To compile string constants and comments.

§1. We now take a wide Unicode string and compile a double-quoted I6 string constant which will print out the same content, or initialise a string array.

A subtly different set of escape characters are used here. It's all trickier than it might be because of some unfortunate design choices in I6. For instance, an @ character can't be escaped as @{40} because that would be expanded too early in the string-reading process, so that it would still be read as an escape-character @, not a literal one. We therefore have to use @@64. This unfortunately goes wrong, however, when the immediately following character is a decimal digit, because then we might construct, e.g., @@647, and I6 will throw an error: no valid character has ZSCII code 647. We therefore use the @{...} escape to represent any digit following an @@ escape. Since no digit is itself an escape character with side-effects, a vicious circle is avoided.

Initialisation of I6 string arrays has different conventions again: these behave more like dictionary words and the @@ escape is not allowed.

define CT_CAPITALISE 1  capitalise first letter of text
define CT_EXPAND_APOSTROPHES 2  sometimes regard ' as "
define CT_RECOGNISE_APOSTROPHE_SUBSTITUTION 4  recognise ['] as a literal '
define CT_RECOGNISE_UNICODE_SUBSTITUTION 8  recognise [unicode N] as a literal char
define CT_DEQUOTE 16  ignore initial and terminal " pair, e.g., render "fish" as fish
define CT_FOR_ARRAY 32  force use of @{xx} form not @@ddd
define CT_BOX_QUOTATION 64  format line breaks into text for an I6 box statement
define CT_RAW 128  ignore everything except capitalisation and dequoting
define CT_I6 256  ignore everything except capitalisation and dequoting
define CT_EXPAND_APOSTROPHES_RAWLY 512  sometimes regard ' as "

void CompiledText::from_wide_string?Usage of CompiledText::from_wide_string:
§5
Text Literals - §13.1
Runtime Support for Kinds - §27.3
Literal Patterns at Run Time - §2.3.1
Naming - §2(OUTPUT_STREAM, wchar_t *p, int options) {
    int i, from = 0, to = Wide::len(p), esc_digit = FALSE;
    if ((options & CT_DEQUOTE) && (p[0] == '"') && (p[to-1] == '"')) {
        from++; to--;
    }
    if (options & CT_RAW) {
        for (i=from; i<to; i++) {
            if ((i == from) && (options & CT_CAPITALISE))
                PUT(Characters::toupper(p[i]));
            else
                PUT(p[i]);
        }
    } else {
        for (i=from; i<to; i++) {
            switch(p[i]) {
                case '\n':
                    if (options & CT_BOX_QUOTATION) WRITE("\n");
                    else WRITE(" ");
                    break;
                case '\t':
                    if (options & CT_BOX_QUOTATION) WRITE("\n");
                    else WRITE(" ");
                    break;
                case NEWLINE_IN_STRING:
                    if (options & CT_BOX_QUOTATION) WRITE("\n");
                    else WRITE("^"); break;
                case '"': WRITE("~"); break;
                case '@':
                    if (options & CT_FOR_ARRAY) WRITE("@{40}");
                    else { WRITE("@@64"); esc_digit = TRUE; continue; }
                    break;
                case '^':
                    if (options & CT_BOX_QUOTATION) WRITE("^");
                    else if (options & CT_FOR_ARRAY) WRITE("@{5E}");
                    else { WRITE("@@94"); esc_digit = TRUE; continue; }
                    break;
                case '~':
                    if (options & CT_FOR_ARRAY) WRITE("@{7E}");
                    else { WRITE("@@126"); esc_digit = TRUE; continue; }
                    break;
                case '\\': WRITE("@{5C}"); break;
                case '\'':
                    if (options & CT_EXPAND_APOSTROPHES)
                        Apply Inform 7's convention on interpreting single quotation marks1.1
                    else WRITE("'");
                    break;
                case '[':
                    if ((options & CT_RECOGNISE_APOSTROPHE_SUBSTITUTION) &&
                        (p[i+1] == '\'') && (p[i+2] == ']')) { i += 2; WRITE("'"); }
                    else if (options & CT_RECOGNISE_UNICODE_SUBSTITUTION) {
                        int n = CompiledText::expand_unisub(OUT, p, i);
                        if (n == -1) WRITE("["); else i = n;
                    } else WRITE("[");
                    break;
                default:
                    if ((i==from) && (options & CT_CAPITALISE))
                        WRITE("%c", Characters::toupper(p[i]));
                    else if ((esc_digit) && (Characters::isdigit(p[i])))
                        WRITE("@{%02x}", p[i]);
                    else
                        WRITE("%c", p[i]);
                    break;
            }
            esc_digit = FALSE;
        }
    }
}

void CompiledText::from_wide_string_for_emission(OUTPUT_STREAM, wchar_t *p) {
    int i, from = 0, to = Wide::len(p);
    if ((p[0] == '"') && (p[to-1] == '"')) {
        from++; to--;
    }
    for (i=from; i<to; i++) {
        switch(p[i]) {
            case '\n':
                WRITE(" ");
                break;
            case '\t':
                WRITE(" ");
                break;
            case NEWLINE_IN_STRING:
                WRITE("\n");
                break;
            case '\'':
                Rawly apply Inform 7's convention on interpreting single quotation marks1.2
                break;
            default:
                WRITE("%c", p[i]);
                break;
        }
    }
}

void CompiledText::bq_from_wide_string?Usage of CompiledText::bq_from_wide_string:
Text Literals - §13.2(OUTPUT_STREAM, wchar_t *p) {
    int i, from = 0, to = Wide::len(p), esc_digit = FALSE;
    if ((p[0] == '"') && (p[to-1] == '"')) {
        from++; to--;
    }
    for (i=from; i<to; i++) {
        switch(p[i]) {
			case '\n':
                WRITE("\n");
                break;
            case '\t':
                WRITE("\n");
                break;
            case NEWLINE_IN_STRING:
                WRITE("\n");
                break;
            case '"': WRITE("~"); break;
            case '@':
                WRITE("@@64"); esc_digit = TRUE; continue;
            case '^':
                WRITE("^");
                break;
            case '~':
                WRITE("@@126"); esc_digit = TRUE; continue;
            case '\\': WRITE("@{5C}"); break;
            case '\'':
                WRITE("'");
                break;
*/
            case '[': {
                int n = CompiledText::expand_unisub(OUT, p, i);
                if (n == -1) WRITE("["); else i = n;
                break;
            }
            default:
                if ((esc_digit) && (Characters::isdigit(p[i])))
                    WRITE("@{%02x}", p[i]);
                else
                    WRITE("%c", p[i]);
                break;
        }
        esc_digit = FALSE;
    }
}

§1.1. This is where Inform's convention on expanding single quotation marks to double, provided they appear to be quoting text rather than used as apostrophes in contractions such as "don't", is implemented. Note the exceptional case.

Apply Inform 7's convention on interpreting single quotation marks1.1 =

    if ((i==from) && (p[i+1] == 's') && ((to == 3) || (p[i+2] == ' ')))
        WRITE("'");  allow apostrophe if appending e.g. "'s nose" to "Jane"
    else if ((i>0) && (p[i+1]) &&
        (CompiledText::alphabetic(p[i-1])) &&
        (CompiledText::alphabetic(p[i+1])))
        WRITE("'");  allow apostrophe sandwiched between two letters
    else {
        WRITE("~");  and otherwise convert to double-quote
    }

• This code is used in §1.

§1.2. Rawly apply Inform 7's convention on interpreting single quotation marks1.2 =

    if ((i==from) && (p[i+1] == 's') && ((to == 3) || (p[i+2] == ' ')))
        WRITE("'");  allow apostrophe if appending e.g. "'s nose" to "Jane"
    else if ((i>0) && (p[i+1]) &&
        (CompiledText::alphabetic(p[i-1])) &&
        (CompiledText::alphabetic(p[i+1])))
        WRITE("'");  allow apostrophe sandwiched between two letters
    else {
        WRITE("\"");  and otherwise convert to double-quote
    }

• This code is used in §1.

§2.

void CompiledText::from_stream?Usage of CompiledText::from_stream:
Test Scripts - §3
Literal Patterns at Run Time - §2.3.2(OUTPUT_STREAM, text_stream *p, int options) {
    int i, from = 0, to = Str::len(p), esc_digit = FALSE;
    if ((options & CT_DEQUOTE) && (Str::get_at(p, from) == '"') && (Str::get_at(p, to-1) == '"')) {
        from++; to--;
    }
    if (options & CT_RAW) {
        for (i=from; i<to; i++) {
            wchar_t c = Str::get_at(p, i);
            if ((i == from) && (options & CT_CAPITALISE))
                WRITE("%c", (int) Characters::toupper(c));
            else
                WRITE("%c", (int) c);
        }
    } else {
        for (i=from; i<to; i++) {
            wchar_t c = Str::get_at(p, i);
            switch(c) {
                case '\n':
                    if (options & CT_BOX_QUOTATION) WRITE("\"\n\"");
                    else WRITE(" ");
                    break;
                case '\t':
                    if (options & CT_BOX_QUOTATION) WRITE("\"\n\"");
                    else WRITE(" ");
                    break;
                case NEWLINE_IN_STRING:
                    if (options & CT_BOX_QUOTATION) WRITE("\"\n\"");
                    else WRITE("^"); break;
                case '"':
                    if (options & CT_I6) WRITE("~");
                    else WRITE("%c", (int) c);
                    break;
                case '@':
                    if (options & CT_I6) {
                        if (options & CT_FOR_ARRAY) WRITE("@{40}");
                        else { WRITE("@@64"); esc_digit = TRUE; continue; }
                    } else WRITE("%c", (int) c);
                    break;
                case '^':
                    if (options & CT_I6) {
                        if (options & CT_BOX_QUOTATION) WRITE("\"\n\"");
                        else if (options & CT_FOR_ARRAY) WRITE("@{5E}");
                        else { WRITE("@@94"); esc_digit = TRUE; continue; }
                    } else WRITE("%c", (int) c);
                    break;
                case '~':
                    if (options & CT_I6) {
                        if (options & CT_FOR_ARRAY) WRITE("@{7E}");
                        else { WRITE("@@126"); esc_digit = TRUE; continue; }
                    } else WRITE("%c", (int) c);
                    break;
                case '\\':
                    if (options & CT_I6) {
                        WRITE("@{5C}");
                    } else WRITE("%c", (int) c);
                    break;
                case '\'':
                    if (options & CT_EXPAND_APOSTROPHES)
                        Apply Inform 7's convention on interpreting single quotation marks, stream version2.1
                    else WRITE("'");
                    break;
                case '[':
                    if ((options & CT_RECOGNISE_APOSTROPHE_SUBSTITUTION) &&
                        (Str::get_at(p, i+1) == '\'') && (Str::get_at(p, i+2) == ']')) {
                            i += 2; WRITE("'");
                    } else if (options & CT_RECOGNISE_UNICODE_SUBSTITUTION) {
                        int n = CompiledText::expand_unisub_S(OUT, p, i);
                        if (n == -1) WRITE("["); else i = n;
                    } else WRITE("[");
                    break;
                default:
                    if ((i==from) && (options & CT_CAPITALISE))
                        WRITE("%c", (int) Characters::toupper(c));
                    else if ((esc_digit) && (Characters::isdigit(c)))
                        WRITE("@{%02x}", (int) c);
                    else
                        WRITE("%c", (int) c);
                    break;
            }
            esc_digit = FALSE;
        }
    }
}

§2.1. This is where Inform's convention on expanding single quotation marks to double, provided they appear to be quoting text rather than used as apostrophes in contractions such as "don't", is implemented. Note the exceptional case.

Apply Inform 7's convention on interpreting single quotation marks, stream version2.1 =

    if ((i==from) && (Str::get_at(p, i+1) == 's') && ((to == 3) || (Str::get_at(p, i+2) == ' ')))
        WRITE("'");  allow apostrophe if appending e.g. "'s nose" to "Jane"
    else if ((i>0) && (Str::get_at(p, i+1)) &&
        (CompiledText::alphabetic(Str::get_at(p, i-1))) &&
        (CompiledText::alphabetic(Str::get_at(p, i+1))))
        WRITE("'");  allow apostrophe sandwiched between two letters
    else {
        if (options & CT_I6) WRITE("~");  and otherwise convert to double-quote
        else WRITE("\"");
    }

• This code is used in §2.

§3. Where we must tiresomely use this:

int CompiledText::alphabetic?Usage of CompiledText::alphabetic:
§1.1, §1.2, §2.1(int letter) {
    return Characters::isalpha((wchar_t) Characters::remove_accent(letter));
}

§4. This looks for "[unicode 8212]" and turns it into an em-dash, for example.

define MAX_UNISUB_LENGTH 128

int CompiledText::expand_unisub?Usage of CompiledText::expand_unisub:
§1(OUTPUT_STREAM, wchar_t *p, int i) {
    if ((p[i+1] == 'u') && (p[i+2] == 'n') && (p[i+3] == 'i') && (p[i+4] == 'c')
        && (p[i+5] == 'o') && (p[i+6] == 'd') && (p[i+7] == 'e') && (p[i+8] == ' ')) {
        TEMPORARY_TEXT(substitution_buffer)
        int j = i+9;
        while (p[j] == ' ') j++;
        while ((p[j]) && (p[j] != ']'))
            PUT_TO(substitution_buffer, p[j++]);
        if (p[j] == ']') {
            wording XW = Feeds::feed_text(substitution_buffer);
            if (<s-unicode-character>(XW) == FALSE) return -1;
            PUT(Rvalues::to_Unicode_point(<<rp>>));
            return j;
        } else return -1;
    } else return -1;
}

int CompiledText::expand_unisub_S?Usage of CompiledText::expand_unisub_S:
§2(OUTPUT_STREAM, text_stream *p, int i) {
    if (Str::includes_wide_string_at(p, L"unicode ", i+1)) {
        TEMPORARY_TEXT(substitution_buffer)
        int j = i+9;
        while (Str::get_at(p, j) == ' ') j++;
        while ((Str::get_at(p, j)) && (Str::get_at(p, j) != ']'))
            PUT_TO(substitution_buffer, Str::get_at(p, j++));
        if (Str::get_at(p, j) == ']') {
            wording XW = Feeds::feed_text(substitution_buffer);
            if (<s-unicode-character>(XW) == FALSE) return -1;
            PUT(Rvalues::to_Unicode_point(<<rp>>));
            return j;
        } else return -1;
    } else return -1;
}

§5. A convenient package for the above:

void CompiledText::from_text_with_options?Usage of CompiledText::from_text_with_options:
§6(OUTPUT_STREAM, wording W, int opts, int raw) {
    LOOP_THROUGH_WORDING(j, W) {
        wchar_t *p;
        if (raw) p = Lexer::word_raw_text(j); else p = Lexer::word_text(j);
        CompiledText::from_wide_string(OUT, p, opts);
        if (j<Wordings::last_wn(W)) WRITE(" ");
    }
}

§6. Whence:

void CompiledText::comment(OUTPUT_STREAM, wording W) {
    CompiledText::from_text_with_options(OUT, W, 0, FALSE);
}

void CompiledText::from_text?Usage of CompiledText::from_text:
Rules - §5.2.1
Relations at Run Time - §4.1.5(OUTPUT_STREAM, wording W) {
    CompiledText::from_text_with_options(OUT, W, 0, TRUE);
}

§7. This doesn't really belong here, or anywhere else, really:

void CompiledText::divider_comment(void) {
    Produce::comment(Emit::tree(), I"----------------------------------------------------------------------------------------------------");
}