process_url: strip illegal characters in filename

There are edge cases when a retrived title contains illegal characters
(e.g. '/').
Replace it with a space to ensure a valid filename.

Reproducible case:
https://www.bilibili.com/video/av20827366/
(title="【2K/60fps】这可能是我做过最美的miku了【boomclap布料解算版】")

1 files changed, 8 insertions, 0 deletions

diff --git a/src/process_url.c b/src/process_url.c
index 4bfce8d..7a92013 100644
--- a/src/process_url.c
+++ b/src/process_url.c
@@ -29,6 +29,9 @@ extern Site_map site_map;
 Options options;
 static queue_t dl_queue;
 
+const char illegal_char[] = {'/', '\\', '|', '<', '>',
+                             ':', '"',  '?', '*', '\0'};
+
 thrd_t tid[MAX_THREAD];
 mtx_t mtx;
 cnd_t cnd;
@@ -516,6 +519,11 @@ void add_url(const char *URL, const char *outdir, const char *fn,
   } else {
     filename = malloc(strlen(fn) + 1);
     strcpy(filename, fn);
+    for (unsigned char i = 0; illegal_char[i] != '\0'; i++) {
+      if (repchr(filename, illegal_char[i], ' '))
+        DEBUG_PRINT("Found illegal character '%c' in filename, replacing ...\n",
+                    illegal_char[i]);
+    }
   }
 
   // Pass our cache (outdir_g) to parse_url()