From 1d95be534b75850a22ffd95dabfd3fd2a5ab180f Mon Sep 17 00:00:00 2001
From: simon987 <me@simon987.net>
Date: Mon, 20 Sep 2021 20:15:35 -0400
Subject: [PATCH] Handle excludes in archive files

---
 libscan/arc/arc.c | 14 +++++++++++---
 libscan/arc/arc.h |  3 ++-
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/libscan/arc/arc.c b/libscan/arc/arc.c
index 6830b7e..c87d965 100644
--- a/libscan/arc/arc.c
+++ b/libscan/arc/arc.c
@@ -142,7 +142,10 @@ int arc_open(scan_arc_ctx_t *ctx, vfile_t *f, struct archive **a, arc_data_t *ar
     }
 }
 
-scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc) {
+static __thread int sub_strings[30];
+#define EXCLUDED(str) (pcre_exec(exclude, exclude_extra, str, strlen(str), 0, 0, sub_strings, sizeof(sub_strings)) >= 0)
+
+scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc, pcre *exclude, pcre_extra *exclude_extra) {
 
     struct archive *a = NULL;
     struct archive_entry *entry = NULL;
@@ -213,14 +216,19 @@ scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc) {
                 }
                 sub_job->base = (int) (strrchr(sub_job->filepath, '/') - sub_job->filepath) + 1;
 
+                // Handle excludes
+                if (exclude != NULL && EXCLUDED(sub_job->filepath)) {
+                    CTX_LOG_DEBUGF("arc.c", "Excluded: %s", sub_job->filepath)
+                    continue;
+                }
+
                 char *p = strrchr(sub_job->filepath, '.');
-                if (p != NULL) {
+                if (p != NULL && (p - sub_job->filepath) > strlen(f->filepath)) {
                     sub_job->ext = (int) (p - sub_job->filepath + 1);
                 } else {
                     sub_job->ext = (int) strlen(sub_job->filepath);
                 }
 
-                memset(&sub_job->vfile.sha1_ctx, 0, sizeof(sub_job->vfile.sha1_ctx));
                 SHA1_Init(&sub_job->vfile.sha1_ctx);
 
                 ctx->parse(sub_job);
diff --git a/libscan/arc/arc.h b/libscan/arc/arc.h
index b38ce98..ea9d570 100644
--- a/libscan/arc/arc.h
+++ b/libscan/arc/arc.h
@@ -4,6 +4,7 @@
 #include <archive.h>
 #include <archive_entry.h>
 #include <fcntl.h>
+#include <pcre.h>
 #include "../scan.h"
 
 # define ARC_SKIPPED (-1)
@@ -68,7 +69,7 @@ int arc_open(scan_arc_ctx_t *ctx, vfile_t *f, struct archive **a, arc_data_t *ar
 
 int should_parse_filtered_file(const char *filepath, int ext);
 
-scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc);
+scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc, pcre *exclude, pcre_extra *exclude_extra);
 
 int arc_read(struct vfile *f, void *buf, size_t size);