simon987/sist2
1
0
Fork 0
mirror of https://github.com/simon987/sist2.git synced 2025-12-14 15:59:03 +00:00
Code Issues Projects Releases Wiki Activity

SQLite search backend

Browse Source
This commit is contained in:
simon987
2023-05-18 14:16:11 -04:00
parent 1cfceba518
commit 944c224904
46 changed files with 3261 additions and 1296 deletions
Download Patch File Download Diff File Expand all files Collapse all files

344
src/web/serve.c
View File

@@ -5,11 +5,24 @@
#include "src/index/web.h"
#include "src/auth0/auth0_c_api.h"
#include "src/web/web_util.h"
#include "src/cli.h"
#include <time.h>
#include <src/ctx.h>
#define HTTP_TEXT_TYPE_HEADER "Content-Type: text/plain;charset=utf-8\r\n"
#define HTTP_REPLY_NOT_FOUND mg_http_reply(nc, 404, HTTP_SERVER_HEADER HTTP_TEXT_TYPE_HEADER, "Not found");
void fts_search_paths(struct mg_connection *nc, struct mg_http_message *hm);
void fts_search_mimetypes(struct mg_connection *nc, struct mg_http_message *hm);
void fts_search_summary_stats(struct mg_connection *nc, struct mg_http_message *hm);
void fts_search(struct mg_connection *nc, struct mg_http_message *hm);
void fts_get_document(struct mg_connection *nc, struct mg_http_message *hm);
void fts_suggest_tag(struct mg_connection *nc, struct mg_http_message *hm);
void fts_get_tags(struct mg_connection *nc, struct mg_http_message *hm);
static struct mg_http_serve_opts DefaultServeOpts = {
.fs = NULL,
@@ -47,12 +60,8 @@ void stats_files(struct mg_connection *nc, struct mg_http_message *hm) {
}
cJSON *json = database_get_stats(db, stat_type);
char *json_str = cJSON_PrintUnformatted(json);
mg_send_json(nc, json);
web_send_headers(nc, 200, strlen(json_str), "Content-Type: application/json");
mg_send(nc, json_str, strlen(json_str));
free(json_str);
cJSON_Delete(json);
}
@@ -93,7 +102,7 @@ void serve_chunk_vendors_css(struct mg_connection *nc, struct mg_http_message *h
}
void serve_thumbnail(struct mg_connection *nc, struct mg_http_message *hm, const char *arg_index,
const char *arg_doc_id, int arg_num) {
const char *arg_doc_id, int arg_num) {
database_t *db = web_get_database(arg_index);
if (db == NULL) {
@@ -252,6 +261,10 @@ void serve_file_from_disk(cJSON *json, index_t *idx, struct mg_connection *nc, s
}
void cache_es_version() {
if (WebCtx.search_backend == SQLITE_SEARCH_BACKEND) {
return;
}
static int is_cached = FALSE;
if (is_cached == TRUE) {
@@ -321,6 +334,12 @@ void index_info(struct mg_connection *nc) {
cJSON_AddItemToArray(arr, idx_json);
}
if (WebCtx.search_backend == SQLITE_SEARCH_BACKEND) {
cJSON_AddStringToObject(json, "searchBackend", "sqlite");
} else {
cJSON_AddStringToObject(json, "searchBackend", "elasticsearch");
}
char *json_str = cJSON_PrintUnformatted(json);
web_send_headers(nc, 200, strlen(json_str), "Content-Type: application/json");
@@ -329,54 +348,63 @@ void index_info(struct mg_connection *nc) {
cJSON_Delete(json);
}
cJSON *get_root_document_by_id(const char *index_id, const char *doc_id) {
database_t *db = web_get_database(index_id);
if (!db) {
return NULL;
}
char next_id[SIST_DOC_ID_LEN];
strcpy(next_id, doc_id);
while (TRUE) {
cJSON *doc = database_get_document(db, next_id);
if (doc == NULL) {
return NULL;
}
cJSON *parent = cJSON_GetObjectItem(doc, "parent");
if (parent == NULL || cJSON_IsNull(parent)) {
return doc;
}
strcpy(next_id, parent->valuestring);
cJSON_Delete(parent);
}
}
void file(struct mg_connection *nc, struct mg_http_message *hm) {
if (hm->uri.len != SIST_DOC_ID_LEN + 2) {
if (hm->uri.len != SIST_INDEX_ID_LEN + SIST_DOC_ID_LEN + 2) {
LOG_DEBUGF("serve.c", "Invalid file path: %.*s", (int) hm->uri.len, hm->uri.ptr);
HTTP_REPLY_NOT_FOUND
return;
}
char arg_doc_id[SIST_DOC_ID_LEN];
memcpy(arg_doc_id, hm->uri.ptr + 3, SIST_DOC_ID_LEN);
char arg_index[SIST_INDEX_ID_LEN];
memcpy(arg_index, hm->uri.ptr + 3, SIST_INDEX_ID_LEN);
*(arg_index + SIST_INDEX_ID_LEN - 1) = '\0';
memcpy(arg_doc_id, hm->uri.ptr + 3 + SIST_INDEX_ID_LEN, SIST_DOC_ID_LEN);
*(arg_doc_id + SIST_DOC_ID_LEN - 1) = '\0';
const char *next = arg_doc_id;
cJSON *doc = NULL;
cJSON *index_id = NULL;
cJSON *source = NULL;
while (true) {
doc = elastic_get_document(next);
source = cJSON_GetObjectItem(doc, "_source");
index_id = cJSON_GetObjectItem(source, "index");
if (index_id == NULL) {
cJSON_Delete(doc);
HTTP_REPLY_NOT_FOUND
return;
}
cJSON *parent = cJSON_GetObjectItem(source, "parent");
if (parent == NULL) {
break;
}
next = parent->valuestring;
}
index_t *idx = web_get_index_by_id(index_id->valuestring);
index_t *idx = web_get_index_by_id(arg_index);
if (idx == NULL) {
cJSON_Delete(doc);
HTTP_REPLY_NOT_FOUND
return;
}
cJSON *source = get_root_document_by_id(arg_index, arg_doc_id);
if (strlen(idx->desc.rewrite_url) == 0) {
serve_file_from_disk(source, idx, nc, hm);
} else {
serve_file_from_url(source, idx, nc);
}
cJSON_Delete(doc);
cJSON_Delete(source);
}
void status(struct mg_connection *nc) {
@@ -398,6 +426,10 @@ typedef struct {
tag_req_t *parse_tag_request(cJSON *json) {
if (json == NULL) {
return NULL;
}
if (!cJSON_IsObject(json)) {
return NULL;
}
@@ -425,115 +457,101 @@ tag_req_t *parse_tag_request(cJSON *json) {
return req;
}
subreq_ctx_t *elastic_delete_tag(const tag_req_t *req) {
char *buf = malloc(sizeof(char) * 8192);
snprintf(buf, 8192,
"{"
" \"script\" : {"
" \"source\": \"if (ctx._source.tag.contains(params.tag)) { ctx._source.tag.remove(ctx._source.tag.indexOf(params.tag)) }\","
" \"lang\": \"painless\","
" \"params\" : {"
" \"tag\" : \"%s\""
" }"
" }"
"}", req->name
);
char url[4096];
snprintf(url, sizeof(url), "%s/%s/_update/%s", WebCtx.es_url, WebCtx.es_index, req->doc_id);
return web_post_async(url, buf, WebCtx.es_insecure_ssl);
}
subreq_ctx_t *elastic_write_tag(const tag_req_t *req) {
char *buf = malloc(sizeof(char) * 8192);
snprintf(buf, 8192,
"{"
" \"script\" : {"
" \"source\": \"if(ctx._source.tag == null) {ctx._source.tag = new ArrayList()} ctx._source.tag.add(params.tag)\","
" \"lang\": \"painless\","
" \"params\" : {"
" \"tag\" : \"%s\""
" }"
" }"
"}", req->name
);
char url[4096];
snprintf(url, sizeof(url), "%s/%s/_update/%s", WebCtx.es_url, WebCtx.es_index, req->doc_id);
return web_post_async(url, buf, WebCtx.es_insecure_ssl);
}
void tag(struct mg_connection *nc, struct mg_http_message *hm) {
// if (hm->uri.len != SIST_INDEX_ID_LEN + 4) {
// LOG_DEBUGF("serve.c", "Invalid tag path: %.*s", (int) hm->uri.len, hm->uri.ptr)
// HTTP_REPLY_NOT_FOUND
// return;
// }
//
// char arg_index[SIST_INDEX_ID_LEN];
// memcpy(arg_index, hm->uri.ptr + 5, SIST_INDEX_ID_LEN);
// *(arg_index + SIST_INDEX_ID_LEN - 1) = '\0';
//
// if (hm->body.len < 2 || hm->method.len != 4 || memcmp(&hm->method, "POST", 4) == 0) {
// LOG_DEBUG("serve.c", "Invalid tag request")
// HTTP_REPLY_NOT_FOUND
// return;
// }
//
// store_t *store = get_tag_store(arg_index);
// if (store == NULL) {
// LOG_DEBUGF("serve.c", "Could not get tag store for index: %s", arg_index)
// HTTP_REPLY_NOT_FOUND
// return;
// }
//
// char *body = malloc(hm->body.len + 1);
// memcpy(body, hm->body.ptr, hm->body.len);
// *(body + hm->body.len) = '\0';
// cJSON *json = cJSON_Parse(body);
//
// tag_req_t *arg_req = parse_tag_request(json);
// if (arg_req == NULL) {
// LOG_DEBUGF("serve.c", "Could not parse tag request", arg_index)
// cJSON_Delete(json);
// free(body);
// mg_http_reply(nc, 400, "", "Invalid request");
// return;
// }
//
// cJSON *arr = NULL;
//
// size_t data_len = 0;
// const char *data = store_read(store, arg_req->doc_id, SIST_DOC_ID_LEN, &data_len);
// if (data_len == 0) {
// arr = cJSON_CreateArray();
// } else {
// arr = cJSON_Parse(data);
// }
//
// if (arg_req->delete) {
//
// if (data_len > 0) {
// cJSON *element = NULL;
// int i = 0;
// cJSON_ArrayForEach(element, arr) {
// if (strcmp(element->valuestring, arg_req->name) == 0) {
// cJSON_DeleteItemFromArray(arr, i);
// break;
// }
// i++;
// }
// }
//
// char *buf = malloc(sizeof(char) * 8192);
// snprintf(buf, 8192,
// "{"
// " \"script\" : {"
// " \"source\": \"if (ctx._source.tag.contains(params.tag)) { ctx._source.tag.remove(ctx._source.tag.indexOf(params.tag)) }\","
// " \"lang\": \"painless\","
// " \"params\" : {"
// " \"tag\" : \"%s\""
// " }"
// " }"
// "}", arg_req->name
// );
//
// char url[4096];
// snprintf(url, sizeof(url), "%s/%s/_update/%s", WebCtx.es_url, WebCtx.es_index, arg_req->doc_id);
// nc->fn_data = web_post_async(url, buf, WebCtx.es_insecure_ssl);
//
// } else {
// cJSON_AddItemToArray(arr, cJSON_CreateString(arg_req->name));
//
// char *buf = malloc(sizeof(char) * 8192);
// snprintf(buf, 8192,
// "{"
// " \"script\" : {"
// " \"source\": \"if(ctx._source.tag == null) {ctx._source.tag = new ArrayList()} ctx._source.tag.add(params.tag)\","
// " \"lang\": \"painless\","
// " \"params\" : {"
// " \"tag\" : \"%s\""
// " }"
// " }"
// "}", arg_req->name
// );
//
// char url[4096];
// snprintf(url, sizeof(url), "%s/%s/_update/%s", WebCtx.es_url, WebCtx.es_index, arg_req->doc_id);
// nc->fn_data = web_post_async(url, buf, WebCtx.es_insecure_ssl);
// }
//
// char *json_str = cJSON_PrintUnformatted(arr);
// store_write(store, arg_req->doc_id, SIST_DOC_ID_LEN, json_str, strlen(json_str) + 1);
// store_flush(store);
//
// free(arg_req);
// free(json_str);
// cJSON_Delete(json);
// cJSON_Delete(arr);
// free(body);
if (hm->uri.len != SIST_INDEX_ID_LEN + 4) {
LOG_DEBUGF("serve.c", "Invalid tag path: %.*s", (int) hm->uri.len, hm->uri.ptr);
HTTP_REPLY_NOT_FOUND
return;
}
char arg_index[SIST_INDEX_ID_LEN];
memcpy(arg_index, hm->uri.ptr + 5, SIST_INDEX_ID_LEN);
*(arg_index + SIST_INDEX_ID_LEN - 1) = '\0';
char *body = malloc(hm->body.len + 1);
memcpy(body, hm->body.ptr, hm->body.len);
*(body + hm->body.len) = '\0';
cJSON *json = cJSON_Parse(body);
free(body);
if (json == NULL) {
HTTP_REPLY_BAD_REQUEST
return;
}
database_t *db = web_get_database(arg_index);
if (db == NULL) {
LOG_DEBUGF("serve.c", "Could not get database for index: %s", arg_index);
HTTP_REPLY_NOT_FOUND
return;
}
tag_req_t *req = parse_tag_request(json);
cJSON_Delete(json);
if (req == NULL) {
LOG_DEBUGF("serve.c", "Could not parse tag request", arg_index);
HTTP_REPLY_BAD_REQUEST
return;
}
if (req->delete) {
database_delete_tag(db, req->doc_id, req->name);
if (WebCtx.search_backend == SQLITE_SEARCH_BACKEND) {
database_delete_tag(WebCtx.search_db, req->doc_id, req->name);
HTTP_REPLY_OK
} else {
nc->fn_data = elastic_delete_tag(req);
}
} else {
database_write_tag(db, req->doc_id, req->name);
if (WebCtx.search_backend == SQLITE_SEARCH_BACKEND) {
database_write_tag(WebCtx.search_db, req->doc_id, req->name);
HTTP_REPLY_OK
} else {
nc->fn_data = elastic_write_tag(req);
}
}
free(req);
}
int validate_auth(struct mg_connection *nc, struct mg_http_message *hm) {
@@ -631,11 +649,39 @@ static void ev_router(struct mg_connection *nc, int ev, void *ev_data, UNUSED(vo
return;
}
if (mg_http_match_uri(hm, "/es")) {
search(nc, hm);
} else if (mg_http_match_uri(hm, "/status")) {
if (WebCtx.search_backend == SQLITE_SEARCH_BACKEND) {
if (mg_http_match_uri(hm, "/fts/paths")) {
fts_search_paths(nc, hm);
return;
} else if (mg_http_match_uri(hm, "/fts/mimetypes")) {
fts_search_mimetypes(nc, hm);
return;
} else if (mg_http_match_uri(hm, "/fts/dateRange")) {
fts_search_summary_stats(nc, hm);
return;
} else if (mg_http_match_uri(hm, "/fts/search")) {
fts_search(nc, hm);
return;
} else if (mg_http_match_uri(hm, "/fts/d/*")) {
fts_get_document(nc, hm);
return;
} else if (mg_http_match_uri(hm, "/fts/suggestTags")) {
fts_suggest_tag(nc, hm);
return;
} else if (mg_http_match_uri(hm, "/fts/tags")) {
fts_get_tags(nc, hm);
return;
}
} else if (WebCtx.search_backend == ES_SEARCH_BACKEND) {
if (mg_http_match_uri(hm, "/es")) {
search(nc, hm);
return;
}
}
if (mg_http_match_uri(hm, "/status")) {
status(nc);
} else if (mg_http_match_uri(hm, "/f/*")) {
} else if (mg_http_match_uri(hm, "/f/*/*")) {
file(nc, hm);
} else if (mg_http_match_uri(hm, "/t/*/*/*")) {
thumbnail_with_num(nc, hm);
@@ -702,14 +748,12 @@ void serve(const char *listen_address) {
struct mg_mgr mgr;
mg_mgr_init(&mgr);
int ok = 1;
struct mg_connection *nc = mg_http_listen(&mgr, listen_address, ev_router, NULL);
if (nc == NULL) {
LOG_FATALF("serve.c", "Couldn't bind web server on address %s", listen_address);
}
while (ok) {
while (TRUE) {
mg_mgr_poll(&mgr, 10);
}
mg_mgr_free(&mgr);