cleaning

datascrapolympedia.ipynb → notebook/datascrapolympedia.ipynb

@@ -348,18 +348,16 @@
         "result_ids_path = os.path.join(dataroot, '_distinct_result_id.csv')\n",
         "csv_save_path = os.path.join(dataroot, 'olympedia_all_results.csv')\n",
         "\n",
-        "# Function to identify if a row is part of the final round based on specific columns\n",
         "def is_final_round(row, headers):\n",
         "    final_columns = [\"Final\", \"Final Round\"]\n",
         "    for col in final_columns:\n",
         "        if col in headers and row.get(col, \"\").strip() and row.get(col).strip() != \"–\":\n",
         "            return True\n",
         "    return False\n",
         "\n",
-        "# Function to parse the event results and filter for final rounds\n",
         "def parse_results_table(table) -> Tuple[List[str], List[Dict[str, str]]]:\n",
         "    headers = [th.get_text().strip() for th in table.find_all('th')]\n",
-        "    rows = table.find_all('tr')[1:]  # skip the header row\n",
+        "    rows = table.find_all('tr')[1:]  \n",
         "    results = []\n",
         "\n",
         "    has_final_column = any(col in headers for col in [\"Final\", \"Final Round\"])\n",
@@ -369,7 +367,6 @@
         "        if cols:\n",
         "            result = {}\n",
         "            for i, col in enumerate(cols):\n",
-        "                # Handle edge cases for cells containing spans or being empty\n",
         "                if i < len(headers):\n",
         "                    text = col.get_text().strip()\n",
         "                    if not text:\n",
@@ -431,7 +428,6 @@
         "        'sport': sport,\n",
         "    }\n",
         "\n",
-        "    # Parsing the event bio table\n",
         "    try:\n",
         "        event_bio_table = result_soup.select('body > div.container > table.biodata')[0]\n",
         "        event_bio_header = [item.get_text() for item in event_bio_table.select('table > tr > th')]\n",
@@ -450,8 +446,6 @@
         "        print(f\"Error parsing event bio table for result ID {result_id}: {e}\")\n",
         "        for key in ['result_date', 'result_location', 'result_participants', 'result_format', 'result_detail']:\n",
         "            event_info[key] = 'na'\n",
-        "\n",
-        "    # Parsing the results table\n",
         "    results_table = result_soup.find('table', class_='table table-striped')\n",
         "    if results_table:\n",
         "        try:\n",
@@ -509,7 +503,6 @@
         "        result_ids = [row[0] for row in reader]\n",
         "    return result_ids\n",
         "\n",
-        "# Run the script\n",
         "if __name__ == \"__main__\":\n",
         "    result_ids = load_result_ids(result_ids_path)\n",
         "    asyncio.run(main(result_ids))\n"

pages/athletics.html

@@ -7,7 +7,7 @@
   <!-- Include D3.js library -->
   <script src="https://d3js.org/d3.v7.min.js"></script>
   <script src="https://cdn.jsdelivr.net/npm/[email protected]/papaparse.min.js"></script>
-  <script src="../scripts/running.js" defer></script>
+  <script src="../scripts/olympics/running.js" defer></script>
   <style>
     body {
       font-family: Arial, sans-serif;

pages/olympics.html

@@ -8,9 +8,9 @@
     <script src="https://d3js.org/d3.v7.min.js"></script>
     <script src="https://cdn.jsdelivr.net/npm/[email protected]/papaparse.min.js"></script>
     <script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
-    <script src="../scripts/medals.js"></script>
+    <script src="../scripts/olympics/medals.js"></script>
     <script src="https://d3js.org/d3.v6.min.js"></script>
-    <script src="../scripts/podium.js"></script>
+    <script src="../scripts/olympics/podium.js"></script>
     <style>
       body {
         display: flex;

pages/podium.html

@@ -6,7 +6,7 @@
   <title>Olympic Podium Visualization</title>
   <script src="https://d3js.org/d3.v6.min.js"></script>
   <script src="https://cdnjs.cloudflare.com/ajax/libs/PapaParse/5.3.0/papaparse.min.js"></script>
-  <script src="../scripts/podium.js"></script>
+  <script src="../scripts/olympics/podium.js"></script>
   <style>
     body {
       display: flex;

pages/swimming.html

@@ -7,7 +7,7 @@
   <!-- Include D3.js library -->
   <script src="https://d3js.org/d3.v7.min.js"></script>
   <script src="https://cdn.jsdelivr.net/npm/[email protected]/papaparse.min.js"></script>
-  <script src="../scripts/swimming.js" defer></script>
+  <script src="../scripts/olympics/swimming.js" defer></script>
   <style>
     body {
       font-family: Arial, sans-serif;

scripts/medals.js → scripts/olympics/medals.js

@@ -1,26 +1,6 @@
- // Sample data format
-/* 
-const data = [
-    { Country: "France", NOC: "FRA", Season: "Summer", Year: "1896", Gold: "0.0", Silver: "1.0", Bronze: "0.0", None: "3.0" },
-    { Country: "USA", NOC: "USA", Season: "Summer", Year: "1896", Gold: "1.0", Silver: "1.0", Bronze: "1.0", None: "3.0" },
-    // Add more data here
-];
- */
 
-let data = [];
 
-// Function to populate year options based on data
-/* function populateYears() {
-    const yearSelect = document.getElementById('yearSelect');
-    const years = [...new Set(data.map(item => item.Year))];
-    console.log("pop data", data);
-    years.forEach(year => {
-        const option = document.createElement('option');
-        option.value = year;
-        option.text = year;
-        yearSelect.appendChild(option);
-    });
-} */
+let data = [];
 
 
 function populateYears() {
@@ -36,18 +16,17 @@ function populateYears() {
 
 
 function updateChart() {
-    //const selectedYear = document.getElementById('yearSelect').value;
+
     const selectedYearSeason = document.getElementById('yearSeasonSelect').value;
     const [selectedYear, selectedSeason] = selectedYearSeason.split(" - ");
-    //const selectedSeason = document.getElementById('seasonSelect').value;
     console.log("chart data", data);
 
-    //let filteredData = data.filter(item => item.Year === selectedYear && item.Season === selectedSeason);
+
     let filteredData = data.filter(item => String(item.Year) === selectedYear && item.Season === selectedSeason);
 
     console.log(filteredData);
 
-    // Aggregate medals by country
+
     const medalCounts = filteredData.reduce((acc, item) => {
         const country = item.Country;
         acc[country] = acc[country] || { Gold: 0, Silver: 0, Bronze: 0 };
@@ -65,10 +44,10 @@ function updateChart() {
 
     const ctx = document.getElementById('medalChart').getContext('2d');
     if (window.myChart) {
-        window.myChart.destroy(); // Destroy existing chart instance if exists
+        window.myChart.destroy(); 
     }
     window.myChart = new Chart(ctx, {
-        type: 'bar', // Correct chart type for horizontal bars in Chart.js 3.x
+        type: 'bar', 
         data: {
             labels: sortedMedals.map(item => item.Country),
             datasets: [{
@@ -92,21 +71,21 @@ function updateChart() {
             }]
         },
         options: {
-            indexAxis: 'y', // Ensures the bar chart is horizontal
+            indexAxis: 'y',
             responsive: false,
             scales: {
                 x: {
                     stacked: true,
                     beginAtZero: true,
                     title: {
-                        display: true, // Ensure the title is displayed
+                        display: true, 
                         text: 'Number of Medals'
                     },
                     ticks: {
-                        stepSize: 1, // Forces the ticks to increment by 1
-                        precision: 0, // Ensures no decimal places
+                        stepSize: 1, 
+                        precision: 0, 
                         callback: function(value) {
-                            if (value % 1 === 0) { // Only display integer values
+                            if (value % 1 === 0) {
                                 return value;
                             }
                         }
@@ -120,11 +99,6 @@ function updateChart() {
     });
 }
 
-/* window.onload = function() {
-    populateYears();
-    updateChart();  // Initial chart display
-    console.log(data);
-}; */
 
 
 window.onload = function() {
@@ -136,8 +110,8 @@ window.onload = function() {
             data = results.data;
             console.log(data);
             populateYears();
-            updateChart(); // Initialize the chart after data is loaded and processed
+            updateChart(); 
         }
     })
-    //console.log("Global Data Check:", data);
+    //console.log(" Data Check:", data);
 };