Version bump

Fix link to scripting.

.github/ISSUE_TEMPLATE/issue-template.md

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

DockerArm64/Dockerfile

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

DockerArm64/build.sh

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

README.md

@@ -2,6 +2,8 @@
 [![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/)
+
 # sist2
 
 sist2 (Simple incremental search tool)
@@ -124,7 +126,7 @@ binaries (GCC 7+ required).
 1. Install compile-time dependencies
 
    ```bash
-   vcpkg install lmdb cjson glib libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 ffmpeg zstd gtest mongoose libuuid libmagic libraw
+   vcpkg install lmdb cjson glib libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 ffmpeg zstd gtest mongoose libuuid libmagic libraw curl[core,ssl] jbig2dec brotli libmupdf
    ```
 
 2. Build

ci/build_arm64.sh

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

docs/USAGE.md

@@ -49,6 +49,7 @@ Scan options
 Index options
     -t, --threads=<int>           Number of threads. DEFAULT=1
     --es-url=<str>                Elasticsearch url with port. DEFAULT=http://localhost:9200
+    --es-index=<str>              Elasticsearch index name. DEFAULT=sist2
     -p, --print                   Just print JSON documents to stdout.
     --script-file=<str>           Path to user script.
     --async-script                Execute user script asynchronously.
@@ -57,11 +58,14 @@ Index options
 
 Web options
     --es-url=<str>                Elasticsearch url. DEFAULT=http://localhost:9200
+    --es-index=<str>              Elasticsearch index name. DEFAULT=sist2
     --bind=<str>                  Listen on this address. DEFAULT=localhost:4090
     --auth=<str>                  Basic auth in user:password format
     --tag-auth=<str>              Basic auth in user:password format for tagging
 
 Exec-script options
+    --es-url=<str>                Elasticsearch url. DEFAULT=http://localhost:9200
+    --es-index=<str>              Elasticsearch index name. DEFAULT=sist2
     --script-file=<str>           Path to user script.
     --async-script                Execute user script asynchronously.
 Made by simon987 <me@simon987.net>. Released under GPL-3.0
@@ -241,6 +245,8 @@ it is currently unsupported and has no guaranties of back/forward compatibility.
  * `--es-url` 
  Elasticsearch url and port. If you are using docker, make sure that both containers are on the
  same network.
+ * `--es-index` 
+    Elasticsearch index name. DEFAULT=sist2
  * `-p, --print` 
     Print index in JSON format to stdout.
  * `--script-file` 
@@ -278,6 +284,8 @@ sist2 index --print ./my_index/ | jq | less
 
 ### Web options
  * `--es-url=<str>` Elasticsearch url.
+ * `--es-index` 
+    Elasticsearch index name. DEFAULT=sist2
  * `--bind=<str>` Listen on this address.
  * `--auth=<str>` Basic auth in user:password format
  * `--tag-auth=<str>` Basic auth in user:password format. Works the same way as the 
@@ -340,4 +348,4 @@ See [Automatic tagging](#automatic-tagging) for information about tag
 
 ### Automatic tagging
 
-See [scripting](docs/scripting.md) documentation.
+See [scripting](scripting.md) documentation.

schema/mappings.json

@@ -54,6 +54,10 @@
       "type": "integer",
       "index": false
     },
+    "pages": {
+      "type": "integer",
+      "index": false
+    },
     "mtime": {
       "type": "integer"
     },

src/cli.c

@@ -9,6 +9,7 @@
 #define DEFAULT_REWRITE_URL ""
 
 #define DEFAULT_ES_URL "http://localhost:9200"
+#define DEFAULT_ES_INDEX "sist2"
 #define DEFAULT_BATCH_SIZE 100
 
 #define DEFAULT_LISTEN_ADDRESS "localhost:4090"
@@ -287,6 +288,10 @@ int index_args_validate(index_args_t *args, int argc, const char **argv) {
         args->es_url = DEFAULT_ES_URL;
     }
 
+    if (args->es_index == NULL) {
+        args->es_index = DEFAULT_ES_INDEX;
+    }
+
     if (args->script_path != NULL) {
         if (load_script(args->script_path, &args->script) != 0) {
             return 1;
@@ -298,6 +303,7 @@ int index_args_validate(index_args_t *args, int argc, const char **argv) {
     }
 
     LOG_DEBUGF("cli.c", "arg es_url=%s", args->es_url)
+    LOG_DEBUGF("cli.c", "arg es_index=%s", args->es_index)
     LOG_DEBUGF("cli.c", "arg index_path=%s", args->index_path)
     LOG_DEBUGF("cli.c", "arg script_path=%s", args->script_path)
     LOG_DEBUGF("cli.c", "arg async_script=%s", args->async_script)
@@ -326,6 +332,10 @@ int web_args_validate(web_args_t *args, int argc, const char **argv) {
         args->listen_address = DEFAULT_LISTEN_ADDRESS;
     }
 
+    if (args->es_index == NULL) {
+        args->es_index = DEFAULT_ES_INDEX;
+    }
+
     if (args->credentials != NULL) {
         char *ptr = strstr(args->credentials, ":");
         if (ptr == NULL) {
@@ -383,6 +393,7 @@ int web_args_validate(web_args_t *args, int argc, const char **argv) {
     }
 
     LOG_DEBUGF("cli.c", "arg es_url=%s", args->es_url)
+    LOG_DEBUGF("cli.c", "arg es_index=%s", args->es_index)
     LOG_DEBUGF("cli.c", "arg listen=%s", args->listen_address)
     LOG_DEBUGF("cli.c", "arg credentials=%s", args->credentials)
     LOG_DEBUGF("cli.c", "arg tag_credentials=%s", args->tag_credentials)
@@ -426,6 +437,10 @@ int exec_args_validate(exec_args_t *args, int argc, const char **argv) {
         args->es_url = DEFAULT_ES_URL;
     }
 
+    if (args->es_index == NULL) {
+        args->es_index = DEFAULT_ES_INDEX;
+    }
+
     if (args->script_path == NULL) {
         LOG_FATAL("cli.c", "--script-file argument is required");
     }

src/cli.h

@@ -35,6 +35,7 @@ int scan_args_validate(scan_args_t *args, int argc, const char **argv);
 
 typedef struct index_args {
     char *es_url;
+    char *es_index;
     const char *index_path;
     const char *script_path;
     char *script;
@@ -47,6 +48,7 @@ typedef struct index_args {
 
 typedef struct web_args {
     char *es_url;
+    char *es_index;
     char *listen_address;
     char *credentials;
     char *tag_credentials;
@@ -60,6 +62,7 @@ typedef struct web_args {
 
 typedef struct exec_args {
     char *es_url;
+    char *es_index;
     const char *index_path;
     const char *script_path;
     int async_script;

src/ctx.h

@@ -58,6 +58,7 @@ typedef struct {
 
 typedef struct {
     char *es_url;
+    char *es_index;
     int batch_size;
     tpool_t *pool;
     store_t *tag_store;
@@ -66,6 +67,7 @@ typedef struct {
 
 typedef struct {
     char *es_url;
+    char *es_index;
     int index_count;
     char *auth_user;
     char *auth_pass;

src/index/elastic.c

@@ -9,6 +9,7 @@
 typedef struct es_indexer {
     int queued;
     char *es_url;
+    char *es_index;
     es_bulk_line_t *line_head;
     es_bulk_line_t *line_tail;
 } es_indexer_t;
@@ -17,11 +18,13 @@ typedef struct es_indexer {
 static __thread es_indexer_t *Indexer;
 
 void delete_queue(int max);
+
 void elastic_flush();
 
 void elastic_cleanup() {
     elastic_flush();
     if (Indexer != NULL) {
+        free(Indexer->es_index);
         free(Indexer->es_url);
         free(Indexer);
     }
@@ -32,7 +35,7 @@ void print_json(cJSON *document, const char uuid_str[UUID_STR_LEN]) {
     cJSON *line = cJSON_CreateObject();
 
     cJSON_AddStringToObject(line, "_id", uuid_str);
-    cJSON_AddStringToObject(line, "_index", "sist2");
+    cJSON_AddStringToObject(line, "_index", IndexCtx.es_index);
     cJSON_AddStringToObject(line, "_type", "_doc");
     cJSON_AddItemReferenceToObject(line, "_source", document);
 
@@ -67,7 +70,7 @@ void index_json(cJSON *document, const char uuid_str[UUID_STR_LEN]) {
 void execute_update_script(const char *script, int async, const char index_id[UUID_STR_LEN]) {
 
     if (Indexer == NULL) {
-        Indexer = create_indexer(IndexCtx.es_url);
+        Indexer = create_indexer(IndexCtx.es_url, IndexCtx.es_index);
     }
 
     cJSON *body = cJSON_CreateObject();
@@ -83,9 +86,10 @@ void execute_update_script(const char *script, int async, const char index_id[UU
 
     char bulk_url[4096];
     if (async) {
-        snprintf(bulk_url, sizeof(bulk_url), "%s/sist2/_update_by_query?wait_for_completion=false", Indexer->es_url);
+        snprintf(bulk_url, sizeof(bulk_url), "%s/%s/_update_by_query?wait_for_completion=false", Indexer->es_url,
+                 Indexer->es_index);
     } else {
-        snprintf(bulk_url, sizeof(bulk_url), "%s/sist2/_update_by_query", Indexer->es_url);
+        snprintf(bulk_url, sizeof(bulk_url), "%s/%s/_update_by_query", Indexer->es_url, Indexer->es_index);
     }
     response_t *r = web_post(bulk_url, str);
     if (!async) {
@@ -113,8 +117,6 @@ void execute_update_script(const char *script, int async, const char index_id[UU
     cJSON_Delete(resp);
 }
 
-#define ACTION_STR_LEN 91
-
 void *create_bulk_buffer(int max, int *count, size_t *buf_len) {
     es_bulk_line_t *line = Indexer->line_head;
     *count = 0;
@@ -126,20 +128,24 @@ 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,
+        snprintf(
-                 "{\"index\":{\"_id\":\"%s\", \"_type\":\"_doc\", \"_index\":\"sist2\"}}\n", line->uuid_str);
+                action_str, 256,
+                "{\"index\":{\"_id\":\"%s\",\"_type\":\"_doc\",\"_index\":\"%s\"}}\n",
+                line->uuid_str, Indexer->es_index
+        );
 
+        size_t action_str_len = strlen(action_str);
         size_t line_len = strlen(line->line);
 
-        while (buf_size + line_len + ACTION_STR_LEN > buf_capacity) {
+        while (buf_size + line_len + action_str_len > buf_capacity) {
             buf_capacity *= 2;
             buf = realloc(buf, buf_capacity);
         }
 
-        buf_size += line_len + ACTION_STR_LEN;
+        buf_size += line_len + action_str_len;
 
-        memcpy(buf + buf_cur, action_str, ACTION_STR_LEN);
+        memcpy(buf + buf_cur, action_str, action_str_len);
-        buf_cur += ACTION_STR_LEN;
+        buf_cur += action_str_len;
         memcpy(buf + buf_cur, line->line, line_len);
         buf_cur += line_len;
 
@@ -177,6 +183,21 @@ void print_errors(response_t *r) {
     free(tmp);
 }
 
+void print_error(response_t *r) {
+    char *tmp = malloc(r->size + 1);
+    memcpy(tmp, r->body, r->size);
+    *(tmp + r->size) = '\0';
+
+    cJSON *ret_json = cJSON_Parse(tmp);
+    if (cJSON_GetObjectItem(ret_json, "error") != NULL) {
+        char *str = cJSON_Print(cJSON_GetObjectItem(ret_json, "error"));
+        LOG_ERRORF("elastic.c", "%s\n", str);
+        cJSON_free(str);
+    }
+    cJSON_Delete(ret_json);
+    free(tmp);
+}
+
 void _elastic_flush(int max) {
 
     if (max == 0) {
@@ -189,7 +210,7 @@ void _elastic_flush(int max) {
     void *buf = create_bulk_buffer(max, &count, &buf_len);
 
     char bulk_url[4096];
-    snprintf(bulk_url, 4096, "%s/sist2/_bulk?pipeline=tie", Indexer->es_url);
+    snprintf(bulk_url, sizeof(bulk_url), "%s/%s/_bulk?pipeline=tie", Indexer->es_url, Indexer->es_index);
     response_t *r = web_post(bulk_url, buf);
 
     if (r->status_code == 0) {
@@ -259,7 +280,7 @@ void delete_queue(int max) {
 void elastic_flush() {
 
     if (Indexer == NULL) {
-        Indexer = create_indexer(IndexCtx.es_url);
+        Indexer = create_indexer(IndexCtx.es_url, IndexCtx.es_index);
     }
 
     _elastic_flush(Indexer->queued);
@@ -268,7 +289,7 @@ void elastic_flush() {
 void elastic_index_line(es_bulk_line_t *line) {
 
     if (Indexer == NULL) {
-        Indexer = create_indexer(IndexCtx.es_url);
+        Indexer = create_indexer(IndexCtx.es_url, IndexCtx.es_index);
     }
 
     if (Indexer->line_head == NULL) {
@@ -286,14 +307,18 @@ void elastic_index_line(es_bulk_line_t *line) {
     }
 }
 
-es_indexer_t *create_indexer(const char *url) {
+es_indexer_t *create_indexer(const char *url, const char *index) {
 
     char *es_url = malloc(strlen(url) + 1);
     strcpy(es_url, url);
 
+    char *es_index = malloc(strlen(index) + 1);
+    strcpy(es_index, index);
+
     es_indexer_t *indexer = malloc(sizeof(es_indexer_t));
 
     indexer->es_url = es_url;
+    indexer->es_index = es_index;
     indexer->queued = 0;
     indexer->line_head = NULL;
     indexer->line_tail = NULL;
@@ -305,7 +330,7 @@ void finish_indexer(char *script, int async_script, char *index_id) {
 
     char url[4096];
 
-    snprintf(url, sizeof(url), "%s/sist2/_refresh", IndexCtx.es_url);
+    snprintf(url, sizeof(url), "%s/%s/_refresh", IndexCtx.es_url, IndexCtx.es_index);
     response_t *r = web_post(url, "");
     LOG_INFOF("elastic.c", "Refresh index <%d>", r->status_code);
     free_response(r);
@@ -314,18 +339,18 @@ void finish_indexer(char *script, int async_script, char *index_id) {
         execute_update_script(script, async_script, index_id);
         free(script);
 
-        snprintf(url, sizeof(url), "%s/sist2/_refresh", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/%s/_refresh", IndexCtx.es_url, IndexCtx.es_index);
         r = web_post(url, "");
         LOG_INFOF("elastic.c", "Refresh index <%d>", r->status_code);
         free_response(r);
     }
 
-    snprintf(url, sizeof(url), "%s/sist2/_forcemerge", IndexCtx.es_url);
+    snprintf(url, sizeof(url), "%s/%s/_forcemerge", IndexCtx.es_url, IndexCtx.es_index);
     r = web_post(url, "");
     LOG_INFOF("elastic.c", "Merge index <%d>", r->status_code);
     free_response(r);
 
-    snprintf(url, sizeof(url), "%s/sist2/_settings", IndexCtx.es_url);
+    snprintf(url, sizeof(url), "%s/%s/_settings", IndexCtx.es_url, IndexCtx.es_index);
     r = web_put(url, "{\"index\":{\"refresh_interval\":\"1s\"}}");
     LOG_INFOF("elastic.c", "Set refresh interval <%d>", r->status_code);
     free_response(r);
@@ -335,7 +360,7 @@ void elastic_init(int force_reset) {
 
     // Check if index exists
     char url[4096];
-    snprintf(url, 4096, "%s/sist2", IndexCtx.es_url);
+    snprintf(url, sizeof(url), "%s/%s", IndexCtx.es_url, IndexCtx.es_index);
     response_t *r = web_get(url, 30);
     int index_exists = r->status_code == 200;
     free_response(r);
@@ -345,32 +370,38 @@ void elastic_init(int force_reset) {
         LOG_INFOF("elastic.c", "Delete index <%d>", r->status_code);
         free_response(r);
 
-        snprintf(url, 4096, "%s/sist2", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/%s", IndexCtx.es_url, IndexCtx.es_index);
         r = web_put(url, "");
+
+        if (r->status_code != 200) {
+            print_error(r);
+            LOG_FATAL("elastic.c", "Could not create index")
+        }
+
         LOG_INFOF("elastic.c", "Create index <%d>", r->status_code);
         free_response(r);
 
-        snprintf(url, 4096, "%s/sist2/_close", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/%s/_close", IndexCtx.es_url, IndexCtx.es_index);
         r = web_post(url, "");
         LOG_INFOF("elastic.c", "Close index <%d>", r->status_code);
         free_response(r);
 
-        snprintf(url, 4096, "%s/_ingest/pipeline/tie", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/_ingest/pipeline/tie", IndexCtx.es_url);
         r = web_put(url, pipeline_json);
         LOG_INFOF("elastic.c", "Create pipeline <%d>", r->status_code);
         free_response(r);
 
-        snprintf(url, 4096, "%s/sist2/_settings", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/%s/_settings", IndexCtx.es_url, IndexCtx.es_index);
         r = web_put(url, settings_json);
         LOG_INFOF("elastic.c", "Update settings <%d>", r->status_code);
         free_response(r);
 
-        snprintf(url, 4096, "%s/sist2/_mappings/_doc?include_type_name=true", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/%s/_mappings/_doc?include_type_name=true", IndexCtx.es_url, IndexCtx.es_index);
         r = web_put(url, mappings_json);
         LOG_INFOF("elastic.c", "Update mappings <%d>", r->status_code);
         free_response(r);
 
-        snprintf(url, 4096, "%s/sist2/_open", IndexCtx.es_url);
+        snprintf(url, sizeof(url), "%s/%s/_open", IndexCtx.es_url, IndexCtx.es_index);
         r = web_post(url, "");
         LOG_INFOF("elastic.c", "Open index <%d>", r->status_code);
         free_response(r);
@@ -379,12 +410,16 @@ void elastic_init(int force_reset) {
 
 cJSON *elastic_get_document(const char *uuid_str) {
     char url[4096];
-    snprintf(url, 4096, "%s/sist2/_doc/%s", WebCtx.es_url, uuid_str);
+    snprintf(url, sizeof(url), "%s/%s/_doc/%s", WebCtx.es_url, WebCtx.es_index, uuid_str);
 
     response_t *r = web_get(url, 3);
     cJSON *json = NULL;
     if (r->status_code == 200) {
-        json = cJSON_Parse(r->body);
+        char *tmp = malloc(r->size + 1);
+        memcpy(tmp, r->body, r->size);
+        *(tmp + r->size) = '\0';
+        json = cJSON_Parse(tmp);
+        free(tmp);
     }
     free_response(r);
     return json;
@@ -392,8 +427,8 @@ cJSON *elastic_get_document(const char *uuid_str) {
 
 char *elastic_get_status() {
     char url[4096];
-    snprintf(url, 4096,
+    snprintf(url, sizeof(url),
-             "%s/_cluster/state/metadata/sist2?filter_path=metadata.indices.*.state", WebCtx.es_url);
+             "%s/_cluster/state/metadata/%s?filter_path=metadata.indices.*.state", WebCtx.es_url, WebCtx.es_index);
 
     response_t *r = web_get(url, 30);
     cJSON *json = NULL;
@@ -401,12 +436,16 @@ char *elastic_get_status() {
     status[0] = '\0';
 
     if (r->status_code == 200) {
-        json = cJSON_Parse(r->body);
+        char *tmp = malloc(r->size + 1);
+        memcpy(tmp, r->body, r->size);
+        *(tmp + r->size) = '\0';
+        json = cJSON_Parse(tmp);
+        free(tmp);
         const cJSON *metadata = cJSON_GetObjectItem(json, "metadata");
         if (metadata != NULL) {
             const cJSON *indices = cJSON_GetObjectItem(metadata, "indices");
-            const cJSON *sist2 = cJSON_GetObjectItem(indices, "sist2");
+            const cJSON *index = cJSON_GetObjectItem(indices, WebCtx.es_index);
-            const cJSON *state = cJSON_GetObjectItem(sist2, "state");
+            const cJSON *state = cJSON_GetObjectItem(index, "state");
             strcpy(status, state->valuestring);
         }
     }

src/index/elastic.h

@@ -20,7 +20,7 @@ void print_json(cJSON *document, const char uuid_str[UUID_STR_LEN]);
 
 void index_json(cJSON *document, const char uuid_str[UUID_STR_LEN]);
 
-es_indexer_t *create_indexer(const char* es_url);
+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);

src/index/web.c

@@ -1,5 +1,6 @@
 #include "web.h"
 #include "src/sist.h"
+#include "src/ctx.h"
 
 #include <mongoose.h>
 #include <pthread.h>
@@ -21,95 +22,82 @@ void free_response(response_t *resp) {
     free(resp);
 }
 
-#define SIST2_HEADERS "User-Agent: sist2\r\nContent-Type: application/json\r\n"
+void web_post_async_poll(subreq_ctx_t* req) {
+    fd_set fdread;
+    fd_set fdwrite;
+    fd_set fdexcep;
+    int maxfd = -1;
 
+    FD_ZERO(&fdread);
+    FD_ZERO(&fdwrite);
+    FD_ZERO(&fdexcep);
 
-void http_req_ev(struct mg_connection *nc, int ev, void *ptr) {
+    CURLMcode mc = curl_multi_fdset(req->multi, &fdread, &fdwrite, &fdexcep, &maxfd);
 
-    http_ev_data_t *ev_data = (http_ev_data_t *) nc->user_data;
+    if(mc != CURLM_OK) {
+        req->done = TRUE;
+        return;
+    }
 
-    switch (ev) {
+    if (maxfd == -1) {
-        case MG_EV_CONNECT: {
+        // no fds ready yet
-            int connect_status = *(int *) ptr;
+        return;
-            if (connect_status != 0) {
+    }
-                ev_data->done = TRUE;
+
-                ev_data->resp->status_code = 0;
+    struct timeval timeout = {1, 0};
-            }
+    int rc = select(maxfd + 1, &fdread, &fdwrite, &fdexcep, &timeout);
+
+    switch(rc) {
+        case -1:
+            req->done = TRUE;
             break;
-        }
+        case 0:
-        case MG_EV_HTTP_REPLY: {
-            struct http_message *hm = (struct http_message *) ptr;
-
-            //TODO: Check errors?
-
-            ev_data->resp->size = hm->body.len;
-            ev_data->resp->status_code = hm->resp_code;
-            ev_data->resp->body = malloc(hm->body.len + 1);
-            memcpy(ev_data->resp->body, hm->body.p, hm->body.len);
-            *(ev_data->resp->body + hm->body.len) = '\0';
-
-            ev_data->done = TRUE;
             break;
-        }
-        case MG_EV_CLOSE: {
-            ev_data->done = TRUE;
-            break;
-        }
         default:
+            curl_multi_perform(req->multi, &req->running_handles);
             break;
     }
+
+    if (req->running_handles == 0) {
+        req->done = TRUE;
+        req->response->body = req->response_buf.buf;
+        req->response->size = req->response_buf.cur;
+        curl_easy_getinfo(req->handle, CURLINFO_RESPONSE_CODE, &req->response->status_code);
+
+        curl_multi_cleanup(req->multi);
+        curl_easy_cleanup(req->handle);
+        curl_slist_free_all(req->headers);
+        return;
+    }
 }
 
-subreq_ctx_t *http_req(const char *url, const char *extra_headers, const char *post_data, const char *method) {
+subreq_ctx_t *web_post_async(const char *url, char *data) {
+    subreq_ctx_t *req = calloc(1, sizeof(subreq_ctx_t));
+    req->response = calloc(1, sizeof(response_t));
+    req->data = data;
+    req->response_buf = dyn_buffer_create();
 
-    struct mg_str scheme;
+    req->handle = curl_easy_init();
-    struct mg_str user_info;
+    CURL *curl = req->handle;
-    struct mg_str host;
+    curl_easy_setopt(curl, CURLOPT_URL, url);
-    unsigned int port;
+    curl_easy_setopt(curl, CURLOPT_WRITEDATA, (void *) (&req->response_buf));
-    struct mg_str path;
+    curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_cb);
-    struct mg_str query;
+    curl_easy_setopt(curl, CURLOPT_POST, 1);
-    struct mg_str fragment;
+    curl_easy_setopt(curl, CURLOPT_USERAGENT, "sist2");
 
-    if (post_data == NULL) post_data = "";
+    struct curl_slist *headers = NULL;
-    if (extra_headers == NULL) extra_headers = "";
+    headers = curl_slist_append(headers, "Content-Type: application/json");
-    if (path.len == 0) path = mg_mk_str("/");
+    curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
-    if (host.len == 0) host = mg_mk_str("");
 
-    // [scheme://[user_info@]]host[:port][/path][?query][#fragment]
+    curl_easy_setopt(curl, CURLOPT_POSTFIELDS, data);
-    mg_parse_uri(mg_mk_str(url), &scheme, &user_info, &host, &port, &path, &query, &fragment);
 
-    if (query.len > 0) path.len += query.len + 1;
+    req->multi = curl_multi_init();
+    curl_multi_add_handle(req->multi, curl);
+    curl_multi_perform(req->multi, &req->running_handles);
 
-    subreq_ctx_t *ctx = malloc(sizeof(subreq_ctx_t));
+    LOG_DEBUGF("web.c", "async request POST %s", url)
-    mg_mgr_init(&ctx->mgr, NULL);
 
-    char address[8192];
+    return req;
-    snprintf(address, sizeof(address), "tcp://%.*s:%u", (int) host.len, host.p, port);
-    struct mg_connection *nc = mg_connect(&ctx->mgr, address, http_req_ev);
-    nc->user_data = &ctx->ev_data;
-    mg_set_protocol_http_websocket(nc);
-
-    ctx->ev_data.resp = calloc(1, sizeof(response_t));
-    ctx->ev_data.done = FALSE;
-
-    mg_printf(
-            nc, "%s %.*s HTTP/1.1\r\n"
-                  "Host: %.*s\r\n"
-                  "Content-Length: %zu\r\n"
-                  "%s\r\n"
-                  "%s",
-            method, (int) path.len, path.p,
-            (int) (path.p - host.p), host.p,
-            strlen(post_data),
-            extra_headers,
-            post_data
-    );
-
-    return ctx;
-}
-
-subreq_ctx_t *web_post_async(const char *url, const char *data) {
-    return http_req(url, SIST2_HEADERS, data, "POST");
 }
 
 response_t *web_get(const char *url, int timeout) {
@@ -125,7 +113,8 @@ response_t *web_get(const char *url, int timeout) {
     curl_easy_setopt(curl, CURLOPT_USERAGENT, "sist2");
     curl_easy_setopt(curl, CURLOPT_TIMEOUT, timeout);
 
-    struct curl_slist *headers = curl_slist_append(headers, "Content-Type: application/json");
+    struct curl_slist *headers = NULL;
+    headers = curl_slist_append(headers, "Content-Type: application/json");
     curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 
     curl_easy_perform(curl);
@@ -153,7 +142,8 @@ response_t *web_post(const char *url, const char *data) {
     curl_easy_setopt(curl, CURLOPT_POST, 1);
     curl_easy_setopt(curl, CURLOPT_USERAGENT, "sist2");
 
-    struct curl_slist *headers = curl_slist_append(headers, "Content-Type: application/json");
+    struct curl_slist *headers = NULL;
+    headers = curl_slist_append(headers, "Content-Type: application/json");
     curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 
     curl_easy_setopt(curl, CURLOPT_POSTFIELDS, data);
@@ -187,7 +177,8 @@ response_t *web_put(const char *url, const char *data) {
     curl_easy_setopt(curl, CURLOPT_DNS_USE_GLOBAL_CACHE, 0);
     curl_easy_setopt(curl, CURLOPT_IPRESOLVE, CURLOPT_DNS_LOCAL_IP4 );
 
-    struct curl_slist *headers = curl_slist_append(headers, "Content-Type: application/json");
+    struct curl_slist *headers = NULL;
+    headers = curl_slist_append(headers, "Content-Type: application/json");
     curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 
     curl_easy_setopt(curl, CURLOPT_POSTFIELDS, data);
@@ -218,7 +209,8 @@ response_t *web_delete(const char *url) {
     curl_easy_setopt(curl, CURLOPT_USERAGENT, "sist2");
 
     curl_easy_setopt(curl, CURLOPT_POSTFIELDS, "");
-    struct curl_slist *headers = curl_slist_append(headers, "Content-Type: application/json");
+    struct curl_slist *headers = NULL;
+    headers = curl_slist_append(headers, "Content-Type: application/json");
     curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 
     curl_easy_perform(curl);

src/index/web.h

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

src/io/serialize.c

@@ -150,6 +150,8 @@ char *get_meta_key_text(enum metakey meta_key) {
             return "modified_by";
         case MetaThumbnail:
             return "thumbnail";
+        case MetaPages:
+            return "pages";
         default:
             return NULL;
     }
@@ -278,6 +280,7 @@ void read_index_bin(const char *path, const char *index_id, index_func func) {
         size_t ret = 0;
         while (key != '\n') {
             switch (key) {
+                case MetaPages:
                 case MetaWidth:
                 case MetaHeight: {
                     int value;

src/log.c

@@ -4,8 +4,7 @@
 #include <stdarg.h>
 
 const char *log_colors[] = {
-        "\033[34m", "\033[01;34m", "\033[0m",
+        "\033[34m", "\033[01;34m", "\033[01;33m", "\033[0m", "\033[31m", "\033[01;31m"
-        "\033[01;33m", "\033[31m", "\033[01;31m"
 };
 
 const char *log_levels[] = {

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 = "2.8.0";
+static const char *const Version = "2.8.4";
 static const char *const usage[] = {
         "sist2 scan [OPTION]... PATH",
         "sist2 index [OPTION]... INDEX",
@@ -259,6 +259,7 @@ void sist2_scan(scan_args_t *args) {
 void sist2_index(index_args_t *args) {
 
     IndexCtx.es_url = args->es_url;
+    IndexCtx.es_index = args->es_index;
     IndexCtx.batch_size = args->batch_size;
 
     if (!args->print) {
@@ -347,6 +348,7 @@ void sist2_exec_script(exec_args_t *args) {
 void sist2_web(web_args_t *args) {
 
     WebCtx.es_url = args->es_url;
+    WebCtx.es_index = args->es_index;
     WebCtx.index_count = args->index_count;
     WebCtx.auth_user = args->auth_user;
     WebCtx.auth_pass = args->auth_pass;
@@ -390,6 +392,7 @@ int main(int argc, const char *argv[]) {
     int arg_version = 0;
 
     char *common_es_url = NULL;
+    char *common_es_index = NULL;
     char *common_script_path = NULL;
     int common_async_script = 0;
     int common_threads = 0;
@@ -432,6 +435,7 @@ int main(int argc, const char *argv[]) {
             OPT_GROUP("Index options"),
             OPT_INTEGER('t', "threads", &common_threads, "Number of threads. DEFAULT=1"),
             OPT_STRING(0, "es-url", &common_es_url, "Elasticsearch url with port. DEFAULT=http://localhost:9200"),
+            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_BOOLEAN(0, "async-script", &common_async_script, "Execute user script asynchronously."),
@@ -441,11 +445,14 @@ int main(int argc, const char *argv[]) {
 
             OPT_GROUP("Web options"),
             OPT_STRING(0, "es-url", &common_es_url, "Elasticsearch url. DEFAULT=http://localhost:9200"),
+            OPT_STRING(0, "es-index", &common_es_index, "Elasticsearch index name. DEFAULT=sist2"),
             OPT_STRING(0, "bind", &web_args->listen_address, "Listen on this address. DEFAULT=localhost:4090"),
             OPT_STRING(0, "auth", &web_args->credentials, "Basic auth in user:password format"),
             OPT_STRING(0, "tag-auth", &web_args->tag_credentials, "Basic auth in user:password format for tagging"),
 
             OPT_GROUP("Exec-script options"),
+            OPT_STRING(0, "es-url", &common_es_url, "Elasticsearch url. DEFAULT=http://localhost:9200"),
+            OPT_STRING(0, "es-index", &common_es_index, "Elasticsearch index name. DEFAULT=sist2"),
             OPT_STRING(0, "script-file", &common_script_path, "Path to user script."),
             OPT_BOOLEAN(0, "async-script", &common_async_script, "Execute user script asynchronously."),
 
@@ -469,6 +476,11 @@ int main(int argc, const char *argv[]) {
     web_args->es_url = common_es_url;
     index_args->es_url = common_es_url;
     exec_args->es_url = common_es_url;
+
+    web_args->es_index = common_es_index;
+    index_args->es_index = common_es_index;
+    exec_args->es_index = common_es_index;
+
     index_args->script_path = common_script_path;
     exec_args->script_path = common_script_path;
     index_args->threads = common_threads;

src/static/css/dark.css

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

src/static/css/light.css

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

src/static/js/dom.js

@@ -75,6 +75,7 @@ function shouldPlayVideo(hit) {
         hit["_source"]["extension"] !== "mkv" &&
         hit["_source"]["extension"] !== "avi" &&
         videoc !== "hevc" &&
+        videoc !== "mpeg1video" &&
         videoc !== "mpeg2video" &&
         videoc !== "wmv3";
 }
@@ -194,10 +195,21 @@ function makeUserTag(tag, hit) {
 function infoButtonCb(hit) {
     return () => {
         getDocumentInfo(hit["_id"]).then(doc => {
+            $("#modal-body").empty()
+
             $("#modal-title").text(doc["name"] + ext(hit));
 
+            if (doc["mime"]) {
+                const mimeCategory = doc["mime"].split("/")[0];
+                const imgWrapper = document.createElement("div");
+                imgWrapper.setAttribute("style", "position: relative");
+                imgWrapper.setAttribute("class", "img-wrapper");
+                makeThumbnail(mimeCategory, hit, imgWrapper, false);
+                $("#modal-body").append(imgWrapper);
+            }
+
             const tbody = $("<tbody>");
-            $("#modal-body").empty()
+            $("#modal-body")
                 .append($("<table class='table table-sm'>")
                     .append($("<thead>")
                         .append($("<tr>")
@@ -208,10 +220,19 @@ function infoButtonCb(hit) {
                     .append(tbody)
                 );
 
+            tbody.append($("<tr>")
+                .append($("<td>").text("index"))
+                .append($("<td>").text(`[${indexMap[doc["index"]]}]`))
+            ).append($("<tr>")
+                .append($("<td>").text("mtime"))
+                .append($("<td>")
+                    .text(new Date(doc["mtime"] * 1000).toISOString().split(".")[0].replace("T", " "))
+                    .attr("title", doc["mtime"]))
+            );
             const displayFields = new Set([
-                "mime", "size", "mtime", "path", "title", "width", "height", "duration", "audioc", "videoc",
+                "mime", "size", "path", "title", "width", "height", "duration", "audioc", "videoc",
                 "bitrate", "artist", "album", "album_artist", "genre", "title", "font_name", "tag", "author",
-                "modified_by"
+                "modified_by", "pages"
             ]);
             Object.keys(doc)
                 .filter(key => key.startsWith("_keyword.") || key.startsWith("_text.") || displayFields.has(key) || key.startsWith("exif_"))

src/static/search.html

@@ -12,7 +12,7 @@
 
 <nav class="navbar navbar-expand-lg">
     <a class="navbar-brand" href="/">sist2</a>
-    <span class="badge badge-pill version">2.8.0</span>
+    <span class="badge badge-pill version">2.8.4</span>
     <span class="tagline">Lightning-fast file system indexer and search tool </span>
     <a class="btn ml-auto" href="/stats">Stats</a>
     <button class="btn" type="button" data-toggle="modal" data-target="#settings" onclick="loadSettings()">Settings

src/static/stats.html

@@ -10,7 +10,7 @@
 
 <nav class="navbar navbar-expand-lg">
     <a class="navbar-brand" href="/">sist2</a>
-    <span class="badge badge-pill version">2.8.0</span>
+    <span class="badge badge-pill version">2.8.4</span>
     <span class="tagline">Lightning-fast file system indexer and search tool </span>
     <a style="margin-left: auto" class="btn" href="/">Back</a>
     <button class="btn" type="button" data-toggle="modal" data-target="#settings"

src/web/serve.c

@@ -10,8 +10,6 @@
 
 #include <mongoose.h>
 
-#define CHUNK_SIZE 1024 * 1024 * 10
-
 
 static int has_prefix(const struct mg_str *str, const struct mg_str *prefix) {
     return str->len > prefix->len && memcmp(str->p, prefix->p, prefix->len) == 0;
@@ -237,13 +235,12 @@ void search(struct mg_connection *nc, struct http_message *hm) {
     *(body + hm->body.len) = '\0';
 
     char url[4096];
-    snprintf(url, 4096, "%s/sist2/_search", WebCtx.es_url);
+    snprintf(url, 4096, "%s/%s/_search", WebCtx.es_url, WebCtx.es_index);
 
     nc->user_data = web_post_async(url, body);
-    free(body);
 }
 
-int serve_file_from_url(cJSON *json, index_t *idx, struct mg_connection *nc) {
+void serve_file_from_url(cJSON *json, index_t *idx, struct mg_connection *nc) {
 
     const char *path = cJSON_GetObjectItem(json, "path")->valuestring;
     const char *name = cJSON_GetObjectItem(json, "name")->valuestring;
@@ -538,8 +535,8 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
             }
         }
 
-        char buf[8192];
+        char *buf = malloc(sizeof(char) * 8192);
-        snprintf(buf, sizeof(buf),
+        snprintf(buf, 8192,
                  "{"
                  "    \"script\" : {"
                  "        \"source\": \"if (ctx._source.tag.contains(params.tag)) { ctx._source.tag.remove(ctx._source.tag.indexOf(params.tag)) }\","
@@ -552,14 +549,14 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
         );
 
         char url[4096];
-        snprintf(url, sizeof(url), "%s/sist2/_update/%s", WebCtx.es_url, arg_req->doc_id);
+        snprintf(url, sizeof(url), "%s/%s/_update/%s", WebCtx.es_url, WebCtx.es_index, arg_req->doc_id);
         nc->user_data = web_post_async(url, buf);
 
     } else {
         cJSON_AddItemToArray(arr, cJSON_CreateString(arg_req->name));
 
-        char buf[8192];
+        char *buf = malloc(sizeof(char) * 8192);
-        snprintf(buf, sizeof(buf),
+        snprintf(buf, 8192,
                  "{"
                  "    \"script\" : {"
                  "        \"source\": \"if(ctx._source.tag == null) {ctx._source.tag = new ArrayList()} ctx._source.tag.add(params.tag)\","
@@ -572,7 +569,7 @@ void tag(struct mg_connection *nc, struct http_message *hm, struct mg_str *path)
         );
 
         char url[4096];
-        snprintf(url, sizeof(url), "%s/sist2/_update/%s", WebCtx.es_url, arg_req->doc_id);
+        snprintf(url, sizeof(url), "%s/%s/_update/%s", WebCtx.es_url, WebCtx.es_index, arg_req->doc_id);
         nc->user_data = web_post_async(url, buf);
     }
 
@@ -668,11 +665,11 @@ static void ev_router(struct mg_connection *nc, int ev, void *p) {
         if (nc->user_data != NULL) {
             //Waiting for ES reply
             subreq_ctx_t *ctx = (subreq_ctx_t *) nc->user_data;
-            mg_mgr_poll(&ctx->mgr, 0);
+            web_post_async_poll(ctx);
 
-            if (ctx->ev_data.done == TRUE) {
+            if (ctx->done == TRUE) {
 
-                response_t *r = ctx->ev_data.resp;
+                response_t *r = ctx->response;
 
                 if (r->status_code == 200) {
                     send_response_line(nc, 200, r->size, "Content-Type: application/json");
@@ -695,6 +692,8 @@ static void ev_router(struct mg_connection *nc, int ev, void *p) {
                 }
 
                 free_response(r);
+                free(ctx->data);
+                free(ctx);
                 nc->flags |= MG_F_SEND_AND_CLOSE;
                 nc->user_data = NULL;
             }