Add .doc support

CMakeLists.txt

@@ -5,6 +5,7 @@ project(sist2 C)
 
 option(SIST_DEBUG "Build a debug executable" on)
 
+set(BUILD_TESTS on)
 add_subdirectory(third-party/libscan)
 set(ARGPARSE_SHARED off)
 add_subdirectory(third-party/argparse)

src/ctx.h

@@ -13,6 +13,7 @@
 #include "libscan/text/text.h"
 #include "libscan/mobi/scan_mobi.h"
 #include "libscan/raw/raw.h"
+#include "libscan/msdoc/msdoc.h"
 #include "src/io/store.h"
 
 #include <glib.h>
@@ -48,6 +49,7 @@ typedef struct {
     scan_text_ctx_t text_ctx;
     scan_mobi_ctx_t mobi_ctx;
     scan_raw_ctx_t raw_ctx;
+    scan_msdoc_ctx_t msdoc_ctx;
 } ScanCtx_t;
 
 typedef struct {

src/main.c

@@ -149,6 +149,14 @@ void initialize_scan_context(scan_args_t *args) {
     ScanCtx.text_ctx.log = _log;
     ScanCtx.text_ctx.logf = _logf;
 
+    // MSDOC
+    ScanCtx.msdoc_ctx.tn_size = args->size;
+    ScanCtx.msdoc_ctx.content_size = args->content_size;
+    ScanCtx.msdoc_ctx.log = _log;
+    ScanCtx.msdoc_ctx.logf = _logf;
+    ScanCtx.msdoc_ctx.store = _store;
+    ScanCtx.msdoc_ctx.msdoc_mime = mime_get_mime_by_string(ScanCtx.mime_table, "application/msword");
+
     ScanCtx.threads = args->threads;
     ScanCtx.depth = args->depth;
 

src/parsing/parse.c

@@ -162,6 +162,8 @@ void parse(void *arg) {
         parse_sidecar(&job->vfile, &doc);
         CLOSE_FILE(job->vfile)
         return;
+    } else if (is_msdoc(&ScanCtx.msdoc_ctx, doc.mime)) {
+        parse_msdoc(&ScanCtx.msdoc_ctx, &job->vfile, &doc);
     }
 
     abort: