messageformat2.h

Go to the documentation of this file.

// © 2024 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html

#include "unicode/utypes.h"

#ifndef MESSAGEFORMAT2_H
#define MESSAGEFORMAT2_H

#if U_SHOW_CPLUSPLUS_API

#if !UCONFIG_NO_NORMALIZATION

#if !UCONFIG_NO_FORMATTING

#if !UCONFIG_NO_MF2

#include "unicode/messageformat2_arguments.h"
#include "unicode/messageformat2_data_model.h"
#include "unicode/messageformat2_function_registry.h"
#include "unicode/normalizer2.h"
#include "unicode/unistr.h"

#ifndef U_HIDE_DEPRECATED_API

U_NAMESPACE_BEGIN

namespace message2 {

    class Environment;
    class MessageContext;
    class StaticErrors;
    class InternalValue;

    class U_I18N_API MessageFormatter : public UObject {
        // Note: This class does not currently inherit from the existing
        // `Format` class.
    public:
        MessageFormatter& operator=(MessageFormatter&&) noexcept;
        virtual ~MessageFormatter();

        UnicodeString formatToString(const MessageArguments& arguments, UErrorCode &status);

        FormattedMessage format(const MessageArguments& arguments, UErrorCode &status) const {
            (void) arguments;
            if (U_SUCCESS(status)) {
                status = U_UNSUPPORTED_ERROR;
            }
            return FormattedMessage(status);
        }

        const Locale& getLocale() const { return locale; }

        UnicodeString getPattern() const;

        const MFDataModel& getDataModel() const;

        typedef enum UMFErrorHandlingBehavior {
            U_MF_BEST_EFFORT = 0,
            U_MF_STRICT
        } UMFErrorHandlingBehavior;

        class U_I18N_API Builder : public UObject {
        private:
            friend class MessageFormatter;

            // The pattern to be parsed to generate the formatted message
            UnicodeString pattern;
            bool hasPattern = false;
            bool hasDataModel = false;
            // The data model to be used to generate the formatted message
            // Initialized either by `setDataModel()`, or by the parser
            // through a call to `setPattern()`
            MFDataModel dataModel;
            // Normalized representation of the pattern;
            // ignored if `setPattern()` wasn't called
            UnicodeString normalizedInput;
            // Errors (internal representation of parse errors)
            // Ignored if `setPattern()` wasn't called
            StaticErrors* errors;
            Locale locale;
            // Not owned
            const MFFunctionRegistry* customMFFunctionRegistry;
            // Error behavior; see comment in `MessageFormatter` class
            bool signalErrors = false;

            void clearState();
        public:
            Builder& setLocale(const Locale& locale);
            Builder& setPattern(const UnicodeString& pattern, UParseError& parseError, UErrorCode& status);
            Builder& setFunctionRegistry(const MFFunctionRegistry& functionRegistry);
            Builder& setDataModel(MFDataModel&& dataModel);
            Builder& setErrorHandlingBehavior(UMFErrorHandlingBehavior type);
            MessageFormatter build(UErrorCode& status) const;
            Builder(UErrorCode& status);
            virtual ~Builder();
        }; // class MessageFormatter::Builder

        // TODO: Shouldn't be public; only used for testing
        const UnicodeString& getNormalizedPattern() const { return normalizedInput; }

    private:
        friend class Builder;
        friend class Checker;
        friend class MessageArguments;
        friend class MessageContext;

        MessageFormatter(const MessageFormatter::Builder& builder, UErrorCode &status);

        MessageFormatter() = delete; // default constructor not implemented

        // Do not define default assignment operator
        const MessageFormatter &operator=(const MessageFormatter &) = delete;

        // Selection methods

        // Takes a vector of FormattedPlaceholders
        void resolveSelectors(MessageContext&, const Environment& env, UErrorCode&, UVector&) const;
        // Takes a vector of vectors of strings (input) and a vector of PrioritizedVariants (output)
        void filterVariants(const UVector&, UVector&, UErrorCode&) const;
        // Takes a vector of vectors of strings (input) and a vector of PrioritizedVariants (input/output)
        void sortVariants(const UVector&, UVector&, UErrorCode&) const;
        // Takes a vector of strings (input) and a vector of strings (output)
        void matchSelectorKeys(const UVector&, MessageContext&, InternalValue* rv, UVector&, UErrorCode&) const;
        // Takes a vector of FormattedPlaceholders (input),
        // and a vector of vectors of strings (output)
        void resolvePreferences(MessageContext&, UVector&, UVector&, UErrorCode&) const;

        // Formatting methods

        // Used for normalizing variable names and keys for comparison
        UnicodeString normalizeNFC(const UnicodeString&) const;
        [[nodiscard]] FormattedPlaceholder formatLiteral(const data_model::Literal&) const;
        void formatPattern(MessageContext&, const Environment&, const data_model::Pattern&, UErrorCode&, UnicodeString&) const;
        // Evaluates a function call
        // Dispatches on argument type
        [[nodiscard]] InternalValue* evalFunctionCall(FormattedPlaceholder&& argument,
                                                     MessageContext& context,
                                                     UErrorCode& status) const;
        // Dispatches on function name
        [[nodiscard]] InternalValue* evalFunctionCall(const FunctionName& functionName,
                                                     InternalValue* argument,
                                                     FunctionOptions&& options,
                                                     MessageContext& context,
                                                     UErrorCode& status) const;
        // Formats an expression that appears in a pattern or as the definition of a local variable
        [[nodiscard]] InternalValue* formatExpression(const Environment&,
                                                     const data_model::Expression&,
                                                     MessageContext&,
                                                     UErrorCode&) const;
        [[nodiscard]] FunctionOptions resolveOptions(const Environment& env, const OptionMap&, MessageContext&, UErrorCode&) const;
        [[nodiscard]] InternalValue* formatOperand(const Environment&, const data_model::Operand&, MessageContext&, UErrorCode&) const;
        [[nodiscard]] FormattedPlaceholder evalArgument(const data_model::VariableName&, MessageContext&, UErrorCode&) const;
        void formatSelectors(MessageContext& context, const Environment& env, UErrorCode &status, UnicodeString& result) const;

        // Function registry methods
        bool hasCustomMFFunctionRegistry() const {
            return (customMFFunctionRegistry != nullptr);
        }

        // Precondition: custom function registry exists
        // Note: this is non-const because the values in the MFFunctionRegistry are mutable
        // (a FormatterFactory can have mutable state)
        const MFFunctionRegistry& getCustomMFFunctionRegistry() const;

        bool isCustomFormatter(const FunctionName&) const;
        FormatterFactory* lookupFormatterFactory(const FunctionName&, UErrorCode& status) const;
        bool isBuiltInSelector(const FunctionName&) const;
        bool isBuiltInFormatter(const FunctionName&) const;
        bool isCustomSelector(const FunctionName&) const;
        const SelectorFactory* lookupSelectorFactory(MessageContext&, const FunctionName&, UErrorCode&) const;
        bool isSelector(const FunctionName& fn) const { return isBuiltInSelector(fn) || isCustomSelector(fn); }
        bool isFormatter(const FunctionName& fn) const { return isBuiltInFormatter(fn) || isCustomFormatter(fn); }
        const Formatter* lookupFormatter(const FunctionName&, UErrorCode&) const;

        Selector* getSelector(MessageContext&, const FunctionName&, UErrorCode&) const;
        Formatter* getFormatter(const FunctionName&, UErrorCode&) const;
        bool getDefaultFormatterNameByType(const UnicodeString&, FunctionName&) const;

        // Checking for resolution errors
        void checkDeclarations(MessageContext&, Environment*&, UErrorCode&) const;
        void check(MessageContext&, const Environment&, const data_model::Expression&, UErrorCode&) const;
        void check(MessageContext&, const Environment&, const data_model::Operand&, UErrorCode&) const;
        void check(MessageContext&, const Environment&, const OptionMap&, UErrorCode&) const;

        void initErrors(UErrorCode&);
        void clearErrors() const;
        void cleanup() noexcept;

        // The locale this MessageFormatter was created with
        /* const */ Locale locale;

        // Registry for built-in functions
        MFFunctionRegistry standardMFFunctionRegistry;
        // Registry for custom functions; may be null if no custom registry supplied
        // Note: this is *not* owned by the MessageFormatter object
        // The reason for this choice is to have a non-destructive MessageFormatter::Builder,
        // while also not requiring the function registry to be deeply-copyable. Making the
        // function registry copyable would impose a requirement on any implementations
        // of the FormatterFactory and SelectorFactory interfaces to implement a custom
        // clone() method, which is necessary to avoid sharing between copies of the
        // function registry (and thus double-frees)
        // Not deeply immutable (the values in the function registry are mutable,
        // as a FormatterFactory can have mutable state
        const MFFunctionRegistry* customMFFunctionRegistry;

        // Data model, representing the parsed message
        MFDataModel dataModel;

        // Normalized version of the input string (optional whitespace removed)
        UnicodeString normalizedInput;

        // Errors -- only used while parsing and checking for data model errors; then
        // the MessageContext keeps track of errors
        // Must be a raw pointer to avoid including the internal header file
        // defining StaticErrors
        // Owned by `this`
        StaticErrors* errors = nullptr;

        // Error handling behavior.
        // If true, then formatting methods set their UErrorCode arguments
        // to signal MessageFormat errors, and no useful output is returned.
        // If false, then MessageFormat errors are not signaled and the
        // formatting methods return best-effort output.
        // The default is false.
        bool signalErrors = false;

        // Used for implementing normalizeNFC()
        const Normalizer2* nfcNormalizer = nullptr;

    }; // class MessageFormatter

} // namespace message2

U_NAMESPACE_END

#endif // U_HIDE_DEPRECATED_API

#endif /* #if !UCONFIG_NO_MF2 */

#endif /* #if !UCONFIG_NO_FORMATTING */

#endif /* #if !UCONFIG_NO_NORMALIZATION */

#endif /* U_SHOW_CPLUSPLUS_API */

#endif // MESSAGEFORMAT2_H

// eof