JSON exports

.gitignore

@@ -8,3 +8,4 @@ whotracksme.egg-info/
 .DS_Store
 venv/
 whotracksme/data/assets/**/*.csv
+whotracksme.db

.tool-versions

@@ -1 +1 @@
-python 3.10.6
+python 3.11.6

Dockerfile

@@ -1,5 +1,5 @@
 # Set base image to build upon
-FROM python:3.10-slim
+FROM python:3.11-slim
 
 # Set arg and env
 ARG VERSION

README.md

@@ -45,7 +45,7 @@ Python 3.11 is needed to build the site. We recommend creating a
 [virtualenv](http://docs.python-guide.org/en/latest/dev/virtualenvs/)
 to install the dependencies, or use `pipenv` or )
 
-to 
+to
 
 ```sh
 python -m venv venv
@@ -54,6 +54,13 @@ python -m venv venv
 
 After the initial setup, you can proceed with installing whotracks.me.
 
+For nushell:
+
+```nushell
+python -m virtualenv venv
+overlay use venv/bin/activate.nu
+```
+
 ## With Pip
 
 ```sh

deploy_to_s3.py

@@ -142,6 +142,7 @@ def upload_file_to_s3(path, filename):
                         f'{site_dir}/websites/',
                         f'{site_dir}/companies/',
                         f'{site_dir}/data/',
+                        f'{site_dir}/api/',
                     ]
                     for directory in directories_to_skip:
                         if local_path.startswith(directory):

requirements.txt

@@ -10,3 +10,4 @@ requests==2.31.0
 six==1.15.0
 urllib3==1.26.18
 wheel==0.38.1
+simplejson==3.19.2

setup.py

@@ -29,7 +29,7 @@
 # This allows us to inject a custom version into setuptools using the
 # environment variable: WTM_VERSION. If it is not specified, a default value is
 # used, but you will not be able to run sdist in this case.
-DEFAULT_VERSION = "dev"
+DEFAULT_VERSION = "0.dev"
 VERSION = os.environ.get("WTM_VERSION", DEFAULT_VERSION)
 
 

whotracksme/website/build/companies.py

@@ -2,7 +2,7 @@
 from collections import defaultdict
 from jinja2 import Markup
 
-from whotracksme.website.utils import print_progress
+from whotracksme.website.utils import print_progress, write_json
 from whotracksme.website.templates import (
     get_template,
     render_template,
@@ -96,6 +96,10 @@ def company_page(template, company_data, data):
     company_id = company_data['overview']['id']
 
     company_name = get_company_name(company_data)
+    write_json(f'_site/api/v2/organizations/{data.url_for("company", company_id)}.json',
+        demographics=company_data,
+        initials=company_name[:2]
+    )
     with open(f'_site/{data.url_for("company", company_id)}', 'w') as output:
         output.write(render_template(
             path_to_root='..',
@@ -118,7 +122,10 @@ def build_company_reach_chart_page(data):
     top100 = company_reach(data.companies, n=100)
     chart = Markup(overview_bars(top100, highlight=10, height=3000))
     template = get_template(data, name='reach-chart-page.html', path_to_root='..')
-
+    write_json('_site/api/v2/organizations.json',
+        top100=top100,
+        organizations=data.companies.sort_by('name', descending=False)
+    )
     with open('_site/companies/reach-chart.html', 'w') as output:
         output.write(render_template(
             path_to_root='..',

whotracksme/website/build/home.py

@@ -2,7 +2,7 @@
 
 from whotracksme.website.plotting.companies import overview_bars
 from whotracksme.website.build.companies import company_reach
-from whotracksme.website.utils import print_progress
+from whotracksme.website.utils import print_progress, write_json
 from whotracksme.website.templates import get_template, render_template
 from whotracksme.website.build.blog import load_blog_posts
 
@@ -13,16 +13,33 @@ def build_home(data):
 
     posts = load_blog_posts()[:3]
 
+    tracker_list = data.trackers.sort_by(metric="reach")[:20]
+    trackers_list_company = data.trackers.sort_by(metric="company_id")[:20]
+    most_tracked_sites = data.sites.sort_by(metric='trackers')[:20]
+    least_tracked_sites = data.sites.sort_by(metric='trackers', descending=False)[:20]
+    websites = data.sites.summary_stats()
+    tracker_stats = data.trackers.summary_stats()
+
+    write_json('_site/api/v2/index.json',
+        tracker_list=tracker_list,
+        trackers_list_company=trackers_list_company,
+        most_tracked_sites=most_tracked_sites,
+        least_tracked_sites=least_tracked_sites,
+        websites=websites,
+        tracker_stats=tracker_stats,
+        top10=top10,
+    )
+
     with open('_site/index.html', 'w') as output:
         output.write(render_template(
             template=get_template(data, "index.html"),
             ts=header_graph,
-            tracker_list=data.trackers.sort_by(metric="reach")[:20],
-            trackers_list_company=data.trackers.sort_by(metric="company_id")[:20],
-            most_tracked_sites=data.sites.sort_by(metric='trackers')[:20],
-            least_tracked_sites=data.sites.sort_by(metric='trackers', descending=False)[:20],
-            websites=data.sites.summary_stats(),
-            tracker_stats=data.trackers.summary_stats(),
+            tracker_list=tracker_list,
+            trackers_list_company=trackers_list_company,
+            most_tracked_sites=most_tracked_sites,
+            least_tracked_sites=least_tracked_sites,
+            websites=websites,
+            tracker_stats=tracker_stats,
             top10=top10,
             posts=posts
         ))

whotracksme/website/build/trackers.py

@@ -2,7 +2,7 @@
 from jinja2 import Markup
 
 from whotracksme.data.loader import DataSource
-from whotracksme.website.utils import print_progress
+from whotracksme.website.utils import print_progress, write_json, without_keys
 from whotracksme.website.templates import (
     get_template,
     render_template,
@@ -71,17 +71,24 @@ def site_summary(row):
 
 
 def build_trackers_list(data):
+    tracker_list = data.trackers.sort_by(metric="reach")
+    tracker_list_company = data.trackers.sort_by(
+        metric="company_id",
+        descending=False
+    )
+    header_stats = data.trackers.summary_stats()
     with open('_site/trackers.html', 'w') as output:
         output.write(render_template(
             template=get_template(data, name="trackers.html"),
-            tracker_list=data.trackers.sort_by(metric="reach"),
-            trackers_list_company=data.trackers.sort_by(
-                metric="company_id",
-                descending=False
-            ),
-            header_stats=data.trackers.summary_stats()
+            tracker_list=tracker_list,
+            trackers_list_company=tracker_list_company,
+            header_stats=header_stats
         ))
-
+    write_json('_site/api/v2/trackers.json',
+        tracker_list=tracker_list,
+        trackers_list_company=tracker_list_company,
+        header_stats=header_stats
+    )
     print_progress(text="Generate tracker list")
 
 
@@ -143,4 +150,5 @@ def build_tracker_page_batch(batch):
             page_data = tracker_page_data(tracker_id,
                                         data.trackers.get_datapoint(tracker_id),
                                         data)
+            write_json(f'_site/api/v2/trackers/{tracker_id}.json', **without_keys(page_data, "app"))
             tracker_page(template, page_data)

whotracksme/website/build/websites.py

@@ -6,7 +6,7 @@
 from jinja2 import Markup
 
 from whotracksme.data.loader import DataSource
-from whotracksme.website.utils import print_progress
+from whotracksme.website.utils import print_progress, write_json
 from whotracksme.website.build.companies import (
     tracker_map_data,
     website_doughnout,
@@ -25,6 +25,12 @@ def build_website_list(data):
     sorted_websites = data.sites.sort_by(metric='popularity', descending=True)
     sorted_websites_cat = data.sites.sort_by(metric='category', descending=True)
 
+    write_json('_site/api/v2/websites.json',
+        website_list=sorted_websites,
+        website_list_cat=sorted_websites_cat,
+        header_numbers=header_numbers
+    )
+
     with open('_site/websites.html', 'w') as output:
         output.write(render_template(
             template=get_template(data, "websites.html"),
@@ -61,6 +67,22 @@ def website_page(template, site, rank, data):
     # apps per site data
     tracker_table = list(data.sites.get_tracker_list(site_id))
 
+    write_json('_site/api/v2/websites/{}.json'.format(site.site),
+        site={
+            'overview': site._asdict()
+        },
+        profile=profile,
+        methods=methods,
+        sankey_data=sankey_data,
+        doughnout={
+            "d_values": d_values,
+            "d_labels": d_labels,
+            "d_total": d_total,
+        },
+        tracker_categories=d_labels,
+        tracker_list=tracker_table,
+    )
+
     with open('_site/websites/{}.html'.format(site.site), 'w') as output:
         output.write(render_template(
             path_to_root='..',

whotracksme/website/utils.py

@@ -1,5 +1,20 @@
+import datetime
+import simplejson
+import pathlib
+import os
 
-def print_progress(text, default_space=40):
-    print("{} {:{}} done".format(text, "." * (default_space - len(text)), default_space - len(text)))
+def write_json(path, **data):
+    def myconverter(o):
+        if isinstance(o, datetime.datetime):
+            return o.isoformat()
+
+    pathlib.Path(os.path.dirname(path)).mkdir(parents=True, exist_ok=True)
+    json = simplejson.dumps(data, indent=2, default = myconverter)
+    with open(path, "w") as file:
+        file.write(json)
 
+def without_keys(d, keys):
+    return {k: d[k] for k in d.keys() - keys}
 
+def print_progress(text, default_space=40):
+    print("{} {:{}} done".format(text, "." * (default_space - len(text)), default_space - len(text)))