From 40659b5e9d345309515873f61e07c213f6b21ac8 Mon Sep 17 00:00:00 2001
From: notevenaperson <66701832+notevenaperson@users.noreply.github.com>
Date: Sun, 11 Sep 2022 17:23:15 +0000
Subject: [PATCH] singlefile.py: Code to ensure options are deduplicated

---
 archivebox/extractors/singlefile.py | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/archivebox/extractors/singlefile.py b/archivebox/extractors/singlefile.py
index 80ad90b1..f29d59c3 100644
--- a/archivebox/extractors/singlefile.py
+++ b/archivebox/extractors/singlefile.py
@@ -46,11 +46,31 @@ def save_singlefile(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEO
 
     # SingleFile CLI Docs: https://github.com/gildas-lormeau/SingleFile/tree/master/cli
     browser_args = '--browser-args={}'.format(json.dumps(browser_args[1:]))
-    cmd = [
-        DEPENDENCIES['SINGLEFILE_BINARY']['path'],
+    options = [
         *SINGLEFILE_ARGS,
         '--browser-executable-path={}'.format(CHROME_BINARY),
         browser_args,
+    ]
+
+    # Deduplicate options (single-file doesn't like when you use the same option two times)
+    #
+    # NOTE: Options names that come first clobber conflicting names that come later
+    # My logic is SINGLEFILE_ARGS is the option that affects the singlefile command with most 
+    # specificity, therefore the user sets it with a lot intent, therefore it should take precedence 
+    # kind of like the ergonomic principle of lexical scope in programming languages.
+    seen_option_names = []
+    def test_seen(argument):
+        option_name = argument.split("=")[0]
+        if option_name in seen_option_names:
+            return False
+        else:
+            seen_option_names.append(option_name)
+            return True
+    deduped_options = list(filter(test_seen, options))
+
+    cmd = [
+        DEPENDENCIES['SINGLEFILE_BINARY']['path'],
+        *deduped_options
         link.url,
         output,
     ]