'Is there a better way to write this incredibly long regex, or perform this error check?
I was trying to find a way to error/style correct a non-standard custom menu file from a decade old video game that I was working on in Notepad++ and this was the best I could come up with.
The below returns any curly brackets that aren't followed by an EOL character or are preceded by anything other than line start and 1-4 tabs, it works fine but seems like it could be a lot more elegant. Any returned brackets are incorrect unless they're the first or last in the file. More tabs are technically okay highly unlikely.
(?<!^\t)(?<!^\t\t)(?<!^\t\t\t)(?<!^\t\t\t\t)[{}]|[{}](?!\R)
Properly formatted:
Menu "MenuName"
{
Menu "SubMenuName"
{
Option "OptionName" "Command"
Option "OptionName" "Command"
}
}
// This is a comment line
// [ curly brackets in comment lines are made square so they don't get counted as balancing
All curly brackets should be on a separate line by themselves with nothing but preceding tabs. They should also be paired but I've got a plugin handling that.
Improperly formatted:
Menu "MenuName"{
Menu "SubMenuName"
{
Option "OptionName" "Command"
Option "OptionName" "Command" }
}Menu "That bracket would be wrong since the line should end afterwards.
{ //this would also be wrong
// Nothing should ever follow a bracket except a End Of Line character.
Is there some better way to implement this search/check, considering Notepad++ uses Boost regex and doesn't allow variable-length lookbehinds? Also perhaps keeping in mind that I learned everything I know about regex last night.
The expression also returns the first (no preceding tab) and last (no EOL character) but I'm okay with that particular behavior.
The full content of a file I use as a template:
It loads from a loose file in the data folder, completely as is.
//DO NOT DELETE, needs a line here for some reason.
Menu "MenuNameTEMPLATE"
{
Title "TitleName"
Option "OptionName" "Command"
Divider
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
PowerReady "PowerIdentifiers"
}
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
Badge "BadgeIdentifiers"
}
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
}
Menu "SubMenuName"
{
Title "TitleName"
Option "OptionName" "Command"
Option "OptionName" "Command"
}
}
Solution 1:[1]
- Ctrl+F
- Find what:
^\h+[{}](?=\h*\R)(*SKIP)(*FAIL)|[{}] - CHECK Wrap around
- CHECK Regular expression
- Find All in Current Document
Explanation:
^ # beginning of line
\h+ # 1 or more horizontal spaces, you can use \t{1,4} if you only want tabulations
[{}] # open or close brace
(?= # positive lookahead, make sure we have after:
\h* # 0 or more horizontal spaces
\R # any kind of linebreak
) # end lookahead
(*SKIP)(*FAIL) # skip this match and consider that fails
| # OR
[{}] # open or close brace
Screenshot:
Solution 2:[2]
I want to start by saying that regex is 100% the wrong tool for this, you want a custom parser to handle both validating your file and parsing it into a model you can then use.
However, with the limitations imposed by your question, the following should do it:
^(?:[^{}]*|\t{1,4}[{}])$
Rather than worry about look-arounds, simply match what you expect to find. See it in action here: https://regex101.com/r/nYNqHw/1
Solution 3:[3]
Since you tagged this Boost, and others rightfully remarked you need to not use regexes together with some editor facilities here, here's a starting point for a proper parser using Boost Spirit.
We'll parse into some types:
struct Option { std::string name, command; };
struct Divider { };
struct LockedOption {
struct Property { std::string key, value; };
using Properties = std::vector<Property>;
Properties properties; // DisplayName, Command, Icon, PowerReady, Badge...?
};
struct Menu;
using MenuItem = boost::variant<Option, Divider, LockedOption,
boost::recursive_wrapper<Menu>>;
struct Menu {
std::string id;
boost::optional<std::string> title;
std::vector<MenuItem> items;
};
Now, we can define a parser:
namespace Parser {
using namespace boost::spirit::x3;
rule<struct menu_rule, Menu> const menu{"menu"};
auto const qstring = rule<void, std::string>{"quoted string"} = //
lexeme['"' > *('"' >> char_('"') | ~char_('"')) > '"'];
auto const option = rule<void, Option>{"option"} = //
"Option" > qstring > qstring;
auto property = (!lit('}')) > lexeme[+graph] > qstring;
auto properties = rule<void, LockedOption::Properties>{"properties"} =
*(property > eol);
auto const lockedoption = rule<void, LockedOption>{"lockedoption"} = //
"LockedOption" > eol //
> '{' > eol //
> properties //
> '}';
auto divider = rule<void, Divider>{"divider"} = //
lit("Divider") >> attr(Divider{});
auto item = rule<void, MenuItem>{"menu|option|lockedoption|divider"} =
menu | option | lockedoption | divider;
auto title = "Title" > qstring;
auto menu_def = //
"Menu" > qstring > eol //
> '{' > eol //
> -(title > eol) //
> *(item > eol) //
> '}';
auto ignore = blank | "//" >> *~char_("\r\n") >> (eol|eoi);
BOOST_SPIRIT_DEFINE(menu)
Menu parseMenu(std::string const& text) try {
Menu result;
parse(begin(text), end(text), skip(ignore)[expect[menu] > *eol > eoi],
result);
return result;
} catch (expectation_failure<std::string::const_iterator> const& ef) {
throw std::runtime_error(
"At " + std::to_string(std::distance(begin(text), ef.where())) +
": Expected " + ef.which() + " (Got '" +
std::string(ef.where(), std::find(ef.where(), end(text), '\n')) +
"')");
}
} // namespace Parser
All we need to use is the parseMenu function, which returns the parsed Menu. So if we wire up the samples from your question:
for (std::string const& sample :
{
// ...
})
{
static int i = 0;
fmt::print("----- {}\n", ++i);
try {
fmt::print("Parsed: {}\n", Parser::parseMenu(sample));
} catch (std::exception const& e) {
std::cout << "Parse failed: " << e.what() << "\n";
}
}
We can get the output (see live demo below):
----- 1
Parsed: Menu "MenuName"
{
Title ""
Menu "SubMenuName"
{
Title ""
Option "OptionName" "Command"
Option "OptionName" "Command"
}
}
----- 2
Parse failed: At 19: Expected eol (Got '{')
----- 3
Parsed: Menu "MenuNameTEMPLATE"
{
Title "TitleName"
Option "OptionName" "Command"
Divider
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
PowerReady "PowerIdentifiers"
}
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
Badge "BadgeIdentifiers"
}
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
}
Menu "SubMenuName"
{
Title "TitleName"
Option "OptionName" "Command"
Option "OptionName" "Command"
}
}
----- 4
Parse failed: At 70: Expected menu|option|lockedoption|divider (Got ' Road Rage')
LIVE DEMO
On Compiler Explorer: https://godbolt.org/z/sW3Y5z9nq
//#define BOOST_SPIRIT_X3_DEBUG
#include <boost/fusion/adapted.hpp>
#include <boost/spirit/home/x3.hpp>
#include <iostream>
#include <map>
#include <fmt/ranges.h>
struct Option { std::string name, command; };
struct Divider { };
struct LockedOption {
#if 1
struct Property { std::string key, value; };
using Properties = std::vector<Property>;
#else
using Properties = std::map<std::string, std::string>;
#endif
Properties properties; // DisplayName, Command, Icon, PowerReady, Badge...?
};
struct Menu;
using MenuItem = boost::variant<Option, Divider, LockedOption,
boost::recursive_wrapper<Menu>>;
struct Menu {
std::string id;
boost::optional<std::string> title;
std::vector<MenuItem> items;
};
#ifdef BOOST_SPIRIT_X3_DEBUG
[[maybe_unused]] std::ostream& operator<<(std::ostream& os, Option) { return os << "Option"; }
[[maybe_unused]] std::ostream& operator<<(std::ostream& os, Divider) { return os << "Divider"; }
[[maybe_unused]] std::ostream& operator<<(std::ostream& os, LockedOption) { return os << "LockedOption"; }
[[maybe_unused]] std::ostream& operator<<(std::ostream& os, Menu) { return os << "Menu"; }
#endif
struct MenuItemFormatter : fmt::formatter<std::string> {
template <typename... Ts>
auto format(boost::variant<Ts...> const& var, auto& ctx) const {
return boost::apply_visitor(
[&](auto const& el) {
return format_to(ctx.out(), fmt::runtime("{}"), el);
}, var);
}
auto format(LockedOption const& lo, auto& ctx) const {
auto out = fmt::format_to(ctx.out(), "LockedOption\n{{\n");
for (auto const& [k, v] : lo.properties)
out = fmt::format_to(out, " {} \"{}\"\n", k, v);
return fmt::format_to(out, "}}");
}
auto format(Divider const&, auto& ctx) const {
return fmt::format_to(ctx.out(), "Divider");
}
auto format(Option const& o, auto& ctx) const {
return fmt::format_to(ctx.out(), "Option \"{}\" \"{}\"", o.name, o.command);
}
auto format(Menu const& m, auto& ctx) const {
return fmt::format_to(
ctx.out(), "Menu \"{}\"\n{{\n Title \"{}\"\n {}\n}}", m.id,
m.title.value_or(""), fmt::join(m.items, "\n "));
}
};
template <> struct fmt::formatter<MenuItem> : MenuItemFormatter{};
template <> struct fmt::formatter<LockedOption> : MenuItemFormatter{};
template <> struct fmt::formatter<Divider> : MenuItemFormatter{};
template <> struct fmt::formatter<Option> : MenuItemFormatter{};
template <> struct fmt::formatter<Menu> : MenuItemFormatter{};
BOOST_FUSION_ADAPT_STRUCT(Option, name, command)
BOOST_FUSION_ADAPT_STRUCT(LockedOption, properties)
BOOST_FUSION_ADAPT_STRUCT(LockedOption::Property, key, value)
BOOST_FUSION_ADAPT_STRUCT(Menu, id, title, items)
namespace Parser {
using namespace boost::spirit::x3;
rule<struct menu_rule, Menu> const menu{"menu"};
auto const qstring = rule<void, std::string>{"quoted string"} = //
lexeme['"' > *('"' >> char_('"') | ~char_('"')) > '"'];
auto const option = rule<void, Option>{"option"} = //
"Option" > qstring > qstring;
auto property = lexeme[+graph] > qstring;
auto properties = rule<void, LockedOption::Properties>{"properties"} =
*((!lit('}')) > property > eol);
auto const lockedoption = rule<void, LockedOption>{"lockedoption"} = //
"LockedOption" > eol //
> '{' > eol //
> properties //
> '}';
auto divider = rule<void, Divider>{"divider"} = //
lit("Divider") >> attr(Divider{});
auto item = rule<void, MenuItem>{"menu|option|lockedoption|divider"} =
menu | option | lockedoption | divider;
auto title = "Title" > qstring;
auto menu_def = //
"Menu" > qstring > eol //
> '{' > eol //
> -(title > eol) //
> *((!lit('}')) > item > eol) //
> '}';
auto ignore = blank | "//" >> *~char_("\r\n") >> (eol|eoi);
BOOST_SPIRIT_DEFINE(menu)
Menu parseMenu(std::string const& text) try {
Menu result;
parse(begin(text), end(text), skip(ignore)[expect[menu] > *eol > eoi],
result);
return result;
} catch (expectation_failure<std::string::const_iterator> const& ef) {
throw std::runtime_error(
"At " + std::to_string(std::distance(begin(text), ef.where())) +
": Expected " + ef.which() + " (Got '" +
std::string(ef.where(), std::find(ef.where(), end(text), '\n')) +
"')");
}
} // namespace Parser
int main() {
for (std::string const& sample :
{
R"( Menu "MenuName"
{
Menu "SubMenuName"
{
Option "OptionName" "Command"
Option "OptionName" "Command"
}
}
// This is a comment line
// // [ curly brackets in comment lines are made square so they don't get counted as balancing)",
R"( Menu "MenuName"{
Menu "SubMenuName"
{
Option "OptionName" "Command"
Option "OptionName" "Command" }
}Menu "That bracket would be wrong since the line should end afterwards.
{ //this would also be wrong
// Nothing should ever follow a bracket except a End Of Line character.)",
R"(//DO NOT DELETE, needs a line here for some reason.
Menu "MenuNameTEMPLATE"
{
Title "TitleName"
Option "OptionName" "Command"
Divider
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
PowerReady "PowerIdentifiers"
}
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
Badge "BadgeIdentifiers"
}
LockedOption
{
DisplayName "OptionName"
Command "Command"
Icon "IconName"
}
Menu "SubMenuName"
{
Title "TitleName"
Option "OptionName" "Command"
Option "OptionName" "Command"
}
})",
R"(Menu "Not So Good"
{
Title "Uhoh"
Road Rage
}
)",
}) //
{
static int i = 0;
fmt::print("----- {}\n", ++i);
try {
fmt::print("Parsed: {}\n", Parser::parseMenu(sample));
} catch (std::exception const& e) {
std::cout << "Parse failed: " << e.what() << "\n";
}
}
}
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|---|
| Solution 1 | |
| Solution 2 | Blindy |
| Solution 3 |