From a997bff877bade8ebdd97d6a5f220fe24b7f30df Mon Sep 17 00:00:00 2001
From: Neelima J <78197684+neelima-j@users.noreply.github.com>
Date: Mon, 22 Dec 2025 13:44:26 +0000
Subject: [PATCH] - Closes
 https://github.com/OpenDataServices/notebooks-oc4ids/issues/70 - Adds an if
 condition to download logic to build URLs for each year, download zip files,
 convert to JSON, extract package information and combine to one package. -
 Refactoring for similar logic in Uganda will be done in a separate PR

---
 OC4IDS_Database_Data_Import.ipynb | 65 +++++++++++++++++++++++++++++--
 1 file changed, 61 insertions(+), 4 deletions(-)

diff --git a/OC4IDS_Database_Data_Import.ipynb b/OC4IDS_Database_Data_Import.ipynb
index 29119b6..86e0c6e 100644
--- a/OC4IDS_Database_Data_Import.ipynb
+++ b/OC4IDS_Database_Data_Import.ipynb
@@ -243,11 +243,11 @@
         "\n",
         "# @markdown After running this cell manually, it will auto-run when you change the source_id.\n",
         "\n",
-        "source_id = 'ukraine_cost_ukraine' #@param [ 'ghana_cost_sekondi_takoradi', 'indonesia_cost_ntb', 'indonesia_cost_west_lombok', 'malawi_cost_malawi', 'mexico_cost_jalisco', 'mexico_nuevo_leon', 'uganda_gpp', 'ukraine_cost_ukraine']\n",
+        "source_id = 'ecuador_cost_ecuador' #@param [ 'ecuador_cost_ecuador','ghana_cost_sekondi_takoradi', 'indonesia_cost_ntb', 'indonesia_cost_west_lombok', 'malawi_cost_malawi', 'mexico_cost_jalisco', 'mexico_nuevo_leon', 'uganda_gpp', 'ukraine_cost_ukraine']\n",
         "\n",
         "print('Source selected:', source_id)"
       ],
-      "execution_count": 9,
+      "execution_count": null,
       "outputs": [
         {
           "output_type": "stream",
@@ -292,6 +292,7 @@
       },
       "source": [
         "sources = {\n",
+        "  'ecuador_cost_ecuador':'https://costecuador.org/storage/json/json_',\n",
         "  'mexico_cost_jalisco': 'http://www.costjalisco.org.mx/jsonprojects',\n",
         "  'ghana_cost_sekondi_takoradi': 'https://costsekondi-takoradigh.org/uploads/projectJson.json',\n",
         "  'mexico_nuevo_leon': 'http://si.nl.gob.mx/siasi_ws/api/edcapi/DescargarProjectPackage',\n",
@@ -313,6 +314,63 @@
         "  with open('project_package.json', 'wb') as f:\n",
         "    f.write(response.content)\n",
         "\n",
+        "elif source_id == 'ecuador_cost_ecuador':\n",
+        "  packages = []\n",
+        "  start_year = 2020\n",
+        "  current_year = datetime.now().year\n",
+        "  years = list(range(start_year, current_year + 1))\n",
+        "\n",
+        "  for year in years:\n",
+        "        url = f\"{sources[source_id]}{year}.zip\"\n",
+        "        try:\n",
+        "            response = requests.get(url, verify=False, stream=True)\n",
+        "            if response.status_code != 200:\n",
+        "                print(f\"Could not download {url}: Status {response.status_code}\"\n",
+        "                )\n",
+        "                continue\n",
+        "            zip_data = io.BytesIO(response.content)\n",
+        "            with zipfile.ZipFile(zip_data, \"r\") as zip_ref:\n",
+        "                for filename in zip_ref.namelist():\n",
+        "                    if filename.endswith(\".json\"):\n",
+        "                        with zip_ref.open(filename) as f:\n",
+        "                            package = json.load(f)\n",
+        "                            packages.append(package)\n",
+        "        except (zipfile.BadZipFile, requests.RequestException) as e:\n",
+        "            print(f\"Error processing {url}: {e}\")\n",
+        "            continue\n",
+        "\n",
+        "  if not packages:\n",
+        "    raise RuntimeError(f\"No valid packages found at {sources[source_id]}\")\n",
+        "\n",
+        "  # Arguments to combine packages\n",
+        "  combine_args = {}\n",
+        "\n",
+        "  if packages[-1].get(\"uri\"):\n",
+        "        combine_args[\"uri\"] = packages[-1].get(\"uri\")\n",
+        "\n",
+        "  versions = {\n",
+        "        package.get(\"version\") for package in packages if package.get(\"version\")\n",
+        "    }\n",
+        "  if len(versions) > 1:\n",
+        "        print(f\"Packages declare more than one version: {versions}\")\n",
+        "  if versions:\n",
+        "        combine_args[\"version\"] = list(versions)[0]\n",
+        "\n",
+        "  published_dates = {\n",
+        "        package.get(\"publishedDate\")\n",
+        "        for package in packages\n",
+        "        if package.get(\"publishedDate\")\n",
+        "    }\n",
+        "  if published_dates:\n",
+        "        combine_args[\"published_date\"] = max(published_dates)\n",
+        "\n",
+        "  print(f\"Combining {len(packages)} packages for Ecuador\")\n",
+        "  combined_package = combine_project_packages(packages, **combine_args)\n",
+        "\n",
+        "  with open('project_package.json', 'w') as f:\n",
+        "      json.dump(combined_package, f, indent=2)  # Use json.dump() to write the dictionary as JSON\n",
+        "\n",
+        "\n",
         "elif source_id == 'uganda_gpp':\n",
         "    packages = []\n",
         "    start_year = 2022\n",
@@ -687,8 +745,7 @@
   "metadata": {
     "colab": {
       "name": " OC4IDS Database - Data Import",
-      "provenance": [],
-      "toc_visible": true
+      "provenance": []
     },
     "kernelspec": {
       "display_name": "Python 3",