Word Unperfect - XT-Emporium

File editor

#include "wp_control_codes.h"
#include "wp_record_parser.h"
#include <string.h>

WpDecompilerTokenClass wp_control_decompiled_token_class(uint8_t code)
{
    /* Exact bucket logic recovered from parser_process_token_stream around
     * decompiled_wp_exe.c 1000:2109:
     *   i=1 for >=C0, i=2 for 80-BF, i=3 for <=1F, i=4 for 20-7F.
     */
    if (code >= 0xC0U) {
        return WP_DECOMP_TOKEN_PACKET;
    }
    if (code >= 0x80U) {
        return WP_DECOMP_TOKEN_SINGLE_BYTE_FORMAT;
    }
    if (code > 0x1FU) {
        return WP_DECOMP_TOKEN_TEXT;
    }
    return WP_DECOMP_TOKEN_ASCII_CONTROL;
}

const char *wp_control_decompiled_token_class_name(WpDecompilerTokenClass token_class)
{
    switch (token_class) {
        case WP_DECOMP_TOKEN_ASCII_CONTROL:
            return "ASCII/control byte";
        case WP_DECOMP_TOKEN_TEXT:
            return "printable text byte";
        case WP_DECOMP_TOKEN_SINGLE_BYTE_FORMAT:
            return "single-byte format code";
        case WP_DECOMP_TOKEN_PACKET:
            return "C0+/D0+ packet code";
        default:
            return "unknown byte class";
    }
}

static bool wp_control_in_range(uint8_t code, uint8_t first, uint8_t last)
{
    return code >= first && code <= last;
}

void wp_control_decode_byte(uint8_t code, WpControlInfo *out_info)
{
    WpControlInfo info;

    if (out_info == NULL) {
        return;
    }
    memset(&info, 0, sizeof(info));
    info.code = code;
    info.token_class = wp_control_decompiled_token_class(code);
    info.label = wp_control_decompiled_token_class_name(info.token_class);
    info.marker = NULL;

    switch (info.token_class) {
        case WP_DECOMP_TOKEN_ASCII_CONTROL:
            info.effects |= WP_CONTROL_EFFECT_ASCII_CONTROL;
            info.label = "ASCII/control byte";
            if (code == '\n' || code == '\r') {
                info.effects |= WP_CONTROL_EFFECT_LINE_BREAK | WP_CONTROL_EFFECT_WHITESPACE_GATE;
                info.label = "Host newline/control";
                info.marker = "[NL]";
            } else if (code == '\t') {
                info.effects |= WP_CONTROL_EFFECT_TAB_STOP | WP_CONTROL_EFFECT_WHITESPACE_GATE;
                info.label = "Host tab/control";
                info.marker = "[Tab]";
            }
            break;

        case WP_DECOMP_TOKEN_TEXT:
            info.effects |= WP_CONTROL_EFFECT_TEXT | WP_CONTROL_EFFECT_PRINTABLE_HOST_BYTE;
            info.display_columns = 1U;
            if (code == 0x20U) {
                info.effects |= WP_CONTROL_EFFECT_WHITESPACE_GATE;
                info.label = "ASCII space / whitespace gate";
                info.marker = "[Space]";
            } else if (code == 0x7FU) {
                info.effects &= ~WP_CONTROL_EFFECT_PRINTABLE_HOST_BYTE;
                info.effects |= WP_CONTROL_EFFECT_ASCII_CONTROL;
                info.display_columns = 0U;
                info.label = "ASCII DEL/control byte";
                info.marker = "[DEL]";
            } else {
                info.label = "Text character";
            }
            break;

        case WP_DECOMP_TOKEN_SINGLE_BYTE_FORMAT:
            info.effects |= WP_CONTROL_EFFECT_SINGLE_BYTE_FORMAT;
            info.label = "Single-byte format code";
            break;

        case WP_DECOMP_TOKEN_PACKET:
            info.effects |= WP_CONTROL_EFFECT_PACKET;
            info.label = "C0+/D0+ packet code";
            break;
    }

    switch (code) {
        case 0x80U:
            info.label = "Hard return";
            info.marker = "[HRt]";
            info.effects |= WP_CONTROL_EFFECT_LINE_BREAK | WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x81U:
            info.label = "Soft return";
            info.marker = "[SRt]";
            info.effects |= WP_CONTROL_EFFECT_LINE_BREAK | WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x82U:
            info.label = "Hard page";
            info.marker = "[HPg]";
            info.effects |= WP_CONTROL_EFFECT_PAGE_BREAK | WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x83U:
            info.label = "Soft page";
            info.marker = "[SPg]";
            info.effects |= WP_CONTROL_EFFECT_PAGE_BREAK |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE |
                            WP_CONTROL_EFFECT_DIRTY_EXEMPT;
            break;
        case 0x84U:
            info.label = "Tab";
            info.marker = "[Tab]";
            info.effects |= WP_CONTROL_EFFECT_TAB_STOP | WP_CONTROL_EFFECT_WHITESPACE_GATE;
            info.tab_columns = 8U;
            break;
        case 0x85U:
            info.label = "Dormant hard return / extension-scan stop";
            info.marker = "[DormHRt]";
            info.effects |= WP_CONTROL_EFFECT_LINE_BREAK |
                            WP_CONTROL_EFFECT_DORMANT |
                            WP_CONTROL_EFFECT_EXTENSION_SCAN_STOP |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x86U:
            info.label = "Dormant soft return";
            info.marker = "[DormSRt]";
            info.effects |= WP_CONTROL_EFFECT_LINE_BREAK |
                            WP_CONTROL_EFFECT_DORMANT |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x87U:
            info.label = "Dormant hard page";
            info.marker = "[DormHPg]";
            info.effects |= WP_CONTROL_EFFECT_PAGE_BREAK |
                            WP_CONTROL_EFFECT_DORMANT |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x88U:
            info.label = "Dormant soft page";
            info.marker = "[DormSPg]";
            info.effects |= WP_CONTROL_EFFECT_PAGE_BREAK |
                            WP_CONTROL_EFFECT_DORMANT |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x8AU:
            info.label = "Text Services Manager highlight off";
            info.marker = "[HighliteOff]";
            info.effects |= WP_CONTROL_EFFECT_TSM_HIGHLIGHT;
            break;
        case 0x8CU:
            info.label = "Whitespace/gate control 8C";
            info.marker = "[Gate8C]";
            info.effects |= WP_CONTROL_EFFECT_WHITESPACE_GATE;
            break;
        case 0x8DU:
            info.label = "Indent";
            info.marker = "[Indent]";
            info.effects |= WP_CONTROL_EFFECT_INDENT | WP_CONTROL_EFFECT_TAB_STOP | WP_CONTROL_EFFECT_WHITESPACE_GATE;
            info.tab_columns = 8U;
            break;
        case 0x96U:
            info.label = "Dirty-neutral parser control 96";
            info.marker = "[Ctrl96]";
            info.effects |= WP_CONTROL_EFFECT_DIRTY_NEUTRAL;
            break;
        case 0x97U:
            info.label = "Dirty-neutral parser control 97";
            info.marker = "[Ctrl97]";
            info.effects |= WP_CONTROL_EFFECT_DIRTY_NEUTRAL;
            break;
        case 0x9AU:
            info.label = "Cancel hyphenation for word";
            info.marker = "[HyphCancel]";
            info.effects |= WP_CONTROL_EFFECT_HYPHENATION_SUPPRESS;
            break;
        case 0x9EU:
            info.label = "Hide functions on";
            info.marker = "[HideOn]";
            info.effects |= WP_CONTROL_EFFECT_HIDDEN_FUNCTION_GATE;
            break;
        case 0x9FU:
            info.label = "Hide functions off";
            info.marker = "[HideOff]";
            info.effects |= WP_CONTROL_EFFECT_HIDDEN_FUNCTION_GATE;
            break;
        case 0xA0U:
            info.label = "Scanner-normalized space";
            info.marker = "[Space]";
            info.effects |= WP_CONTROL_EFFECT_NORMALIZED_SPACE |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE |
                            WP_CONTROL_EFFECT_PRINTABLE_HOST_BYTE;
            info.display_columns = 1U;
            break;
        case 0xA9U:
            info.label = "Scanner-normalized hyphen";
            info.marker = "[Hyphen]";
            info.effects |= WP_CONTROL_EFFECT_NORMALIZED_HYPHEN |
                            WP_CONTROL_EFFECT_WHITESPACE_GATE |
                            WP_CONTROL_EFFECT_RENDER_DIRTY_GATE |
                            WP_CONTROL_EFFECT_PRINTABLE_HOST_BYTE;
            info.display_columns = 1U;
            break;
        case 0xC6U:
            info.label = "C6 fixed dirty-exempt command packet";
            info.marker = "[C6]";
            info.effects |= WP_CONTROL_EFFECT_DIRTY_EXEMPT;
            break;
        case 0xD4U:
            info.label = "D4 dirty-exempt layout-state packet";
            info.marker = "[D4]";
            info.effects |= WP_CONTROL_EFFECT_DIRTY_EXEMPT;
            break;
        default:
            break;
    }

    if (wp_control_in_range(code, 0x93U, 0x95U)) {
        info.label = "Layout whitespace/gate control 93-95";
        info.marker = "[Gate93-95]";
        info.effects |= WP_CONTROL_EFFECT_WHITESPACE_GATE;
    }
    if (wp_control_in_range(code, 0xA9U, 0xAEU)) {
        info.effects |= WP_CONTROL_EFFECT_WHITESPACE_GATE | WP_CONTROL_EFFECT_RENDER_DIRTY_GATE;
        if (code != 0xA9U) {
            info.label = "Layout render-dirty gate control A9-AE";
            info.marker = "[DirtyGate]";
        }
    } else if (wp_control_in_range(code, 0xAFU, 0xB0U)) {
        info.label = "Layout whitespace/gate control AF-B0";
        info.marker = "[GateAF-B0]";
        info.effects |= WP_CONTROL_EFFECT_WHITESPACE_GATE;
    }

    *out_info = info;
}

const char *wp_control_code_label(uint8_t code)
{
    WpControlInfo info;
    wp_control_decode_byte(code, &info);
    return info.label;
}

const char *wp_control_code_marker(uint8_t code)
{
    WpControlInfo info;
    wp_control_decode_byte(code, &info);
    return info.marker;
}

bool wp_control_has_effect(uint8_t code, unsigned long effect_mask)
{
    WpControlInfo info;
    wp_control_decode_byte(code, &info);
    return (info.effects & effect_mask) != 0UL;
}

uint8_t wp_control_display_columns(uint8_t code)
{
    WpControlInfo info;
    wp_control_decode_byte(code, &info);
    return info.display_columns;
}

void wp_control_summary_stats_clear(WpControlSummaryStats *stats)
{
    if (stats != NULL) {
        memset(stats, 0, sizeof(*stats));
    }
}

bool wp_control_summary_add_record(WpControlSummaryStats *stats, const WpRecord *rec)
{
    WpControlInfo info;

    if (stats == NULL || rec == NULL) {
        return false;
    }

    wp_control_decode_byte(rec->code, &info);
    stats->records_seen++;
    stats->bytes_consumed += rec->length;
    stats->code_histogram[rec->code]++;

    switch (info.token_class) {
        case WP_DECOMP_TOKEN_ASCII_CONTROL:
            stats->ascii_control_records++;
            break;
        case WP_DECOMP_TOKEN_TEXT:
            stats->printable_text_records++;
            break;
        case WP_DECOMP_TOKEN_SINGLE_BYTE_FORMAT:
            stats->single_byte_format_records++;
            break;
        case WP_DECOMP_TOKEN_PACKET:
            stats->packet_records++;
            break;
    }

    if ((info.effects & WP_CONTROL_EFFECT_LINE_BREAK) != 0UL) {
        stats->line_break_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_PAGE_BREAK) != 0UL) {
        stats->page_break_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_WHITESPACE_GATE) != 0UL) {
        stats->whitespace_gate_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_RENDER_DIRTY_GATE) != 0UL) {
        stats->render_dirty_gate_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_DIRTY_EXEMPT) != 0UL) {
        stats->dirty_exempt_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_DIRTY_NEUTRAL) != 0UL) {
        stats->dirty_neutral_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_EXTENSION_SCAN_STOP) != 0UL) {
        stats->extension_scan_stop_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_TSM_HIGHLIGHT) != 0UL) {
        stats->tsm_highlight_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_HYPHENATION_SUPPRESS) != 0UL) {
        stats->hyphenation_suppression_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_HIDDEN_FUNCTION_GATE) != 0UL) {
        stats->hidden_function_gate_records++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_NORMALIZED_SPACE) != 0UL) {
        stats->normalized_spaces++;
    }
    if ((info.effects & WP_CONTROL_EFFECT_NORMALIZED_HYPHEN) != 0UL) {
        stats->normalized_hyphens++;
    }
    stats->display_columns += info.display_columns;

    switch (rec->code) {
        case 0x80U: stats->hard_returns++; break;
        case 0x81U: stats->soft_returns++; break;
        case 0x82U: stats->hard_pages++; break;
        case 0x83U: stats->soft_pages++; break;
        case 0x84U: stats->tabs++; break;
        case 0x85U:
        case 0x86U: stats->dormant_returns++; break;
        case 0x87U:
        case 0x88U: stats->dormant_pages++; break;
        case 0x8DU: stats->indents++; break;
        default:
            if (rec->type == WP_CODE_SINGLE_BYTE && info.marker == NULL) {
                stats->unknown_single_byte_records++;
            }
            break;
    }

    return true;
}

bool wp_control_summary_run_stream(WpLayoutGlobals *wl, WpControlSummaryStats *stats)
{
    WpLayoutGlobals cursor;
    WpControlSummaryStats local;
    bool ok = true;

    if (wl == NULL || stats == NULL) {
        return false;
    }

    wp_control_summary_stats_clear(&local);
    cursor = *wl;

    while (cursor.record_used_bytes > 0) {
        WpRecord rec;
        int before = cursor.record_used_bytes;

        wp_parser_consume_record(&cursor, &rec);
        if (rec.length == 0U || cursor.record_used_bytes >= before) {
            wp_record_free(&rec);
            ok = false;
            break;
        }
        if (!wp_control_summary_add_record(&local, &rec)) {
            ok = false;
        }
        wp_record_free(&rec);
    }

    *stats = local;
    return ok;
}

bool wp_control_summary_run_loaded_file(WpLoadedFile *file, WpControlSummaryStats *stats)
{
    WpLayoutGlobals wl;

    if (file == NULL || stats == NULL) {
        return false;
    }
    memset(&wl, 0, sizeof(wl));
    if (!wp_file_bind_primary_stream(file, &wl, 4096U)) {
        return false;
    }
    return wp_control_summary_run_stream(&wl, stats);
}

bool wp_control_summary_run_file(const char *filename, WpControlSummaryStats *stats)
{
    WpLoadedFile file;
    bool ok;

    if (filename == NULL || stats == NULL) {
        return false;
    }
    if (!wp_file_load_body(filename, &file)) {
        return false;
    }
    ok = wp_control_summary_run_loaded_file(&file, stats);
    wp_file_free(&file);
    return ok;
}
Commit message
This repository is read-only for this account.
Repository snapshot

Current branch
main
Visibility
public
Your access
Read
Remote
None
File activity

View file history