From 8899fe0b9259748da2ef19d37028c317f39f37d3 Mon Sep 17 00:00:00 2001
From: renaisun <43090234+renaisun@users.noreply.github.com>
Date: Thu, 9 Jun 2022 14:35:48 +0800
Subject: [PATCH 1/3] Add SINGLEFILE_ARGS to control single-file arguments

---
 archivebox/config.py                | 2 ++
 archivebox/extractors/singlefile.py | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/archivebox/config.py b/archivebox/config.py
index cfe41b53..bd3a1688 100644
--- a/archivebox/config.py
+++ b/archivebox/config.py
@@ -177,6 +177,7 @@ CONFIG_SCHEMA: Dict[str, ConfigDefaultDict] = {
                                                                 '--compressed'
                                                                ]},
         'GIT_ARGS':                 {'type': list,  'default': ['--recursive']},
+        'SINGLEFILE_ARGS':          {'type': list,  'default' : None}
     },
 
     'SEARCH_BACKEND_CONFIG' : {
@@ -391,6 +392,7 @@ DYNAMIC_CONFIG_SCHEMA: ConfigDefaultDict = {
 
     'USE_SINGLEFILE':           {'default': lambda c: c['USE_SINGLEFILE'] and c['SAVE_SINGLEFILE']},
     'SINGLEFILE_VERSION':       {'default': lambda c: bin_version(c['SINGLEFILE_BINARY']) if c['USE_SINGLEFILE'] else None},
+    'SINGLEFILE_ARGS':          {'default': lambda c: c['SINGLEFILE_ARGS'] or []},
 
     'USE_READABILITY':          {'default': lambda c: c['USE_READABILITY'] and c['SAVE_READABILITY']},
     'READABILITY_VERSION':      {'default': lambda c: bin_version(c['READABILITY_BINARY']) if c['USE_READABILITY'] else None},
diff --git a/archivebox/extractors/singlefile.py b/archivebox/extractors/singlefile.py
index 3279960e..80ad90b1 100644
--- a/archivebox/extractors/singlefile.py
+++ b/archivebox/extractors/singlefile.py
@@ -17,6 +17,7 @@ from ..config import (
     SAVE_SINGLEFILE,
     DEPENDENCIES,
     SINGLEFILE_VERSION,
+    SINGLEFILE_ARGS,
     CHROME_BINARY,
 )
 from ..logging_util import TimedProgress
@@ -47,6 +48,7 @@ def save_singlefile(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEO
     browser_args = '--browser-args={}'.format(json.dumps(browser_args[1:]))
     cmd = [
         DEPENDENCIES['SINGLEFILE_BINARY']['path'],
+        *SINGLEFILE_ARGS,
         '--browser-executable-path={}'.format(CHROME_BINARY),
         browser_args,
         link.url,

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 2/3] 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,
     ]

From 0ea955b3edbb9d8fdf60f40b448b7653ff20ada3 Mon Sep 17 00:00:00 2001
From: renaisun <43090234+renaisun@users.noreply.github.com>
Date: Mon, 12 Sep 2022 08:54:15 +0800
Subject: [PATCH 3/3] add a missing comma

---
 archivebox/extractors/singlefile.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/archivebox/extractors/singlefile.py b/archivebox/extractors/singlefile.py
index f29d59c3..f7b1b686 100644
--- a/archivebox/extractors/singlefile.py
+++ b/archivebox/extractors/singlefile.py
@@ -70,7 +70,7 @@ def save_singlefile(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEO
 
     cmd = [
         DEPENDENCIES['SINGLEFILE_BINARY']['path'],
-        *deduped_options
+        *deduped_options,
         link.url,
         output,
     ]