This is a bit of a successor to #270. This commit adds four improvements to the Cocktail Party scraper:
First, some Cocktail Party recipes use units that the recipe-utils parser doesn't understand. For example, the
Manhattan Bianco uses a "piece." Such ingredients would've simply been presented to the user without the unit and the user had to fill that in themselves. Now the code will fall back to whatever the parser didn't parse, which is a fairly good default for Cocktail Party.
Next, the Cocktail Party website uses "parts" for lots of ingredients, but they actually mean fluid ounces (i.e. the same recipes in their mobile app show up with fluid ounces instead of parts). Thus the scraper now maps parts to fluid ounces.
Next, the scraper now reads the links in the "post info" part of the page as tags. The links usually provide categories or names of the cocktail's creator, so this works out nicely.
Finally, I've fixed an oversight introduced in #270. The code for parsing the cocktail's description goes through all the paragraphs and then joins them up to form proper Markdown paragraphs. However, those paragraphs were then squashed together within the toArray() function in the clean up process. That's obviously undesirable. So now the paragraphs are cleaned up before they're joined together, which produces nice Markdown with multiple paragraphs.
karlomikus
commented
2 months ago
This is a bit of a successor to #270. This commit adds four improvements to the Cocktail Party scraper:
First, some Cocktail Party recipes use units that the recipe-utils parser doesn't understand. For example, the Manhattan Bianco uses a "piece." Such ingredients would've simply been presented to the user without the unit and the user had to fill that in themselves. Now the code will fall back to whatever the parser didn't parse, which is a fairly good default for Cocktail Party.
Next, the Cocktail Party website uses "parts" for lots of ingredients, but they actually mean fluid ounces (i.e. the same recipes in their mobile app show up with fluid ounces instead of parts). Thus the scraper now maps parts to fluid ounces.
Next, the scraper now reads the links in the "post info" part of the page as tags. The links usually provide categories or names of the cocktail's creator, so this works out nicely.
Finally, I've fixed an oversight introduced in #270. The code for parsing the cocktail's description goes through all the paragraphs and then joins them up to form proper Markdown paragraphs. However, those paragraphs were then squashed together within the toArray() function in the clean up process. That's obviously undesirable. So now the paragraphs are cleaned up before they're joined together, which produces nice Markdown with multiple paragraphs.