Read .raw thumbnails #80, fix media probing for some formats

libscan/comic/comic.c

@@ -30,7 +30,7 @@ void parse_comic(scan_comic_ctx_t *ctx, vfile_t *f, document_t *doc) {
                 void* buf = malloc(entry_size);
                 archive_read_data(a, buf, entry_size);
 
-                ret = store_image_thumbnail((scan_media_ctx_t*)ctx, buf, entry_size, doc);
+                ret = store_image_thumbnail((scan_media_ctx_t*)ctx, buf, entry_size, doc, file_path);
                 free(buf);
 
                 if (ret == TRUE) {

libscan/media/media.c

@@ -396,7 +396,14 @@ typedef struct {
 
 int memfile_read(void *ptr, uint8_t *buf, int buf_size) {
     memfile_t *mem = ptr;
-    return (int) fread(buf, 1, buf_size, mem->file);
+
+    size_t ret = fread(buf, 1, buf_size, mem->file);
+
+    if (ret != buf_size) {
+        return AVERROR_EOF;
+    }
+
+    return buf_size;
 }
 
 long memfile_seek(void *ptr, long offset, int whence) {
@@ -406,7 +413,12 @@ long memfile_seek(void *ptr, long offset, int whence) {
         return mem->info.st_size;
     }
 
-    return fseek(mem->file, offset, whence);
+    int ret = fseek(mem->file, offset, whence);
+    if (ret != 0) {
+        return AVERROR_EOF;
+    }
+
+    return ftell(mem->file);
 }
 
 int memfile_open(vfile_t *f, memfile_t *mem) {
@@ -465,23 +477,19 @@ void parse_media_vfile(scan_media_ctx_t *ctx, struct vfile *f, document_t *doc)
     }
 
     pFormatCtx->pb = io_ctx;
+    pFormatCtx->flags = AVFMT_FLAG_CUSTOM_IO;
 
-    int res = avformat_open_input(&pFormatCtx, "", NULL, NULL);
+    int res = avformat_open_input(&pFormatCtx, f->filepath, NULL, NULL);
-    if (res == -5) {
+    if (res < 0) {
-        // Tried to parse media that requires seek
+        if (res != -5) {
-        av_free(io_ctx->buffer);
-        memfile_close(&memfile);
-        avio_context_free(&io_ctx);
-        avformat_close_input(&pFormatCtx);
-        avformat_free_context(pFormatCtx);
-        return;
-    } else if (res < 0) {
             CTX_LOG_ERRORF(doc->filepath, "(media.c) avformat_open_input() returned [%d] %s", res, av_err2str(res))
+        }
         av_free(io_ctx->buffer);
         memfile_close(&memfile);
         avio_context_free(&io_ctx);
         avformat_close_input(&pFormatCtx);
         avformat_free_context(pFormatCtx);
+        printf("%s\n", av_err2str(res));
         return;
     }
 
@@ -504,7 +512,7 @@ void init_media() {
     av_log_set_level(AV_LOG_QUIET);
 }
 
-int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, document_t *doc) {
+int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, document_t *doc, const char *url) {
     memfile_t memfile;
     AVIOContext *io_ctx = NULL;
 
@@ -531,7 +539,7 @@ int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, docu
 
     pFormatCtx->pb = io_ctx;
 
-    int res = avformat_open_input(&pFormatCtx, "", NULL, NULL);
+    int res = avformat_open_input(&pFormatCtx, url, NULL, NULL);
     if (res != 0) {
         av_free(io_ctx->buffer);
         avformat_close_input(&pFormatCtx);

libscan/media/media.h

@@ -45,6 +45,6 @@ static AVCodecContext *alloc_jpeg_encoder(int w, int h, float qscale) {
 void parse_media(scan_media_ctx_t *ctx, vfile_t *f, document_t *doc);
 void init_media();
 
-int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, document_t *doc);
+int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, document_t *doc, const char *url);
 
 #endif

libscan/raw/raw.c

@@ -2,89 +2,17 @@
 #include <libraw/libraw.h>
 
 #include "../media/media.h"
+#include <fcntl.h>
+#include <unistd.h>
 
 
 #define MIN_SIZE 32
 
-void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc) {
+int store_thumbnail_jpeg(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, document_t *doc) {
-    libraw_data_t *libraw_lib = libraw_init(0);
+    return store_image_thumbnail((scan_media_ctx_t*)ctx, img->data, img->data_size, doc, "x.jpeg");
-
-    if (!libraw_lib) {
-        CTX_LOG_ERROR("raw.c", "Cannot create libraw handle")
-        return;
 }
 
-    size_t buf_len = 0;
+int store_thumbnail_rgb24(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, document_t *doc) {
-    void *buf = read_all(f, &buf_len);
-
-    int ret = libraw_open_buffer(libraw_lib, buf, buf_len);
-    if (ret != 0) {
-        CTX_LOG_ERROR(f->filepath, "Could not open raw file")
-        free(buf);
-        return;
-    }
-    ret = libraw_unpack(libraw_lib);
-    if (ret != 0) {
-        CTX_LOG_ERROR(f->filepath, "Could not unpack raw file")
-        free(buf);
-        libraw_close(libraw_lib);
-        return;
-    }
-
-    libraw_dcraw_process(libraw_lib);
-
-    if (*libraw_lib->idata.model != '\0') {
-        APPEND_STR_META(doc, MetaExifModel, libraw_lib->idata.model)
-    }
-    if (*libraw_lib->idata.make != '\0') {
-        APPEND_STR_META(doc, MetaExifMake, libraw_lib->idata.make)
-    }
-    if (*libraw_lib->idata.software != '\0') {
-        APPEND_STR_META(doc, MetaExifSoftware, libraw_lib->idata.software)
-    }
-    APPEND_INT_META(doc, MetaWidth, libraw_lib->sizes.width)
-    APPEND_INT_META(doc, MetaHeight, libraw_lib->sizes.height)
-    char tmp[1024];
-    snprintf(tmp, sizeof(tmp), "%g", libraw_lib->other.iso_speed);
-    APPEND_STR_META(doc, MetaExifIsoSpeedRatings, tmp)
-
-    if (*libraw_lib->other.desc != '\0') {
-        APPEND_STR_META(doc, MetaContent, libraw_lib->other.desc)
-    }
-    if (*libraw_lib->other.artist != '\0') {
-        APPEND_STR_META(doc, MetaArtist, libraw_lib->other.artist)
-    }
-
-    struct tm *time = localtime(&libraw_lib->other.timestamp);
-    strftime(tmp, sizeof(tmp), "%Y:%m:%d %H:%M:%S", time);
-    APPEND_STR_META(doc, MetaExifDateTime, tmp)
-
-    snprintf(tmp, sizeof(tmp), "%.1f", libraw_lib->other.focal_len);
-    APPEND_STR_META(doc, MetaExifFocalLength, tmp)
-
-    snprintf(tmp, sizeof(tmp), "%.1f", libraw_lib->other.aperture);
-    APPEND_STR_META(doc, MetaExifFNumber, tmp)
-
-    int denominator = (int)roundf(1 / libraw_lib->other.shutter);
-    snprintf(tmp, sizeof(tmp), "1/%d", denominator);
-    APPEND_STR_META(doc, MetaExifExposureTime, tmp)
-
-    APPEND_STR_META(doc, MetaMediaVideoCodec, "raw")
-
-    if (ctx->tn_size <= 0) {
-        free(buf);
-        libraw_close(libraw_lib);
-        return;
-    }
-
-    int errc = 0;
-    libraw_processed_image_t *img = libraw_dcraw_make_mem_image(libraw_lib, &errc);
-    if (errc != 0) {
-        free(buf);
-        libraw_dcraw_clear_mem(img);
-        libraw_close(libraw_lib);
-        return;
-    }
 
     int dstW;
     int dstH;
@@ -104,10 +32,7 @@ void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc) {
     }
 
     if (dstW <= MIN_SIZE || dstH <= MIN_SIZE) {
-        free(buf);
+        return FALSE;
-        libraw_dcraw_clear_mem(img);
-        libraw_close(libraw_lib);
-        return;
     }
 
     AVFrame *scaled_frame = av_frame_alloc();
@@ -153,6 +78,119 @@ void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc) {
     av_frame_free(&scaled_frame);
     avcodec_free_context(&jpeg_encoder);
 
+    return TRUE;
+}
+
+void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc) {
+    libraw_data_t *libraw_lib = libraw_init(0);
+
+    if (!libraw_lib) {
+        CTX_LOG_ERROR("raw.c", "Cannot create libraw handle")
+        return;
+    }
+
+    size_t buf_len = 0;
+    void *buf = read_all(f, &buf_len);
+
+    int ret = libraw_open_buffer(libraw_lib, buf, buf_len);
+    if (ret != 0) {
+        CTX_LOG_ERROR(f->filepath, "Could not open raw file")
+        free(buf);
+        return;
+    }
+
+    if (*libraw_lib->idata.model != '\0') {
+        APPEND_STR_META(doc, MetaExifModel, libraw_lib->idata.model)
+    }
+    if (*libraw_lib->idata.make != '\0') {
+        APPEND_STR_META(doc, MetaExifMake, libraw_lib->idata.make)
+    }
+    if (*libraw_lib->idata.software != '\0') {
+        APPEND_STR_META(doc, MetaExifSoftware, libraw_lib->idata.software)
+    }
+    APPEND_INT_META(doc, MetaWidth, libraw_lib->sizes.width)
+    APPEND_INT_META(doc, MetaHeight, libraw_lib->sizes.height)
+    char tmp[1024];
+    snprintf(tmp, sizeof(tmp), "%g", libraw_lib->other.iso_speed);
+    APPEND_STR_META(doc, MetaExifIsoSpeedRatings, tmp)
+
+    if (*libraw_lib->other.desc != '\0') {
+        APPEND_STR_META(doc, MetaContent, libraw_lib->other.desc)
+    }
+    if (*libraw_lib->other.artist != '\0') {
+        APPEND_STR_META(doc, MetaArtist, libraw_lib->other.artist)
+    }
+
+    struct tm *time = localtime(&libraw_lib->other.timestamp);
+    strftime(tmp, sizeof(tmp), "%Y:%m:%d %H:%M:%S", time);
+    APPEND_STR_META(doc, MetaExifDateTime, tmp)
+
+    snprintf(tmp, sizeof(tmp), "%.1f", libraw_lib->other.focal_len);
+    APPEND_STR_META(doc, MetaExifFocalLength, tmp)
+
+    snprintf(tmp, sizeof(tmp), "%.1f", libraw_lib->other.aperture);
+    APPEND_STR_META(doc, MetaExifFNumber, tmp)
+
+    int denominator = (int)roundf(1 / libraw_lib->other.shutter);
+    snprintf(tmp, sizeof(tmp), "1/%d", denominator);
+    APPEND_STR_META(doc, MetaExifExposureTime, tmp)
+
+    APPEND_STR_META(doc, MetaMediaVideoCodec, "raw")
+
+    if (ctx->tn_size <= 0) {
+        free(buf);
+        libraw_close(libraw_lib);
+        return;
+    }
+
+    libraw_unpack_thumb(libraw_lib);
+
+    int errc = 0;
+    libraw_processed_image_t *thumb = libraw_dcraw_make_mem_thumb(libraw_lib, &errc);
+    if (errc != 0) {
+        free(buf);
+        libraw_dcraw_clear_mem(thumb);
+        libraw_close(libraw_lib);
+        return;
+    }
+
+    int tn_ok = 0;
+    if (libraw_lib->thumbnail.tformat == LIBRAW_THUMBNAIL_JPEG) {
+//        tn_ok = store_thumbnail_jpeg(ctx, thumb, doc);
+    } else if (libraw_lib->thumbnail.tformat == LIBRAW_THUMBNAIL_BITMAP) {
+        // TODO: technically this should work but is currently untested
+        tn_ok = store_thumbnail_rgb24(ctx, thumb, doc);
+    }
+
+    libraw_dcraw_clear_mem(thumb);
+
+    if (tn_ok == TRUE) {
+        free(buf);
+        libraw_close(libraw_lib);
+        return;
+    }
+
+    ret = libraw_unpack(libraw_lib);
+    if (ret != 0) {
+        CTX_LOG_ERROR(f->filepath, "Could not unpack raw file")
+        free(buf);
+        libraw_close(libraw_lib);
+        return;
+    }
+
+    libraw_dcraw_process(libraw_lib);
+
+    errc = 0;
+    libraw_processed_image_t *img = libraw_dcraw_make_mem_image(libraw_lib, &errc);
+    if (errc != 0) {
+        free(buf);
+        libraw_dcraw_clear_mem(img);
+        libraw_close(libraw_lib);
+        return;
+    }
+
+    store_thumbnail_rgb24(ctx, img, doc);
+
     libraw_dcraw_clear_mem(img);
     libraw_close(libraw_lib);
 

libscan/raw/raw.h

@@ -4,12 +4,12 @@
 #include "../scan.h"
 
 typedef struct {
-    int tn_size;
-    float tn_qscale;
-
     log_callback_t log;
     logf_callback_t logf;
     store_callback_t store;
+
+    int tn_size;
+    float tn_qscale;
 } scan_raw_ctx_t;
 
 void parse_raw(scan_raw_ctx_t *ctx, vfile_t *f, document_t *doc);

test/main.cpp

@@ -268,6 +268,20 @@ TEST(MediaImage, Exif1) {
     cleanup(&doc, &f);
 }
 
+TEST(MediaImage, Mem1) {
+    vfile_t f;
+    document_t doc;
+    load_doc_file("libscan-test-files/test_files/media/test.jpeg.tar", &f, &doc);
+
+    size_t size_before = store_size;
+
+    parse_archive(&arc_recurse_media_ctx, &f, &doc);
+
+    ASSERT_NE(size_before, store_size);
+
+    cleanup(&doc, &f);
+}
+
 TEST(MediaVideo, Vid3Mp4) {
     vfile_t f;
     document_t doc;
@@ -320,13 +334,15 @@ TEST(MediaVideoVfile, Vid3Ogv) {
     document_t doc;
     load_doc_file("libscan-test-files/test_files/arc/vid3.tar", &f, &doc);
 
+    size_t size_before = store_size;
+
     parse_archive(&arc_recurse_media_ctx, &f, &doc);
 
-    ASSERT_STREQ(get_meta(&LastSubDoc, MetaMediaVideoCodec)->str_val, "theora");
+//    ASSERT_STREQ(get_meta(&LastSubDoc, MetaMediaVideoCodec)->str_val, "theora");
-    ASSERT_EQ(get_meta(&LastSubDoc, MetaMediaBitrate)->long_val, 590261);
+    ASSERT_EQ(get_meta(&LastSubDoc, MetaMediaBitrate)->long_val, 600758);
     ASSERT_EQ(get_meta(&LastSubDoc, MetaMediaDuration)->long_val, 10);
+    ASSERT_NE(size_before, store_size);
 
-    //TODO: Check that thumbnail was generated correctly
     cleanup(&doc, &f);
 }
 
@@ -493,6 +509,8 @@ TEST(RAW, Panasonic) {
     document_t doc;
     load_doc_file("libscan-test-files/test_files/raw/Panasonic.RW2", &f, &doc);
 
+    size_t size_before = store_size;
+
     parse_raw(&raw_ctx, &f, &doc);
 
     ASSERT_STREQ(get_meta(&doc, MetaMediaVideoCodec)->str_val, "raw");
@@ -504,7 +522,7 @@ TEST(RAW, Panasonic) {
     ASSERT_STREQ(get_meta(&doc, MetaExifFNumber)->str_val, "2.0");
     ASSERT_EQ(get_meta(&doc, MetaWidth)->int_val, 5200);
     ASSERT_EQ(get_meta(&doc, MetaHeight)->int_val, 3904);
-
+    ASSERT_NE(size_before, store_size);
 
     cleanup(&doc, &f);
 }
@@ -514,6 +532,8 @@ TEST(RAW, Nikon) {
     document_t doc;
     load_doc_file("libscan-test-files/test_files/raw/Nikon.NEF", &f, &doc);
 
+    size_t size_before = store_size;
+
     parse_raw(&raw_ctx, &f, &doc);
 
     ASSERT_STREQ(get_meta(&doc, MetaMediaVideoCodec)->str_val, "raw");
@@ -521,6 +541,7 @@ TEST(RAW, Nikon) {
     ASSERT_STREQ(get_meta(&doc, MetaExifMake)->str_val, "Nikon");
     ASSERT_EQ(get_meta(&doc, MetaWidth)->int_val, 6032);
     ASSERT_EQ(get_meta(&doc, MetaHeight)->int_val, 4032);
+    ASSERT_NE(size_before, store_size);
 
     cleanup(&doc, &f);
 }
@@ -530,6 +551,8 @@ TEST(RAW, Sony) {
     document_t doc;
     load_doc_file("libscan-test-files/test_files/raw/Sony.ARW", &f, &doc);
 
+    size_t size_before = store_size;
+
     parse_raw(&raw_ctx, &f, &doc);
 
     ASSERT_STREQ(get_meta(&doc, MetaMediaVideoCodec)->str_val, "raw");
@@ -537,6 +560,7 @@ TEST(RAW, Sony) {
     ASSERT_STREQ(get_meta(&doc, MetaExifMake)->str_val, "Sony");
     ASSERT_EQ(get_meta(&doc, MetaWidth)->int_val, 7968);
     ASSERT_EQ(get_meta(&doc, MetaHeight)->int_val, 5320);
+    ASSERT_NE(size_before, store_size);
 
     cleanup(&doc, &f);
 }
@@ -546,6 +570,8 @@ TEST(RAW, Olympus) {
     document_t doc;
     load_doc_file("libscan-test-files/test_files/raw/Olympus.ORF", &f, &doc);
 
+    size_t size_before = store_size;
+
     parse_raw(&raw_ctx, &f, &doc);
 
     ASSERT_STREQ(get_meta(&doc, MetaMediaVideoCodec)->str_val, "raw");
@@ -553,6 +579,7 @@ TEST(RAW, Olympus) {
     ASSERT_STREQ(get_meta(&doc, MetaExifMake)->str_val, "Olympus");
     ASSERT_EQ(get_meta(&doc, MetaWidth)->int_val, 4640);
     ASSERT_EQ(get_meta(&doc, MetaHeight)->int_val, 3472);
+    ASSERT_NE(size_before, store_size);
 
     cleanup(&doc, &f);
 }
@@ -562,6 +589,7 @@ TEST(RAW, Fuji) {
     document_t doc;
     load_doc_file("libscan-test-files/test_files/raw/Fuji.RAF", &f, &doc);
 
+    size_t size_before = store_size;
     parse_raw(&raw_ctx, &f, &doc);
 
     ASSERT_STREQ(get_meta(&doc, MetaMediaVideoCodec)->str_val, "raw");
@@ -569,6 +597,7 @@ TEST(RAW, Fuji) {
     ASSERT_STREQ(get_meta(&doc, MetaExifMake)->str_val, "Fujifilm");
     ASSERT_EQ(get_meta(&doc, MetaWidth)->int_val, 6032);
     ASSERT_EQ(get_meta(&doc, MetaHeight)->int_val, 4028);
+    ASSERT_NE(size_before, store_size);
 
     cleanup(&doc, &f);
 }
@@ -614,7 +643,7 @@ int main(int argc, char **argv) {
     media_ctx.store = counter_store;
     media_ctx.tn_size = 500;
     media_ctx.tn_qscale = 1.0;
-    media_ctx.max_media_buffer = (long)2000 * 1024 * 1024;
+    media_ctx.max_media_buffer = (long)2000 * (long)1024 * (long)1024;
 
     ooxml_500_ctx.content_size = 500;
     ooxml_500_ctx.log = noop_log;