Remove UUID dep, fix incremental scan, use MD5(path) as unique id, version bump

libscan/arc/arc.c

@@ -5,8 +5,6 @@
 #include <string.h>
 #include <fcntl.h>
 
-#define MAX_SUBFILE_SIZE (long)(10000000000)
-
 
 int should_parse_filtered_file(const char *filepath, int ext) {
     char tmp[PATH_MAX * 2];
@@ -110,7 +108,7 @@ scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc) {
         meta_line_t *meta_list = malloc(sizeof(meta_line_t) + buf.cur);
         meta_list->key = MetaContent;
         strcpy(meta_list->str_val, buf.buf);
-        APPEND_META(doc, meta_list);
+        APPEND_META(doc, meta_list)
         dyn_buffer_destroy(&buf);
 
     } else {
@@ -125,7 +123,7 @@ scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc) {
         sub_job->vfile.is_fs_file = FALSE;
         sub_job->vfile.log = ctx->log;
         sub_job->vfile.logf = ctx->logf;
-        memcpy(sub_job->parent, doc->uuid, sizeof(uuid_t));
+        memcpy(sub_job->parent, doc->path_md5, MD5_DIGEST_LENGTH);
 
         while (archive_read_next_header(a, &entry) == ARCHIVE_OK) {
             sub_job->vfile.info = *archive_entry_stat(entry);

libscan/ebook/ebook.c

@@ -157,7 +157,7 @@ int render_cover(scan_ebook_ctx_t *ctx, fz_context *fzctx, document_t *doc, fz_d
     avcodec_receive_packet(jpeg_encoder, &jpeg_packet);
 
     APPEND_TN_META(doc, pixmap->w, pixmap->h)
-    ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) jpeg_packet.data, jpeg_packet.size);
+    ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) jpeg_packet.data, jpeg_packet.size);
 
     av_packet_unref(&jpeg_packet);
     av_free(*scaled_frame->data);
@@ -174,14 +174,14 @@ void fz_err_callback(void *user, const char *message) {
     document_t *doc = (document_t *) user;
 
     const scan_ebook_ctx_t *ctx = &thread_ctx;
-    CTX_LOG_WARNINGF(doc->filepath, "FZ: %s", message);
+    CTX_LOG_WARNINGF(doc->filepath, "FZ: %s", message)
 }
 
 void fz_warn_callback(void *user, const char *message) {
     document_t *doc = (document_t *) user;
 
     const scan_ebook_ctx_t *ctx = &thread_ctx;
-    CTX_LOG_DEBUGF(doc->filepath, "FZ: %s", message);
+    CTX_LOG_DEBUGF(doc->filepath, "FZ: %s", message)
 }
 
 static void init_fzctx(fz_context *fzctx, document_t *doc) {

libscan/font/font.c

@@ -229,7 +229,7 @@ void parse_font(scan_font_ctx_t *ctx, vfile_t *f, document_t *doc) {
     bmp_format(&bmp_data, dimensions, bitmap);
 
     APPEND_TN_META(doc, dimensions.width, dimensions.height)
-    ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) bmp_data.buf, bmp_data.cur);
+    ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) bmp_data.buf, bmp_data.cur);
 
     dyn_buffer_destroy(&bmp_data);
     free(bitmap);

libscan/media/media.c

@@ -1,8 +1,4 @@
 #include "media.h"
-
-#include "../util.h"
-
-
 #include <ctype.h>
 
 #define MIN_SIZE 32
@@ -323,7 +319,7 @@ void parse_media_format_ctx(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx,
 
         //Seek
         if (stream->nb_frames > 1 && stream->codecpar->codec_id != AV_CODEC_ID_GIF) {
-            int seek_ret = 0;
+            int seek_ret;
             for (int i = 20; i >= 0; i--) {
                 seek_ret = av_seek_frame(pFormatCtx, video_stream,
                                          stream->duration * 0.10, 0);
@@ -356,7 +352,7 @@ void parse_media_format_ctx(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx,
 
         if (scaled_frame == STORE_AS_IS) {
             APPEND_TN_META(doc, frame_and_packet->frame->width, frame_and_packet->frame->height)
-            ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) frame_and_packet->packet->data, frame_and_packet->packet->size);
+            ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) frame_and_packet->packet->data, frame_and_packet->packet->size);
         } else {
             // Encode frame to jpeg
             AVCodecContext *jpeg_encoder = alloc_jpeg_encoder(scaled_frame->width, scaled_frame->height,
@@ -369,7 +365,7 @@ void parse_media_format_ctx(scan_media_ctx_t *ctx, AVFormatContext *pFormatCtx,
 
             // Save thumbnail
             APPEND_TN_META(doc, scaled_frame->width, scaled_frame->height)
-            ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) jpeg_packet.data, jpeg_packet.size);
+            ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) jpeg_packet.data, jpeg_packet.size);
 
             avcodec_free_context(&jpeg_encoder);
             av_packet_unref(&jpeg_packet);
@@ -608,7 +604,7 @@ int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, docu
 
     if (scaled_frame == STORE_AS_IS) {
         APPEND_TN_META(doc, frame_and_packet->frame->width, frame_and_packet->frame->height)
-        ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) frame_and_packet->packet->data, frame_and_packet->packet->size);
+        ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) frame_and_packet->packet->data, frame_and_packet->packet->size);
     } else {
         // Encode frame to jpeg
         AVCodecContext *jpeg_encoder = alloc_jpeg_encoder(scaled_frame->width, scaled_frame->height,
@@ -621,7 +617,7 @@ int store_image_thumbnail(scan_media_ctx_t *ctx, void* buf, size_t buf_len, docu
 
         // Save thumbnail
         APPEND_TN_META(doc, scaled_frame->width, scaled_frame->height)
-        ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) jpeg_packet.data, jpeg_packet.size);
+        ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) jpeg_packet.data, jpeg_packet.size);
 
         av_packet_unref(&jpeg_packet);
         avcodec_free_context(&jpeg_encoder);

libscan/msdoc/msdoc.c

@@ -6,7 +6,7 @@
 
 #include "../ebook/ebook.h"
 
-void parse_msdoc_text(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc, FILE *file_in, void* buf, size_t buf_len) {
+void parse_msdoc_text(scan_msdoc_ctx_t *ctx, document_t *doc, FILE *file_in, void* buf, size_t buf_len) {
 
     // Open word doc
     options_type *opts = direct_vGetOptions();
@@ -73,7 +73,7 @@ void parse_msdoc_text(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc, FILE *
     free(out_buf);
 }
 
-void parse_msdoc_pdf(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc, FILE *file, void* buf, size_t buf_len) {
+void parse_msdoc_pdf(scan_msdoc_ctx_t *ctx, document_t *doc, FILE *file, void* buf, size_t buf_len) {
 
     scan_ebook_ctx_t ebook_ctx = {
             .content_size = ctx->content_size,
@@ -115,7 +115,7 @@ void parse_msdoc_pdf(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc, FILE *f
         return;
     }
 
-    int ret = bWordDecryptor(file, buf_len, diag);
+    bWordDecryptor(file, buf_len, diag);
     vDestroyDiagram(diag);
 
     fclose(file_out);
@@ -144,8 +144,8 @@ void parse_msdoc(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc) {
     }
 
     if (ctx->tn_size > 0) {
-        parse_msdoc_pdf(ctx, f, doc, file, buf, buf_len);
+        parse_msdoc_pdf(ctx, doc, file, buf, buf_len);
     } else {
-        parse_msdoc_text(ctx, f, doc, file, buf, buf_len);
+        parse_msdoc_text(ctx, doc, file, buf, buf_len);
     }
 }

libscan/msdoc/msdoc.h

@@ -19,6 +19,6 @@ static int is_msdoc(scan_msdoc_ctx_t *ctx, unsigned int mime) {
 
 void parse_msdoc(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc);
 
-void parse_msdoc_text(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc, FILE *file_in, void* buf, size_t buf_len);
+void parse_msdoc_text(scan_msdoc_ctx_t *ctx, document_t *doc, FILE *file_in, void* buf, size_t buf_len);
 
 #endif

libscan/ooxml/ooxml.c

@@ -1,6 +1,5 @@
 #include "ooxml.h"
 
-#include "../util.h"
 #include <archive.h>
 #include <archive_entry.h>
 #include <libxml/xmlstring.h>
@@ -106,7 +105,7 @@ static int read_part(scan_ooxml_ctx_t *ctx, struct archive *a, text_buffer_t *bu
 }
 
 __always_inline
-static int read_doc_props(scan_ooxml_ctx_t *ctx, struct archive *a, text_buffer_t *buf, document_t *doc) {
+static int read_doc_props(scan_ooxml_ctx_t *ctx, struct archive *a, document_t *doc) {
     xmlDoc *xml = xmlReadIO(xml_io_read, xml_io_close, a, "/", NULL,
                             XML_PARSE_RECOVER | XML_PARSE_NOWARNING | XML_PARSE_NOERROR | XML_PARSE_NONET);
 
@@ -158,7 +157,7 @@ void read_thumbnail(scan_ooxml_ctx_t *ctx, document_t *doc, struct archive *a, s
     archive_read_data(a, buf, entry_size);
 
     APPEND_TN_META(doc, 1, 1) // Size unknown
-    ctx->store((char *) doc->uuid, sizeof(doc->uuid), buf, entry_size);
+    ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), buf, entry_size);
     free(buf);
 }
 
@@ -198,7 +197,7 @@ void parse_ooxml(scan_ooxml_ctx_t *ctx, vfile_t *f, document_t *doc) {
                     buffer_full = TRUE;
                 }
             } else if (strcmp(path, "docProps/core.xml") == 0) {
-                if (read_doc_props(ctx, a, &tex, doc) != 0) {
+                if (read_doc_props(ctx, a, doc) != 0) {
                     break;
                 }
             } else if (strcmp(path, "docProps/thumbnail.jpeg") == 0) {

libscan/raw/raw.c

@@ -2,7 +2,6 @@
 #include <libraw/libraw.h>
 
 #include "../media/media.h"
-#include <fcntl.h>
 #include <unistd.h>
 
 
@@ -71,7 +70,7 @@ int store_thumbnail_rgb24(scan_raw_ctx_t *ctx, libraw_processed_image_t *img, do
     avcodec_receive_packet(jpeg_encoder, &jpeg_packet);
 
     APPEND_TN_META(doc, scaled_frame->width, scaled_frame->height)
-    ctx->store((char *) doc->uuid, sizeof(doc->uuid), (char *) jpeg_packet.data, jpeg_packet.size);
+    ctx->store((char *) doc->path_md5, sizeof(doc->path_md5), (char *) jpeg_packet.data, jpeg_packet.size);
 
     av_packet_unref(&jpeg_packet);
     av_free(*scaled_frame->data);

libscan/scan.h

@@ -7,7 +7,7 @@
 
 #include <stdio.h>
 #include <sys/stat.h>
-#include <uuid/uuid.h>
+#include <openssl/md5.h>
 
 #include "macros.h"
 
@@ -25,7 +25,7 @@
 
 #define IS_META_INT(key) (key & META_INT_MASK) == META_INT_MASK
 #define IS_META_LONG(key) (key & META_LONG_MASK) == META_LONG_MASK
-#define IS_META_STR(meta) (meta->key & META_STR_MASK) == META_STR_MASK
+#define IS_META_STR(key) (key & META_STR_MASK) == META_STR_MASK
 
 typedef void (*store_callback_t)(char *key, size_t key_len, char *buf, size_t buf_len);
 typedef void (*logf_callback_t)(const char *filepath, int level, char *format, ...);
@@ -99,13 +99,13 @@ typedef struct meta_line {
 
 
 typedef struct document {
-    unsigned char uuid[16];
-    unsigned long ino;
+    unsigned char path_md5[MD5_DIGEST_LENGTH];
     unsigned long size;
     unsigned int mime;
     int mtime;
     short base;
     short ext;
+    char has_parent;
     meta_line_t *meta_head;
     meta_line_t *meta_tail;
     char *filepath;
@@ -146,7 +146,7 @@ typedef struct parse_job_t {
     int base;
     int ext;
     struct vfile vfile;
-    uuid_t parent;
+    unsigned char parent[MD5_DIGEST_LENGTH];
     char filepath[1];
 } parse_job_t;
 

test/main.cpp

@@ -801,7 +801,7 @@ TEST(Msdoc, TestFuzz1) {
 
         fuzz_buffer(buf_copy, &buf_len_copy, 3, 8, 5);
         FILE *file = fmemopen(buf_copy, buf_len_copy, "rb");
-        parse_msdoc_text(&msdoc_text_ctx, &f, &doc, file, buf_copy, buf_len_copy);
+        parse_msdoc_text(&msdoc_text_ctx, &doc, file, buf_copy, buf_len_copy);
     }
     free(buf);
     cleanup(&doc, &f);

test/test_util.h

@@ -4,7 +4,6 @@
 #include "../libscan/scan.h"
 #include <fcntl.h>
 #include <unistd.h>
-#include <uuid/uuid.h>
 
 void load_file(const char *filepath, vfile_t *f);
 void load_mem(void *mem, size_t size, vfile_t *f);