Question

I am trying to build a dictionary with two keys but am getting a KeyError when assigning items. I don’t get the error when using each of the keys separately, and the syntax seems pretty straightforward so I’m stumped.

searchIndices = ['Books', 'DVD']
allProducts = {}
for index in searchIndices:
    res = amazon.ItemSearch(Keywords = entity, SearchIndex = index, ResponseGroup = 'Large', ItemPage = 1, Sort = "salesrank", Version = '2010-11-01')
    products = feedparser.parse(res)
    for x in range(10):
        allProducts[index][x] = { 'price' : products['entries'][x]['formattedprice'],  
                                  'url'   : products['entries'][x]['detailpageurl'], 
                                  'title' : products['entries'][x]['title'], 
                                  'img'   : products['entries'][x]['href'],
                                  'rank'  : products['entries'][x]['salesrank'] 
                                }

I don’t believe the issue lies with feedparser (which converts xml to dict) or with the results I’m getting from amazon, as I have no issues building a dict when either using ‘allProducts[x]’ or ‘allProducts[index]’, but not both.

What am I missing?

Answer 1

In order to assign to allProducts[index][x], first a lookup is done on allProducts[index] to get a dict, then the value you’re assigning is stored at index x in that dict.

However, the first time through the loop, allProducts[index] doesn’t exist yet. Try this:

for x in range(10):
    if index not in allProducts:
        allProducts[index] = {  }    # or dict() if you prefer
    allProducts[index][x] = ...

Since you know all the indices that are supposed to be in allProducts in advance, you can initialize it before hand like this instead:

map(lambda i: allProducts[i] = {  }, searchIndices)
for index in searchIndices:
    # ... rest of loop does not need to be modified