Mirrors/fish-shell
2
0
Fork 0
mirror of https://github.com/fish-shell/fish-shell synced 2024-11-15 17:28:19 +00:00
Code Issues Projects Releases Packages Wiki Activity

Clean up, debug, optimize some command description generation

Browse source 
Use some more move semantics to reduce allocations.

Correctly handle the case where the completion is empty. For example, if
you type:

    ls<tab>

we get an empty completion (since ls is already a valid command), but we
still want to show its description.

Remove some unsafe statics - these are unsafe today in weird cases where
completions might invoke complete recursively, and also will soon be
unsafe with concurrent execution.
This commit is contained in:
ridiculousfish 2020-01-25 18:14:38 -08:00
parent 38f4330683
commit 86897cafd6
1 changed files with 22 additions and 23 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

45
src/complete.cpp
View file

@ -606,47 +606,46 @@ void completer_t::complete_cmd_desc(const wcstring &str) {
wcstring lookup_cmd(L"__fish_describe_command ");
lookup_cmd.append(escape_string(cmd, ESCAPE_ALL));
// See the ASSERT_IS_MAIN_THREAD() above, making it safe to make this a static variable.
// This lets us reuse the heap-allocated memory across calls.
static std::unordered_map<wcstring, wcstring> lookup;
static wcstring_list_t list;
// First locate a list of possible descriptions using a single call to apropos or a direct
// search if we know the location of the whatis database. This can take some time on slower
// systems with a large set of manuals, but it should be ok since apropos is only called once.
lookup.clear();
list.clear();
wcstring_list_t list;
(void)exec_subshell(lookup_cmd, *ctx.parser, list, false /* don't apply exit status */);
// Then discard anything that is not a possible completion and put the result into a
// hashtable with the completion as key and the description as value.
//
// Should be reasonably fast, since no memory allocations are needed.
// mqudsi: I don't know if the above were ever true, but it's certainly not any more.
// Plenty of allocations below.
std::unordered_map<wcstring, wcstring> lookup;
// A typical entry is the command name, followed by a tab, followed by a description.
for (const wcstring &elstr : list) {
if (elstr.length() < cmd.length()) continue;
const wcstring fullkey(elstr, cmd.length());
// Skip keys that are too short.
if (elstr.size() < cmd.size()) continue;
size_t tab_idx = fullkey.find(L'\t');
if (tab_idx == wcstring::npos) continue;
// Skip cases without a tab, or without a description, or bizarre cases where the tab is
// part of the command.
size_t tab_idx = elstr.find(L'\t');
if (tab_idx == wcstring::npos || tab_idx + 1 >= elstr.size() || tab_idx < cmd.size())
continue;
const wcstring key(fullkey, 0, tab_idx);
wcstring val(fullkey, tab_idx + 1);
// Make the key. This is the stuff after the command.
// For example:
// elstr = lsmod
// cmd = ls
// key = mod
// Note an empty key is common and natural, if 'cmd' were already valid.
wcstring key(elstr, cmd.size(), tab_idx - cmd.size());
wcstring val(elstr, tab_idx + 1);
assert(!val.empty() && "tab index should not have been at the end.");
// And once again I make sure the first character is uppercased because I like it that
// way, and I get to decide these things.
if (!val.empty()) val[0] = towupper(val[0]);
lookup[key] = val;
val.at(0) = towupper(val.at(0));
lookup.insert(std::make_pair(std::move(key), std::move(val)));
}
// Then do a lookup on every completion and if a match is found, change to the new
// description.
//
// This needs to do a reallocation for every description added, but there shouldn't be that
// many completions, so it should be ok.
for (auto &completion : completions) {
const wcstring &el = completion.completion;
if (el.empty()) continue;
auto new_desc_iter = lookup.find(el);
if (new_desc_iter != lookup.end()) completion.description = new_desc_iter->second;
}