Improve bookmarklet and backend to upload bookmarks with direct html (#…

…287)

archivy/models.py

@@ -89,7 +89,7 @@ class DataObj:
     fullpath: Optional[str] = attrib(validator=optional(instance_of(str)), default=None)
     error: Optional[str] = attrib(validator=optional(instance_of(str)), default=None)
 
-    def process_bookmark_url(self):
+    def process_bookmark_url(self, raw_html=None):
         """Process url to get content for bookmark"""
         if self.type not in ("bookmark", "pocket_bookmark") or not validators.url(
             self.url
@@ -107,17 +107,20 @@ def process_bookmark_url(self):
                 return
 
         try:
-            url_request = requests.get(
-                self.url,
-                headers={"User-agent": f"Archivy/v{require('archivy')[0].version}"},
+            page_html = (
+                raw_html
+                or requests.get(
+                    self.url,
+                    headers={"User-agent": f"Archivy/v{require('archivy')[0].version}"},
+                ).text
             )
         except Exception:
             self.error = f"Could not retrieve {self.url}\n"
             self.wipe()
             return
 
         try:
-            document = Document(url_request.text)
+            document = Document(page_html)
             self.title = document.short_title() or self.url
             parsed_html = BeautifulSoup(document.summary(), features="html.parser")
         except Exception:

archivy/routes.py

@@ -18,7 +18,7 @@
 from werkzeug.security import check_password_hash, generate_password_hash
 
 from archivy.models import DataObj, User
-from archivy import data, app, forms
+from archivy import data, app, forms, csrf
 from archivy.helpers import get_db, write_config
 from archivy.tags import get_all_tags
 from archivy.search import search, search_frontmatter_tags
@@ -396,3 +396,24 @@ def update_config_value(key, val, dictionary):
     return render_template(
         "config.html", conf=form, default=default, title="Edit Config"
     )
+
+
+@csrf.exempt  # exempt from CSRF to be able to submit info directly from bookmarklet
+@app.route("/save_from_bookmarklet", methods=["POST"])
+def save_raw_url():
+    """
+    Used in the bookmarklet - Saves a URL by taking its raw HTML.
+
+    POST parameters:
+    - html
+    - url
+    """
+    html = request.form.get("html")
+    if not html:
+        return "No HTML provided", 400
+    bookmark = DataObj(url=request.form.get("url"), type="bookmark")
+    bookmark.process_bookmark_url(html)
+    if bookmark.insert():
+        return redirect(f"/dataobj/{bookmark.id}")
+    else:
+        return "Could not save bookmark", 500

archivy/templates/bookmarklet.html

@@ -7,6 +7,18 @@ <h2>Bookmarklet</h2>
 
 All you need to do is drage the link below to your browser toolbar. Then, when you find an interesting link, just click the button on your toolbar named "Add to archivy".
 
-<h3><a href="javascript:window.location = `{{ request.host_url | safe  }}bookmarks/new?url=${window.location}`;">Add to {{ config.SITE_TITLE }}</a></h3>
+<h3><a href="javascript: (async function() {
+  document.body.insertAdjacentHTML('beforeend', `
+    <form style='display: none' method='post' target='_blank' id='archivy_bookmarklet' action='{{ request.host_url | safe }}/save_from_bookmarklet'>
+      <input type='hidden' name='url'>
+      <input type='text' name='html'>
+    </form>
+  `);
+  const form = document.getElementById('archivy_bookmarklet');
+  form.querySelector('input[name=url]').value = window.location.href;
+  form.querySelector('input[name=html]').value = document.documentElement.outerHTML;
+  form.submit();})();">
+	Add to {{ config.SITE_TITLE }}
+</a></h3>
 
 {% endblock %}

tests/functional/test_routes.py

@@ -398,3 +398,18 @@ def test_getting_matches_for_specific_tag(test_app, client, bookmark_fixture):
     assert resp.status_code == 200
     assert bookmark_fixture.title in str(resp.data)
     assert str(bookmark_fixture.id) in str(resp.data)
+
+
+def test_bookmarklet_upload(test_app, client):
+    title = "bookmarklet bookmark"
+    resp = client.post(
+        "/save_from_bookmarklet",
+        data={
+            "url": "https://example.com",
+            "html": f"<html><title>{title}</title><html>",
+        },
+        follow_redirects=True,
+    )
+    assert resp.status_code == 200
+    assert title in str(resp.data)
+    assert "/dataobj/" in request.path

tests/integration/test_api.py

@@ -5,7 +5,7 @@
 from flask import Flask
 from flask.testing import FlaskClient
 from tinydb import Query
-from archivy.data import create_dir, get_items, create_dir
+from archivy.data import create_dir, get_items, create_dir, get_item
 from archivy.models import DataObj
 from archivy.helpers import get_db