Overhaul completions, redo #6693 (#8131)

As I complained in
https://github.com/NixOS/nix/pull/6784#issuecomment-1421777030 (a
comment on the wrong PR, sorry again!), #6693 introduced a second
completions mechanism to fix a bug. Having two completion mechanisms
isn't so nice.

As @thufschmitt also pointed out, it was a bummer to go from `FlakeRef`
to `std::string` when collecting flake refs. Now it is `FlakeRefs`
again.

The underlying issue that sought to work around was that completion of
arguments not at the end can still benefit from the information from
latter arguments.

To fix this better, we rip out that change and simply defer all
completion processing until after all the (regular, already-complete)
arguments have been passed.

In addition, I noticed the original completion logic used some global
variables. I do not like global variables, because even if they save
lines of code, they also obfuscate the architecture of the code.

I got rid of them  moved them to a new `RootArgs` class, which now has
`parseCmdline` instead of `Args`. The idea is that we have many argument
parsers from subcommands and what-not, but only one root args that owns
the other per actual parsing invocation. The state that was global is
now part of the root args instead.

This did, admittedly, add a bunch of new code. And I do feel bad about
that. So I went and added a lot of API docs to try to at least make the
current state of things clear to the next person.

--

This is needed for RFC 134 (tracking issue #7868). It was very hard to
modularize `Installable` parsing when there were two completion
arguments. I wouldn't go as far as to say it is *easy* now, but at least
it is less hard (and the completions test finally passed).

Co-authored-by: Valentin Gagarin <valentin.gagarin@tweag.io>

src/libutil/args/root.hh

@@ -0,0 +1,72 @@
+#pragma once
+
+#include "args.hh"
+
+namespace nix {
+
+/**
+ * The concrete implementation of a collection of completions.
+ *
+ * This is exposed so that the main entry point can print out the
+ * collected completions.
+ */
+struct Completions final : AddCompletions
+{
+    std::set<Completion> completions;
+    Type type = Type::Normal;
+
+    void setType(Type type) override;
+    void add(std::string completion, std::string description = "") override;
+};
+
+/**
+ * The outermost Args object. This is the one we will actually parse a command
+ * line with, whereas the inner ones (if they exists) are subcommands (and this
+ * is also a MultiCommand or something like it).
+ *
+ * This Args contains completions state shared between it and all of its
+ * descendent Args.
+ */
+class RootArgs : virtual public Args
+{
+public:
+    /** Parse the command line, throwing a UsageError if something goes
+     * wrong.
+     */
+    void parseCmdline(const Strings & cmdline);
+
+    std::shared_ptr<Completions> completions;
+
+protected:
+
+    friend class Args;
+
+    /**
+     * A pointer to the completion and its two arguments; a thunk;
+     */
+    struct DeferredCompletion {
+        const CompleterClosure & completer;
+        size_t n;
+        std::string prefix;
+    };
+
+    /**
+     * Completions are run after all args and flags are parsed, so completions
+     * of earlier arguments can benefit from later arguments.
+     */
+    std::vector<DeferredCompletion> deferredCompletions;
+
+    /**
+     * Experimental features needed when parsing args. These are checked
+     * after flag parsing is completed in order to support enabling
+     * experimental features coming after the flag that needs the
+     * experimental feature.
+     */
+    std::set<ExperimentalFeature> flagExperimentalFeatures;
+
+private:
+
+    std::optional<std::string> needsCompletion(std::string_view s);
+};
+
+}