If you want a bunch of main algorithms in "recipe" form, with minimal (though not zero) performance analysis - though not really lacking explanation - I quite like Niklaus Wirths "Algorithms and Data Structures". I have the 1986 (Modula 2) edition in dead tree, but a more recent edition is available as a free PDF download...
Search for "Algorithms and Data Structures (1985) (Oberon version: August 2004)" on the page.
Niklaus Wirth designed many Pascal-family languages (including Pascal itself), and there's also a book on compiler design.
There are some weaknesses, even for practical rather than theory purposes. Some of the terminology isn't standard - for example "Balanced trees" rather than "AVL trees" (there's more than one way to balance a tree). Some algorithms aren't the obvious choice - e.g. "Binary B-Trees" are closely related to, but not quite the same, as "Red-Black trees". The "choice of hash function" chapter for hash tables (sorry, "Key Transformation") only suggests the modulo-a-prime-number method, which is pretty limited in practice. The reason is probably that growing/shrinking the hashtable isn't discussed that I remember - for a fixed-size table, modulo-prime is a fair option.
Even so, there's a lot of good stuff in a relatively small and convenient (288 pages for the Modula 2 edition) package. The number of times I've looked up even the lower-bound binary search is huge.
And if you top up with e.g. a view of the MIT OpenCourseware algorithms course notes, and look up some of the terms in Wikipedia, you shouldn't have any serious confusion issues.