It's a fact that any groupoid is equivalent to a disjoint union of (deloopings of) groups. See, e.g. Proposition 4.3 of https://ncatlab.org/nlab/show/groupoid#PropertiesEquivalencesOfGroupoids.
Does this situation carry over to topological groupoids? I.e., is every topological groupoid equivalent to a disjoint union of topological groups?
No. For instance, any groupoid can be made into a topological groupoid $G$ by giving the spaces of objects and morphisms the indiscrete topology. Any map from $G$ to a disjoint union of topological groups must be constant on objects, since the space of objects of $G$ is connected. It follows that if $G$ has more than one different isomorphism class of objects, then $G$ cannot be equivalent to a disjoint union of topological groups.
If you want a nice (say, Hausdorff) counterexample, you can take your favorite space $X$ and make a topological groupoid where the space of objects is $X$ and there are no non-identity morphisms (so the space of morphisms is also $X$). By a similar argument to the previous paragraph, this topological groupoid is not equivalent to a disjoint union of topological groups unless $X$ is discrete.