| 1 |
*********** |
| 2 |
Collections |
| 3 |
*********** |
| 4 |
|
| 5 |
lazr.restful makes collections of data available through Pythonic |
| 6 |
mechanisms like slices. |
| 7 |
|
| 8 |
>>> from lazr.restfulclient.tests.example import CookbookWebServiceClient |
| 9 |
>>> service = CookbookWebServiceClient() |
| 10 |
|
| 11 |
You can iterate through all the items in a collection. |
| 12 |
|
| 13 |
>>> names = sorted([recipe.dish.name for recipe in service.recipes]) |
| 14 |
>>> len(names) |
| 15 |
5 |
| 16 |
>>> names |
| 17 |
[u'Baked beans', ..., u'Roast chicken'] |
| 18 |
|
| 19 |
But it's almost always better to slice them. |
| 20 |
|
| 21 |
>>> sorted([recipe.dish.name for recipe in service.recipes[:2]]) |
| 22 |
[u'Roast chicken', u'Roast chicken'] |
| 23 |
|
| 24 |
You can get a slice of any collection, so long as you provide start |
| 25 |
and end points keyed to the beginning of the list. You can't key a |
| 26 |
slice to the end of the list because it might be expensive to |
| 27 |
calculate how big the list is. |
| 28 |
|
| 29 |
This set-up code creates a regular Python list of all recipes on the |
| 30 |
site, for comparison with a lazr.restful Collection object |
| 31 |
representing the same list. |
| 32 |
|
| 33 |
>>> all_recipes = [recipe for recipe in service.recipes] |
| 34 |
>>> recipes = service.recipes |
| 35 |
|
| 36 |
Calling len() on the Collection object makes sure that the first page |
| 37 |
of representations is cached, which forces this test to test an |
| 38 |
optimization. |
| 39 |
|
| 40 |
>>> ignored = len(recipes) |
| 41 |
|
| 42 |
These tests demonstrate that slicing the collection resource gives the |
| 43 |
same results as collecting all the entries in the collection, and |
| 44 |
slicing an ordinary list. |
| 45 |
|
| 46 |
>>> def slices_match(slice): |
| 47 |
... """Slice two lists of recipes, then make sure they're the same.""" |
| 48 |
... list1 = recipes[slice] |
| 49 |
... list2 = all_recipes[slice] |
| 50 |
... if len(list1) != len(list2): |
| 51 |
... raise ("Lists are different sizes: %d vs. %d" % |
| 52 |
... (len(list1), len(list2))) |
| 53 |
... for index in range(0, len(list1)): |
| 54 |
... if list1[index].id != list2[index].id: |
| 55 |
... raise ("%s doesn't match %s in position %d" % |
| 56 |
... (list1[index].id, list2[index].id, index)) |
| 57 |
... return True |
| 58 |
|
| 59 |
>>> slices_match(slice(3)) |
| 60 |
True |
| 61 |
>>> slices_match(slice(50)) |
| 62 |
True |
| 63 |
>>> slices_match(slice(1,2)) |
| 64 |
True |
| 65 |
>>> slices_match(slice(2,21)) |
| 66 |
True |
| 67 |
>>> slices_match(slice(2,21,3)) |
| 68 |
True |
| 69 |
|
| 70 |
>>> slices_match(slice(0, 200)) |
| 71 |
True |
| 72 |
>>> slices_match(slice(30, 200)) |
| 73 |
True |
| 74 |
>>> slices_match(slice(60, 100)) |
| 75 |
True |
| 76 |
|
| 77 |
>>> recipes[5:] |
| 78 |
Traceback (most recent call last): |
| 79 |
... |
| 80 |
ValueError: Collection slices must have a definite, nonnegative end point. |
| 81 |
|
| 82 |
>>> recipes[10:-1] |
| 83 |
Traceback (most recent call last): |
| 84 |
... |
| 85 |
ValueError: Collection slices must have a definite, nonnegative end point. |
| 86 |
|
| 87 |
>>> recipes[-1:] |
| 88 |
Traceback (most recent call last): |
| 89 |
... |
| 90 |
ValueError: Collection slices must have a nonnegative start point. |
| 91 |
|
| 92 |
>>> recipes[:] |
| 93 |
Traceback (most recent call last): |
| 94 |
... |
| 95 |
ValueError: Collection slices must have a definite, nonnegative end point. |
| 96 |
|
| 97 |
You can slice a collection that's the return value of a named |
| 98 |
operation. |
| 99 |
|
| 100 |
>>> e_recipes = service.cookbooks.find_recipes(search='e') |
| 101 |
>>> len(e_recipes[1:3]) |
| 102 |
2 |
| 103 |
|
| 104 |
You can also access individual items in this collection by index. |
| 105 |
|
| 106 |
>>> print e_recipes[1].dish.name |
| 107 |
Foies de voilaille en aspic |
| 108 |
|
| 109 |
>>> e_recipes[1000] |
| 110 |
Traceback (most recent call last): |
| 111 |
... |
| 112 |
IndexError: list index out of range |
| 113 |
|
| 114 |
When are representations fetched? |
| 115 |
================================= |
| 116 |
|
| 117 |
To avoid unnecessary HTTP requests, a representation of a collection |
| 118 |
is fetched at the last possible moment. Let's see what that means. |
| 119 |
|
| 120 |
>>> import httplib2 |
| 121 |
>>> httplib2.debuglevel = 1 |
| 122 |
|
| 123 |
>>> service = CookbookWebServiceClient() |
| 124 |
send: ... |
| 125 |
... |
| 126 |
|
| 127 |
Just accessing a top-level collection doesn't trigger an HTTP request. |
| 128 |
|
| 129 |
>>> recipes = service.recipes |
| 130 |
>>> dishes = service.dishes |
| 131 |
>>> cookbooks = service.cookbooks |
| 132 |
|
| 133 |
Getting the length of the collection, or any entry from the |
| 134 |
collection, triggers an HTTP request. |
| 135 |
|
| 136 |
>>> len(recipes) |
| 137 |
send: 'GET /1.0/recipes... |
| 138 |
... |
| 139 |
|
| 140 |
>>> dish = dishes[1] |
| 141 |
send: 'GET /1.0/dishes... |
| 142 |
... |
| 143 |
|
| 144 |
Invoking a named operation will also trigger an HTTP request. |
| 145 |
|
| 146 |
>>> cookbooks.find_recipes(search="foo") |
| 147 |
send: ... |
| 148 |
... |
| 149 |
|
| 150 |
Scoped collections work the same way: just getting a reference to the |
| 151 |
collection doesn't trigger an HTTP request. |
| 152 |
|
| 153 |
>>> recipes = dish.recipes |
| 154 |
|
| 155 |
But getting any information about the collection triggers an HTTP request. |
| 156 |
|
| 157 |
>>> len(recipes) |
| 158 |
send: 'GET /1.0/dishes/.../recipes ... |
| 159 |
... |
| 160 |
|
| 161 |
Cleanup. |
| 162 |
|
| 163 |
>>> httplib2.debuglevel = None |