better data folder checking on startup

archivebox/cli/archivebox_add.py

@@ -7,6 +7,7 @@ __description__ = 'Add a new URL or list of URLs to your archive'
 import sys
 import argparse
 
+from ..legacy.config import stderr, check_dependencies, check_data_folder
 from ..legacy.util import (
     handle_stdin_import,
     handle_file_import,
@@ -14,7 +15,7 @@ from ..legacy.util import (
 from ..legacy.main import update_archive_data
 
 
-def main(args=None):
+def main(args=None, stdin=None):
     args = sys.argv[1:] if args is None else args
 
     parser = argparse.ArgumentParser(
@@ -53,13 +54,16 @@ def main(args=None):
     )
     command = parser.parse_args(args)
 
+    check_dependencies()
+    check_data_folder()
+
     ### Handle ingesting urls piped in through stdin
     # (.e.g if user does cat example_urls.txt | archivebox add)
     import_path = None
-    if not sys.stdin.isatty():
+    if stdin or not sys.stdin.isatty():
-        stdin_raw_text = sys.stdin.read()
+        stdin_raw_text = stdin or sys.stdin.read()
         if stdin_raw_text and command.url:
-            print(
+            stderr(
                 '[X] You should pass either a path as an argument, '
                 'or pass a list of links via stdin, but not both.\n'
             )

archivebox/cli/archivebox_init.py

@@ -9,59 +9,7 @@ import sys
 import argparse
 
 from ..legacy.util import reject_stdin
-from ..legacy.index import write_links_index
+from ..legacy.main import init
-from ..legacy.config import (
-    OUTPUT_DIR,
-    SOURCES_DIR,
-    ARCHIVE_DIR,
-    DATABASE_DIR,
-    ANSI,
-    stderr,
-)
-
-
-def init(output_dir: str=OUTPUT_DIR):
-    if not os.path.exists(output_dir):
-        os.makedirs(output_dir)
-
-    harmless_files = {'.DS_Store', '.venv', 'venv', 'virtualenv', '.virtualenv'}
-    is_empty = not len(set(os.listdir(output_dir)) - harmless_files)
-    existing_index = os.path.exists(os.path.join(output_dir, 'index.json'))
-
-    if not is_empty:
-        if existing_index:
-            stderr('{green}[√] You already have an archive index in this folder.{reset}'.format(**ANSI))
-            stderr('    To add new links, you can run:')
-            stderr("        archivebox add 'https://example.com'")
-            stderr()
-            stderr('    For more usage and examples, run:')
-            stderr('        archivebox help')
-            # TODO: import old archivebox version's archive data folder
-
-            raise SystemExit(1)
-        else:
-            stderr(
-                ("{red}[X] This folder already has files in it. You must run init inside a completely empty directory.{reset}"
-                "\n\n"
-                "    {lightred}Hint:{reset} To import a data folder created by an older version of ArchiveBox, \n"
-                "    just cd into the folder and run the archivebox command to pick up where you left off.\n\n"
-                "    (Always make sure your data folder is backed up first before updating ArchiveBox)"
-                ).format(output_dir, **ANSI)
-            )
-            raise SystemExit(1)
-
-
-    stderr('{green}[+] Initializing new archive directory: {}{reset}'.format(output_dir, **ANSI))
-    os.makedirs(SOURCES_DIR)
-    stderr(f'    > {SOURCES_DIR}')
-    os.makedirs(ARCHIVE_DIR)
-    stderr(f'    > {ARCHIVE_DIR}')
-    os.makedirs(DATABASE_DIR)
-    stderr(f'    > {DATABASE_DIR}')
-
-    write_links_index([], out_dir=OUTPUT_DIR, finished=True)
-
-    stderr('{green}[√] Done.{reset}'.format(**ANSI))
 
 
 def main(args=None):

archivebox/legacy/__init__.py

@@ -0,0 +1 @@
+__package__ = 'archivebox.legacy'