normalizers.cpp

Go to the documentation of this file.

#include <algorithm>
#include <iterator>
#include <numeric>
#include <ranges>
#include <vector>
 
#include <automaton/range_helper.h>
#include <fe/assert.h>
 
#include "mim/axm.h"
#include "mim/def.h"
#include "mim/tuple.h"
#include "mim/world.h"
 
#include "mim/util/dbg.h"
#include "mim/util/log.h"
 
#include "mim/plug/regex/autogen.h"
#include "mim/plug/regex/regex.h"
 
using Range  = automaton::Range;
using Ranges = mim::Vector<Range>;
 
namespace mim::plug::regex {
 
template<quant id>
const Def* normalize_quant(const Def* type, const Def* callee, const Def* arg) {
    auto& world = type->world();
 
    // quantifiers are idempotent
    if (Axm::isa(id, arg)) return arg;
 
    if constexpr (id == quant::plus) {
        // (\d?)+ and (\d*)+ == \d*
        if (auto optional_app = Axm::isa(quant::optional, arg))
            return world.call(quant::star, optional_app->arg());
        else if (auto star_app = Axm::isa(quant::star, arg))
            return arg;
    } else if constexpr (id == quant::star) {
        // (\d?)* and (\d+)* == \d*
        if (auto quant_app = Axm::isa<quant>(arg)) return world.app(callee, quant_app->arg());
    } else if constexpr (id == quant::optional) {
        // (\d*)? and (\d+)? == \d*
        if (auto star_app = Axm::isa(quant::star, arg))
            return arg;
        else if (auto plus_app = Axm::isa(quant::plus, arg))
            return world.call(quant::star, plus_app->arg());
    }
 
    return {};
}
 
template<class ConjOrDisj>
const Def* make_binary_tree(Defs args) {
    assert(!args.empty());
    auto& world = args.front()->world();
    return std::accumulate(args.begin() + 1, args.end(), args.front(), [&world](const Def* lhs, const Def* rhs) {
        return world.call<ConjOrDisj, false>(Defs{lhs, rhs});
    });
}
 
const Def* normalize_conj(const Def* type, const Def* callee, const Def* arg) {
    auto& world = type->world();
    world.DLOG("conj {}:{} ({})", type, callee, arg);
 
    if (auto a = Lit::isa(arg->arity())) {
        switch (*a) {
            case 0: return world.lit_tt();
            case 1: return arg;
            default:
                if (auto args = detail::flatten_in_arg<conj>(arg); !args.empty())
                    return make_binary_tree<conj>(args);
                else
                    return world.annex<empty>();
        }
    }
 
    return {};
}
 
bool compare_re(const Def* lhs, const Def* rhs) {
    auto lhs_range = Axm::isa<range>(lhs);
    auto rhs_range = Axm::isa<range>(rhs);
    // sort ranges by increasing lower bound
    if (lhs_range && rhs_range) return Lit::as(lhs_range->arg()->proj(0)) < Lit::as(rhs_range->arg()->proj(0));
    // ranges to the end
    if (lhs_range) return false;
    if (rhs_range) return true;
 
    return lhs->gid() < rhs->gid(); // make irreflexive
}
 
void make_vector_unique(DefVec& args) {
    std::stable_sort(args.begin(), args.end(), &compare_re);
    {
        auto new_end = std::unique(args.begin(), args.end());
        args.erase(new_end, args.end());
    }
}
 
bool is_in_range(Range range, nat_t needle) { return needle >= range.first && needle <= range.second; }
 
auto get_range(const Def* rng) -> Range {
    auto rng_match = Axm::isa<range, false>(rng);
    return {Lit::as<std::uint8_t>(rng_match->arg(0)), Lit::as<std::uint8_t>(rng_match->arg(1))};
}
 
struct app_range {
    World& w;
    const Def* operator()(Range rng) { return w.call<range>(Defs{w.lit_i8(rng.first), w.lit_i8(rng.second)}); }
};
 
void merge_ranges(DefVec& args) {
    auto ranges_begin = args.begin();
    while (ranges_begin != args.end() && !Axm::isa<range>(*ranges_begin))
        ranges_begin++;
    if (ranges_begin == args.end()) return;
 
    std::set<const Def*> to_remove;
    Ranges old_ranges;
    auto& world = (*ranges_begin)->world();
 
    std::transform(ranges_begin, args.end(), std::back_inserter(old_ranges), get_range);
 
    auto new_ranges = automaton::merge_ranges(old_ranges, [&world](std::string_view msg) { world.DLOG("{}", msg); });
 
    // invalidates ranges_begin
    args.erase(ranges_begin, args.end());
    std::transform(new_ranges.begin(), new_ranges.end(), std::back_inserter(args), app_range{world});
 
    make_vector_unique(args);
}
 
template<cls A, cls B>
bool equals_any(const Def* cls0, const Def* cls1) {
    return (Axm::isa(A, cls0) && Axm::isa(B, cls1)) || (Axm::isa(A, cls1) && Axm::isa(B, cls0));
}
 
bool equals_any(const Def* lhs, const Def* rhs) {
    auto check_arg_equiv = [](const Def* lhs, const Def* rhs) {
        if (auto rng_lhs = Axm::isa<range>(lhs))
            if (auto not_rhs = Axm::isa<not_>(rhs)) {
                if (auto rng_rhs = Axm::isa<range>(not_rhs->arg())) return rng_lhs == rng_rhs;
            }
        return false;
    };
 
    return check_arg_equiv(lhs, rhs) || check_arg_equiv(rhs, lhs);
}
 
bool equals_any(Defs lhs, Defs negated_rhs) {
    auto is_range = [](const Def* d) { return Axm::isa<range>(d); };
    auto to_range = std::views::filter(is_range) | std::views::transform(get_range);
    auto rhs_view = negated_rhs | to_range;
 
    if (std::ranges::distance(rhs_view) != std::ranges::distance(negated_rhs)) return false;
 
    return std::ranges::includes(lhs | to_range, rhs_view);
}
 
const Def* normalize_disj(const Def* type, const Def*, const Def* arg) {
    auto& world = type->world();
    if (auto a = Lit::isa(arg->arity())) {
        switch (*a) {
            case 0: return world.lit_ff();
            case 1: return arg;
            default:
                auto new_args = detail::flatten_in_arg<disj>(arg);
 
                const bool contains_any
                    = std::ranges::find_if(new_args, [](const Def* ax) -> bool { return Axm::isa<any>(ax); })
                   != new_args.end();
                const bool contains_empty
                    = std::ranges::find_if(new_args, [](const Def* ax) -> bool { return Axm::isa<empty>(ax); })
                   != new_args.end();
 
                auto make_any = [&world, contains_empty]() {
                    if (contains_empty)
                        // (any|) matches everything, including empty string
                        // don't normalize again, as we'd just run into this normalizer again..
                        return world.call<disj, false>(Defs{world.annex<any>(), world.annex<empty>()});
                    else
                        return world.annex<any>();
                };
 
                if (contains_any) return make_any();
                make_vector_unique(new_args);
                merge_ranges(new_args);
 
                const Def* to_remove = nullptr;
                for (const auto* cls0 : new_args) {
                    for (const auto* cls1 : new_args)
                        if (equals_any(cls0, cls1)) return make_any();
 
                    if (auto not_rhs = Axm::isa<not_>(cls0)) {
                        if (auto disj_rhs = Axm::isa<disj>(not_rhs->arg())) {
                            auto rngs = detail::flatten_in_arg<disj>(disj_rhs->arg());
                            make_vector_unique(rngs);
                            if (equals_any(new_args, rngs)) return make_any();
                        }
                    }
                }
 
                erase(new_args, to_remove);
                world.DLOG("final ranges {}", fe::Join(new_args));
 
                if (new_args.size() > 2) return make_binary_tree<disj>(new_args);
                if (new_args.size() > 1) return world.call<disj, false>(new_args);
                return new_args.back();
        }
    }
    return {};
}
 
const Def* normalize_range(const Def* type, const Def* callee, const Def* arg) {
    auto& world     = type->world();
    auto [lhs, rhs] = arg->projs<2>();
 
    if (!lhs->isa<Var>() && !rhs->isa<Var>()) // before first PE.
        if (lhs->as<Lit>()->get() > rhs->as<Lit>()->get()) return world.raw_app(type, callee, {rhs, lhs});
 
    return {};
}
 
const Def* any_unwanted_for_not(const Def* arg) {
    if (auto ax = Axm::isa<regex::range>(arg)) return nullptr;
    if (auto ax = Axm::isa<regex::not_>(arg)) return nullptr;
    if (auto ax = Axm::isa<regex::any>(arg)) return nullptr;
    if (auto disj = Axm::isa<regex::disj>(arg)) {
        for (const auto* disj_arg : disj->args())
            if (auto ret = any_unwanted_for_not(disj_arg)) return ret;
        return nullptr;
    }
    return arg;
}
 
const Def* normalize_not(const Def*, const Def* callee, const Def* arg) {
    if (auto unwanted = any_unwanted_for_not(arg)) {
        throw Error()
            .error(arg->loc(),
                   "regex.not_ must only be used with regex.disj, regex.range, regex.any and regex.not_: {} {}", callee,
                   arg)
            .error(unwanted->loc(), "found unwanted: {}", unwanted);
    }
    return {};
}
 
MIM_regex_NORMALIZER_IMPL
 
} // namespace mim::plug::regex