From f3a3d76439ccfb822197d086c94520686756aa5a Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Sat, 27 Mar 2021 04:30:15 -0400
Subject: [PATCH] support adding urls with tags directly via CLI and add page

---
 archivebox/core/forms.py            |  1 +
 archivebox/core/models.py           |  6 +++++-
 archivebox/core/views.py            |  2 ++
 archivebox/main.py                  | 26 +++++++++++++++-----------
 archivebox/templates/core/add.html  |  2 +-
 archivebox/templates/static/add.css |  2 +-
 6 files changed, 25 insertions(+), 14 deletions(-)

diff --git a/archivebox/core/forms.py b/archivebox/core/forms.py
index 5521ff1d..e3e904df 100644
--- a/archivebox/core/forms.py
+++ b/archivebox/core/forms.py
@@ -20,6 +20,7 @@ ARCHIVE_METHODS = [
 
 class AddLinkForm(forms.Form):
     url = forms.RegexField(label="URLs (one per line)", regex=URL_REGEX, min_length='6', strip=True, widget=forms.Textarea, required=True)
+    tag = forms.CharField(label="Tags (comma separated tag1,tag2,tag3)", strip=True, required=False)
     depth = forms.ChoiceField(label="Archive depth", choices=CHOICES, initial='0', widget=forms.RadioSelect(attrs={"class": "depth-selection"}))
     archive_methods = forms.MultipleChoiceField(
         label="Archive methods (select at least 1, otherwise all will be used by default)",
diff --git a/archivebox/core/models.py b/archivebox/core/models.py
index d3495826..e7741920 100644
--- a/archivebox/core/models.py
+++ b/archivebox/core/models.py
@@ -116,9 +116,13 @@ class Snapshot(models.Model):
         from ..index import load_link_details
         return load_link_details(self.as_link())
 
-    def tags_str(self) -> str:
+    def tags_str(self, nocache=True) -> str:
         cache_key = f'{self.id}-{(self.updated or self.added).timestamp()}-tags'
         calc_tags_str = lambda: ','.join(self.tags.order_by('name').values_list('name', flat=True))
+        if nocache:
+            tags_str = calc_tags_str()
+            cache.set(cache_key, tags_str)
+            return tags_str
         return cache.get_or_set(cache_key, calc_tags_str)
 
     @cached_property
diff --git a/archivebox/core/views.py b/archivebox/core/views.py
index 801676b3..36794a8d 100644
--- a/archivebox/core/views.py
+++ b/archivebox/core/views.py
@@ -267,10 +267,12 @@ class AddView(UserPassesTestMixin, FormView):
     def form_valid(self, form):
         url = form.cleaned_data["url"]
         print(f'[+] Adding URL: {url}')
+        tag = form.cleaned_data["tag"]
         depth = 0 if form.cleaned_data["depth"] == "0" else 1
         extractors = ','.join(form.cleaned_data["archive_methods"])
         input_kwargs = {
             "urls": url,
+            "tag": tag,
             "depth": depth,
             "update_all": False,
             "out_dir": OUTPUT_DIR,
diff --git a/archivebox/main.py b/archivebox/main.py
index 3b52a179..884f5b46 100644
--- a/archivebox/main.py
+++ b/archivebox/main.py
@@ -611,17 +611,6 @@ def add(urls: Union[str, List[str]],
     write_main_index(links=new_links, out_dir=out_dir)
     all_links = load_main_index(out_dir=out_dir)
 
-    # add any tags to imported links
-    tags = [
-        Tag.objects.get_or_create(name=name.strip())
-        for name in tag.split(',')
-        if name.strip()
-    ]
-    if tags:
-        for link in imported_links:
-            link.as_snapshot().tags.add(*tags)
-
-    
     if index_only:
         # mock archive all the links using the fake index_only extractor method in order to update their state
         if overwrite:
@@ -644,6 +633,21 @@ def add(urls: Union[str, List[str]],
             archive_links(new_links, overwrite=False, **archive_kwargs)
 
 
+    # add any tags to imported links
+    tags = [
+        Tag.objects.get_or_create(name=name.strip())[0]
+        for name in tag.split(',')
+        if name.strip()
+    ]
+    if tags:
+        for link in imported_links:
+            snapshot = link.as_snapshot()
+            snapshot.tags.add(*tags)
+            tags_str = snapshot.tags_str(nocache=True)
+            snapshot.save()
+        # print(f'    √ Tagged {len(imported_links)} Snapshots with {len(tags)} tags {tags_str}')
+
+
     return all_links
 
 @enforce_types
diff --git a/archivebox/templates/core/add.html b/archivebox/templates/core/add.html
index 0f161885..8123ef7a 100644
--- a/archivebox/templates/core/add.html
+++ b/archivebox/templates/core/add.html
@@ -15,7 +15,7 @@
 {% endblock %}
 
 {% block body %}
-    <div style="max-width: 550px; margin: auto; float: none">
+    <div style="max-width: 1440px; margin: auto; float: none">
         <br/><br/>
         {% if stdout %}
             <h1>Add new URLs to your archive: results</h1>
diff --git a/archivebox/templates/static/add.css b/archivebox/templates/static/add.css
index 875c61bc..21645e70 100644
--- a/archivebox/templates/static/add.css
+++ b/archivebox/templates/static/add.css
@@ -42,7 +42,7 @@ header {
   background-color: #f5dd5d;
 }
 #stdout {
-  background-color: #ded;
+  background-color: #fbfbfb;
   padding: 10px 10px;
   border-radius: 4px;
   white-space: normal;