utf8 update + bug fixes

.gitmodules

@@ -16,3 +16,12 @@
 [submodule "lmdb"]
 	path = lmdb
 	url = https://github.com/LMDB/lmdb
+[submodule "utf8.h"]
+	path = utf8.h
+	url = https://github.com/sheredom/utf8.h
+[submodule "lib/openjpeg"]
+	path = lib/openjpeg
+	url = https://github.com/uclouvain/openjpeg
+[submodule "lib/harfbuzz"]
+	path = lib/harfbuzz
+	url = https://github.com/harfbuzz/harfbuzz

CMakeLists.txt

@@ -37,6 +37,9 @@ if (WITH_SIST2)
             lmdb/libraries/liblmdb/lmdb.h lmdb/libraries/liblmdb/mdb.c
             lmdb/libraries/liblmdb/midl.h lmdb/libraries/liblmdb/midl.c
             src/cli.c src/cli.h
+
+            # utf8.h
+            utf8.h/utf8.h
     )
 endif ()
 
@@ -67,6 +70,9 @@ if (WITH_SIST2_SCAN)
             lmdb/libraries/liblmdb/lmdb.h lmdb/libraries/liblmdb/mdb.c
             lmdb/libraries/liblmdb/midl.h lmdb/libraries/liblmdb/midl.c
             src/cli.c src/cli.h
+
+            # utf8.h
+            utf8.h/utf8.h
     )
 endif ()
 
@@ -116,10 +122,10 @@ if (WITH_SIST2)
 
     target_compile_options(sist2
             PRIVATE
-#            -O3
+            -Ofast
 #                    -march=native
-#            -fno-stack-protector
+            -fno-stack-protector
-#            -fomit-frame-pointer
+            -fomit-frame-pointer
             )
 
     TARGET_LINK_LIBRARIES(
@@ -150,6 +156,9 @@ if (WITH_SIST2)
             m
             bz2
             magic
+            harfbuzz
+            openjp2
+            freetype
     )
 
 endif ()
@@ -187,7 +196,7 @@ if (WITH_SIST2_SCAN)
     )
     target_compile_options(sist2_scan
             PRIVATE
-            -O3
+            -Ofast
             #        -march=native
             -fno-stack-protector
             -fomit-frame-pointer
@@ -215,6 +224,9 @@ if (WITH_SIST2_SCAN)
 
             pthread
             m
+            ${PROJECT_SOURCE_DIR}/lib/libharfbuzz.a
+            ${PROJECT_SOURCE_DIR}/lib/libopenjp2.a
+            freetype
     )
 endif ()
 

README.md

@@ -58,7 +58,7 @@ File type | Library | Content | Thumbnail | Metadata
 :---|:---|:---|:---|:---
 pdf,xps,cbz,cbr,fb2,epub | MuPDF | yes | yes, `png` | title |
 `audio/*` | libav | - | yes, `jpeg` | ID3 tags |
-`video/*` | libav | - | yes, `jpeg` | *planned* |
+`video/*` | libav | - | yes, `jpeg` | title, comment |
 `image/*` | libav | - | yes, `jpeg` | *planned* |
 ttf,ttc,cff,woff,fnt,otf | Freetype2 | - | yes, `bmp` | Name & style |
 `text/plain` | *(none)* | yes | no | - |

cJSON

@@ -1 +1 @@
-Subproject commit 2de7d04aaf67598e23d06573dcb4e370ebbad410
+Subproject commit 533ff8a783be0d5c81581ab17cd2aeba3f0044c1

lib/harfbuzz

@@ -0,0 +1 @@
+Subproject commit 7cde68f10cdf2c3ff77c1d9077475c0fc034c75c

lib/openjpeg

@@ -0,0 +1 @@
+Subproject commit 5875a6b44618fb7dfd5cd6d742533eaee2014060

mime.csv

@@ -254,6 +254,7 @@ text/mcf, mcf
 text/pascal, pas
 text/plain, com|cmd|conf|def|g|idc|list|lst|mar|sdml|text|txt|md|groovy|license|properties|desktop|ini|rst|cmake|ipynb|readme|less|lo|go|yml|d|cs|hpp|srt
 text/richtext, rt|rtf|rtx
+text/rtf,
 text/scriplet, wsc
 text/x-awk, awk
 !video/x-jng, jng
@@ -263,7 +264,7 @@ image/x-xwindowdump, xwd
 !image/vnd.adobe.photoshop, psd
 text/tab-separated-values, tsv
 text/troff, man|me|ms|roff|t|tr
-text/uri-list, uni|unis|uri|uris
+text/uri-list, uji|unis|uri|uris
 text/vnd.abc, abc
 text/vnd.fmi.flexstor, flx
 text/vnd.wap.wmlscript, wmls
@@ -360,3 +361,53 @@ application/x-wine-extension-ini,
 application/x-cbz, cbz
 application/x-cbr, cbr
 application/x-ms-compress-szdd, fon
+application/x-atari-7800-rom, a78
+application/x-nes-rom, nes
+application/x-font-pfm, pfm
+application/x-gettext-translation,
+image/wmf,
+application/pgp-keys,
+image/x-3ds, 3ds
+application/x-lz4, lz4
+application/vnd.openxmlformats-officedocument.presentationml.presentation, pptx
+application/vnd.oasis.opendocument.presentation, odp
+application/x-msaccess, accdb
+application/vnd.oasis.opendocument.spreadsheet, ods
+audio/x-aiff, aiff|aif
+text/x-ms-regedit, reg
+application/x-gamecube-rom,
+application/x-nintendo-ds-rom,
+text/x-objective-c,
+application/x-font-gdos,
+application/x-apple-diskimage,
+application/x-zstd, zst
+video/x-m4v, m4v
+message/news,
+application/vnd.symbian.install,
+application/x-lzh-compressed,
+application/x-dosdriver,
+application/vnd.tcpdump.pcap, pcap
+x-epoc/x-sisx-app,
+application/x-avira-qua,
+video/MP2T,
+application/x-snappy-framed,
+application/x-lz4+json, jsonlz4
+application/x-dmp, dmp
+application/zlib, z
+application/x-pgp-keyring,
+application/x-gdbm,
+application/x-font-pf2, pf2
+application/x-zip,
+application/x-coredump,
+application/x-java-jmod, jmod
+application/x-terminfo,
+application/x-terminfo2,
+application/x-arc,
+application/vnd.lotus-1-2-3,
+image/x-win-bitmap,
+application/x-maxis-dbpf,
+text/PGP,
+audio/x-hx-aac-adts,
+application/x-chrome-extension,
+image/heic, heic
+image/x-gem,

scripts/get_static_libs.sh

@@ -2,12 +2,28 @@
 cd lib
 
 cd mupdf
-HAVE_X11=no HAVE_GLUT=no make -j 4
+USE_SYSTEM_HARFBUZZ=yes USE_SYSTEM_OPENJPEG=yes HAVE_X11=no HAVE_GLUT=no make -j 4
 cd ..
 
 mv mupdf/build/release/libmupdf.a .
 mv mupdf/build/release/libmupdf-third.a .
 
+# openjp2
+cd openjpeg
+#cmake . -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS="-O3 -march=native -DNDEBUG"
+cmake . -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS="-O3"
+make -j 4
+cd ..
+mv openjpeg/bin/libopenjp2.a .
+
+# harfbuzz
+cd harfbuzz
+./autogen.sh
+./configure --disable-shared --enable-static
+make -j 4
+cd ..
+mv harfbuzz/src/.libs/libharfbuzz.a .
+
 # ffmpeg
 cd ffmpeg
 ./configure --disable-shared --enable-static --disable-ffmpeg --disable-ffplay \

scripts/get_static_libs_freebsd.sh

@@ -9,6 +9,22 @@ cd ..
 mv mupdf/build/release/libmupdf.a .
 mv mupdf/build/release/libmupdf-third.a .
 
+# openjp2
+cd openjpeg
+#cmake . -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS="-O3 -march=native -DNDEBUG"
+cmake . -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS="-O3"
+gmake -j 4
+cd ..
+mv openjpeg/bin/libopenjp2.a .
+
+# harfbuzz
+cd harfbuzz
+./autogen.sh
+./configure --disable-shared --enable-static
+gmake -j 4
+cd ..
+mv harfbuzz/src/.libs/libharfbuzz.a .
+
 # ffmpeg
 cd ffmpeg
 ./configure --disable-shared --enable-static --disable-ffmpeg --disable-ffplay \

scripts/mime.py

@@ -12,7 +12,8 @@ major_mime = {
     "audio": 7,
     "image": 8,
     "text": 9,
-    "application": 10
+    "application": 10,
+    "x-epoc": 11,
 }
 
 pdf = (

src/index/elastic.c

@@ -102,7 +102,14 @@ void elastic_flush() {
 
     cJSON *ret_json = cJSON_Parse(r->body);
     if (cJSON_GetObjectItem(ret_json, "errors")->valueint != 0) {
-        fprintf(stderr, "%s\n", r->body);
+        cJSON *err;
+        cJSON_ArrayForEach(err, cJSON_GetObjectItem(ret_json, "items")) {
+            if (cJSON_GetObjectItem(cJSON_GetObjectItem(err, "index"), "status")->valueint != 201) {
+                char* str = cJSON_Print(err);
+                fprintf(stderr, "%s\n", str);
+                cJSON_free(str);
+            }
+        }
     }
 
     cJSON_Delete(ret_json);

src/io/serialize.c

@@ -236,7 +236,7 @@ void read_index(const char *path, const char index_id[UUID_STR_LEN], index_func
                 case MetaTitle: {
                     buf.cur = 0;
                     while ((c = getc(file)) != 0) {
-                        if (!(SHOULD_IGNORE_CHAR(c)) || c == ' ') {
+                        if (SHOULD_KEEP_CHAR(c) || c == ' ') {
                             dyn_buffer_write_char(&buf, (char) c);
                         }
                     }
@@ -244,6 +244,9 @@ void read_index(const char *path, const char index_id[UUID_STR_LEN], index_func
                     cJSON_AddStringToObject(document, get_meta_key_text(key), buf.buf);
                     break;
                 }
+                default:
+                    fprintf(stderr, "Invalid meta key (corrupt index): %x", key);
+                    break;
             }
 
             key = getc(file);

src/io/walk.c

@@ -3,10 +3,9 @@
 
 parse_job_t *create_parse_job(const char *filepath, const struct stat *info, int base) {
     int len = (int) strlen(filepath);
-
     parse_job_t *job = malloc(sizeof(parse_job_t) + len);
 
-    memcpy(&(job->filepath), filepath, len + 1);
+    strcpy(job->filepath, filepath);
     job->base = base;
     char *p = strrchr(filepath + base, '.');
     if (p != NULL) {

src/main.c

@@ -10,7 +10,7 @@
 #define EPILOG "Made by simon987 <me@simon987.net>. Released under GPL-3.0"
 
 
-static const char *const Version = "1.0.14";
+static const char *const Version = "1.1.0";
 static const char *const usage[] = {
         "sist2 scan [OPTION]... PATH",
         "sist2 index [OPTION]... INDEX",
@@ -52,11 +52,10 @@ void sist2_scan(scan_args_t *args) {
     ScanCtx.tn_qscale = args->quality;
     ScanCtx.tn_size = args->size;
     ScanCtx.content_size = args->content_size;
-    ScanCtx.pool = tpool_create(args->threads, serializer_cleanup);
     ScanCtx.threads = args->threads;
     strncpy(ScanCtx.index.path, args->output, sizeof(ScanCtx.index.path));
     strncpy(ScanCtx.index.desc.name, args->name, sizeof(ScanCtx.index.desc.name));
-    strcpy(ScanCtx.index.desc.root, args->path);
+    strncpy(ScanCtx.index.desc.root, args->path, sizeof(ScanCtx.index.desc.root));
     ScanCtx.index.desc.root_len = (short) strlen(ScanCtx.index.desc.root);
 
     init_dir(ScanCtx.index.path);
@@ -93,6 +92,8 @@ void sist2_scan(scan_args_t *args) {
         printf("Loaded %d items in to mtime table.", g_hash_table_size(ScanCtx.original_table));
     }
 
+    ScanCtx.pool = tpool_create(args->threads, serializer_cleanup);
+    tpool_start(ScanCtx.pool);
     walk_directory_tree(ScanCtx.index.desc.root);
     tpool_wait(ScanCtx.pool);
     tpool_destroy(ScanCtx.pool);

src/parsing/font.c

@@ -184,10 +184,14 @@ void parse_font(const char *buf, size_t buf_len, document_t *doc) {
     for (int i = 0; i < num_chars; i++) {
         char c = font_name[i];
 
+        err = FT_Load_Char(face, c, FT_LOAD_NO_HINTING | FT_LOAD_RENDER);
+        if (err != 0) {
+            c = c >= 'a' && c <= 'z' ? c - 32 : c + 32;
             err = FT_Load_Char(face, c, FT_LOAD_NO_HINTING | FT_LOAD_RENDER);
             if (err != 0) {
                 continue;
             }
+        }
         glyph_t glyph = ft_glyph_to_glyph(face->glyph);
 
         pen.x += kerning_offset(c, pc, face);

src/parsing/media.c

@@ -1,6 +1,9 @@
 #include "src/sist.h"
 #include "src/ctx.h"
 
+#define MIN_SIZE 32
+
+__always_inline
 AVCodecContext *alloc_jpeg_encoder(int dstW, int dstH, float qscale) {
 
     AVCodec *jpeg_codec = avcodec_find_encoder(AV_CODEC_ID_MJPEG);
@@ -22,8 +25,8 @@ AVCodecContext *alloc_jpeg_encoder(int dstW, int dstH, float qscale) {
     return jpeg;
 }
 
+__always_inline
 AVFrame *scale_frame(const AVCodecContext *decoder, const AVFrame *frame, int size) {
-    AVFrame *scaled_frame = av_frame_alloc();
 
     int dstW;
     int dstH;
@@ -41,16 +44,22 @@ AVFrame *scale_frame(const AVCodecContext *decoder, const AVFrame *frame, int si
         }
     }
 
+    if (dstW <= MIN_SIZE || dstH <= MIN_SIZE) {
+        return NULL;
+    }
+
+    AVFrame *scaled_frame = av_frame_alloc();
+
     struct SwsContext *ctx = sws_getContext(
             decoder->width, decoder->height, decoder->pix_fmt,
             dstW, dstH, AV_PIX_FMT_YUVJ420P,
             SWS_FAST_BILINEAR, 0, 0, 0
     );
 
-    int dst_buf_len = avpicture_get_size(AV_PIX_FMT_YUVJ420P, dstW, dstH);
+    int dst_buf_len = av_image_get_buffer_size(AV_PIX_FMT_YUV420P, dstW, dstH, 1);
     uint8_t *dst_buf = (uint8_t *) av_malloc(dst_buf_len);
 
-    avpicture_fill((AVPicture *) scaled_frame, dst_buf, AV_PIX_FMT_YUVJ420P, dstW, dstH);
+    av_image_fill_arrays(scaled_frame->data, scaled_frame->linesize, dst_buf, AV_PIX_FMT_YUV420P, dstW, dstH, 1);
 
     sws_scale(ctx,
               (const uint8_t *const *) frame->data, frame->linesize,
@@ -81,7 +90,7 @@ AVFrame *read_frame(AVFormatContext *pFormatCtx, AVCodecContext *decoder, int st
 
             if (read_frame_ret != 0) {
                 if (read_frame_ret != AVERROR_EOF) {
-                    fprintf(stderr, "Error reading frame: %s\n", av_err2str(read_frame_ret));
+                    fprintf(stderr, "Error reading frame: %d\n", read_frame_ret);
                 }
                 av_frame_free(&frame);
                 av_packet_unref(&avPacket);
@@ -108,35 +117,40 @@ AVFrame *read_frame(AVFormatContext *pFormatCtx, AVCodecContext *decoder, int st
 }
 
 #define APPEND_TAG_META(doc, tag, keyname) \
-    text_buffer_t tex = text_buffer_create(4096); \
+    text_buffer_t tex = text_buffer_create(-1); \
-    text_buffer_append_string(&tex, tag->value); \
+    text_buffer_append_string0(&tex, tag->value); \
     meta_line_t *meta_tag = malloc(sizeof(meta_line_t) + tex.dyn_buffer.cur); \
     meta_tag->key = keyname; \
     strcpy(meta_tag->strval, tex.dyn_buffer.buf); \
     APPEND_META(doc, meta_tag) \
     text_buffer_destroy(&tex);
 
+__always_inline
 void append_audio_meta(AVFormatContext *pFormatCtx, document_t *doc) {
 
     AVDictionaryEntry *tag = NULL;
     while ((tag = av_dict_get(pFormatCtx->metadata, "", tag, AV_DICT_IGNORE_SUFFIX))) {
-        char *key = tag->key;
+        char key[32];
-        for (; *key; ++key) *key = (char) tolower(*key);
+        strncpy(key, tag->key, sizeof(key));
 
-        if (strcmp(tag->key, "artist") == 0) {
+        char *ptr = key;
+        for (; *ptr; ++ptr) *ptr = (char) tolower(*ptr);
+
+        if (strcmp(key, "artist") == 0) {
             APPEND_TAG_META(doc, tag, MetaArtist)
-        } else if (strcmp(tag->key, "genre") == 0) {
+        } else if (strcmp(key, "genre") == 0) {
             APPEND_TAG_META(doc, tag, MetaGenre)
-        } else if (strcmp(tag->key, "title") == 0) {
+        } else if (strcmp(key, "title") == 0) {
             APPEND_TAG_META(doc, tag, MetaTitle)
-        } else if (strcmp(tag->key, "album_artist") == 0) {
+        } else if (strcmp(key, "album_artist") == 0) {
             APPEND_TAG_META(doc, tag, MetaAlbumArtist)
-        } else if (strcmp(tag->key, "album") == 0) {
+        } else if (strcmp(key, "album") == 0) {
             APPEND_TAG_META(doc, tag, MetaAlbum)
         }
     }
 }
 
+__always_inline
 void append_video_meta(AVFormatContext *pFormatCtx, document_t *doc, int include_audio_tags) {
 
     meta_line_t *meta_duration = malloc(sizeof(meta_line_t));
@@ -146,17 +160,20 @@ void append_video_meta(AVFormatContext *pFormatCtx, document_t *doc, int include
 
     meta_line_t *meta_bitrate = malloc(sizeof(meta_line_t));
     meta_bitrate->key = MetaMediaBitrate;
-    meta_bitrate->intval = pFormatCtx->bit_rate;
+    meta_bitrate->longval = pFormatCtx->bit_rate;
     APPEND_META(doc, meta_bitrate)
 
     AVDictionaryEntry *tag = NULL;
     while ((tag = av_dict_get(pFormatCtx->metadata, "", tag, AV_DICT_IGNORE_SUFFIX))) {
-        char *key = tag->key;
+        char key[32];
-        for (; *key; ++key) *key = (char) tolower(*key);
+        strncpy(key, tag->key, sizeof(key));
 
-        if (strcmp(tag->key, "title") == 0 && include_audio_tags) {
+        char *ptr = key;
+        for (; *ptr; ++ptr) *ptr = (char) tolower(*ptr);
+
+        if (strcmp(key, "title") == 0 && include_audio_tags) {
             APPEND_TAG_META(doc, tag, MetaTitle)
-        } else if (strcmp(tag->key, "comment") == 0) {
+        } else if (strcmp(key, "comment") == 0) {
             APPEND_TAG_META(doc, tag, MetaContent)
         }
     }
@@ -174,7 +191,7 @@ void parse_media(const char *filepath, document_t *doc) {
     }
     int res = avformat_open_input(&pFormatCtx, filepath, NULL, NULL);
     if (res < 0) {
-        printf("ERR%s %s\n", filepath, av_err2str(res));
+        fprintf(stderr, "media error: %s %s\n", filepath, av_err2str(res));
         return;
     }
 
@@ -224,7 +241,7 @@ void parse_media(const char *filepath, document_t *doc) {
             append_video_meta(pFormatCtx, doc, audio_stream == -1);
         }
 
-        if (stream->codecpar->width <= 20 || stream->codecpar->height <= 20) {
+        if (stream->codecpar->width <= MIN_SIZE || stream->codecpar->height <= MIN_SIZE) {
             avformat_close_input(&pFormatCtx);
             avformat_free_context(pFormatCtx);
             return;
@@ -259,6 +276,14 @@ void parse_media(const char *filepath, document_t *doc) {
         // Scale frame
         AVFrame *scaled_frame = scale_frame(decoder, frame, ScanCtx.tn_size);
 
+        if (scaled_frame == NULL) {
+            av_frame_free(&frame);
+            avcodec_free_context(&decoder);
+            avformat_close_input(&pFormatCtx);
+            avformat_free_context(pFormatCtx);
+            return;
+        }
+
         // Encode frame to jpeg
         AVCodecContext *jpeg_encoder = alloc_jpeg_encoder(scaled_frame->width, scaled_frame->height, ScanCtx.tn_qscale);
         avcodec_send_frame(jpeg_encoder, scaled_frame);
@@ -268,7 +293,8 @@ void parse_media(const char *filepath, document_t *doc) {
         avcodec_receive_packet(jpeg_encoder, &jpeg_packet);
 
         // Save thumbnail
-        store_write(ScanCtx.index.store, (char *) doc->uuid, sizeof(doc->uuid), (char *) jpeg_packet.data, jpeg_packet.size);
+        store_write(ScanCtx.index.store, (char *) doc->uuid, sizeof(doc->uuid), (char *) jpeg_packet.data,
+                    jpeg_packet.size);
 
         av_packet_unref(&jpeg_packet);
         av_frame_free(&frame);

src/parsing/media.h

@@ -5,6 +5,7 @@
 #include "src/sist.h"
 
 #define MIN_VIDEO_SIZE 1024 * 64
+#define MIN_IMAGE_SIZE 1024 * 2
 
 void parse_media(const char * filepath, document_t *doc);
 

src/parsing/mime.c

@@ -1,10 +1,12 @@
 #include "mime.h"
 
 unsigned int mime_get_mime_by_ext(GHashTable *ext_table, const char * ext) {
-    char lower[64];
+    char lower[8];
     char *p = lower;
-    while ((*ext)) {
+    int cnt = 0;
+    while ((*ext) != '\0' && cnt + 1 < sizeof(lower)) {
         *p++ = (char)tolower(*ext++);
+        cnt++;
     }
     *p = '\0';
     return (size_t) g_hash_table_lookup(ext_table, lower);

src/parsing/mime_generated.c

@@ -39,334 +39,385 @@ enum mime {
     application_oda=655391,
     application_ogg=655392,
     application_pdf=655393 | 0x40000000,
-    application_pgp_signature=655394,
+    application_pgp_keys=655394,
-    application_pkcs7_signature=655395,
+    application_pgp_signature=655395,
-    application_pkix_cert=655396,
+    application_pkcs7_signature=655396,
-    application_postscript=655397,
+    application_pkix_cert=655397,
-    application_pro_eng=655398,
+    application_postscript=655398,
-    application_ringing_tones=655399,
+    application_pro_eng=655399,
-    application_smil=655400,
+    application_ringing_tones=655400,
-    application_solids=655401,
+    application_smil=655401,
-    application_sounder=655402,
+    application_solids=655402,
-    application_step=655403,
+    application_sounder=655403,
-    application_streamingmedia=655404,
+    application_step=655404,
-    application_vda=655405,
+    application_streamingmedia=655405,
-    application_vnd_fdf=655406,
+    application_vda=655406,
-    application_vnd_font_fontforge_sfd=655407,
+    application_vnd_fdf=655407,
-    application_vnd_hp_hpgl=655408,
+    application_vnd_font_fontforge_sfd=655408,
-    application_vnd_iccprofile=655409,
+    application_vnd_hp_hpgl=655409,
-    application_vnd_ms_cab_compressed=655410,
+    application_vnd_iccprofile=655410,
-    application_vnd_ms_excel=655411,
+    application_vnd_lotus_1_2_3=655411,
-    application_vnd_ms_fontobject=655412,
+    application_vnd_ms_cab_compressed=655412,
-    application_vnd_ms_opentype=655413 | 0x20000000,
+    application_vnd_ms_excel=655413,
-    application_vnd_ms_pki_certstore=655414,
+    application_vnd_ms_fontobject=655414,
-    application_vnd_ms_pki_pko=655415,
+    application_vnd_ms_opentype=655415 | 0x20000000,
-    application_vnd_ms_pki_seccat=655416,
+    application_vnd_ms_pki_certstore=655416,
-    application_vnd_ms_powerpoint=655417,
+    application_vnd_ms_pki_pko=655417,
-    application_vnd_ms_project=655418,
+    application_vnd_ms_pki_seccat=655418,
-    application_vnd_oasis_opendocument_base=655419,
+    application_vnd_ms_powerpoint=655419,
-    application_vnd_oasis_opendocument_formula=655420,
+    application_vnd_ms_project=655420,
-    application_vnd_oasis_opendocument_graphics=655421,
+    application_vnd_oasis_opendocument_base=655421,
-    application_vnd_oasis_opendocument_text=655422,
+    application_vnd_oasis_opendocument_formula=655422,
-    application_vnd_openxmlformats_officedocument_spreadsheetml_sheet=655423,
+    application_vnd_oasis_opendocument_graphics=655423,
-    application_vnd_openxmlformats_officedocument_wordprocessingml_document=655424,
+    application_vnd_oasis_opendocument_presentation=655424,
-    application_vnd_wap_wmlc=655425,
+    application_vnd_oasis_opendocument_spreadsheet=655425,
-    application_vnd_wap_wmlscriptc=655426,
+    application_vnd_oasis_opendocument_text=655426,
-    application_vnd_xara=655427,
+    application_vnd_openxmlformats_officedocument_presentationml_presentation=655427,
-    application_vocaltec_media_desc=655428,
+    application_vnd_openxmlformats_officedocument_spreadsheetml_sheet=655428,
-    application_vocaltec_media_file=655429,
+    application_vnd_openxmlformats_officedocument_wordprocessingml_document=655429,
-    application_winhelp=655430,
+    application_vnd_symbian_install=655430,
-    application_wordperfect=655431,
+    application_vnd_tcpdump_pcap=655431,
-    application_wordperfect6_0=655432,
+    application_vnd_wap_wmlc=655432,
-    application_wordperfect6_1=655433,
+    application_vnd_wap_wmlscriptc=655433,
-    application_x_123=655434,
+    application_vnd_xara=655434,
-    application_x_7z_compressed=655435,
+    application_vocaltec_media_desc=655435,
-    application_x_aim=655436,
+    application_vocaltec_media_file=655436,
-    application_x_archive=655437,
+    application_winhelp=655437,
-    application_x_authorware_bin=655438,
+    application_wordperfect=655438,
-    application_x_authorware_map=655439,
+    application_wordperfect6_0=655439,
-    application_x_authorware_seg=655440,
+    application_wordperfect6_1=655440,
-    application_x_bcpio=655441,
+    application_x_123=655441,
-    application_x_bittorrent=655442,
+    application_x_7z_compressed=655442,
-    application_x_bsh=655443,
+    application_x_aim=655443,
-    application_x_bytecode_python=655444,
+    application_x_apple_diskimage=655444,
-    application_x_bzip=655445,
+    application_x_arc=655445,
-    application_x_bzip2=655446,
+    application_x_archive=655446,
-    application_x_cbr=655447,
+    application_x_atari_7800_rom=655447,
-    application_x_cbz=655448 | 0x40000000,
+    application_x_authorware_bin=655448,
-    application_x_cdlink=655449,
+    application_x_authorware_map=655449,
-    application_x_chat=655450,
+    application_x_authorware_seg=655450,
-    application_x_cocoa=655451,
+    application_x_avira_qua=655451,
-    application_x_conference=655452,
+    application_x_bcpio=655452,
-    application_x_cpio=655453,
+    application_x_bittorrent=655453,
-    application_x_dbf=655454,
+    application_x_bsh=655454,
-    application_x_dbt=655455,
+    application_x_bytecode_python=655455,
-    application_x_debian_package=655456,
+    application_x_bzip=655456,
-    application_x_deepv=655457,
+    application_x_bzip2=655457,
-    application_x_director=655458,
+    application_x_cbr=655458,
-    application_x_dosexec=655459,
+    application_x_cbz=655459 | 0x40000000,
-    application_x_dvi=655460,
+    application_x_cdlink=655460,
-    application_x_elc=655461,
+    application_x_chat=655461,
+    application_x_chrome_extension=655462,
+    application_x_cocoa=655463,
+    application_x_conference=655464,
+    application_x_coredump=655465,
+    application_x_cpio=655466,
+    application_x_dbf=655467,
+    application_x_dbt=655468,
+    application_x_debian_package=655469,
+    application_x_deepv=655470,
+    application_x_director=655471,
+    application_x_dmp=655472,
+    application_x_dosdriver=655473,
+    application_x_dosexec=655474,
+    application_x_dvi=655475,
+    application_x_elc=655476,
     application_x_empty=1,
-    application_x_envoy=655463,
+    application_x_envoy=655478,
-    application_x_esrehber=655464,
+    application_x_esrehber=655479,
-    application_x_excel=655465,
+    application_x_excel=655480,
-    application_x_executable=655466,
+    application_x_executable=655481,
-    application_x_font_sfn=655467,
+    application_x_font_gdos=655482,
-    application_x_font_ttf=655468 | 0x20000000,
+    application_x_font_pf2=655483,
-    application_x_freelance=655469,
+    application_x_font_pfm=655484,
-    application_x_git=655470,
+    application_x_font_sfn=655485,
-    application_x_gsp=655471,
+    application_x_font_ttf=655486 | 0x20000000,
-    application_x_gss=655472,
+    application_x_freelance=655487,
-    application_x_gtar=655473,
+    application_x_gamecube_rom=655488,
-    application_x_gzip=655474,
+    application_x_gdbm=655489,
-    application_x_hdf=655475,
+    application_x_gettext_translation=655490,
-    application_x_helpfile=655476,
+    application_x_git=655491,
-    application_x_httpd_imap=655477,
+    application_x_gsp=655492,
-    application_x_ima=655478,
+    application_x_gss=655493,
-    application_x_innosetup=655479,
+    application_x_gtar=655494,
-    application_x_internett_signup=655480,
+    application_x_gzip=655495,
-    application_x_inventor=655481,
+    application_x_hdf=655496,
-    application_x_ip2=655482,
+    application_x_helpfile=655497,
-    application_x_java_applet=655483,
+    application_x_httpd_imap=655498,
-    application_x_java_commerce=655484,
+    application_x_ima=655499,
-    application_x_java_image=655485,
+    application_x_innosetup=655500,
-    application_x_java_keystore=655486,
+    application_x_internett_signup=655501,
-    application_x_kdelnk=655487,
+    application_x_inventor=655502,
-    application_x_koan=655488,
+    application_x_ip2=655503,
-    application_x_latex=655489,
+    application_x_java_applet=655504,
-    application_x_livescreen=655490,
+    application_x_java_commerce=655505,
-    application_x_lotus=655491,
+    application_x_java_image=655506,
-    application_x_lzh=655492,
+    application_x_java_jmod=655507,
-    application_x_lzx=655493,
+    application_x_java_keystore=655508,
-    application_x_mach_binary=655494,
+    application_x_kdelnk=655509,
-    application_x_mach_executable=655495,
+    application_x_koan=655510,
-    application_x_magic_cap_package_1_0=655496,
+    application_x_latex=655511,
-    application_x_mathcad=655497,
+    application_x_livescreen=655512,
-    application_x_meme=655498,
+    application_x_lotus=655513,
-    application_x_midi=655499,
+    application_x_lz4=655514,
-    application_x_mif=655500,
+    application_x_lz4_json=655515,
-    application_x_mix_transfer=655501,
+    application_x_lzh=655516,
-    application_x_mobipocket_ebook=655502,
+    application_x_lzh_compressed=655517,
-    application_x_ms_compress_szdd=655503,
+    application_x_lzx=655518,
-    application_x_ms_pdb=655504,
+    application_x_mach_binary=655519,
-    application_x_ms_reader=655505,
+    application_x_mach_executable=655520,
-    application_x_navi_animation=655506,
+    application_x_magic_cap_package_1_0=655521,
-    application_x_navidoc=655507,
+    application_x_mathcad=655522,
-    application_x_navimap=655508,
+    application_x_maxis_dbpf=655523,
-    application_x_navistyle=655509,
+    application_x_meme=655524,
-    application_x_netcdf=655510,
+    application_x_midi=655525,
-    application_x_newton_compatible_pkg=655511,
+    application_x_mif=655526,
-    application_x_object=655512,
+    application_x_mix_transfer=655527,
-    application_x_omc=655513,
+    application_x_mobipocket_ebook=655528,
-    application_x_omcdatamaker=655514,
+    application_x_ms_compress_szdd=655529,
-    application_x_omcregerator=655515,
+    application_x_ms_pdb=655530,
-    application_x_pagemaker=655516,
+    application_x_ms_reader=655531,
-    application_x_pcl=655517,
+    application_x_msaccess=655532,
-    application_x_pixclscript=655518,
+    application_x_navi_animation=655533,
-    application_x_pkcs7_certreqresp=655519,
+    application_x_navidoc=655534,
-    application_x_pkcs7_signature=655520,
+    application_x_navimap=655535,
-    application_x_project=655521,
+    application_x_navistyle=655536,
-    application_x_qpro=655522,
+    application_x_nes_rom=655537,
-    application_x_rar=655523,
+    application_x_netcdf=655538,
-    application_x_rpm=655524,
+    application_x_newton_compatible_pkg=655539,
-    application_x_sdp=655525,
+    application_x_nintendo_ds_rom=655540,
-    application_x_sea=655526,
+    application_x_object=655541,
-    application_x_seelogo=655527,
+    application_x_omc=655542,
-    application_x_setupscript=655528,
+    application_x_omcdatamaker=655543,
-    application_x_shar=655529,
+    application_x_omcregerator=655544,
-    application_x_sharedlib=655530,
+    application_x_pagemaker=655545,
-    application_x_shockwave_flash=655531,
+    application_x_pcl=655546,
-    application_x_sprite=655532,
+    application_x_pgp_keyring=655547,
-    application_x_sqlite3=655533,
+    application_x_pixclscript=655548,
-    application_x_sv4cpio=655534,
+    application_x_pkcs7_certreqresp=655549,
-    application_x_sv4crc=655535,
+    application_x_pkcs7_signature=655550,
-    application_x_tar=655536,
+    application_x_project=655551,
-    application_x_tbook=655537,
+    application_x_qpro=655552,
-    application_x_tex_tfm=655538,
+    application_x_rar=655553,
-    application_x_texinfo=655539,
+    application_x_rpm=655554,
-    application_x_ustar=655540,
+    application_x_sdp=655555,
-    application_x_visio=655541,
+    application_x_sea=655556,
-    application_x_vnd_audioexplosion_mzz=655542,
+    application_x_seelogo=655557,
-    application_x_vnd_ls_xpix=655543,
+    application_x_setupscript=655558,
-    application_x_vrml=655544,
+    application_x_shar=655559,
-    application_x_wais_source=655545,
+    application_x_sharedlib=655560,
-    application_x_wine_extension_ini=655546,
+    application_x_shockwave_flash=655561,
-    application_x_wintalk=655547,
+    application_x_snappy_framed=655562,
-    application_x_world=655548,
+    application_x_sprite=655563,
-    application_x_wri=655549,
+    application_x_sqlite3=655564,
-    application_x_x509_ca_cert=655550,
+    application_x_sv4cpio=655565,
-    application_x_xz=655551,
+    application_x_sv4crc=655566,
-    application_xml=655552,
+    application_x_tar=655567,
-    application_zip=655553,
+    application_x_tbook=655568,
-    audio_it=458946,
+    application_x_terminfo=655569,
-    audio_make=458947,
+    application_x_terminfo2=655570,
-    audio_mid=458948,
+    application_x_tex_tfm=655571,
-    audio_midi=458949,
+    application_x_texinfo=655572,
-    audio_mp4=458950,
+    application_x_ustar=655573,
-    audio_mpeg=458951,
+    application_x_visio=655574,
-    audio_ogg=458952,
+    application_x_vnd_audioexplosion_mzz=655575,
-    audio_s3m=458953,
+    application_x_vnd_ls_xpix=655576,
-    audio_tsp_audio=458954,
+    application_x_vrml=655577,
-    audio_tsplayer=458955,
+    application_x_wais_source=655578,
-    audio_vnd_qcelp=458956,
+    application_x_wine_extension_ini=655579,
-    audio_voxware=458957,
+    application_x_wintalk=655580,
-    audio_x_flac=458958,
+    application_x_world=655581,
-    audio_x_gsm=458959,
+    application_x_wri=655582,
-    audio_x_jam=458960,
+    application_x_x509_ca_cert=655583,
-    audio_x_liveaudio=458961,
+    application_x_xz=655584,
-    audio_x_m4a=458962,
+    application_x_zip=655585,
-    audio_x_midi=458963,
+    application_x_zstd=655586,
-    audio_x_mod=458964,
+    application_xml=655587,
-    audio_x_mp4a_latm=458965,
+    application_zip=655588,
-    audio_x_mpeg_3=458966,
+    application_zlib=655589,
-    audio_x_mpequrl=458967,
+    audio_it=458982,
-    audio_x_nspaudio=458968,
+    audio_make=458983,
-    audio_x_pn_realaudio=458969,
+    audio_mid=458984,
-    audio_x_psid=458970,
+    audio_midi=458985,
-    audio_x_realaudio=458971,
+    audio_mp4=458986,
-    audio_x_twinvq=458972,
+    audio_mpeg=458987,
-    audio_x_twinvq_plugin=458973,
+    audio_ogg=458988,
-    audio_x_voc=458974,
+    audio_s3m=458989,
-    audio_x_wav=458975,
+    audio_tsp_audio=458990,
-    audio_xm=458976,
+    audio_tsplayer=458991,
-    font_otf=327905 | 0x20000000,
+    audio_vnd_qcelp=458992,
-    font_sfnt=327906 | 0x20000000,
+    audio_voxware=458993,
-    font_woff=327907 | 0x20000000,
+    audio_x_aiff=458994,
-    font_woff2=327908 | 0x20000000,
+    audio_x_flac=458995,
-    image_cmu_raster=524517,
+    audio_x_gsm=458996,
-    image_fif=524518,
+    audio_x_hx_aac_adts=458997,
-    image_florian=524519,
+    audio_x_jam=458998,
-    image_g3fax=524520,
+    audio_x_liveaudio=458999,
-    image_gif=524521,
+    audio_x_m4a=459000,
-    image_ief=524522,
+    audio_x_midi=459001,
-    image_jpeg=524523,
+    audio_x_mod=459002,
-    image_jutvision=524524,
+    audio_x_mp4a_latm=459003,
-    image_naplps=524525,
+    audio_x_mpeg_3=459004,
-    image_pict=524526,
+    audio_x_mpequrl=459005,
-    image_png=524527,
+    audio_x_nspaudio=459006,
-    image_svg=524528 | 0x80000000,
+    audio_x_pn_realaudio=459007,
-    image_svg_xml=524529 | 0x80000000,
+    audio_x_psid=459008,
-    image_tiff=524530,
+    audio_x_realaudio=459009,
-    image_vnd_adobe_photoshop=524531 | 0x80000000,
+    audio_x_twinvq=459010,
-    image_vnd_djvu=524532 | 0x80000000,
+    audio_x_twinvq_plugin=459011,
-    image_vnd_fpx=524533,
+    audio_x_voc=459012,
-    image_vnd_microsoft_icon=524534,
+    audio_x_wav=459013,
-    image_vnd_rn_realflash=524535,
+    audio_xm=459014,
-    image_vnd_rn_realpix=524536,
+    font_otf=327943 | 0x20000000,
-    image_vnd_wap_wbmp=524537,
+    font_sfnt=327944 | 0x20000000,
-    image_vnd_xiff=524538,
+    font_woff=327945 | 0x20000000,
-    image_webp=524539,
+    font_woff2=327946 | 0x20000000,
-    image_x_cmu_raster=524540,
+    image_cmu_raster=524555,
-    image_x_cur=524541,
+    image_fif=524556,
-    image_x_dwg=524542,
+    image_florian=524557,
-    image_x_eps=524543,
+    image_g3fax=524558,
-    image_x_exr=524544,
+    image_gif=524559,
-    image_x_icns=524545,
+    image_heic=524560,
-    image_x_icon=524546 | 0x80000000,
+    image_ief=524561,
-    image_x_jg=524547,
+    image_jpeg=524562,
-    image_x_jps=524548,
+    image_jutvision=524563,
-    image_x_ms_bmp=524549,
+    image_naplps=524564,
-    image_x_niff=524550,
+    image_pict=524565,
-    image_x_pcx=524551,
+    image_png=524566,
-    image_x_pict=524552,
+    image_svg=524567 | 0x80000000,
-    image_x_portable_bitmap=524553,
+    image_svg_xml=524568 | 0x80000000,
-    image_x_portable_graymap=524554,
+    image_tiff=524569,
-    image_x_portable_pixmap=524555,
+    image_vnd_adobe_photoshop=524570 | 0x80000000,
-    image_x_quicktime=524556,
+    image_vnd_djvu=524571 | 0x80000000,
-    image_x_rgb=524557,
+    image_vnd_fpx=524572,
-    image_x_tga=524558,
+    image_vnd_microsoft_icon=524573,
-    image_x_tiff=524559,
+    image_vnd_rn_realflash=524574,
-    image_x_xcf=524560 | 0x80000000,
+    image_vnd_rn_realpix=524575,
-    image_x_xpixmap=524561 | 0x80000000,
+    image_vnd_wap_wbmp=524576,
-    image_x_xwindowdump=524562,
+    image_vnd_xiff=524577,
-    message_rfc822=196883,
+    image_webp=524578,
-    model_vnd_dwf=65812,
+    image_wmf=524579,
-    model_vnd_gdl=65813,
+    image_x_3ds=524580,
-    model_vnd_gs_gdl=65814,
+    image_x_cmu_raster=524581,
-    model_vrml=65815,
+    image_x_cur=524582,
-    model_x_pov=65816,
+    image_x_dwg=524583,
-    text_asp=590105,
+    image_x_eps=524584,
-    text_css=590106,
+    image_x_exr=524585,
-    text_html=590107,
+    image_x_gem=524586,
-    text_javascript=590108,
+    image_x_icns=524587,
-    text_mcf=590109,
+    image_x_icon=524588 | 0x80000000,
-    text_pascal=590110,
+    image_x_jg=524589,
-    text_plain=590111,
+    image_x_jps=524590,
-    text_richtext=590112,
+    image_x_ms_bmp=524591,
-    text_scriplet=590113,
+    image_x_niff=524592,
-    text_tab_separated_values=590114,
+    image_x_pcx=524593,
-    text_troff=590115,
+    image_x_pict=524594,
-    text_uri_list=590116,
+    image_x_portable_bitmap=524595,
-    text_vnd_abc=590117,
+    image_x_portable_graymap=524596,
-    text_vnd_fmi_flexstor=590118,
+    image_x_portable_pixmap=524597,
-    text_vnd_wap_wml=590119,
+    image_x_quicktime=524598,
-    text_vnd_wap_wmlscript=590120,
+    image_x_rgb=524599,
-    text_webviewhtml=590121,
+    image_x_tga=524600,
-    text_x_Algol68=590122,
+    image_x_tiff=524601,
-    text_x_asm=590123,
+    image_x_win_bitmap=524602,
-    text_x_audiosoft_intra=590124,
+    image_x_xcf=524603 | 0x80000000,
-    text_x_awk=590125,
+    image_x_xpixmap=524604 | 0x80000000,
-    text_x_bcpl=590126,
+    image_x_xwindowdump=524605,
-    text_x_c=590127,
+    message_news=196926,
-    text_x_c__=590128,
+    message_rfc822=196927,
-    text_x_component=590129,
+    model_vnd_dwf=65856,
-    text_x_diff=590130,
+    model_vnd_gdl=65857,
-    text_x_fortran=590131,
+    model_vnd_gs_gdl=65858,
-    text_x_java=590132,
+    model_vrml=65859,
-    text_x_la_asf=590133,
+    model_x_pov=65860,
-    text_x_lisp=590134,
+    text_PGP=590149,
-    text_x_m=590135,
+    text_asp=590150,
-    text_x_m4=590136,
+    text_css=590151,
-    text_x_makefile=590137,
+    text_html=590152,
-    text_x_msdos_batch=590138,
+    text_javascript=590153,
-    text_x_pascal=590139,
+    text_mcf=590154,
-    text_x_perl=590140,
+    text_pascal=590155,
-    text_x_php=590141,
+    text_plain=590156,
-    text_x_po=590142,
+    text_richtext=590157,
-    text_x_python=590143,
+    text_rtf=590158,
-    text_x_ruby=590144,
+    text_scriplet=590159,
-    text_x_sass=590145,
+    text_tab_separated_values=590160,
-    text_x_scss=590146,
+    text_troff=590161,
-    text_x_server_parsed_html=590147,
+    text_uri_list=590162,
-    text_x_setext=590148,
+    text_vnd_abc=590163,
-    text_x_sgml=590149,
+    text_vnd_fmi_flexstor=590164,
-    text_x_shellscript=590150,
+    text_vnd_wap_wml=590165,
-    text_x_speech=590151,
+    text_vnd_wap_wmlscript=590166,
-    text_x_tcl=590152,
+    text_webviewhtml=590167,
-    text_x_tex=590153,
+    text_x_Algol68=590168,
-    text_x_uil=590154,
+    text_x_asm=590169,
-    text_x_uuencode=590155,
+    text_x_audiosoft_intra=590170,
-    text_x_vcalendar=590156,
+    text_x_awk=590171,
-    text_x_vcard=590157,
+    text_x_bcpl=590172,
-    text_xml=590158,
+    text_x_c=590173,
-    video_animaflex=393551,
+    text_x_c__=590174,
-    video_avi=393552,
+    text_x_component=590175,
-    video_avs_video=393553,
+    text_x_diff=590176,
-    video_mp4=393554,
+    text_x_fortran=590177,
-    video_mpeg=393555,
+    text_x_java=590178,
-    video_quicktime=393556,
+    text_x_la_asf=590179,
-    video_vdo=393557,
+    text_x_lisp=590180,
-    video_vivo=393558,
+    text_x_m=590181,
-    video_vnd_rn_realvideo=393559,
+    text_x_m4=590182,
-    video_vosaic=393560,
+    text_x_makefile=590183,
-    video_webm=393561,
+    text_x_ms_regedit=590184,
-    video_x_amt_demorun=393562,
+    text_x_msdos_batch=590185,
-    video_x_amt_showrun=393563,
+    text_x_objective_c=590186,
-    video_x_atomic3d_feature=393564,
+    text_x_pascal=590187,
-    video_x_dl=393565,
+    text_x_perl=590188,
-    video_x_dv=393566,
+    text_x_php=590189,
-    video_x_fli=393567,
+    text_x_po=590190,
-    video_x_flv=393568,
+    text_x_python=590191,
-    video_x_isvideo=393569,
+    text_x_ruby=590192,
-    video_x_jng=393570 | 0x80000000,
+    text_x_sass=590193,
-    video_x_matroska=393571,
+    text_x_scss=590194,
-    video_x_mng=393572,
+    text_x_server_parsed_html=590195,
-    video_x_motion_jpeg=393573,
+    text_x_setext=590196,
-    video_x_ms_asf=393574,
+    text_x_sgml=590197,
-    video_x_msvideo=393575,
+    text_x_shellscript=590198,
-    video_x_qtc=393576,
+    text_x_speech=590199,
-    video_x_sgi_movie=393577,
+    text_x_tcl=590200,
+    text_x_tex=590201,
+    text_x_uil=590202,
+    text_x_uuencode=590203,
+    text_x_vcalendar=590204,
+    text_x_vcard=590205,
+    text_xml=590206,
+    video_MP2T=393599,
+    video_animaflex=393600,
+    video_avi=393601,
+    video_avs_video=393602,
+    video_mp4=393603,
+    video_mpeg=393604,
+    video_quicktime=393605,
+    video_vdo=393606,
+    video_vivo=393607,
+    video_vnd_rn_realvideo=393608,
+    video_vosaic=393609,
+    video_webm=393610,
+    video_x_amt_demorun=393611,
+    video_x_amt_showrun=393612,
+    video_x_atomic3d_feature=393613,
+    video_x_dl=393614,
+    video_x_dv=393615,
+    video_x_fli=393616,
+    video_x_flv=393617,
+    video_x_isvideo=393618,
+    video_x_jng=393619 | 0x80000000,
+    video_x_m4v=393620,
+    video_x_matroska=393621,
+    video_x_mng=393622,
+    video_x_motion_jpeg=393623,
+    video_x_ms_asf=393624,
+    video_x_msvideo=393625,
+    video_x_qtc=393626,
+    video_x_sgi_movie=393627,
+    x_epoc_x_sisx_app=721308,
 };
 char *mime_get_mime_text(unsigned int mime_id) {switch (mime_id) {
 case application_arj: return "application/arj";
@@ -625,6 +676,7 @@ case text_mcf: return "text/mcf";
 case text_pascal: return "text/pascal";
 case text_plain: return "text/plain";
 case text_richtext: return "text/richtext";
+case text_rtf: return "text/rtf";
 case text_scriplet: return "text/scriplet";
 case text_x_awk: return "text/x-awk";
 case video_x_jng: return "video/x-jng";
@@ -730,6 +782,56 @@ case application_x_wine_extension_ini: return "application/x-wine-extension-ini"
 case application_x_cbz: return "application/x-cbz";
 case application_x_cbr: return "application/x-cbr";
 case application_x_ms_compress_szdd: return "application/x-ms-compress-szdd";
+case application_x_atari_7800_rom: return "application/x-atari-7800-rom";
+case application_x_nes_rom: return "application/x-nes-rom";
+case application_x_font_pfm: return "application/x-font-pfm";
+case application_x_gettext_translation: return "application/x-gettext-translation";
+case image_wmf: return "image/wmf";
+case application_pgp_keys: return "application/pgp-keys";
+case image_x_3ds: return "image/x-3ds";
+case application_x_lz4: return "application/x-lz4";
+case application_vnd_openxmlformats_officedocument_presentationml_presentation: return "application/vnd.openxmlformats-officedocument.presentationml.presentation";
+case application_vnd_oasis_opendocument_presentation: return "application/vnd.oasis.opendocument.presentation";
+case application_x_msaccess: return "application/x-msaccess";
+case application_vnd_oasis_opendocument_spreadsheet: return "application/vnd.oasis.opendocument.spreadsheet";
+case audio_x_aiff: return "audio/x-aiff";
+case text_x_ms_regedit: return "text/x-ms-regedit";
+case application_x_gamecube_rom: return "application/x-gamecube-rom";
+case application_x_nintendo_ds_rom: return "application/x-nintendo-ds-rom";
+case text_x_objective_c: return "text/x-objective-c";
+case application_x_font_gdos: return "application/x-font-gdos";
+case application_x_apple_diskimage: return "application/x-apple-diskimage";
+case application_x_zstd: return "application/x-zstd";
+case video_x_m4v: return "video/x-m4v";
+case message_news: return "message/news";
+case application_vnd_symbian_install: return "application/vnd.symbian.install";
+case application_x_lzh_compressed: return "application/x-lzh-compressed";
+case application_x_dosdriver: return "application/x-dosdriver";
+case application_vnd_tcpdump_pcap: return "application/vnd.tcpdump.pcap";
+case x_epoc_x_sisx_app: return "x-epoc/x-sisx-app";
+case application_x_avira_qua: return "application/x-avira-qua";
+case video_MP2T: return "video/MP2T";
+case application_x_snappy_framed: return "application/x-snappy-framed";
+case application_x_lz4_json: return "application/x-lz4+json";
+case application_x_dmp: return "application/x-dmp";
+case application_zlib: return "application/zlib";
+case application_x_pgp_keyring: return "application/x-pgp-keyring";
+case application_x_gdbm: return "application/x-gdbm";
+case application_x_font_pf2: return "application/x-font-pf2";
+case application_x_zip: return "application/x-zip";
+case application_x_coredump: return "application/x-coredump";
+case application_x_java_jmod: return "application/x-java-jmod";
+case application_x_terminfo: return "application/x-terminfo";
+case application_x_terminfo2: return "application/x-terminfo2";
+case application_x_arc: return "application/x-arc";
+case application_vnd_lotus_1_2_3: return "application/vnd.lotus-1-2-3";
+case image_x_win_bitmap: return "image/x-win-bitmap";
+case application_x_maxis_dbpf: return "application/x-maxis-dbpf";
+case text_PGP: return "text/PGP";
+case audio_x_hx_aac_adts: return "audio/x-hx-aac-adts";
+case application_x_chrome_extension: return "application/x-chrome-extension";
+case image_heic: return "image/heic";
+case image_x_gem: return "image/x-gem";
 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);
@@ -1097,7 +1199,7 @@ g_hash_table_insert(ext_table, "ms", (gpointer)text_troff);
 g_hash_table_insert(ext_table, "roff", (gpointer)text_troff);
 g_hash_table_insert(ext_table, "t", (gpointer)text_troff);
 g_hash_table_insert(ext_table, "tr", (gpointer)text_troff);
-g_hash_table_insert(ext_table, "uni", (gpointer)text_uri_list);
+g_hash_table_insert(ext_table, "uji", (gpointer)text_uri_list);
 g_hash_table_insert(ext_table, "unis", (gpointer)text_uri_list);
 g_hash_table_insert(ext_table, "uri", (gpointer)text_uri_list);
 g_hash_table_insert(ext_table, "uris", (gpointer)text_uri_list);
@@ -1211,6 +1313,27 @@ g_hash_table_insert(ext_table, "hlp", (gpointer)application_winhelp);
 g_hash_table_insert(ext_table, "cbz", (gpointer)application_x_cbz);
 g_hash_table_insert(ext_table, "cbr", (gpointer)application_x_cbr);
 g_hash_table_insert(ext_table, "fon", (gpointer)application_x_ms_compress_szdd);
+g_hash_table_insert(ext_table, "a78", (gpointer)application_x_atari_7800_rom);
+g_hash_table_insert(ext_table, "nes", (gpointer)application_x_nes_rom);
+g_hash_table_insert(ext_table, "pfm", (gpointer)application_x_font_pfm);
+g_hash_table_insert(ext_table, "3ds", (gpointer)image_x_3ds);
+g_hash_table_insert(ext_table, "lz4", (gpointer)application_x_lz4);
+g_hash_table_insert(ext_table, "pptx", (gpointer)application_vnd_openxmlformats_officedocument_presentationml_presentation);
+g_hash_table_insert(ext_table, "odp", (gpointer)application_vnd_oasis_opendocument_presentation);
+g_hash_table_insert(ext_table, "accdb", (gpointer)application_x_msaccess);
+g_hash_table_insert(ext_table, "ods", (gpointer)application_vnd_oasis_opendocument_spreadsheet);
+g_hash_table_insert(ext_table, "aiff", (gpointer)audio_x_aiff);
+g_hash_table_insert(ext_table, "aif", (gpointer)audio_x_aiff);
+g_hash_table_insert(ext_table, "reg", (gpointer)text_x_ms_regedit);
+g_hash_table_insert(ext_table, "zst", (gpointer)application_x_zstd);
+g_hash_table_insert(ext_table, "m4v", (gpointer)video_x_m4v);
+g_hash_table_insert(ext_table, "pcap", (gpointer)application_vnd_tcpdump_pcap);
+g_hash_table_insert(ext_table, "jsonlz4", (gpointer)application_x_lz4_json);
+g_hash_table_insert(ext_table, "dmp", (gpointer)application_x_dmp);
+g_hash_table_insert(ext_table, "z", (gpointer)application_zlib);
+g_hash_table_insert(ext_table, "pf2", (gpointer)application_x_font_pf2);
+g_hash_table_insert(ext_table, "jmod", (gpointer)application_x_java_jmod);
+g_hash_table_insert(ext_table, "heic", (gpointer)image_heic);
 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);
@@ -1469,6 +1592,7 @@ g_hash_table_insert(mime_table, "text/mcf", (gpointer)text_mcf);
 g_hash_table_insert(mime_table, "text/pascal", (gpointer)text_pascal);
 g_hash_table_insert(mime_table, "text/plain", (gpointer)text_plain);
 g_hash_table_insert(mime_table, "text/richtext", (gpointer)text_richtext);
+g_hash_table_insert(mime_table, "text/rtf", (gpointer)text_rtf);
 g_hash_table_insert(mime_table, "text/scriplet", (gpointer)text_scriplet);
 g_hash_table_insert(mime_table, "text/x-awk", (gpointer)text_x_awk);
 g_hash_table_insert(mime_table, "video/x-jng", (gpointer)video_x_jng);
@@ -1574,5 +1698,55 @@ g_hash_table_insert(mime_table, "application/x-wine-extension-ini", (gpointer)ap
 g_hash_table_insert(mime_table, "application/x-cbz", (gpointer)application_x_cbz);
 g_hash_table_insert(mime_table, "application/x-cbr", (gpointer)application_x_cbr);
 g_hash_table_insert(mime_table, "application/x-ms-compress-szdd", (gpointer)application_x_ms_compress_szdd);
+g_hash_table_insert(mime_table, "application/x-atari-7800-rom", (gpointer)application_x_atari_7800_rom);
+g_hash_table_insert(mime_table, "application/x-nes-rom", (gpointer)application_x_nes_rom);
+g_hash_table_insert(mime_table, "application/x-font-pfm", (gpointer)application_x_font_pfm);
+g_hash_table_insert(mime_table, "application/x-gettext-translation", (gpointer)application_x_gettext_translation);
+g_hash_table_insert(mime_table, "image/wmf", (gpointer)image_wmf);
+g_hash_table_insert(mime_table, "application/pgp-keys", (gpointer)application_pgp_keys);
+g_hash_table_insert(mime_table, "image/x-3ds", (gpointer)image_x_3ds);
+g_hash_table_insert(mime_table, "application/x-lz4", (gpointer)application_x_lz4);
+g_hash_table_insert(mime_table, "application/vnd.openxmlformats-officedocument.presentationml.presentation", (gpointer)application_vnd_openxmlformats_officedocument_presentationml_presentation);
+g_hash_table_insert(mime_table, "application/vnd.oasis.opendocument.presentation", (gpointer)application_vnd_oasis_opendocument_presentation);
+g_hash_table_insert(mime_table, "application/x-msaccess", (gpointer)application_x_msaccess);
+g_hash_table_insert(mime_table, "application/vnd.oasis.opendocument.spreadsheet", (gpointer)application_vnd_oasis_opendocument_spreadsheet);
+g_hash_table_insert(mime_table, "audio/x-aiff", (gpointer)audio_x_aiff);
+g_hash_table_insert(mime_table, "text/x-ms-regedit", (gpointer)text_x_ms_regedit);
+g_hash_table_insert(mime_table, "application/x-gamecube-rom", (gpointer)application_x_gamecube_rom);
+g_hash_table_insert(mime_table, "application/x-nintendo-ds-rom", (gpointer)application_x_nintendo_ds_rom);
+g_hash_table_insert(mime_table, "text/x-objective-c", (gpointer)text_x_objective_c);
+g_hash_table_insert(mime_table, "application/x-font-gdos", (gpointer)application_x_font_gdos);
+g_hash_table_insert(mime_table, "application/x-apple-diskimage", (gpointer)application_x_apple_diskimage);
+g_hash_table_insert(mime_table, "application/x-zstd", (gpointer)application_x_zstd);
+g_hash_table_insert(mime_table, "video/x-m4v", (gpointer)video_x_m4v);
+g_hash_table_insert(mime_table, "message/news", (gpointer)message_news);
+g_hash_table_insert(mime_table, "application/vnd.symbian.install", (gpointer)application_vnd_symbian_install);
+g_hash_table_insert(mime_table, "application/x-lzh-compressed", (gpointer)application_x_lzh_compressed);
+g_hash_table_insert(mime_table, "application/x-dosdriver", (gpointer)application_x_dosdriver);
+g_hash_table_insert(mime_table, "application/vnd.tcpdump.pcap", (gpointer)application_vnd_tcpdump_pcap);
+g_hash_table_insert(mime_table, "x-epoc/x-sisx-app", (gpointer)x_epoc_x_sisx_app);
+g_hash_table_insert(mime_table, "application/x-avira-qua", (gpointer)application_x_avira_qua);
+g_hash_table_insert(mime_table, "video/MP2T", (gpointer)video_MP2T);
+g_hash_table_insert(mime_table, "application/x-snappy-framed", (gpointer)application_x_snappy_framed);
+g_hash_table_insert(mime_table, "application/x-lz4+json", (gpointer)application_x_lz4_json);
+g_hash_table_insert(mime_table, "application/x-dmp", (gpointer)application_x_dmp);
+g_hash_table_insert(mime_table, "application/zlib", (gpointer)application_zlib);
+g_hash_table_insert(mime_table, "application/x-pgp-keyring", (gpointer)application_x_pgp_keyring);
+g_hash_table_insert(mime_table, "application/x-gdbm", (gpointer)application_x_gdbm);
+g_hash_table_insert(mime_table, "application/x-font-pf2", (gpointer)application_x_font_pf2);
+g_hash_table_insert(mime_table, "application/x-zip", (gpointer)application_x_zip);
+g_hash_table_insert(mime_table, "application/x-coredump", (gpointer)application_x_coredump);
+g_hash_table_insert(mime_table, "application/x-java-jmod", (gpointer)application_x_java_jmod);
+g_hash_table_insert(mime_table, "application/x-terminfo", (gpointer)application_x_terminfo);
+g_hash_table_insert(mime_table, "application/x-terminfo2", (gpointer)application_x_terminfo2);
+g_hash_table_insert(mime_table, "application/x-arc", (gpointer)application_x_arc);
+g_hash_table_insert(mime_table, "application/vnd.lotus-1-2-3", (gpointer)application_vnd_lotus_1_2_3);
+g_hash_table_insert(mime_table, "image/x-win-bitmap", (gpointer)image_x_win_bitmap);
+g_hash_table_insert(mime_table, "application/x-maxis-dbpf", (gpointer)application_x_maxis_dbpf);
+g_hash_table_insert(mime_table, "text/PGP", (gpointer)text_PGP);
+g_hash_table_insert(mime_table, "audio/x-hx-aac-adts", (gpointer)audio_x_hx_aac_adts);
+g_hash_table_insert(mime_table, "application/x-chrome-extension", (gpointer)application_x_chrome_extension);
+g_hash_table_insert(mime_table, "image/heic", (gpointer)image_heic);
+g_hash_table_insert(mime_table, "image/x-gem", (gpointer)image_x_gem);
 return mime_table;}
 #endif

src/parsing/parse.c

@@ -1,7 +1,7 @@
 #include "src/sist.h"
 #include "src/ctx.h"
 
-__thread magic_t Magic;
+__thread magic_t Magic = NULL;
 
 void *read_all(parse_job_t *job, const char *buf, int bytes_read, int *fd) {
 
@@ -62,7 +62,7 @@ void parse(void *arg) {
 
     if (job->info.st_size == 0) {
         doc.mime = MIME_EMPTY;
-    } else if (*(job->filepath + job->ext) != '\0') {
+    } else if (*(job->filepath + job->ext) != '\0' && (job->ext - job->base != 1)) {
         doc.mime = mime_get_mime_by_ext(ScanCtx.ext_table, job->filepath + job->ext);
     }
 
@@ -80,11 +80,18 @@ void parse(void *arg) {
 
         bytes_read = read(fd, buf, PARSE_BUF_SIZE);
 
+        if (bytes_read == -1) {
+            perror("read");
+            close(fd);
+            free(job);
+            return;
+        }
+
         const char *magic_mime_str = magic_buffer(Magic, buf, bytes_read);
         if (magic_mime_str != NULL) {
             doc.mime = mime_get_mime_by_string(ScanCtx.mime_table, magic_mime_str);
             if (doc.mime == 0) {
-                fprintf(stderr, "Couldn't find mime %s, %s!\n", magic_mime_str, job->filepath + job->base);
+                fprintf(stderr, "Couldn't find mime %s, %s\n", magic_mime_str, job->filepath + job->base);
             }
         }
     }
@@ -93,7 +100,8 @@ void parse(void *arg) {
 
     if (!(SHOULD_PARSE(doc.mime))) {
 
-    } else if ((mmime == MimeVideo && doc.size >= MIN_VIDEO_SIZE) || mmime == MimeAudio || mmime == MimeImage) {
+    } else if ((mmime == MimeVideo && doc.size >= MIN_VIDEO_SIZE) ||
+    (mmime == MimeImage && doc.size >= MIN_IMAGE_SIZE) || mmime == MimeAudio) {
         parse_media(job->filepath, &doc);
 
     } else if (IS_PDF(doc.mime)) {

src/parsing/pdf.c

@@ -1,10 +1,22 @@
-#include <src/ctx.h>
 #include "pdf.h"
 #include "src/ctx.h"
 
 fz_page *render_cover(fz_context *ctx, document_t *doc, fz_document *fzdoc) {
 
-    fz_page *cover = fz_load_page(ctx, fzdoc, 0);
+    int err = 0;
+    fz_page *cover = NULL;
+
+    fz_var(cover);
+    fz_try(ctx)
+        cover = fz_load_page(ctx, fzdoc, 0);
+    fz_catch(ctx)
+        err = 1;
+
+    if (err != 0) {
+        fz_drop_page(ctx, cover);
+        return NULL;
+    }
+
     fz_rect bounds = fz_bound_page(ctx, cover);
 
     float scale;
@@ -24,24 +36,49 @@ fz_page *render_cover(fz_context *ctx, document_t *doc, fz_document *fzdoc) {
     fz_clear_pixmap_with_value(ctx, pixmap, 0xFF);
     fz_device *dev = fz_new_draw_device(ctx, m, pixmap);
 
-    pthread_mutex_lock(&ScanCtx.mupdf_mu);
+    fz_var(err);
     fz_try(ctx)
+    {
+        pthread_mutex_lock(&ScanCtx.mupdf_mu);
         fz_run_page(ctx, cover, dev, fz_identity, NULL);
+    }
     fz_always(ctx)
-        pthread_mutex_unlock(&ScanCtx.mupdf_mu);
+    {
-    fz_catch(ctx)
+        fz_close_device(ctx, dev);
-        fz_rethrow(ctx);
-
         fz_drop_device(ctx, dev);
+        pthread_mutex_unlock(&ScanCtx.mupdf_mu);
+    }
+    fz_catch(ctx)
+        err = ctx->error.errcode;
 
-    fz_buffer *fzbuf = fz_new_buffer_from_pixmap_as_png(ctx, pixmap, fz_default_color_params);
+    if (err != 0) {
+        fz_drop_page(ctx, cover);
+        fz_drop_pixmap(ctx, pixmap);
+        return NULL;
+    }
+
+    fz_buffer *fzbuf = NULL;
+    fz_var(fzbuf);
+    fz_var(err);
+
+    fz_try(ctx)
+        fzbuf = fz_new_buffer_from_pixmap_as_png(ctx, pixmap, fz_default_color_params);
+    fz_catch(ctx)
+        err = ctx->error.errcode;
+
+    if (err == 0) {
         unsigned char *tn_buf;
         size_t tn_len = fz_buffer_storage(ctx, fzbuf, &tn_buf);
-
         store_write(ScanCtx.index.store, (char *) doc->uuid, sizeof(doc->uuid), (char *) tn_buf, tn_len);
+    }
 
-    fz_drop_pixmap(ctx, pixmap);
     fz_drop_buffer(ctx, fzbuf);
+    fz_drop_pixmap(ctx, pixmap);
+
+    if (err != 0) {
+        fz_drop_page(ctx, cover);
+        return NULL;
+    }
 
     return cover;
 }
@@ -49,6 +86,32 @@ fz_page *render_cover(fz_context *ctx, document_t *doc, fz_document *fzdoc) {
 void fz_noop_callback(__attribute__((unused)) void *user, __attribute__((unused)) const char *message) {}
 
 
+void init_ctx(fz_context *ctx) {
+    fz_disable_icc(ctx);
+    fz_register_document_handlers(ctx);
+    ctx->warn.print = fz_noop_callback;
+    ctx->error.print = fz_noop_callback;
+}
+
+int read_stext_block(fz_stext_block *block, text_buffer_t *tex) {
+    if (block->type != FZ_STEXT_BLOCK_TEXT) {
+        return 0;
+    }
+
+    fz_stext_line *line = block->u.t.first_line;
+    while (line != NULL) {
+        fz_stext_char *c = line->first_char;
+        while (c != NULL) {
+            if (text_buffer_append_char(tex, c->c) == TEXT_BUF_FULL) {
+                return TEXT_BUF_FULL;
+            }
+            c = c->next;
+        }
+        line = line->next;
+    }
+    return 0;
+}
+
 void parse_pdf(void *buf, size_t buf_len, document_t *doc) {
 
     static int mu_is_initialized = 0;
@@ -57,105 +120,140 @@ void parse_pdf(void *buf, size_t buf_len, document_t *doc) {
         mu_is_initialized = 1;
     }
     fz_context *ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
-    fz_stream *stream = NULL;
-    fz_document *fzdoc = NULL;
 
-    fz_var(stream);
+    init_ctx(ctx);
+
+    int err = 0;
+
+    fz_document *fzdoc = NULL;
+    fz_stream *stream = NULL;
     fz_var(fzdoc);
+    fz_var(stream);
+    fz_var(err);
 
     fz_try(ctx)
     {
-        fz_disable_icc(ctx);
-        fz_register_document_handlers(ctx);
-
-        //disable warnings
-        ctx->warn.print = fz_noop_callback;
-        ctx->error.print = fz_noop_callback;
-
         stream = fz_open_memory(ctx, buf, buf_len);
         fzdoc = fz_open_document_with_stream(ctx, mime_get_mime_text(doc->mime), stream);
+    }
+    fz_catch(ctx)
+        err = ctx->error.errcode;
+
+    if (err) {
+        fz_drop_stream(ctx, stream);
+        fz_drop_document(ctx, fzdoc);
+        fz_drop_context(ctx);
+        return;
+    }
 
     char title[4096] = {'\0',};
+    fz_try(ctx)
         fz_lookup_metadata(ctx, fzdoc, FZ_META_INFO_TITLE, title, sizeof(title));
-        printf("Title: %s\n", title); //todo rmv
+    fz_catch(ctx)
+        ;
 
     if (strlen(title) > 0) {
-            meta_line_t *meta_content = malloc(sizeof(meta_line_t) + strlen(title) + 1);
+        meta_line_t *meta_content = malloc(sizeof(meta_line_t) + strlen(title));
         meta_content->key = MetaTitle;
         strcpy(meta_content->strval, title);
         APPEND_META(doc, meta_content)
     }
 
-        int page_count = fz_count_pages(ctx, fzdoc);
+    int page_count = -1;
+    fz_var(err);
+    fz_try(ctx)
+        page_count = fz_count_pages(ctx, fzdoc);
+    fz_catch(ctx)
+        err = ctx->error.errcode;
+
+    if (err) {
+        fz_drop_stream(ctx, stream);
+        fz_drop_document(ctx, fzdoc);
+        fz_drop_context(ctx);
+        return;
+    }
 
     fz_page *cover = render_cover(ctx, doc, fzdoc);
+    if (cover == NULL) {
+        fz_drop_stream(ctx, stream);
+        fz_drop_document(ctx, fzdoc);
+        fz_drop_context(ctx);
+        return;
+    }
 
-        fz_stext_options opts;
+    fz_stext_options opts = {0};
-
     text_buffer_t text_buf = text_buffer_create(ScanCtx.content_size);
 
     for (int current_page = 0; current_page < page_count; current_page++) {
-            fz_page *page; if (current_page == 0) {
+        fz_page *page = NULL;
+        if (current_page == 0) {
             page = cover;
         } else {
+            fz_var(err);
+            fz_try(ctx)
                 page = fz_load_page(ctx, fzdoc, current_page);
+            fz_catch(ctx)
+                err = ctx->error.errcode;
+            if (err != 0) {
+                text_buffer_destroy(&text_buf);
+                fz_drop_page(ctx, page);
+                fz_drop_stream(ctx, stream);
+                fz_drop_document(ctx, fzdoc);
+                fz_drop_context(ctx);
+                return;
+            }
         }
 
         fz_stext_page *stext = fz_new_stext_page(ctx, fz_bound_page(ctx, page));
         fz_device *dev = fz_new_stext_device(ctx, stext, &opts);
 
-            pthread_mutex_lock(&ScanCtx.mupdf_mu);
+        fz_var(err);
         fz_try(ctx)
-                fz_run_page_contents(ctx, page, dev, fz_identity, NULL);
+            fz_run_page(ctx, page, dev, fz_identity, NULL);
         fz_always(ctx)
-                pthread_mutex_unlock(&ScanCtx.mupdf_mu);
+        {
-            fz_catch(ctx)
+            fz_close_device(ctx, dev);
-                fz_rethrow(ctx);
-
             fz_drop_device(ctx, dev);
+        }
+        fz_catch(ctx)
+            err = ctx->error.errcode;
+
+        if (err != 0) {
+            text_buffer_destroy(&text_buf);
+            fz_drop_page(ctx, page);
+            fz_drop_stext_page(ctx, stext);
+            fz_drop_stream(ctx, stream);
+            fz_drop_document(ctx, fzdoc);
+            fz_drop_context(ctx);
+            return;
+        }
 
         fz_stext_block *block = stext->first_block;
         while (block != NULL) {
-
+            int ret = read_stext_block(block, &text_buf);
-                if (block->type != FZ_STEXT_BLOCK_TEXT) {
+            if (ret == TEXT_BUF_FULL) {
-                    block = block->next;
+                break;
-                    continue;
-                }
-
-                fz_stext_line *line = block->u.t.first_line;
-                while (line != NULL) {
-                    fz_stext_char *c = line->first_char;
-                    while (c != NULL) {
-                        if (text_buffer_append_char(&text_buf, c->c) == TEXT_BUF_FULL) {
-                            fz_drop_page(ctx, page);
-                            fz_drop_stext_page(ctx, stext);
-                            goto write_loop_end;
-                        }
-                        c = c->next;
-                    }
-                    line = line->next;
             }
             block = block->next;
         }
-            fz_drop_page(ctx, page);
         fz_drop_stext_page(ctx, stext);
+        fz_drop_page(ctx, page);
+
+        if (text_buf.dyn_buffer.cur >= text_buf.dyn_buffer.size) {
+            break;
+        }
     }
-        write_loop_end:;
     text_buffer_terminate_string(&text_buf);
 
     meta_line_t *meta_content = malloc(sizeof(meta_line_t) + text_buf.dyn_buffer.cur);
     meta_content->key = MetaContent;
     memcpy(meta_content->strval, text_buf.dyn_buffer.buf, text_buf.dyn_buffer.cur);
-        text_buffer_destroy(&text_buf);
     APPEND_META(doc, meta_content)
-    }
+
-    fz_always(ctx)
-    {
     fz_drop_stream(ctx, stream);
     fz_drop_document(ctx, fzdoc);
     fz_drop_context(ctx);
-    } fz_catch(ctx) {
+
-        fprintf(stderr, "Error %s %s\n", doc->filepath, ctx->error.message);
+    text_buffer_destroy(&text_buf);
-    }
 }
 

src/parsing/text.c

@@ -27,17 +27,14 @@ void parse_text(int bytes_read, int *fd, char *buf, document_t *doc) {
 
         read(*fd, intermediate_buf + bytes_read, to_read);
     }
+    text_buffer_t tex = text_buffer_create(ScanCtx.content_size);
+    text_buffer_append_string(&tex, intermediate_buf, intermediate_buf_len);
 
-    text_buffer_t text_buf = text_buffer_create(ScanCtx.content_size);
+    meta_line_t *meta = malloc(sizeof(meta_line_t) + tex.dyn_buffer.cur);
-    for (int i = 0; i < intermediate_buf_len; i++) {
-        text_buffer_append_char(&text_buf, *(intermediate_buf + i));
-    }
-    text_buffer_terminate_string(&text_buf);
-
-    meta_line_t *meta = malloc(sizeof(meta_line_t) + text_buf.dyn_buffer.cur);
     meta->key = MetaContent;
-    strcpy(meta->strval, text_buf.dyn_buffer.buf);
+    strcpy(meta->strval, tex.dyn_buffer.buf);
-    text_buffer_destroy(&text_buf);
-    free(intermediate_buf);
     APPEND_META(doc, meta)
+
+    free(intermediate_buf);
+    text_buffer_destroy(&tex);
 }

src/sist.h

@@ -16,6 +16,7 @@
 #include <libswscale/swscale.h>
 #include <libswresample/swresample.h>
 #include <libavcodec/avcodec.h>
+#include <libavutil/imgutils.h>
 #include <ctype.h>
 #include <mupdf/fitz.h>
 #include <mupdf/pdf.h>
@@ -49,6 +50,7 @@
 #include "parsing/media.h"
 #include "parsing/font.h"
 #include "cli.h"
+#include "utf8.h/utf8.h"
 
 #ifndef SIST_SCAN_ONLY
 #include "src/index/elastic.h"

src/tpool.c

@@ -25,6 +25,7 @@ typedef struct tpool {
     int done_cnt;
 
     int stop;
+
     void (*cleanup_func)();
 } tpool_t;
 
@@ -100,7 +101,7 @@ static void *tpool_worker(void *arg) {
     tpool_t *pool = arg;
 
     while (1) {
-        pthread_mutex_lock(&(pool->work_mutex));
+        pthread_mutex_lock(&pool->work_mutex);
         if (pool->stop) {
             break;
         }
@@ -192,7 +193,7 @@ tpool_t *tpool_create(size_t thread_cnt, void cleanup_func()) {
     pool->done_cnt = 0;
     pool->stop = 0;
     pool->cleanup_func = cleanup_func;
-    pool->threads = malloc(sizeof(pthread_t) * thread_cnt);
+    pool->threads = calloc(sizeof(pthread_t), thread_cnt);
 
     pthread_mutex_init(&(pool->work_mutex), NULL);
 
@@ -202,11 +203,14 @@ tpool_t *tpool_create(size_t thread_cnt, void cleanup_func()) {
     pool->work_head = NULL;
     pool->work_tail = NULL;
 
-    for (size_t i = 0; i < thread_cnt; i++) {
+    return pool;
+}
+
+void tpool_start(tpool_t *pool) {
+
+    for (size_t i = 0; i < pool->thread_cnt; i++) {
         pthread_t thread = pool->threads[i];
         pthread_create(&thread, NULL, tpool_worker, pool);
         pthread_detach(thread);
     }
-
-    return pool;
 }

src/tpool.h

@@ -9,6 +9,7 @@ typedef struct tpool tpool_t;
 typedef void (*thread_func_t)(void *arg);
 
 tpool_t *tpool_create(size_t num, void (*cleanup_func)());
+void tpool_start(tpool_t *pool);
 void tpool_destroy(tpool_t *tm);
 
 int tpool_add_work(tpool_t *pool, thread_func_t func, void *arg);

src/util.c

@@ -89,10 +89,71 @@ void text_buffer_terminate_string(text_buffer_t *buf) {
     dyn_buffer_write_char(&buf->dyn_buffer, '\0');
 }
 
-int text_buffer_append_string(text_buffer_t *buf, char * str) {
+__always_inline
-    char * ptr = str;
+int utf8_validchr(const char* s) {
-    while (*ptr) {
+    if (0x00 == (0x80 & *s)) {
-        text_buffer_append_char(buf, *ptr++);
+        return TRUE;
+    } else if (0xf0 == (0xf8 & *s)) {
+        if ((0x80 != (0xc0 & s[1])) || (0x80 != (0xc0 & s[2])) ||
+            (0x80 != (0xc0 & s[3]))) {
+            return FALSE;
+        }
+
+        if (0x80 == (0xc0 & s[4])) {
+            return FALSE;
+        }
+
+        if ((0 == (0x07 & s[0])) && (0 == (0x30 & s[1]))) {
+            return FALSE;
+        }
+    } else if (0xe0 == (0xf0 & *s)) {
+        if ((0x80 != (0xc0 & s[1])) || (0x80 != (0xc0 & s[2]))) {
+            return FALSE;
+        }
+
+        if (0x80 == (0xc0 & s[3])) {
+            return FALSE;
+        }
+
+        if ((0 == (0x0f & s[0])) && (0 == (0x20 & s[1]))) {
+            return FALSE;
+        }
+    } else if (0xc0 == (0xe0 & *s)) {
+        if (0x80 != (0xc0 & s[1])) {
+            return FALSE;
+        }
+
+        if (0x80 == (0xc0 & s[2])) {
+            return FALSE;
+        }
+
+        if (0 == (0x1e & s[0])) {
+            return FALSE;
+        }
+    } else  {
+        return FALSE;
+    }
+
+    return TRUE;
+}
+
+int text_buffer_append_string(text_buffer_t *buf, char *str, size_t len) {
+
+    utf8_int32_t c;
+    for (void *v = utf8codepoint(str, &c); c != '\0' && ((char*)v - str + 4) < len; v = utf8codepoint(v, &c)) {
+        if (utf8_validchr(v)) {
+            text_buffer_append_char(buf, c);
+        }
+    }
+    text_buffer_terminate_string(buf);
+}
+
+int text_buffer_append_string0(text_buffer_t *buf, char *str) {
+    utf8_int32_t c;
+    for (void *v = utf8codepoint(str, &c); c != '\0'; v = utf8codepoint(v, &c)) {
+        if (utf8_validchr(v)) {
+            text_buffer_append_char(buf, c);
+        }
     }
     text_buffer_terminate_string(buf);
 }
@@ -104,15 +165,31 @@ int text_buffer_append_char(text_buffer_t *buf, int c) {
             dyn_buffer_write_char(&buf->dyn_buffer, ' ');
             buf->last_char_was_whitespace = TRUE;
 
-            if (buf->dyn_buffer.cur >= buf->max_size) {
+            if (buf->max_size > 0 && buf->dyn_buffer.cur >= buf->max_size) {
                 return TEXT_BUF_FULL;
             }
         }
     } else {
         buf->last_char_was_whitespace = FALSE;
-        dyn_buffer_write_char(&buf->dyn_buffer, (char) c);
+        grow_buffer_small(&buf->dyn_buffer);
 
-        if (buf->dyn_buffer.cur >= buf->max_size) {
+        if (0 == ((utf8_int32_t) 0xffffff80 & c)) {
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = (char) c;
+        } else if (0 == ((utf8_int32_t) 0xfffff800 & c)) {
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0xc0 | (char) (c >> 6);
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0x80 | (char) (c & 0x3f);
+        } else if (0 == ((utf8_int32_t) 0xffff0000 & c)) {
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0xe0 | (char) (c >> 12);
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0x80 | (char) ((c >> 6) & 0x3f);
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0x80 | (char) (c & 0x3f);
+        } else {
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0xf0 | (char) (c >> 18);
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0x80 | (char) ((c >> 12) & 0x3f);
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0x80 | (char) ((c >> 6) & 0x3f);
+            *(buf->dyn_buffer.buf + buf->dyn_buffer.cur++) = 0x80 | (char) (c & 0x3f);
+        }
+
+        if (buf->max_size > 0 && buf->dyn_buffer.cur >= buf->max_size) {
             return TEXT_BUF_FULL;
         }
     }

src/util.h

@@ -5,7 +5,10 @@
 
 #define TEXT_BUF_FULL -1
 #define INITIAL_BUF_SIZE 1024 * 16
-#define SHOULD_IGNORE_CHAR(c) c < '0' || c > 'z'
+
+#define SHOULD_IGNORE_CHAR(c) !(SHOULD_KEEP_CHAR(c))
+#define SHOULD_KEEP_CHAR(c) (c >= (int)'!')
+
 
 typedef struct dyn_buffer {
     char *buf;
@@ -22,7 +25,9 @@ typedef struct text_buffer {
 } text_buffer_t;
 
 char *abspath(const char *path);
+
 char *expandpath(const char *path);
+
 dyn_buffer_t url_escape(char *str);
 
 void progress_bar_print(double percentage, size_t tn_size, size_t index_size);
@@ -56,14 +61,16 @@ text_buffer_t text_buffer_create(int max_size);
 
 void text_buffer_terminate_string(text_buffer_t *buf);
 
-int text_buffer_append_string(text_buffer_t *buf, char * str);
+int text_buffer_append_string(text_buffer_t *buf, char *str, size_t len);
+int text_buffer_append_string0(text_buffer_t *buf, char *str);
+
 int text_buffer_append_char(text_buffer_t *buf, int c);
 
 void incremental_put(GHashTable *table, unsigned long inode_no, int mtime);
 
 int incremental_get(GHashTable *table, unsigned long inode_no);
+
 int incremental_mark_file_for_copy(GHashTable *table, unsigned long inode_no);
 
 
-
 #endif

utf8.h

@@ -0,0 +1 @@
+Subproject commit 2a7c5bfa952816cd1c674e604d31c6e0268ba770