From c48717b0a66b646dd54163c57309b03087d00208 Mon Sep 17 00:00:00 2001
From: simon987 <me@simon987.net>
Date: Thu, 26 Mar 2020 09:40:04 -0400
Subject: [PATCH] cbr component

---
 CMakeLists.txt        |  1 +
 libscan/cbr/cbr.c     | 65 +++++++++++++++++++++++++++++++++++++++++++
 libscan/cbr/cbr.h     | 17 +++++++++++
 libscan/ebook/ebook.c |  2 +-
 libscan/ebook/ebook.h |  4 +--
 5 files changed, 86 insertions(+), 3 deletions(-)
 create mode 100644 libscan/cbr/cbr.c
 create mode 100644 libscan/cbr/cbr.h

diff --git a/CMakeLists.txt b/CMakeLists.txt
index f7d4856..3ef7dd8 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -40,6 +40,7 @@ add_library(
         libscan/text/text.c libscan/text/text.h
         libscan/arc/arc.c libscan/arc/arc.h
         libscan/ebook/ebook.c libscan/ebook/ebook.h
+        libscan/cbr/cbr.c libscan/cbr/cbr.h
 
         third-party/utf8.h
 )
diff --git a/libscan/cbr/cbr.c b/libscan/cbr/cbr.c
new file mode 100644
index 0000000..970b0a5
--- /dev/null
+++ b/libscan/cbr/cbr.c
@@ -0,0 +1,65 @@
+#include "cbr.h"
+#include "../scan.h"
+#include "../util.h"
+#include "../arc/arc.h"
+#include "../ebook/ebook.h"
+
+#include <stdlib.h>
+#include <archive.h>
+
+unsigned int cbr_mime;
+unsigned int cbz_mime;
+
+void cbr_init() {
+    //TODO: get mime str
+//    cbr_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/x-cbr");
+//    cbz_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/x-cbz");
+}
+
+int is_cbr(unsigned int mime) {
+    return mime == cbr_mime;
+}
+
+void parse_cbr(scan_cbr_ctx_t *ctx, vfile_t *f, document_t *doc) {
+
+    size_t buf_len;
+    void *buf = read_all(f, &buf_len);
+
+    char *out_buf = malloc(buf_len * 2); // TODO: we probably only need 1.2x or 1.5x, even better would be a dynamic buffer
+    size_t out_buf_used = 0;
+
+    struct archive *rar_in = archive_read_new();
+    archive_read_support_filter_none(rar_in);
+    archive_read_support_format_rar(rar_in);
+
+    archive_read_open_memory(rar_in, buf, buf_len);
+
+    struct archive *zip_out = archive_write_new();
+    archive_write_set_format_zip(zip_out);
+    archive_write_open_memory(zip_out, out_buf, buf_len * 2, &out_buf_used);
+
+    struct archive_entry *entry;
+    while (archive_read_next_header(rar_in, &entry) == ARCHIVE_OK) {
+        archive_write_header(zip_out, entry);
+
+        char arc_buf[ARC_BUF_SIZE];
+        int len = archive_read_data(rar_in, arc_buf, ARC_BUF_SIZE);
+        while (len > 0) {
+            archive_write_data(zip_out, arc_buf, len);
+            len = archive_read_data(rar_in, arc_buf, ARC_BUF_SIZE);
+        }
+    }
+
+    archive_write_close(zip_out);
+    archive_write_free(zip_out);
+
+    archive_read_close(rar_in);
+    archive_read_free(rar_in);
+
+    doc->mime = cbz_mime;
+
+    //TODO: get mime string
+//    parse_ebook(out_buf, out_buf_used, doc);
+    doc->mime = cbr_mime;
+    free(out_buf);
+}
diff --git a/libscan/cbr/cbr.h b/libscan/cbr/cbr.h
new file mode 100644
index 0000000..97d258d
--- /dev/null
+++ b/libscan/cbr/cbr.h
@@ -0,0 +1,17 @@
+#ifndef SCAN_CBR_H
+#define SCAN_CBR_H
+
+#include <stdlib.h>
+#include "../scan.h"
+
+typedef struct {
+
+} scan_cbr_ctx_t;
+
+void cbr_init();
+
+int is_cbr(unsigned int mime);
+
+void parse_cbr(scan_cbr_ctx_t *ctx, vfile_t *f, document_t *doc);
+
+#endif
diff --git a/libscan/ebook/ebook.c b/libscan/ebook/ebook.c
index 10db675..b4c1d68 100644
--- a/libscan/ebook/ebook.c
+++ b/libscan/ebook/ebook.c
@@ -159,7 +159,7 @@ void fill_image(fz_context *fzctx, UNUSED(fz_device *dev),
             if (len >= MIN_OCR_LEN) {
                 text_buffer_append_string(&thread_buffer, text, len - 1);
 //                LOG_DEBUGF(
-//                        "pdf.c",
+//                        "ebook.c",
 //                        "(OCR) %dx%d got %dB from tesseract (%s), buffer:%dB",
 //                        pix->w, pix->h, len, ScanCtx.tesseract_lang, thread_buffer.dyn_buffer.cur
 //                )
diff --git a/libscan/ebook/ebook.h b/libscan/ebook/ebook.h
index 7621e7f..d0bfbc0 100644
--- a/libscan/ebook/ebook.h
+++ b/libscan/ebook/ebook.h
@@ -1,5 +1,5 @@
-#ifndef SCAN_PDF_H
-#define SCAN_PDF_H
+#ifndef SCAN_EBOOK_H
+#define SCAN_EBOOK_H
 
 #include "../scan.h"