libcaca/cucul/import.c

/*
 *  libcucul      Canvas for ultrafast compositing of Unicode letters
 *  Copyright (c) 2002-2006 Sam Hocevar <sam@zoy.org>
 *                All Rights Reserved
 *
 *  $Id$
 *
 *  This library is free software; you can redistribute it and/or
 *  modify it under the terms of the Do What The Fuck You Want To
 *  Public License, Version 2, as published by Sam Hocevar. See
 *  http://sam.zoy.org/wtfpl/COPYING for more details.
 */

/*
 *  This file contains various import functions.
 */

#include "config.h"
#include "common.h"

#if !defined(__KERNEL__)
#   if defined(HAVE_ERRNO_H)
#       include <errno.h>
#   endif
#   include <stdio.h>
#   include <stdlib.h>
#   include <string.h>
#endif

#include "cucul.h"
#include "cucul_internals.h"

/* ANSI Graphic Rendition Combination Mode */
struct ansi_grcm
{
    uint8_t fg, bg;   /* ANSI-context fg/bg */
    uint8_t efg, ebg; /* Effective (libcucul) fg/bg */
    uint8_t bold, negative, concealed;
};

static cucul_canvas_t *import_caca(void const *, unsigned int);
static cucul_canvas_t *import_text(void const *, unsigned int);
static cucul_canvas_t *import_ansi(void const *, unsigned int, int);

static void ansi_parse_grcm(cucul_canvas_t *, struct ansi_grcm *,
                            unsigned int, unsigned int const *);

/** \brief Import a buffer into a canvas
 *
 *  Import a libcucul buffer as returned by cucul_load_memory()
 *  or cucul_load_file() into an internal libcucul canvas.
 *
 *  Valid values for \c format are:
 *  - \c "": attempt to autodetect the file format.
 *  - \c "text": import ASCII text files.
 *  - \c "ansi": import ANSI files.
 *  - \c "utf8": import UTF-8 files with ANSI colour codes.
 *  - \c "caca": import native libcaca files.
 *
 *  If an error occurs, NULL is returned and \b errno is set accordingly:
 *  - \c ENOMEM Not enough memory to allocate canvas.
 *  - \c EINVAL Invalid format requested.
 *
 *  \param buffer A \e libcucul buffer containing the data to be loaded
 *         into a canvas.
 *  \param format A string describing the input format.
 *  \return A libcucul canvas, or NULL in case of error.
 */
cucul_canvas_t * cucul_import_canvas(cucul_buffer_t *buffer, char const *format)
{
    char const *buf = (char const*)buffer->data;

    if(!strcasecmp("caca", format))
        return import_caca(buffer->data, buffer->size);
    if(!strcasecmp("utf8", format))
        return import_ansi(buffer->data, buffer->size, 1);
    if(!strcasecmp("text", format))
        return import_text(buffer->data, buffer->size);
    if(!strcasecmp("ansi", format))
        return import_ansi(buffer->data, buffer->size, 0);

    /* Autodetection */
    if(!strcasecmp("", format))
    {
        unsigned int i;

        /* If 4 first letters are CACA */
        if(buffer->size >= 4 &&
           buf[0] == 'C' && buf[1] == 'A' && buf[2] == 'C' && buf[3] != 'A')
            return import_caca(buffer->data, buffer->size);

        /* If we find ESC[ argv, we guess it's an ANSI file */
        for(i = 0; i + 1 < buffer->size; i++)
            if((buf[i] == 0x1b) && (buf[i + 1] == '['))
                return import_ansi(buffer->data, buffer->size, 0);

        /* Otherwise, import it as text */
        return import_text(buffer->data, buffer->size);
    }

#if defined(HAVE_ERRNO_H)
    errno = EINVAL;
#endif
    return NULL;
}

/** \brief Get available import formats
 *
 *  Return a list of available import formats. The list is a NULL-terminated
 *  array of strings, interleaving a string containing the internal value for
 *  the import format, to be used with cucul_import_canvas(), and a string
 *  containing the natural language description for that import format.
 *
 *  This function never fails.
 *
 *  \return An array of strings.
 */
char const * const * cucul_get_import_list(void)
{
    static char const * const list[] =
        {
            "", "autodetect",
            "text", "plain text",
            "caca", "native libcaca format",
            "ansi", "ANSI coloured text",
            NULL, NULL
        };

    return list;
}

/*
 * XXX: the following functions are local.
 */

static cucul_canvas_t *import_caca(void const *data, unsigned int size)
{
    cucul_canvas_t *cv;
    uint8_t const *buf = (uint8_t const *)data;
    unsigned int width, height, n;

    if(size < 16)
        goto invalid_caca;

    if(buf[0] != 'C' || buf[1] != 'A' || buf[2] != 'C' || buf[3] != 'A')
        goto invalid_caca;

    if(buf[4] != 'C' || buf[5] != 'A' || buf[6] != 'N' || buf[7] != 'V')
        goto invalid_caca;

    width = ((uint32_t)buf[8] << 24) | ((uint32_t)buf[9] << 16)
        | ((uint32_t)buf[10] << 8) | (uint32_t)buf[11];
    height = ((uint32_t)buf[12] << 24) | ((uint32_t)buf[13] << 16)
        | ((uint32_t)buf[14] << 8) | (uint32_t)buf[15];

    if(size != 16 + width * height * 8)
        goto invalid_caca;

    cv = cucul_create_canvas(width, height);

    if(!cv)
    {
#if defined(HAVE_ERRNO_H)
        errno = ENOMEM;
#endif
        return NULL;
    }

    for(n = height * width; n--; )
    {
        cv->chars[n] = ((uint32_t)buf[16 + 0 + 8 * n] << 24)
            | ((uint32_t)buf[16 + 1 + 8 * n] << 16)
            | ((uint32_t)buf[16 + 2 + 8 * n] << 8)
            | (uint32_t)buf[16 + 3 + 8 * n];
        cv->attr[n] = ((uint32_t)buf[16 + 4 + 8 * n] << 24)
            | ((uint32_t)buf[16 + 5 + 8 * n] << 16)
            | ((uint32_t)buf[16 + 6 + 8 * n] << 8)
            | (uint32_t)buf[16 + 7 + 8 * n];
    }

    return cv;

invalid_caca:
#if defined(HAVE_ERRNO_H)
    errno = EINVAL;
#endif
    return NULL;
}

static cucul_canvas_t *import_text(void const *data, unsigned int size)
{
    cucul_canvas_t *cv;
    char const *text = (char const *)data;
    unsigned int width = 0, height = 0, x = 0, y = 0, i;

    cv = cucul_create_canvas(width, height);
    if(!cv)
    {
#if defined(HAVE_ERRNO_H)
        errno = ENOMEM;
#endif
        return NULL;
    }

    cucul_set_color(cv, CUCUL_COLOR_DEFAULT, CUCUL_COLOR_TRANSPARENT);

    for(i = 0; i < size; i++)
    {
        unsigned char ch = *text++;

        if(ch == '\r')
            continue;

        if(ch == '\n')
        {
            x = 0;
            y++;
            continue;
        }

        if(x >= width || y >= height)
        {
            if(x >= width)
                width = x + 1;

            if(y >= height)
                height = y + 1;

            cucul_set_canvas_size(cv, width, height);
        }

        cucul_putchar(cv, x, y, ch);
        x++;
    }

    if(y > height)
        cucul_set_canvas_size(cv, width, height = y);

    return cv;
}

static cucul_canvas_t *import_ansi(void const *data, unsigned int size,
                                   int utf8)
{
    struct ansi_grcm grcm;
    unsigned char const *buffer = (unsigned char const*)data;
    cucul_canvas_t *cv;
    unsigned int i, j, skip, dummy = 0;
    unsigned int width = 0, height = 0, wch = 1;
    unsigned long int ch;
    int x = 0, y = 0, save_x = 0, save_y = 0;

    cv = cucul_create_canvas(width, height);
    if(!cv)
    {
#if defined(HAVE_ERRNO_H)
        errno = ENOMEM;
#endif
        return NULL;
    }

    ansi_parse_grcm(cv, &grcm, 1, &dummy);

    for(i = 0; i < size; i += skip)
    {
        skip = 1;

        /* Wrap long lines */
        if((unsigned int)x >= 80)
        {
            x = 0;
            y++;
        }

        if(buffer[i] == '\x1a' && size - i >= 8
           && !memcmp(buffer + i + 1, "SAUCE00", 7))
            break; /* End before SAUCE data */

        if(buffer[i] == '\r')
            continue; /* DOS sucks */

        if(buffer[i] == '\n')
        {
            x = 0;
            y++;
            continue;
        }

        /* Interpret escape commands, as per Standard ECMA-48 "Control
         * Functions for Coded Character Sets", 5.4. Control sequences. */
        if(buffer[i] == '\x1b' && buffer[i + 1] == '[')
        {
            unsigned int argc = 0, argv[101];
            unsigned int param, inter, final;

        /* Compute offsets to parameter bytes, intermediate bytes and
         * to the final byte. Only the final byte is mandatory, there
         * can be zero of the others.
         * 0  param=2             inter                 final           final+1
         * +-----+------------------+---------------------+-----------------+
         * | CSI | parameter bytes  | intermediate bytes  |   final byte    |
         * |     |   0x30 - 0x3f    |    0x20 - 0x2f      |   0x40 - 0x7e   |
         * | ^[[ | 0123456789:;<=>? | SPC !"#$%&'()*+,-./ | azAZ@[\]^_`{|}~ |
         * +-----+------------------+---------------------+-----------------+
         */
            param = 2;

            for(inter = param; i + inter < size; inter++)
                if(buffer[i + inter] < 0x30 || buffer[i + inter] > 0x3f)
                    break;

            for(final = inter; i + final < size; final++)
                if(buffer[i + final] < 0x20 || buffer[i + final] > 0x2f)
                    break;

            if(buffer[i + final] < 0x40 || buffer[i + final] > 0x7e)
                break; /* Invalid Final Byte */

            skip += final;

            /* Sanity checks */
            if(param < inter && buffer[i + param] >= 0x3c)
            {
                fprintf(stderr, "private sequence \"^[[%.*s\"\n",
                        final - param + 1, buffer + i + param);
                continue; /* Private sequence, skip it entirely */
            }

            if(final - param > 100)
                continue; /* Suspiciously long sequence, skip it */

            /* Parse parameter bytes as per ECMA-48 5.4.2: Parameter string
             * format */
            if(param < inter)
            {
                argv[0] = 0;
                for(j = param; j < inter; j++)
                {
                    if(buffer[i + j] == ';')
                        argv[++argc] = 0;
                    else if(buffer[i + j] >= '0' && buffer[i + j] <= '9')
                        argv[argc] = 10 * argv[argc] + (buffer[i + j] - '0');
                }
                argc++;
            }

            /* Interpret final byte. The code representations are given in
             * ECMA-48 5.4: Control sequences, and the code definitions are
             * given in ECMA-48 8.3: Definition of control functions. */
            switch(buffer[i + final])
            {
            case 'f': /* CUP - Cursor Position */
            case 'H': /* HVP - Character And Line Position */
                x = (argc > 1 && argv[1] > 0) ? argv[1] - 1 : 0;
                y = (argc > 0 && argv[0] > 0) ? argv[0] - 1 : 0;
                break;
            case 'A': /* CUU - Cursor Up */
                y -= argc ? argv[0] : 1;
                if(y < 0)
                    y = 0;
                break;
            case 'B': /* CUD - Cursor Down */
                y += argc ? argv[0] : 1;
                break;
            case 'C': /* CUF - Cursor Right */
                x += argc ? argv[0] : 1;
                break;
            case 'D': /* CUB - Cursor Left */
                x -= argc ? argv[0] : 1;
                if(x < 0)
                    x = 0;
                break;
            case 's': /* Private (save cursor position) */
                save_x = x;
                save_y = y;
                break;
            case 'u': /* Private (reload cursor position) */
                x = save_x;
                y = save_y;
                break;
            case 'J': /* ED - Erase In Page */
                if(argv[0] == 2)
                    x = y = 0;
                break;
            case 'K': /* EL - Erase In Line */
                if(width < 80)
                    cucul_set_color(cv, CUCUL_COLOR_DEFAULT,
                                        CUCUL_COLOR_TRANSPARENT);
                    cucul_set_canvas_size(cv, width = 80, height);
                for(j = x; j < 80; j++)
                    cucul_putchar(cv, j, y, ' ');
                x = 80;
                break;
            case 'm': /* SGR - Select Graphic Rendition */
                ansi_parse_grcm(cv, &grcm, argc, argv);
                break;
            default:
                fprintf(stderr, "unknown command %c\n", buffer[i + final]);
                break;
            }

            continue;
        }

        /* Get the character we’re going to paste */
        if(utf8)
        {
            unsigned int bytes;
            ch = cucul_utf8_to_utf32((char const *)(buffer + i), &bytes);
            wch = cucul_utf32_is_fullwidth(ch) ? 2 : 1;
            skip += bytes - 1;
        }
        else
        {
            ch = cucul_cp437_to_utf32(buffer[i]);
        }

        /* Make sure the canvas is big enough. */
        if((unsigned int)x + wch > width)
        {
            cucul_set_color(cv, CUCUL_COLOR_DEFAULT, CUCUL_COLOR_TRANSPARENT);
            cucul_set_canvas_size(cv, width = x + wch, height);
        }

        if((unsigned int)y >= height)
        {
            cucul_set_color(cv, CUCUL_COLOR_DEFAULT, CUCUL_COLOR_TRANSPARENT);
            cucul_set_canvas_size(cv, width, height = y + 1);
        }

        /* Now paste our character */
        cucul_set_color(cv, grcm.efg, grcm.ebg);
        cucul_putchar(cv, x, y, ch);
        x += wch;
    }

    if((unsigned int)y > height)
    {
        cucul_set_color(cv, CUCUL_COLOR_DEFAULT, CUCUL_COLOR_TRANSPARENT);
        cucul_set_canvas_size(cv, width, height = y);
    }

    return cv;
}

/* XXX : ANSI loader helper */

static void ansi_parse_grcm(cucul_canvas_t *cv, struct ansi_grcm *g,
                            unsigned int argc, unsigned int const *argv)
{
    static uint8_t const ansi2cucul[] =
    {
        CUCUL_COLOR_BLACK, CUCUL_COLOR_RED,
        CUCUL_COLOR_GREEN, CUCUL_COLOR_BROWN,
        CUCUL_COLOR_BLUE, CUCUL_COLOR_MAGENTA,
        CUCUL_COLOR_CYAN, CUCUL_COLOR_LIGHTGRAY
    };

    unsigned int j;

    for(j = 0; j < argc; j++)
    {
        /* Defined in ECMA-48 8.3.117: SGR - SELECT GRAPHIC RENDITION */
        if(argv[j] >= 30 && argv[j] <= 37)
            g->fg = ansi2cucul[argv[j] - 30];
        else if(argv[j] >= 40 && argv[j] <= 47)
            g->bg = ansi2cucul[argv[j] - 40];
        else if(argv[j] >= 90 && argv[j] <= 97)
            g->fg = ansi2cucul[argv[j] - 90] + 8;
        else if(argv[j] >= 100 && argv[j] <= 107)
            g->bg = ansi2cucul[argv[j] - 100] + 8;
        else switch(argv[j])
        {
        case 0: /* default rendition */
            g->fg = CUCUL_COLOR_DEFAULT;
            g->bg = CUCUL_COLOR_TRANSPARENT;
            g->bold = g->negative = g->concealed = 0;
            break;
        case 1: /* bold or increased intensity */
            g->bold = 1;
            break;
        case 4: /* singly underlined */
            break;
        case 5: /* slowly blinking (less then 150 per minute) */
            break;
        case 7: /* negative image */
            g->negative = 1;
            break;
        case 8: /* concealed characters */
            g->concealed = 1;
            break;
        case 22: /* normal colour or normal intensity (neither bold nor faint) */
            g->bold = 0;
            break;
        case 28: /* revealed characters */
            g->concealed = 0;
            break;
        case 39: /* default display colour (implementation-defined) */
            g->fg = CUCUL_COLOR_DEFAULT;
            break;
        case 49: /* default background colour (implementation-defined) */
            g->bg = CUCUL_COLOR_TRANSPARENT;
            break;
        default:
            fprintf(stderr, "unknown sgr %i\n", argv[j]);
            break;
        }
    }

    if(g->concealed)
    {
        g->efg = g->ebg = CUCUL_COLOR_TRANSPARENT;
    }
    else
    {
        g->efg = g->negative ? g->bg : g->fg;
        g->ebg = g->negative ? g->fg : g->bg;

        if(g->bold)
        {
            if(g->efg < 8)
                g->efg += 8;
            else if(g->efg == CUCUL_COLOR_DEFAULT)
                g->efg = CUCUL_COLOR_WHITE;
        }
    }
}