如何调整爬取www.oddsportal.com的Python爬虫代码以处理赛事无比分场景

Got it, let's resolve that annoying data shifting issue when matches don't have a score yet. The root problem is we're hardcoding TD element indices, but rows without a table-score cell have one fewer column—so our odds end up pulling from the wrong positions. Here's how to adjust the logic properly:

Step 1: Modify the generate_matches Function #

We need to dynamically check if a score exists in each row before pulling data, instead of relying on fixed indices. Update the function like this:

def generate_matches(table):
    tr_tags = table.findAll('tr')
    country = ""
    league = ""
    for tr_tag in tr_tags:
        if 'class' in tr_tag.attrs and 'dark' in tr_tag['class']:
            # This is a league header row, update country/league
            th_tag = tr_tag.find('th', {'class': 'first2 tl'})
            a_tags = th_tag.findAll('a')
            country = a_tags[0].text
            league = a_tags[1].text
        else:
            td_tags = tr_tag.findAll('td')
            if len(td_tags) < 4:  # Skip empty/malformed rows
                continue
            # Check if this row has a score cell
            has_score = any('table-score' in td.get('class', []) for td in td_tags)
            
            # Assign values based on whether score exists
            time = td_tags[0].text.strip()
            game = td_tags[1].text.strip()
            
            if has_score:
                score = td_tags[2].text.strip()
                home_odds = td_tags[3].text.strip()
                draw_odds = td_tags[4].text.strip()
                away_odds = td_tags[5].text.strip()
            else:
                score = None  # Will convert to NaN in DataFrame
                home_odds = td_tags[2].text.strip()
                draw_odds = td_tags[3].text.strip()
                away_odds = td_tags[4].text.strip()
            
            yield time, game, score, home_odds, draw_odds, away_odds, country, league

Step 2: Keep the parse_data Loop Clean #

Since we've already handled the score check in generate_matches, you don't need to change anything in the loop where you populate GameData—it can stay exactly as it was:

for row in generate_matches(table):
    game_data.date.append(game_date)
    game_data.time.append(row[0])
    game_data.game.append(row[1])
    game_data.score.append(row[2])
    game_data.home_odds.append(row[3])
    game_data.draw_odds.append(row[4])
    game_data.away_odds.append(row[5])
    game_data.country.append(row[6])
    game_data.league.append(row[7])

What Changed? #

• We added a check for the table-score class in each row's TD elements to detect if a score exists.
• If no score is found, we set score to None (which Pandas automatically converts to NaN in the DataFrame) and shift the odds indices left by one.
• We also added a check for empty rows (len(td_tags) <4) to avoid errors from malformed rows.
• Added .strip() to clean up any extra whitespace in the text values for cleaner data.

This way, whether a match has a score or not, your home_odds, draw_odds, and away_odds will always pull from the correct positions, and the score column will show NaN for matches without results.

内容的提问来源于stack exchange，提问作者user16304089