Issue 136 - Add "optimap" and timestamp to the download files names

Dockerfile

@@ -2,8 +2,8 @@ ARG UBUNTU_VERSION=22.04
 
 FROM ubuntu:${UBUNTU_VERSION}
 
-ENV PYTHONDONTWRITEBYTECODE 1
-ENV PYTHONUNBUFFERED 1
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
 
 ENV OPTIMAP_DEBUG=False
 ENV OPTIMAP_ALLOWED_HOST=*

optimap/.env.example

@@ -24,3 +24,5 @@ OPTIMAP_EMAIL_USE_TLS=True
 OPTIMAP_EMAIL_IMAP_SENT_FOLDER=""
 
 OPTIMAP_LOGGING_CONSOLE_LEVEL=INFO
+
+OPTIMAP_DATA_DUMP_RETENTION=3

optimap/settings.py

@@ -45,6 +45,8 @@
 
 ROOT_URLCONF = 'optimap.urls'
 
+DATA_DUMP_RETENTION = int(os.getenv("OPTIMAP_DATA_DUMP_RETENTION", 3))
+
 AUTHENTICATION_BACKENDS = [
     'django.contrib.auth.backends.ModelBackend',
     "sesame.backends.ModelBackend",

publications/tasks.py

@@ -28,9 +28,32 @@
 from urllib.parse import quote
 from django_q.tasks import schedule
 from django_q.models import Schedule
+import glob
+from pathlib import Path
+from datetime import datetime, timezone as dt_timezone
 
 BASE_URL = settings.BASE_URL
 DOI_REGEX = re.compile(r'10\.\d{4,9}/[-._;()/:A-Z0-9]+', re.IGNORECASE)
+CACHE_DIR = Path(tempfile.gettempdir()) / 'optimap_cache'
+
+def generate_data_dump_filename(extension: str) -> str:
+    """
+    Returns: optimap_data_dump_YYYYMMDDThhmmss.<extension>
+    """
+    ts = datetime.now(dt_timezone.utc).strftime("%Y%m%dT%H%M%S")
+    return f"optimap_data_dump_{ts}.{extension}"
+
+def cleanup_old_data_dumps(directory: Path, keep: int):
+    """
+    Deletes all files matching optimap_data_dump_* beyond the newest `keep` ones.
+    """
+    pattern = str(directory / "optimap_data_dump_*")
+    files = sorted(glob.glob(pattern), reverse=True)  # newest first
+    for old in files[keep:]:
+        try:
+            os.remove(old)
+        except OSError:
+            logger.warning("Could not delete old dump %s", old)
 
 def extract_geometry_from_html(content):
     for tag in content.find_all("meta"):
@@ -64,8 +87,6 @@ def extract_timeperiod_from_html(content):
     # returning arrays for array field in DB
     return [period[0]], [period[1]]
 
-DOI_REGEX = re.compile(r'10\.\d{4,9}/[-._;()/:A-Z0-9]+', re.IGNORECASE)
-
 def parse_oai_xml_and_save_publications(content, event):
     try:
         DOMTree = xml.dom.minidom.parseString(content)
@@ -324,7 +345,8 @@ def regenerate_geojson_cache():
     cache_dir = os.path.join(tempfile.gettempdir(), "optimap_cache")
     os.makedirs(cache_dir, exist_ok=True)
 
-    json_path = os.path.join(cache_dir, 'geojson_cache.json')
+    json_filename = generate_data_dump_filename("geojson")
+    json_path = os.path.join(cache_dir, json_filename)   
     with open(json_path, 'w') as f:
         serialize(
             'geojson',
@@ -334,31 +356,40 @@ def regenerate_geojson_cache():
             stream=f
         )
 
-    gzip_path = json_path + '.gz'
+    gzip_filename = generate_data_dump_filename("geojson.gz")
+    gzip_path = os.path.join(cache_dir, gzip_filename)  
     with open(json_path, 'rb') as fin, gzip.open(gzip_path, 'wb') as fout:
         fout.writelines(fin)
 
     size = os.path.getsize(json_path)
     logger.info("Cached GeoJSON at %s (%d bytes), gzipped at %s", json_path, size, gzip_path)
+        # remove old dumps beyond retention
+    cleanup_old_data_dumps(Path(cache_dir), settings.DATA_DUMP_RETENTION)
     return json_path
 
 def convert_geojson_to_geopackage(geojson_path):
     cache_dir = os.path.dirname(geojson_path)
-    gpkg_path = os.path.join(cache_dir, 'publications.gpkg')
+    gpkg_filename = generate_data_dump_filename("gpkg")
+    gpkg_path = os.path.join(cache_dir, gpkg_filename)    
     try:
         output = subprocess.check_output(
             ["ogr2ogr", "-f", "GPKG", gpkg_path, geojson_path],
             stderr=subprocess.STDOUT,
             text=True,
         )
         logger.info("ogr2ogr output:\n%s", output)
+            # remove old dumps beyond retention
         return gpkg_path
     except subprocess.CalledProcessError as e:
         return None
         # on success, return the filename so callers can stream it or inspect it
+        # remove old dumps beyond retention
     return gpkg_path
 
 
 def regenerate_geopackage_cache():
     geojson_path = regenerate_geojson_cache()
-    return convert_geojson_to_geopackage(geojson_path)
+    cache_dir = Path(geojson_path).parent
+    gpkg_path = convert_geojson_to_geopackage(geojson_path)
+    cleanup_old_data_dumps(cache_dir, settings.DATA_DUMP_RETENTION)
+    return gpkg_path

publications/templates/data.html

@@ -39,7 +39,7 @@ <h2 class="py-2">OpenAPI UI</h2>
 
     <h2 class="py-2">Download Publication Data</h2>
     <ul class="list-unstyled mb-4">
-      {% if geojson_size %}
+      {% if last_geojson %}
       <li class="mb-3">
         <div class="d-flex align-items-center">
           <a class="btn btn-primary btn-sm" href="{% url 'optimap:download_geojson' %}">
@@ -48,28 +48,32 @@ <h2 class="py-2">Download Publication Data</h2>
           (<a href="https://geojson.org/" target="_blank" class="ms-2 small">GeoJSON spec</a>)
         </div>
         <div class="small text-muted mt-1">
-          File size: {{ geojson_size }}
+          File: {{ last_geojson }}{% if geojson_size %} &middot; Size: {{ geojson_size }}{% endif %}
         </div>
       </li>
       {% endif %}
 
-      {% if geopackage_size %}
+      {% if last_gpkg %}
       <li>
         <div class="d-flex align-items-center">
           <a class="btn btn-primary btn-sm" href="{% url 'optimap:download_geopackage' %}">
             Download GeoPackage
-         </a>
+          </a>
           (<a href="https://www.geopackage.org/" target="_blank" class="ms-2 small">GeoPackage spec</a>)
         </div>
         <div class="small text-muted mt-1">
-         File size: {{ geopackage_size }}
+          File: {{ last_gpkg }}{% if geopackage_size %} &middot; Size: {{ geopackage_size }}{% endif %}
         </div>
       </li>
       {% endif %}
     </ul>
     <p class="small text-muted text-center mb-0">
       Data dumps run every {{ interval }} hour{{ interval|pluralize }}.<br>
-      Last updated: {{ last_updated|naturaltime }}
+      {% if last_updated %}
+        Last updated: {{ last_updated|naturaltime }}
+      {% else %}
+        No dumps have been generated yet.
+      {% endif %}
     </p>
 
   </div>

publications/views.py

@@ -4,7 +4,7 @@
 from django.contrib.auth import login, logout
 from django.shortcuts import render, redirect, get_object_or_404
 from django.core.cache import cache
-from django.http import HttpResponseRedirect, HttpResponse, FileResponse
+from django.http import HttpResponseRedirect, HttpResponse, FileResponse, Http404
 from django.contrib.auth.decorators import login_required
 from django.views.decorators.http import require_GET
 from django.core.mail import EmailMessage, send_mail, get_connection
@@ -26,7 +26,8 @@
 from publications.models import BlockedEmail, BlockedDomain, Subscription, UserProfile, Publication
 from django.contrib.auth import get_user_model
 User = get_user_model()
-import tempfile, os
+import tempfile, os, glob
+from pathlib import Path
 from publications.tasks import regenerate_geojson_cache, regenerate_geopackage_cache
 from osgeo import ogr, osr
 ogr.UseExceptions()
@@ -45,32 +46,34 @@
 @require_GET
 def download_geojson(request):
     """
-    Returns the cached GeoJSON file, gzipped if the client accepts it.
+    Returns the latest GeoJSON dump file, gzipped if the client accepts it.
     """
-    cache_dir = os.path.join(tempfile.gettempdir(), "optimap_cache")
-    os.makedirs(cache_dir, exist_ok=True)
-    json_path = os.path.join(cache_dir, 'geojson_cache.json')
-    gzip_path = os.path.join(cache_dir, 'geojson_cache.json.gz')
-    if not os.path.exists(json_path):
-        json_path = regenerate_geojson_cache()
+    cache_dir = Path(tempfile.gettempdir()) / "optimap_cache"
+    cache_dir.mkdir(exist_ok=True)
+
+    # regenerate and find latest geojson dump
+    path = regenerate_geojson_cache()
+    gzip_path = Path(str(path) + ".gz")
     accept_enc = request.META.get('HTTP_ACCEPT_ENCODING', '')
-    if 'gzip' in accept_enc and os.path.exists(gzip_path):
+
+    if 'gzip' in accept_enc and gzip_path.exists():
         response = FileResponse(
             open(gzip_path, 'rb'),
             content_type="application/json",
             as_attachment=True,
-            filename="publications.geojson"
+            filename=gzip_path.name
         )
         response['Content-Encoding'] = 'gzip'
     else:
         response = FileResponse(
-            open(json_path, 'rb'),
+            open(path, 'rb'),
             content_type="application/json",
             as_attachment=True,
-            filename="publications.geojson"
+            filename=Path(path).name
         )
     return response
 
+
 def generate_geopackage():
     cache_dir = os.path.join(tempfile.gettempdir(), "optimap_cache")
     os.makedirs(cache_dir, exist_ok=True)
@@ -97,7 +100,7 @@ def generate_geopackage():
         feat.SetField("doi", pub.doi or "")
         feat.SetField("source", pub.source or "")
         if pub.geometry:
-            wkb = pub.geometry.wkb  # bytes
+            wkb = pub.geometry.wkb
             geom = ogr.CreateGeometryFromWkb(wkb)
             geom.AssignSpatialReference(srs)
             feat.SetGeometry(geom)
@@ -110,16 +113,16 @@ def generate_geopackage():
 @require_GET
 def download_geopackage(request):
     """
-    Returns the generated GeoPackage file as a downloadable file.
+    Returns the latest GeoPackage dump file.
     """
-    filename = generate_geopackage()
-    if not os.path.exists(filename):
-        return HttpResponse("Error generating GeoPackage.", status=500)
+    path = regenerate_geopackage_cache()
+    if not os.path.exists(path):
+        raise Http404('GeoPackage dump not found')
     return FileResponse(
-        open(filename, 'rb'),
+        open(path, 'rb'),
         content_type="application/geopackage+sqlite3",
         as_attachment=True,
-        filename="publications.gpkg"
+        filename=Path(path).name
     )
 
 
@@ -185,37 +188,54 @@ def loginres(request):
 def privacy(request):
     return render(request, 'privacy.html')
 
-
 @never_cache
 def data(request):
-    cache_dir = os.path.join(tempfile.gettempdir(), "optimap_cache")
-    json_path = os.path.join(cache_dir, "geojson_cache.json")
-    gpkg_path = os.path.join(cache_dir, "publications.gpkg")
-
-    # If dumps don’t exist yet, trigger one synchronously
-    if not os.path.exists(json_path) or not os.path.exists(gpkg_path):
-        regenerate_geopackage_cache()
-
-    if os.path.exists(json_path):
-        geojson_size = humanize.naturalsize(os.path.getsize(json_path), binary=True)
-    else:
-        geojson_size = None
-
-    if os.path.exists(gpkg_path):
-        geopackage_size = humanize.naturalsize(os.path.getsize(gpkg_path), binary=True)
+    """
+    Renders the data page showing links and sizes for the latest dumps.
+    """
+    cache_dir = Path(tempfile.gettempdir()) / "optimap_cache"
+    cache_dir.mkdir(exist_ok=True)
+
+    # scan for existing dumps
+    geojson_files = sorted(cache_dir.glob('optimap_data_dump_*.geojson'), reverse=True)
+    gpkg_files    = sorted(cache_dir.glob('optimap_data_dump_*.gpkg'),   reverse=True)
+
+    last_geo  = geojson_files[0] if geojson_files else None
+    last_gzip = Path(str(last_geo) + ".gz") if last_geo else None
+    last_gpkg = gpkg_files[0]    if gpkg_files    else None
+
+    # — Supervisor check: ensure all dump file times are within 1 hour
+    mtimes = []
+    for p in (last_geo, last_gzip, last_gpkg):
+        if p and p.exists():
+            mtimes.append(p.stat().st_mtime)
+    if mtimes and (max(mtimes) - min(mtimes) > 3600):
+        ts_map = {
+            p.name: datetime.fromtimestamp(p.stat().st_mtime, get_default_timezone())
+            for p in (last_geo, last_gzip, last_gpkg) if p and p.exists()
+        }
+        logger.warning("Data dump timestamps differ by >1h: %s", ts_map)
+
+    # humanized sizes
+    geojson_size    = humanize.naturalsize(last_geo.stat().st_size, binary=True) if last_geo else None
+    geopackage_size = humanize.naturalsize(last_gpkg.stat().st_size, binary=True) if last_gpkg else None
+
+    # last updated timestamp (using JSON file)
+    if last_geo:
+        ts = last_geo.stat().st_mtime
+        last_updated = datetime.fromtimestamp(ts, get_default_timezone())
     else:
-        geopackage_size = None
-    ts = os.path.getmtime(json_path)
-    tz = get_default_timezone()
-    last_updated = datetime.fromtimestamp(ts, tz)
+        last_updated = None
 
     return render(request, 'data.html', {
-        'geojson_size': geojson_size,
+        'geojson_size':    geojson_size,
         'geopackage_size': geopackage_size,
-        'interval': settings.DATA_DUMP_INTERVAL_HOURS,
-        'last_updated': last_updated,
+        'interval':        settings.DATA_DUMP_INTERVAL_HOURS,
+        'last_updated':    last_updated,
+        'last_geojson':    last_geo.name  if last_geo else None,
+        'last_gpkg':       last_gpkg.name if last_gpkg else None,
     })
-    
+
 def Confirmationlogin(request):
     return render(request, 'confirmation_login.html')