remove Seed model in favor of Crawl as template

archivebox/misc/jsonl.py

@@ -27,10 +27,9 @@ TYPE_SNAPSHOT = 'Snapshot'
 TYPE_ARCHIVERESULT = 'ArchiveResult'
 TYPE_TAG = 'Tag'
 TYPE_CRAWL = 'Crawl'
-TYPE_SEED = 'Seed'
 TYPE_INSTALLEDBINARY = 'InstalledBinary'
 
-VALID_TYPES = {TYPE_SNAPSHOT, TYPE_ARCHIVERESULT, TYPE_TAG, TYPE_CRAWL, TYPE_SEED, TYPE_INSTALLEDBINARY}
+VALID_TYPES = {TYPE_SNAPSHOT, TYPE_ARCHIVERESULT, TYPE_TAG, TYPE_CRAWL, TYPE_INSTALLEDBINARY}
 
 
 def parse_line(line: str) -> Optional[Dict[str, Any]]:
@@ -206,7 +205,8 @@ def crawl_to_jsonl(crawl) -> Dict[str, Any]:
     return {
         'type': TYPE_CRAWL,
         'id': str(crawl.id),
-        'seed_id': str(crawl.seed_id),
+        'urls': crawl.urls,
+        'extractor': crawl.extractor,
         'status': crawl.status,
         'max_depth': crawl.max_depth,
         'created_at': crawl.created_at.isoformat() if crawl.created_at else None,

archivebox/misc/logging.py

@@ -13,9 +13,11 @@ from rich.console import Console
 from rich.highlighter import Highlighter
 
 # SETUP RICH CONSOLE / TTY detection / COLOR / PROGRESS BARS
-CONSOLE = Console()
-STDERR = Console(stderr=True)
-IS_TTY = CONSOLE.is_interactive
+# Disable wrapping - use soft_wrap=True and large width so text flows naturally
+# Colors are preserved, just no hard line breaks inserted
+CONSOLE = Console(width=32768, soft_wrap=True, force_terminal=True)
+STDERR = Console(stderr=True, width=32768, soft_wrap=True, force_terminal=True)
+IS_TTY = sys.stdout.isatty()
 
 class RainbowHighlighter(Highlighter):
     def highlight(self, text):

archivebox/misc/logging_util.py

@@ -603,21 +603,17 @@ def log_worker_event(
 
     # Build final message
     error_str = f' {type(error).__name__}: {error}' if error else ''
-    # Build colored message - worker_label needs to be inside color tags
-    # But first we need to format the color tags separately from the worker label
     from archivebox.misc.logging import CONSOLE
     from rich.text import Text
 
     # Create a Rich Text object for proper formatting
     text = Text()
-    text.append(indent)  # Indentation
-    # Append worker label and event with color
+    text.append(indent)
     text.append(f'{worker_label} {event}{error_str}', style=color)
-    # Append metadata without color (add separator if metadata exists)
     if metadata_str:
         text.append(f' | {metadata_str}')
 
-    CONSOLE.print(text)
+    CONSOLE.print(text, soft_wrap=True)
 
 
 @enforce_types

archivebox/misc/monkey_patches.py

@@ -1,7 +1,5 @@
 __package__ = 'archivebox'
 
-import sys
-import shutil
 import django
 import pydantic
 
@@ -20,14 +18,10 @@ timezone.utc = datetime.timezone.utc
 # DjangoSignalWebhooksConfig.verbose_name = 'API'
 
 
-# Install rich for pretty tracebacks in console logs
-# https://rich.readthedocs.io/en/stable/traceback.html#traceback-handler
-
-from rich.traceback import install      # noqa
-
-TERM_WIDTH = (shutil.get_terminal_size((200, 10)).columns - 1) if sys.stdout.isatty() else 200
-# os.environ.setdefault('COLUMNS', str(TERM_WIDTH))
-install(show_locals=True, word_wrap=False, locals_max_length=10, locals_hide_dunder=True, suppress=[django, pydantic], extra_lines=2, width=TERM_WIDTH)
+# Rich traceback handler disabled - it adds frames/boxes that wrap weirdly in log files
+# Standard Python tracebacks are used instead (full width, no frames)
+# from rich.traceback import install
+# install(show_locals=True, word_wrap=False, ...)
 
 
 # Hide site-packages/sonic/client.py:115: SyntaxWarning