From ef865dd76adee21d932bb79e208a911165a5ca0a Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@nicksweeting.com>
Date: Fri, 22 Mar 2019 23:00:43 -0400
Subject: [PATCH] better stats collection and printing

---
 archivebox/archive_methods.py | 23 +++++++++++------------
 archivebox/logs.py            | 32 +++++++++++++-------------------
 2 files changed, 24 insertions(+), 31 deletions(-)

diff --git a/archivebox/archive_methods.py b/archivebox/archive_methods.py
index 9ef0dc83..bf2eec4e 100644
--- a/archivebox/archive_methods.py
+++ b/archivebox/archive_methods.py
@@ -85,29 +85,28 @@ def archive_link(link_dir, link):
 
         link = load_json_link_index(link_dir, link)
         log_link_archiving_started(link_dir, link, is_new)
-        skipped_entirely = True
+        stats = {'skipped': 0, 'succeeded': 0, 'failed': 0}
 
         for method_name, should_run, method_function in ARCHIVE_METHODS:
             if method_name not in link['history']:
                 link['history'][method_name] = []
             
             if should_run(link_dir, link):
-                if skipped_entirely:
-                    skipped_entirely = False
-                    print()
+                log_archive_method_started(method_name)
+
+                result = method_function(link_dir, link)
+                link['history'][method_name].append(result)
+
+                stats[result['status']] += 1
+                log_archive_method_finished(result)
             else:
-                continue
+                stats['skipped'] += 1
 
-            log_archive_method_started(method_name)
-            result = method_function(link_dir, link)
-            log_archive_method_finished(result)
-
-            link['history'][method_name].append(result)
+        # print('    ', stats)
 
         write_link_index(link_dir, link)
         patch_links_index(link)
-
-        log_link_archiving_finished(link_dir, link, is_new, skipped_entirely)
+        log_link_archiving_finished(link_dir, link, is_new, stats)
 
     except Exception as err:
         print('    ! Failed to archive link: {}: {}'.format(err.__class__.__name__, err))
diff --git a/archivebox/logs.py b/archivebox/logs.py
index 3d6a804b..4dc2c051 100644
--- a/archivebox/logs.py
+++ b/archivebox/logs.py
@@ -116,9 +116,9 @@ def log_archiving_finished(num_links):
         duration,
         ANSI['reset'],
     ))
-    print('    - {} entries skipped'.format(_LAST_RUN_STATS['skipped']))
-    print('    - {} entries updated'.format(_LAST_RUN_STATS['succeeded']))
-    print('    - {} errors'.format(_LAST_RUN_STATS['failed']))
+    print('    - {} links skipped'.format(_LAST_RUN_STATS['skipped']))
+    print('    - {} links updated'.format(_LAST_RUN_STATS['succeeded']))
+    print('    - {} links had errors'.format(_LAST_RUN_STATS['failed']))
     print('    To view your archive, open: {}/index.html'.format(OUTPUT_DIR.replace(REPO_DIR + '/', '')))
 
 
@@ -135,26 +135,20 @@ def log_link_archiving_started(link_dir, link, is_new):
         **ANSI,
     ))
     print('    {blue}{url}{reset}'.format(url=link['url'], **ANSI))
-    sys.stdout.write('    > {}{}'.format(
+    print('    {} {}'.format(
+        '>' if is_new else '√',
         pretty_path(link_dir),
-        ' (new)' if is_new else '',
     ))
 
-def log_link_archiving_finished(link_dir, link, is_new, skipped_entirely):
-    from util import latest_output
-    
-    if all(output == 'succeeded' for output in latest_output(link).values()):
-        _LAST_RUN_STATS['succeeded'] += 1
-    elif any(output == 'failed' for output in latest_output(link).values()):
-        _LAST_RUN_STATS['failed'] += 1
-    else:
-        _LAST_RUN_STATS['skipped'] += 1
+def log_link_archiving_finished(link_dir, link, is_new, stats):
+    total = sum(stats.values())
 
-    if skipped_entirely:
-        print('\r    √ {}{}'.format(
-            pretty_path(link_dir),
-            ' (new)' if is_new else '',
-        ))
+    if stats['failed'] > 0 :
+        _LAST_RUN_STATS['failed'] += 1
+    elif stats['skipped'] == total:
+        _LAST_RUN_STATS['skipped'] += 1
+    else:
+        _LAST_RUN_STATS['succeeded'] += 1
 
 
 def log_archive_method_started(method):