relPath)
     }).then(() => {
         tagBar.blur();
         $("#tagModal").modal("hide");
@@ -604,6 +604,7 @@ function search(after = null) {
         hits.forEach(hit => {
             hit["_source"]["name"] = strUnescape(hit["_source"]["name"]);
             hit["_source"]["path"] = strUnescape(hit["_source"]["path"]);
+            hit["_path_md5"] = md5(hit["_source"]["path"] + (hit["_source"]["path"] ? "/" : "") + hit["_source"]["name"] + ext(hit));
         });
 
         if (!after) {

src/static/search.html

@@ -12,9 +12,9 @@
 
 <nav class="navbar navbar-expand-lg">
     <a class="navbar-brand" href="/">sist2</a>
-    <span class="badge badge-pill version">2.8.4</span>
+    <span class="badge badge-pill version">2.9.0</span>
     <span class="tagline">Lightning-fast file system indexer and search tool </span>
-    <a class="btn ml-auto" href="/stats">Stats</a>
+    <a class="btn ml-auto" href="stats">Stats</a>
     <button class="btn" type="button" data-toggle="modal" data-target="#settings" onclick="loadSettings()">Settings
     </button>
     <button class="btn" title="Toggle theme" onclick="toggleTheme()">Theme</button>

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">2.8.4</span>
+    <span class="badge badge-pill version">2.9.0</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"
@@ -727,7 +727,7 @@ function updateStats() {
 
     const indexId = $("#indices").val();
 
-    d3.csv(`/s/${indexId}/1`).then(tabularData => {
+    d3.csv(`./s/${indexId}/1`).then(tabularData => {
         tabularData.forEach(row => {
             row.taxonomy = row.path.split("/");
             row.size = Number(row.size);
@@ -742,16 +742,16 @@ function updateStats() {
         }
     });
 
-    d3.csv(`/s/${indexId}/2`).then(tabularData => {
+    d3.csv(`./s/${indexId}/2`).then(tabularData => {
         mimeBarSize(tabularData.slice(), mimeSvgSize);
         mimeBarCount(tabularData.slice(), mimeSvgCount);
     });
 
-    d3.csv(`/s/${indexId}/3`).then(tabularData => {
+    d3.csv(`./s/${indexId}/3`).then(tabularData => {
         sizeHistogram(tabularData, sizeHistogramSvg);
     });
 
-    d3.csv(`/s/${indexId}/4`).then(tabularData => {
+    d3.csv(`./s/${indexId}/4`).then(tabularData => {
         dateHistogram(tabularData, dateHistogramSvg);
     });
 

src/stats.c

@@ -2,8 +2,6 @@
 #include "io/serialize.h"
 #include "ctx.h"
 
-#include <glib.h>
-
 static GHashTable *FlatTree;
 static GHashTable *BufferTable;
 
@@ -22,7 +20,7 @@ typedef struct {
     long count;
 } agg_t;
 
-void fill_tables(cJSON *document, UNUSED(const char uuid_str[UUID_STR_LEN])) {
+void fill_tables(cJSON *document, UNUSED(const char index_id[MD5_STR_LENGTH])) {
 
     if (cJSON_GetObjectItem(document, "parent") != NULL) {
         return;
@@ -103,8 +101,8 @@ void read_index_into_tables(index_t *index) {
     while ((de = readdir(dir)) != NULL) {
         if (strncmp(de->d_name, "_index_", sizeof("_index_") - 1) == 0) {
             char file_path[PATH_MAX];
-            snprintf(file_path, PATH_MAX, "%s/%s", index->path, de->d_name);
+            snprintf(file_path, PATH_MAX, "%s%s", index->path, de->d_name);
-            read_index(file_path, index->desc.uuid, index->desc.type, fill_tables);
+            read_index(file_path, index->desc.id, index->desc.type, fill_tables);
         }
     }
     closedir(dir);

src/types.h

@@ -6,7 +6,7 @@
 #define INDEX_VERSION_EXTERNAL "_external_v1"
 
 typedef struct index_descriptor {
-    char uuid[UUID_STR_LEN];
+    char id[MD5_STR_LENGTH];
     char version[64];
     long timestamp;
     char root[PATH_MAX];
@@ -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/util.c

@@ -2,7 +2,6 @@
 #include "src/ctx.h"
 
 #include <wordexp.h>
-#include <glib.h>
 
 #define PBSTR "========================================"
 #define PBWIDTH 40
@@ -125,7 +124,7 @@ void progress_bar_print(double percentage, size_t tn_size, size_t index_size) {
 }
 
 GHashTable *incremental_get_table() {
-    GHashTable *file_table = g_hash_table_new(g_direct_hash, g_direct_equal);
+    GHashTable *file_table = g_hash_table_new_full(g_str_hash, g_str_equal, free, NULL);
     return file_table;
 }
 

src/util.h

@@ -10,6 +10,8 @@
 #include "third-party/utf8.h/utf8.h"
 #include "libscan/scan.h"
 
+#define MD5_STR_LENGTH 33
+
 
 char *abspath(const char *path);
 
@@ -21,25 +23,6 @@ void progress_bar_print(double percentage, size_t tn_size, size_t index_size);
 
 GHashTable *incremental_get_table();
 
-__always_inline
-static void incremental_put(GHashTable *table, unsigned long inode_no, int mtime) {
-    g_hash_table_insert(table, (gpointer) inode_no, GINT_TO_POINTER(mtime));
-}
-
-__always_inline
-static int incremental_get(GHashTable *table, unsigned long inode_no) {
-    if (table != NULL) {
-        return GPOINTER_TO_INT(g_hash_table_lookup(table, (gpointer) inode_no));
-    } else {
-        return 0;
-    }
-}
-
-__always_inline
-static int incremental_mark_file_for_copy(GHashTable *table, unsigned long inode_no) {
-    return g_hash_table_insert(table, GINT_TO_POINTER(inode_no), GINT_TO_POINTER(1));
-}
-
 
 const char *find_file_in_paths(const char **paths, const char *filename);
 
@@ -48,4 +31,95 @@ void str_escape(char *dst, const char *str);
 
 void str_unescape(char *dst, const char *str);
 
+static int hex2buf(const char *str, int len, unsigned char *bytes) {
+    static const uint8_t hashmap[] = {
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
+            0x08, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
+    };
+
+    for (int pos = 0; pos < len; pos += 2) {
+        int idx0 = (uint8_t) str[pos + 0];
+        int idx1 = (uint8_t) str[pos + 1];
+        bytes[pos / 2] = (uint8_t) (hashmap[idx0] << 4) | hashmap[idx1];
+    }
+    return TRUE;
+}
+
+__always_inline
+static void buf2hex(const unsigned char *buf, size_t buflen, char *hex_string) {
+    static const char hexdig[] = "0123456789abcdef";
+
+    const unsigned char *p;
+    size_t i;
+
+    char *s = hex_string;
+    for (i = 0, p = buf; i < buflen; i++, p++) {
+        *s++ = hexdig[(*p >> 4) & 0x0f];
+        *s++ = hexdig[*p & 0x0f];
+    }
+    *s = '\0';
+}
+
+
+__always_inline
+static int md5_digest_is_null(const unsigned char digest[MD5_DIGEST_LENGTH]) {
+    return (*(int64_t *) digest) == 0 && (*((int64_t *) digest + 1)) == 0;
+}
+
+
+__always_inline
+static void incremental_put(GHashTable *table, unsigned char path_md5[MD5_DIGEST_LENGTH], int mtime) {
+    char *ptr = malloc(MD5_STR_LENGTH);
+    buf2hex(path_md5, MD5_DIGEST_LENGTH, ptr);
+    g_hash_table_insert(table, ptr, GINT_TO_POINTER(mtime));
+}
+
+__always_inline
+static int incremental_get(GHashTable *table, unsigned char path_md5[MD5_DIGEST_LENGTH]) {
+    if (table != NULL) {
+        char md5_str[MD5_STR_LENGTH];
+        buf2hex(path_md5, MD5_DIGEST_LENGTH, md5_str);
+        return GPOINTER_TO_INT(g_hash_table_lookup(table, md5_str));
+    } else {
+        return 0;
+    }
+}
+
+__always_inline
+static int incremental_mark_file_for_copy(GHashTable *table, unsigned char path_md5[MD5_DIGEST_LENGTH]) {
+    char *ptr = malloc(MD5_STR_LENGTH);
+    buf2hex(path_md5, MD5_DIGEST_LENGTH, ptr);
+    return g_hash_table_insert(table, ptr, GINT_TO_POINTER(1));
+}
+
 #endif

src/web/serve.c

@@ -36,7 +36,7 @@ static void send_response_line(struct mg_connection *nc, int status_code, int le
 
 index_t *get_index_by_id(const char *index_id) {
     for (int i = WebCtx.index_count; i >= 0; i--) {
-        if (strcmp(index_id, WebCtx.indices[i].desc.uuid) == 0) {
+        if (strncmp(index_id, WebCtx.indices[i].desc.id, MD5_STR_LENGTH) == 0) {
             return &WebCtx.indices[i];
         }
     }
@@ -73,17 +73,17 @@ void stats(struct mg_connection *nc) {
 
 void stats_files(struct mg_connection *nc, struct http_message *hm, struct mg_str *path) {
 
-    if (path->len != UUID_STR_LEN + 4) {
+    if (path->len != MD5_STR_LENGTH + 4) {
         mg_http_send_error(nc, 404, NULL);
         nc->flags |= MG_F_SEND_AND_CLOSE;
         return;
     }
 
-    char arg_uuid[UUID_STR_LEN];
+    char arg_md5[MD5_STR_LENGTH];
-    memcpy(arg_uuid, hm->uri.p + 3, UUID_STR_LEN);
+    memcpy(arg_md5, hm->uri.p + 3, MD5_STR_LENGTH);
-    *(arg_uuid + UUID_STR_LEN - 1) = '\0';
+    *(arg_md5 + MD5_STR_LENGTH - 1) = '\0';
 
-    index_t *index = get_index_by_id(arg_uuid);
+    index_t *index = get_index_by_id(arg_md5);
     if (index == NULL) {
         mg_http_send_error(nc, 404, NULL);
         nc->flags |= MG_F_SEND_AND_CLOSE;
@@ -91,7 +91,7 @@ void stats_files(struct mg_connection *nc, struct http_message *hm, struct mg_st
     }
 
     const char *file;
-    switch (atoi(hm->uri.p + 3 + UUID_STR_LEN)) {
+    switch (atoi(hm->uri.p + 3 + MD5_STR_LENGTH)) {
         case 1:
             file = "treemap.csv";
             break;
@@ -179,29 +179,23 @@ void img_sprite_skin_flat(struct mg_connection *nc, struct http_message *hm) {
 
 void thumbnail(struct mg_connection *nc, struct http_message *hm, struct mg_str *path) {
 
-    if (path->len != UUID_STR_LEN * 2 + 2) {
+    if (path->len != 68) {
         LOG_DEBUGF("serve.c", "Invalid thumbnail path: %.*s", (int) path->len, path->p)
         mg_http_send_error(nc, 404, NULL);
         nc->flags |= MG_F_SEND_AND_CLOSE;
         return;
     }
 
-    char arg_uuid[UUID_STR_LEN];
+    char arg_file_md5[MD5_STR_LENGTH];
-    char arg_index[UUID_STR_LEN];
+    char arg_index[MD5_STR_LENGTH];
 
-    memcpy(arg_index, hm->uri.p + 3, UUID_STR_LEN);
+    memcpy(arg_index, hm->uri.p + 3, MD5_STR_LENGTH);
-    *(arg_index + UUID_STR_LEN - 1) = '\0';
+    *(arg_index + MD5_STR_LENGTH - 1) = '\0';
-    memcpy(arg_uuid, hm->uri.p + 3 + UUID_STR_LEN, UUID_STR_LEN);
+    memcpy(arg_file_md5, hm->uri.p + 3 + MD5_STR_LENGTH, MD5_STR_LENGTH);
-    *(arg_uuid + UUID_STR_LEN - 1) = '\0';
+    *(arg_file_md5 + MD5_STR_LENGTH - 1) = '\0';
 
-    uuid_t uuid;
+    unsigned char md5_buf[MD5_DIGEST_LENGTH];
-    int ret = uuid_parse(arg_uuid, uuid);
+    hex2buf(arg_file_md5, MD5_STR_LENGTH - 1, md5_buf);
-    if (ret != 0) {
-        LOG_DEBUGF("serve.c", "Invalid thumbnail UUID: %s", arg_uuid)
-        mg_http_send_error(nc, 404, NULL);
-        nc->flags |= MG_F_SEND_AND_CLOSE;
-        return;
-    }
 
     store_t *store = get_store(arg_index);
     if (store == NULL) {
@@ -212,7 +206,7 @@ void thumbnail(struct mg_connection *nc, struct http_message *hm, struct mg_str
     }
 
     size_t data_len = 0;
-    char *data = store_read(store, (char *) uuid, sizeof(uuid_t), &data_len);
+    char *data = store_read(store, (char *) md5_buf, sizeof(md5_buf), &data_len);
     if (data_len != 0) {
         send_response_line(nc, 200, data_len, "Content-Type: image/jpeg");
         mg_send(nc, data, data_len);
@@ -305,7 +299,7 @@ void index_info(struct mg_connection *nc) {
         cJSON *idx_json = cJSON_CreateObject();
         cJSON_AddStringToObject(idx_json, "name", idx->desc.name);
         cJSON_AddStringToObject(idx_json, "version", idx->desc.version);
-        cJSON_AddStringToObject(idx_json, "id", idx->desc.uuid);
+        cJSON_AddStringToObject(idx_json, "id", idx->desc.id);
         cJSON_AddNumberToObject(idx_json, "timestamp", (double) idx->desc.timestamp);
         cJSON_AddItemToArray(arr, idx_json);
     }
@@ -323,18 +317,18 @@ void index_info(struct mg_connection *nc) {
 
 void document_info(struct mg_connection *nc, struct http_message *hm, struct mg_str *path) {
 
-    if (path->len != UUID_STR_LEN + 2) {
+    if (path->len != MD5_STR_LENGTH + 2) {
         LOG_DEBUGF("serve.c", "Invalid document_info path: %.*s", (int) path->len, path->p)
         mg_http_send_error(nc, 404, NULL);
         nc->flags |= MG_F_SEND_AND_CLOSE;
         return;
     }
 
-    char arg_uuid[UUID_STR_LEN];
+    char arg_md5[MD5_STR_LENGTH];
-    memcpy(arg_uuid, hm->uri.p + 3, UUID_STR_LEN);
+    memcpy(arg_md5, hm->uri.p + 3, MD5_STR_LENGTH);
-    *(arg_uuid + UUID_STR_LEN - 1) = '\0';
+    *(arg_md5 + MD5_STR_LENGTH - 1) = '\0';
 
-    cJSON *doc = elastic_get_document(arg_uuid);
+    cJSON *doc = elastic_get_document(arg_md5);
     cJSON *source = cJSON_GetObjectItem(doc, "_source");
 
     cJSON *index_id = cJSON_GetObjectItem(source, "index");
@@ -364,18 +358,18 @@ void document_info(struct mg_connection *nc, struct http_message *hm, struct mg_
 
 void file(struct mg_connection *nc, struct http_message *hm, struct mg_str *path) {
 
-    if (path->len != UUID_STR_LEN + 2) {
+    if (path->len != MD5_STR_LENGTH + 2) {
         LOG_DEBUGF("serve.c", "Invalid file path: %.*s", (int) path->len, path->p)
         mg_http_send_error(nc, 404, NULL);
         nc->flags |= MG_F_SEND_AND_CLOSE;
         return;
     }
 
-    char arg_uuid[UUID_STR_LEN];
+    char arg_md5[MD5_STR_LENGTH];
-    memcpy(arg_uuid, hm->uri.p + 3, UUID_STR_LEN);
+    memcpy(arg_md5, hm->uri.p + 3, MD5_STR_LENGTH);
-    *(arg_uuid + UUID_STR_LEN - 1) = '\0';
+    *(arg_md5 + MD5_STR_LENGTH - 1) = '\0';
 
-    const char *next = arg_uuid;
+    const char *next = arg_md5;
     cJSON *doc = NULL;
     cJSON *index_id = NULL;
     cJSON *source = NULL;
@@ -430,7 +424,7 @@ void status(struct mg_connection *nc) {
 typedef struct {
     char *name;
     int delete;
-    char *relpath;
+    char *path_md5_str;
     char *doc_id;
 } tag_req_t;
 
@@ -450,8 +444,9 @@ tag_req_t *parse_tag_request(cJSON *json) {
         return NULL;
     }
 
-    cJSON *arg_relpath = cJSON_GetObjectItem(json, "relpath");
+    cJSON *arg_path_md5 = cJSON_GetObjectItem(json, "path_md5");
-    if (arg_relpath == NULL || !cJSON_IsString(arg_relpath)) {
+    if (arg_path_md5 == NULL || !cJSON_IsString(arg_path_md5) ||
+        strlen(arg_path_md5->valuestring) != MD5_STR_LENGTH - 1) {
         return NULL;
     }
 
@@ -463,23 +458,23 @@ tag_req_t *parse_tag_request(cJSON *json) {
     tag_req_t *req = malloc(sizeof(tag_req_t));
     req->delete = arg_delete->valueint;
     req->name = arg_name->valuestring;
-    req->relpath = arg_relpath->valuestring;
+    req->path_md5_str = arg_path_md5->valuestring;
     req->doc_id = arg_doc_id->valuestring;
 
     return req;
 }
 
 void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path) {
-    if (path->len != UUID_STR_LEN + 4) {
+    if (path->len != MD5_STR_LENGTH + 4) {
         LOG_DEBUGF("serve.c", "Invalid tag path: %.*s", (int) path->len, path->p)
         mg_http_send_error(nc, 404, NULL);
         nc->flags |= MG_F_SEND_AND_CLOSE;
         return;
     }
 
-    char arg_index[UUID_STR_LEN];
+    char arg_index[MD5_STR_LENGTH];
-    memcpy(arg_index, hm->uri.p + 5, UUID_STR_LEN);
+    memcpy(arg_index, hm->uri.p + 5, MD5_STR_LENGTH);
-    *(arg_index + UUID_STR_LEN - 1) = '\0';
+    *(arg_index + MD5_STR_LENGTH - 1) = '\0';
 
     if (hm->body.len < 2 || hm->method.len != 4 || memcmp(&hm->method, "POST", 4) == 0) {
         LOG_DEBUG("serve.c", "Invalid tag request")
@@ -514,7 +509,7 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
     cJSON *arr = NULL;
 
     size_t data_len = 0;
-    const char *data = store_read(store, arg_req->relpath, strlen(arg_req->relpath), &data_len);
+    const char *data = store_read(store, arg_req->path_md5_str, MD5_STR_LENGTH, &data_len);
     if (data_len == 0) {
         arr = cJSON_CreateArray();
     } else {
@@ -574,7 +569,8 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
     }
 
     char *json_str = cJSON_PrintUnformatted(arr);
-    store_write(store, arg_req->relpath, strlen(arg_req->relpath) + 1, json_str, strlen(json_str) + 1);
+    store_write(store, arg_req->path_md5_str, MD5_STR_LENGTH, json_str, strlen(json_str) + 1);
+    store_flush(store);
 
     free(arg_req);
     free(json_str);

tests/test_scan.py

@@ -0,0 +1,75 @@
+import unittest
+import subprocess
+import shutil
+import json
+import os
+
+TEST_FILES = "third-party/libscan/libscan-test-files/test_files"
+
+
+def copy_files(files):
+    base = os.path.basename(files)
+    new_path = os.path.join("/tmp/sist2_test/", base)
+
+    shutil.rmtree(new_path, ignore_errors=True)
+    shutil.copytree(files, new_path)
+    return new_path
+
+
+def sist2(*args):
+    return subprocess.check_output(
+        args=["./sist2_debug", *args],
+    )
+
+
+def sist2_index(files, *args):
+    path = copy_files(files)
+
+    shutil.rmtree("i", ignore_errors=True)
+    sist2("scan", path, "-o", "i", *args)
+    return iter(sist2_index_to_dict("i"))
+
+
+def sist2_incremental_index(files, func=None, *args):
+    path = copy_files(files)
+
+    if func:
+        func(path)
+
+    shutil.rmtree("i_inc", ignore_errors=True)
+    sist2("scan", path, "-o", "i_inc", "--incremental", "i", *args)
+    return iter(sist2_index_to_dict("i_inc"))
+
+
+def sist2_index_to_dict(index):
+    res = subprocess.check_output(
+        args=["./sist2_debug", "index", "--print", index],
+    )
+
+    for line in res.splitlines():
+        if line:
+            yield json.loads(line)
+
+
+class ScanTest(unittest.TestCase):
+
+    def test_incremental1(self):
+        def remove_files(path):
+            os.remove(os.path.join(path, "msdoc/test1.doc"))
+            os.remove(os.path.join(path, "msdoc/test2.doc"))
+
+        def add_files(path):
+            with open(os.path.join(path, "newfile1"), "w"):
+                pass
+            with open(os.path.join(path, "newfile2"), "w"):
+                pass
+            with open(os.path.join(path, "newfile3"), "w"):
+                pass
+
+        file_count = sum(1 for _ in sist2_index(TEST_FILES))
+        self.assertEqual(sum(1 for _ in sist2_incremental_index(TEST_FILES, remove_files)), file_count - 2)
+        self.assertEqual(sum(1 for _ in sist2_incremental_index(TEST_FILES, add_files)), file_count + 3)
+
+
+if __name__ == "__main__":
+    unittest.main()

third-party/argparse

@@ -1 +1 @@
-Subproject commit 4ed6099cb33245b06343518b9f3c45ac56e8283c
+Subproject commit d3e67a8070afd0f4f61441342aa43cf4a6b5c2de

third-party/libscan

@@ -1 +1 @@
-Subproject commit 21f1f4b98ad3b29839379e138a8718f192876a25
+Subproject commit ae9fadec473e6e4ade05259fe359c5366c3f3af6