Use 16-bit ints for meta keys (wip)

CMakeLists.txt

@@ -4,6 +4,7 @@ project(scan)
 set(CMAKE_C_STANDARD 11)
 
 option(BUILD_TESTS "Build tests" off)
+option(SIST_FAKE_STORE "Disable IO operations of LMDB stores for debugging purposes" 0)
 
 add_subdirectory(third-party/antiword)
 add_compile_definitions(
@@ -87,25 +88,31 @@ ExternalProject_Add(
 SET(MOBI_LIB_DIR ${CMAKE_CURRENT_BINARY_DIR}/third-party/ext_libmobi/src/libmobi/src/.libs/)
 SET(MOBI_INCLUDE_DIR ${CMAKE_CURRENT_BINARY_DIR}/third-party/ext_libmobi/src/libmobi/src/)
 
+if (SIST_DEBUG)
+    SET(FFMPEG_DEBUG "--enable-debug=3" "--disable-optimizations")
+else()
+    SET(FFMPEG_DEBUG "")
+endif()
+
 ExternalProject_Add(
         ffmpeg
         GIT_REPOSITORY https://git.ffmpeg.org/ffmpeg.git
-        GIT_TAG "master"
+        GIT_TAG "n4.3.2"
 
         UPDATE_COMMAND ""
         PATCH_COMMAND ""
         TEST_COMMAND ""
         CONFIGURE_COMMAND ./configure --disable-shared --enable-static --disable-ffmpeg --disable-ffplay
         --disable-ffprobe --disable-doc --disable-manpages --disable-postproc --disable-avfilter --disable-alsa
-        --disable-lzma --disable-xlib --disable-debug --disable-vdpau --disable-vaapi --disable-sdl2
+        --disable-lzma --disable-xlib --disable-vdpau --disable-vaapi --disable-sdl2
-        --disable-network --extra-cflags=-fPIC
+        --disable-network  ${FFMPEG_DEBUG}
         INSTALL_COMMAND ""
 
         PREFIX "third-party/ext_ffmpeg"
         SOURCE_DIR "third-party/ext_ffmpeg/src/ffmpeg"
         BINARY_DIR "third-party/ext_ffmpeg/src/ffmpeg"
 
-        BUILD_COMMAND ${MAKE_EXE} -j 8 --silent
+        BUILD_COMMAND ${MAKE_EXE} -j33 --silent
 )
 
 SET(FFMPEG_LIB_DIR ${CMAKE_CURRENT_BINARY_DIR}/third-party/ext_ffmpeg/src/ffmpeg)

libscan/ebook/ebook.c

@@ -134,7 +134,7 @@ int render_cover(scan_ebook_ctx_t *ctx, fz_context *fzctx, document_t *doc, fz_d
     av_image_fill_arrays(scaled_frame->data, scaled_frame->linesize, dst_buf, AV_PIX_FMT_YUV420P, pixmap->w, pixmap->h, 1);
 
     const uint8_t *in_data[1] = {pixmap->samples};
-    int in_line_size[1] = {pixmap->stride};
+    int in_line_size[1] = {(int) pixmap->stride};
 
     sws_scale(sws_ctx,
               in_data, in_line_size,
@@ -223,7 +223,7 @@ static int read_stext_block(fz_stext_block *block, text_buffer_t *tex) {
     return 0;
 }
 
-#define IS_VALID_BPP(d) (d==1 || d==2 || d==4 || d==8 || d==16 || d==24 || d==32)
+#define IS_VALID_BPP(d) ((d)==1 || (d)==2 || (d)==4 || (d)==8 || (d)==16 || (d)==24 || (d)==32)
 
 void fill_image(fz_context *fzctx, UNUSED(fz_device *dev),
                 fz_image *img, UNUSED(fz_matrix ctm), UNUSED(float alpha),
@@ -257,7 +257,7 @@ void fill_image(fz_context *fzctx, UNUSED(fz_device *dev),
 
 void parse_ebook_mem(scan_ebook_ctx_t *ctx, void *buf, size_t buf_len, const char *mime_str, document_t *doc, int tn_only) {
 
-    fz_context *fzctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
+    fz_context *fzctx = fz_new_context(NULL, NULL, FZ_STORE_DEFAULT);
     thread_ctx = *ctx;
 
     init_fzctx(fzctx, doc);

libscan/font/font.c

@@ -144,27 +144,28 @@ void parse_font(scan_font_ctx_t *ctx, vfile_t *f, document_t *doc) {
     }
 
     size_t buf_len = 0;
-    void * buf = read_all(f, &buf_len);
+    void *buf = read_all(f, &buf_len);
     if (buf == NULL) {
         CTX_LOG_ERROR(f->filepath, "read_all() failed")
         return;
     }
 
     FT_Face face;
-    FT_Error err = FT_New_Memory_Face(ft_lib, (unsigned char *) buf, buf_len, 0, &face);
+    FT_Error err = FT_New_Memory_Face(ft_lib, (unsigned char *) buf, (int) buf_len, 0, &face);
     if (err != 0) {
-        CTX_LOG_ERRORF(doc->filepath, "(font.c) FT_New_Memory_Face() returned error code [%d] %s", err, FT_Error_String(err))
+        CTX_LOG_ERRORF(doc->filepath, "(font.c) FT_New_Memory_Face() returned error code [%d] %s", err,
+                       FT_Error_String(err))
         free(buf);
         return;
     }
 
-    char font_name[1024];
+    char font_name[4096];
 
-    if (face->style_name == NULL || *(face->style_name) == '?') {
+    if (face->style_name == NULL || (strcmp(face->style_name, "?") == 0)) {
         if (face->family_name == NULL) {
             strcpy(font_name, "(null)");
         } else {
-            strcpy(font_name, face->family_name);
+            strncpy(font_name, face->family_name, sizeof(font_name));
         }
     } else {
         snprintf(font_name, sizeof(font_name), "%s %s", face->family_name, face->style_name);
@@ -186,7 +187,8 @@ void parse_font(scan_font_ctx_t *ctx, vfile_t *f, document_t *doc) {
 
     err = FT_Set_Pixel_Sizes(face, 0, pixel);
     if (err != 0) {
-        CTX_LOG_WARNINGF(doc->filepath, "(font.c) FT_Set_Pixel_Sizes() returned error code [%d] %s", err, FT_Error_String(err))
+        CTX_LOG_WARNINGF(doc->filepath, "(font.c) FT_Set_Pixel_Sizes() returned error code [%d] %s", err,
+                         FT_Error_String(err))
         FT_Done_Face(face);
         free(buf);
         return;
@@ -207,7 +209,8 @@ void parse_font(scan_font_ctx_t *ctx, vfile_t *f, document_t *doc) {
             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) {
-                CTX_LOG_WARNINGF(doc->filepath, "(font.c) FT_Load_Char() returned error code [%d] %s", err, FT_Error_String(err))
+                CTX_LOG_WARNINGF(doc->filepath, "(font.c) FT_Load_Char() returned error code [%d] %s", err,
+                                 FT_Error_String(err))
                 continue;
             }
         }

libscan/media/media.c

@@ -127,12 +127,15 @@ static void read_subtitles(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx, i
             }
             avsubtitle_free(&subtitle);
         }
+
+        av_packet_unref(&packet);
     }
 
     text_buffer_terminate_string(&tex);
 
     APPEND_STR_META(doc, MetaContent, tex.dyn_buffer.buf)
     text_buffer_destroy(&tex);
+    avcodec_free_context(&decoder);
 }
 
 __always_inline
@@ -284,26 +287,34 @@ append_video_meta(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx, AVFrame *f
 
             if (strcmp(key, "artist") == 0) {
                 append_tag_meta_if_not_exists(ctx, doc, tag, MetaArtist);
-            } else if (strcmp(tag->key, "ImageDescription") == 0) {
+            } else if (strcmp(key, "imagedescription") == 0) {
                 APPEND_TAG_META(MetaContent)
-            } else if (strcmp(tag->key, "Make") == 0) {
+            } else if (strcmp(key, "make") == 0) {
                 APPEND_TAG_META(MetaExifMake)
-            } else if (strcmp(tag->key, "Model") == 0) {
+            } else if (strcmp(key, "model") == 0) {
                 APPEND_TAG_META(MetaExifModel)
-            } else if (strcmp(tag->key, "Software") == 0) {
+            } else if (strcmp(key, "software") == 0) {
                 APPEND_TAG_META(MetaExifSoftware)
-            } else if (strcmp(tag->key, "FNumber") == 0) {
+            } else if (strcmp(key, "fnumber") == 0) {
                 APPEND_TAG_META(MetaExifFNumber)
-            } else if (strcmp(tag->key, "FocalLength") == 0) {
+            } else if (strcmp(key, "focallength") == 0) {
                 APPEND_TAG_META(MetaExifFocalLength)
-            } else if (strcmp(tag->key, "UserComment") == 0) {
+            } else if (strcmp(key, "usercomment") == 0) {
                 APPEND_TAG_META(MetaExifUserComment)
-            } else if (strcmp(tag->key, "ISOSpeedRatings") == 0) {
+            } else if (strcmp(key, "isospeedratings") == 0) {
                 APPEND_TAG_META(MetaExifIsoSpeedRatings)
-            } else if (strcmp(tag->key, "ExposureTime") == 0) {
+            } else if (strcmp(key, "exposuretime") == 0) {
                 APPEND_TAG_META(MetaExifExposureTime)
-            } else if (strcmp(tag->key, "DateTime") == 0) {
+            } else if (strcmp(key, "datetime") == 0) {
                 APPEND_TAG_META(MetaExifDateTime)
+            } else if (strcmp(key, "gpslatitude") == 0) {
+                APPEND_TAG_META(MetaExifGpsLatitudeDMS)
+            } else if (strcmp(key, "gpslatituderef") == 0) {
+                APPEND_TAG_META(MetaExifGpsLatitudeRef)
+            } else if (strcmp(key, "gpslongitude") == 0) {
+                APPEND_TAG_META(MetaExifGpsLongitudeDMS)
+            } else if (strcmp(key, "gpslongituderef") == 0) {
+                APPEND_TAG_META(MetaExifGpsLongitudeRef)
             }
         }
     }

libscan/raw/raw.c

@@ -8,7 +8,7 @@
 #define MIN_SIZE 32
 
 int store_thumbnail_jpeg(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, document_t *doc) {
-    return store_image_thumbnail((scan_media_ctx_t*)ctx, img->data, img->data_size, doc, "x.jpeg");
+    return store_image_thumbnail((scan_media_ctx_t *) ctx, img->data, img->data_size, doc, "x.jpeg");
 }
 
 int store_thumbnail_rgb24(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, document_t *doc) {
@@ -36,7 +36,7 @@ int store_thumbnail_rgb24(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, do
 
     AVFrame *scaled_frame = av_frame_alloc();
 
-    struct SwsContext *sws_ctx= sws_getContext(
+    struct SwsContext *sws_ctx = sws_getContext(
             img->width, img->height, AV_PIX_FMT_RGB24,
             dstW, dstH, AV_PIX_FMT_YUVJ420P,
             SIST_SWS_ALGO, 0, 0, 0
@@ -80,6 +80,8 @@ int store_thumbnail_rgb24(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, do
     return TRUE;
 }
 
+#define DMS_REF(ref) (((ref) == 'S' || (ref) == 'W') ? -1 : 1)
+
 void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc) {
     libraw_data_t *libraw_lib = libraw_init(0);
 
@@ -134,10 +136,23 @@ void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc) {
     snprintf(tmp, sizeof(tmp), "%.1f", libraw_lib->other.aperture);
     APPEND_STR_META(doc, MetaExifFNumber, tmp)
 
-    int denominator = (int)roundf(1 / libraw_lib->other.shutter);
+    int denominator = (int) roundf(1 / libraw_lib->other.shutter);
     snprintf(tmp, sizeof(tmp), "1/%d", denominator);
     APPEND_STR_META(doc, MetaExifExposureTime, tmp)
 
+    libraw_gps_info_t gps = libraw_lib->other.parsed_gps;
+    snprintf(
+            tmp, sizeof(tmp), "%.15f",
+            (gps.longtitude[0] + gps.longtitude[1] / 60 + gps.longtitude[2] / 3600) * DMS_REF(gps.longref)
+    );
+    APPEND_STR_META(doc, MetaExifGpsLongitudeDec, tmp)
+
+    snprintf(
+            tmp, sizeof(tmp), "%.15f",
+            (gps.latitude[0] + gps.latitude[1] / 60 + gps.latitude[2] / 3600) * DMS_REF(gps.latref)
+    );
+    APPEND_STR_META(doc, MetaExifGpsLatitudeDec, tmp)
+
     APPEND_STR_META(doc, MetaMediaVideoCodec, "raw")
 
     if (ctx->tn_size <= 0) {

libscan/scan.h

@@ -13,9 +13,9 @@
 
 #define SIST_SWS_ALGO SWS_LANCZOS
 
-#define META_INT_MASK 0x80
+#define META_INT_MASK 0x8000
-#define META_STR_MASK 0x40
+#define META_STR_MASK 0x4000
-#define META_LONG_MASK 0x20
+#define META_LONG_MASK 0x2000
 
 #define UNUSED(x) __attribute__((__unused__))  x
 
@@ -85,6 +85,12 @@ enum metakey {
     MetaModifiedBy = META_STR(25),
     MetaThumbnail = META_STR(26),
     MetaPages = META_INT(27),
+    MetaExifGpsLongitudeDMS = META_STR(28),
+    MetaExifGpsLongitudeRef = META_STR(29),
+    MetaExifGpsLatitudeDMS = META_STR(30),
+    MetaExifGpsLatitudeRef = META_STR(31),
+    MetaExifGpsLatitudeDec = META_STR(32),
+    MetaExifGpsLongitudeDec = META_STR(33),
 };
 
 typedef struct meta_line {

libscan/util.h

@@ -133,11 +133,11 @@ static void dyn_buffer_write_int(dyn_buffer_t *buf, int d) {
     buf->cur += sizeof(int);
 }
 
-static void dyn_buffer_write_short(dyn_buffer_t *buf, short s) {
+static void dyn_buffer_write_short(dyn_buffer_t *buf, uint16_t s) {
     grow_buffer_small(buf);
 
-    *(short *) (buf->buf + buf->cur) = s;
+    *(uint16_t *) (buf->buf + buf->cur) = s;
-    buf->cur += sizeof(short);
+    buf->cur += sizeof(uint16_t);
 }
 
 static void dyn_buffer_write_long(dyn_buffer_t *buf, unsigned long l) {

test/main.cpp

@@ -319,6 +319,22 @@ TEST(Comic, ComicCbrFilters) {
 
 /* Media (image) */
 
+TEST(MediaImage, ExifGps1) {
+    vfile_t f;
+    document_t doc;
+    load_doc_file("libscan-test-files/test_files/media/exif_GPS.jpg", &f, &doc);
+
+    parse_media(&media_ctx, &f, &doc);
+
+    ASSERT_STREQ(get_meta(&doc, MetaExifGpsLatitudeRef)->str_val, "N");
+    ASSERT_STREQ(get_meta(&doc, MetaExifGpsLatitudeDMS)->str_val, "48:1 , 56585399:1000000, 0:1");
+
+    ASSERT_STREQ(get_meta(&doc, MetaExifGpsLongitudeRef)->str_val, "E");
+    ASSERT_STREQ(get_meta(&doc, MetaExifGpsLongitudeDMS)->str_val, "9:1 , 28046900:1000000, 0:1");
+
+    cleanup(&doc, &f);
+}
+
 TEST(MediaImage, Exif1) {
     vfile_t f;
     document_t doc;
@@ -666,6 +682,23 @@ TEST(RAW, Panasonic) {
     cleanup(&doc, &f);
 }
 
+TEST(RAW, ExifGps1) {
+    vfile_t f;
+    document_t doc;
+    load_doc_file("libscan-test-files/test_files/raw/exif_gps.DNG", &f, &doc);
+
+    size_t size_before = store_size;
+
+    parse_raw(&raw_ctx, &f, &doc);
+
+    ASSERT_NE(size_before, store_size);
+
+    ASSERT_STREQ(get_meta(&doc, MetaExifGpsLatitudeDec)->str_val, "48.943088531494141");
+    ASSERT_STREQ(get_meta(&doc, MetaExifGpsLongitudeDec)->str_val, "9.467448234558105");
+
+    cleanup(&doc, &f);
+}
+
 TEST(RAW, Nikon) {
     vfile_t f;
     document_t doc;

third-party/utf8.h

@@ -1 +1 @@
-Subproject commit ee5a7d4beb7755da13e4d4ec3eccfb65a0530456
+Subproject commit a67acc78fd0fc272ad45362b828efdcb24874e64