path: root/devices/vector/gdevpdtc.c

/* Copyright (C) 2001-2019 Artifex Software, Inc.
   All Rights Reserved.

   This software is provided AS-IS with no warranty, either express or
   implied.

   This software is distributed under license and may not be copied,
   modified or distributed except as expressly authorized under the terms
   of the license contained in the file LICENSE in this distribution.

   Refer to licensing information at http://www.artifex.com or contact
   Artifex Software, Inc.,  1305 Grant Avenue - Suite 200, Novato,
   CA 94945, U.S.A., +1(415)492-9861, for further information.
*/


/* Composite and CID-based text processing for pdfwrite. */
#include "memory_.h"
#include "gx.h"
#include "gserrors.h"
#include "gxfcmap.h"
#include "gxfont.h"
#include "gxfont0.h"
#include "gxfont0c.h"
#include "gzpath.h"
#include "gxchar.h"
#include "gdevpsf.h"
#include "gdevpdfx.h"
#include "gdevpdtx.h"
#include "gdevpdtd.h"
#include "gdevpdtf.h"
#include "gdevpdts.h"
#include "gdevpdtt.h"

#include "gximage.h"
#include "gxcpath.h"
/* ---------------- Non-CMap-based composite font ---------------- */

/*
 * Process a text string in a composite font with FMapType != 9 (CMap).
 */
int
process_composite_text(gs_text_enum_t *pte, void *vbuf, uint bsize)
{
    byte *const buf = vbuf;
    pdf_text_enum_t *const penum = (pdf_text_enum_t *)pte;
    int code = 0;
    gs_string str;
    pdf_text_process_state_t text_state;
    pdf_text_enum_t curr, prev, out;
    gs_point total_width;
    const gs_matrix *psmat = 0;
    gs_font *prev_font = 0;
    gs_char chr, char_code = 0x0badf00d, space_char = GS_NO_CHAR;
    int buf_index = 0;
    bool return_width = (penum->text.operation & TEXT_RETURN_WIDTH);

    str.data = buf;
    if (return_width) {
        code = gx_path_current_point(penum->path, &penum->origin);
        if (code < 0)
            return code;
    }
    if (pte->text.operation &
        (TEXT_FROM_ANY - (TEXT_FROM_STRING | TEXT_FROM_BYTES))
        )
        return_error(gs_error_rangecheck);
    if (pte->text.operation & TEXT_INTERVENE) {
        /* Not implemented. (PostScript doesn't even allow this case.) */
        return_error(gs_error_rangecheck);
    }
    total_width.x = total_width.y = 0;
    curr = *penum;
    prev = curr;
    out = curr;
    out.current_font = 0;
    /* Scan runs of characters in the same leaf font. */
    for ( ; ; ) {
        int font_code;
        gs_font *new_font = 0;

        gs_text_enum_copy_dynamic((gs_text_enum_t *)&out,
                                  (gs_text_enum_t *)&curr, false);
        for (;;) {
            gs_glyph glyph;

            gs_text_enum_copy_dynamic((gs_text_enum_t *)&prev,
                                      (gs_text_enum_t *)&curr, false);
            font_code = pte->orig_font->procs.next_char_glyph
                ((gs_text_enum_t *)&curr, &chr, &glyph);
            /*
             * We check for a font change by comparing the current
             * font, rather than testing the return code, because
             * it makes the control structure a little simpler.
             */
            switch (font_code) {
            case 0:		/* no font change */
            case 1:		/* font change */
                curr.returned.current_char = chr;
                char_code = gx_current_char((gs_text_enum_t *)&curr);
                new_font = curr.fstack.items[curr.fstack.depth].font;
                if (new_font != prev_font)
                    break;
                if (chr != (byte)chr)	/* probably can't happen */
                    return_error(gs_error_rangecheck);
                if (buf_index >= bsize)
                    return_error(gs_error_unregistered); /* Must not happen. */
                buf[buf_index] = (byte)chr;
                buf_index++;
                prev_font = new_font;
                psmat = &curr.fstack.items[curr.fstack.depth - 1].font->FontMatrix;
                if ((pte->text.operation & TEXT_ADD_TO_SPACE_WIDTH) &&
                        pte->text.space.s_char == char_code)
                    space_char = chr;
                continue;
            case 2:		/* end of string */
                break;
            default:	/* error */
                return font_code;
            }
            break;
        }
        str.size = buf_index;
        if (buf_index) {
            /* buf_index == 0 is only possible the very first time. */
            /*
             * The FontMatrix of leaf descendant fonts is not updated
             * by scalefont.  Compute the effective FontMatrix now.
             */
            gs_matrix fmat;

            /* set up the base font : */
            out.fstack.depth = 0;
            out.fstack.items[out.fstack.depth].font = out.current_font = prev_font;
            pte->current_font = prev_font;

            /* Provide the decoded space character : */
            out.text.space.s_char = space_char;

            gs_matrix_multiply(&prev_font->FontMatrix, psmat, &fmat);
            out.index = 0; /* Note : we don't reset out.xy_index here. */
            code = pdf_process_string_aux(&out, &str, NULL, &fmat, &text_state);
            if (code < 0) {
                if (code == gs_error_undefined && new_font && new_font->FontType == ft_encrypted2)
                /* Caused by trying to make a CFF font resource for ps2write, which doesn't support CFF, abort now! */
		  return_error(gs_error_invalidfont);
                return code;
            }
            curr.xy_index = out.xy_index; /* pdf_encode_process_string advanced it. */
            if (out.index < str.size) {
                gs_glyph glyph;

                /* Advance *pte exactly for out.index chars,
                   because above we stored bytes into buf. */
                while (out.index--)
                    pte->orig_font->procs.next_char_glyph(pte, &chr, &glyph);
                font_code = 2; /* force exiting the loop */
            } else {
                /* advance *pte past the current substring */
                gs_text_enum_copy_dynamic(pte, (gs_text_enum_t *)&prev, true);
            }
            pte->xy_index = out.xy_index;
            if (return_width) {
                /* This is silly, but its a consequence of the way pdf_process_string
                 * works. When we have TEXT_DO_NONE (stringwidth) we add the width of the
                 * glyph(s) to the enumerator 'returned.total_width' so we keep track
                 * of the total width as we go. However when we are returning the width
                 * but its NOT for a stringwidth, we set the enumerator 'retuerned'
                 * value to just the width of the glyph(s) processed. So when we are *not*
                 * handling a stringwidth we need to keep track of the total width
                 * ourselves. I'd have preferred to alter pdf_process_string, but that
                 * is used in many other places, and those places rely on this behaviour.
                 */
                if (pte->text.operation & TEXT_DO_NONE) {
                    pte->returned.total_width.x = total_width.x = out.returned.total_width.x;
                    pte->returned.total_width.y = total_width.y = out.returned.total_width.y;
                } else {
                    pte->returned.total_width.x = total_width.x +=
                        out.returned.total_width.x;
                    pte->returned.total_width.y = total_width.y +=
                        out.returned.total_width.y;
                }
            }
            pdf_text_release_cgp(penum);
        }
        if (font_code == 2)
            break;
        buf[0] = (byte)chr;
        buf_index = 1;
        space_char = ((pte->text.operation & TEXT_ADD_TO_SPACE_WIDTH) &&
                      pte->text.space.s_char == char_code ? chr : ~0);
        psmat = &curr.fstack.items[curr.fstack.depth - 1].font->FontMatrix;
        prev_font = new_font;
    }
    if (!return_width)
        return 0;
    return pdf_shift_text_currentpoint(penum, &total_width);
}

/* ---------------- CMap-based composite font ---------------- */

/*
 * Process a text string in a composite font with FMapType == 9 (CMap).
 */
static const char *const standard_cmap_names[] = {
    /* The following were added in PDF 1.5. */

    "UniGB-UTF16-H", "UniGB-UTF16-V",

    "GBKp-EUC-H", "GBKp-EUC-V",
    "HKscs-B5-H", "HKscs-B5-V",
    "UniCNS-UTF16-H", "UniCNS-UTF16-V",
    "UniJIS-UTF16-H", "UniJIS-UTF16-V",
    "UniKS-UTF16-H", "UniKS-UTF16-V",
#define END_PDF15_CMAP_NAMES_INDEX 12
    /* The following were added in PDF 1.4. */
    "GBKp-EUC-H", "GBKp-EUC-V",
    "GBK2K-H", "GBK2K-V",
    "HKscs-B5-H", "HKscs-B5-V",
#define END_PDF14_CMAP_NAMES_INDEX 18
    /* The following were added in PDF 1.3. */

    "GBpc-EUC-V",
    "GBK-EUC-H", "GBK-EUC-V",
    "UniGB-UCS2-H", "UniGB-UCS2-V",

    "ETenms-B5-H", "ETenms-B5-V",

    "UniCNS-UCS2-H", "UniCNS-UCS2-V",

    "90msp-RKSJ-H", "90msp-RKSJ-V",
    "EUC-H", "EUC-V",
    "UniJIS-UCS2-H", "UniJIS-UCS2-V",
    "UniJIS-UCS2-HW-H", "UniJIS-UCS2-HW-V",

    "KSCms-UHC-HW-H", "KSCms-UHC-HW-V",
    "UniKS-UCS2-H", "UniKS-UCS2-V",

#define END_PDF13_CMAP_NAMES_INDEX 39
    /* The following were added in PDF 1.2. */

    "GB-EUC-H", "GB-EUC-V",
    "GBpc-EUC-H",

    "B5pc-H", "B5pc-V",
    "ETen-B5-H", "ETen-B5-V",
    "CNS-EUC-H", "CNS-EUC-V",

    "83pv-RKSJ-H",
    "90ms-RKSJ-H", "90ms-RKSJ-V",
    "90pv-RKSJ-H",
    "Add-RKSJ-H", "Add-RKSJ-V",
    "Ext-RKSJ-H", "Ext-RKSJ-V",
    "H", "V",

    "KSC-EUC-H", "KSC-EUC-V",
    "KSCms-UHC-H", "KSCms-UHC-V",
    "KSCpc-EUC-H",

    "Identity-H", "Identity-V",

    0
};

static int
attach_cmap_resource(gx_device_pdf *pdev, pdf_font_resource_t *pdfont,
                const gs_cmap_t *pcmap, int font_index_only)
{
    const char *const *pcmn =
        standard_cmap_names +
        (pdev->CompatibilityLevel < 1.3 ? END_PDF13_CMAP_NAMES_INDEX :
         pdev->CompatibilityLevel < 1.4 ? END_PDF14_CMAP_NAMES_INDEX :
         pdev->CompatibilityLevel < 1.5 ? END_PDF15_CMAP_NAMES_INDEX : 0);
    bool is_identity = false;
    pdf_resource_t *pcmres = 0;	/* CMap */
    int code;

    /* Make sure cmap names is properly initialised. Silences Coverity warning */
    if (!pcmn)
        return_error(gs_error_unknownerror);

    /*
     * If the CMap isn't standard, write it out if necessary.
     */
    for (; *pcmn != 0; ++pcmn)
        if (pcmap->CMapName.size == strlen(*pcmn) &&
            !memcmp(*pcmn, pcmap->CMapName.data, pcmap->CMapName.size))
            break;

    /* For PDF/A we need to write out all non-identity CMaps
     * first force the identity check.
     */
    if (*pcmn == 0 || pdev->PDFA != 0) {
        /*
         * PScript5.dll Version 5.2 creates identity CMaps with
         * instandard name. Check this specially here
         * and later replace with a standard name.
         * This is a temporary fix for SF bug #615994 "CMAP is corrupt".
         */
        is_identity = gs_cmap_is_identity(pcmap, font_index_only);
    }
    /* If the CMap is non-standard, or we are producing PDF/A, and its not
     * an Identity CMap, then we need to emit it.
     */
    if ((*pcmn == 0  || pdev->PDFA != 0) && !is_identity) {		/* not standard */
        pcmres = pdf_find_resource_by_gs_id(pdev, resourceCMap, pcmap->id + font_index_only);
        if (pcmres == 0) {
            /* Create and write the CMap object. */
            code = pdf_cmap_alloc(pdev, pcmap, &pcmres, font_index_only);
            if (code < 0)
                return code;
        }
    }
    if (pcmap->from_Unicode) {
        gs_cmap_ranges_enum_t renum;

        gs_cmap_ranges_enum_init(pcmap, &renum);
        if (gs_cmap_enum_next_range(&renum) == 0 && renum.range.size == 2 &&
            gs_cmap_enum_next_range(&renum) == 1) {
            /*
             * Exactly one code space range, of size 2.  Add an identity
             * ToUnicode CMap.
             */
            if (!pdev->Identity_ToUnicode_CMaps[pcmap->WMode]) {
                /* Create and write an identity ToUnicode CMap now. */
                gs_cmap_t *pidcmap;

                code = gs_cmap_create_char_identity(&pidcmap, 2, pcmap->WMode,
                                                    pdev->memory);
                if (code < 0)
                    return code;
                pidcmap->CMapType = 2;	/* per PDF Reference */
                pidcmap->ToUnicode = true;
                code = pdf_cmap_alloc(pdev, pidcmap,
                                &pdev->Identity_ToUnicode_CMaps[pcmap->WMode], -1);
                if (code < 0)
                    return code;
            }
            pdfont->res_ToUnicode = pdev->Identity_ToUnicode_CMaps[pcmap->WMode];
        }
    }
    if (pcmres || is_identity) {
        uint size = pcmap->CMapName.size;
        byte *chars = gs_alloc_string(pdev->pdf_memory, size,
                                      "pdf_font_resource_t(CMapName)");

        if (chars == 0)
            return_error(gs_error_VMerror);
        memcpy(chars, pcmap->CMapName.data, size);
        if (is_identity)
            strcpy(pdfont->u.type0.Encoding_name,
                    (pcmap->WMode ? "/Identity-V" : "/Identity-H"));
        else
            gs_sprintf(pdfont->u.type0.Encoding_name, "%ld 0 R",
                    pdf_resource_id(pcmres));
        pdfont->u.type0.CMapName.data = chars;
        pdfont->u.type0.CMapName.size = size;
    } else {
        if (!*pcmn)
            /* Should not be possible, if *pcmn is NULL then either
             * is_identity is true or we create pcmres.
             */
            return_error(gs_error_invalidfont);

        gs_sprintf(pdfont->u.type0.Encoding_name, "/%s", *pcmn);
        pdfont->u.type0.CMapName.data = (const byte *)*pcmn;
        pdfont->u.type0.CMapName.size = strlen(*pcmn);
        pdfont->u.type0.cmap_is_standard = true;
    }
    pdfont->u.type0.WMode = pcmap->WMode;
    return 0;
}

static int estimate_fontbbox(pdf_text_enum_t *pte, gs_font_base *font,
                          const gs_matrix *pfmat,
                          gs_rect *text_bbox)
{
    gs_matrix m;
    gs_point p0, p1, p2, p3;

    if (font->FontBBox.p.x == font->FontBBox.q.x ||
        font->FontBBox.p.y == font->FontBBox.q.y)
        return_error(gs_error_undefined);
    if (pfmat == 0)
        pfmat = &font->FontMatrix;
    m = ctm_only(pte->pgs);
    m.tx = fixed2float(pte->origin.x);
    m.ty = fixed2float(pte->origin.y);
    gs_matrix_multiply(pfmat, &m, &m);

    gs_point_transform(font->FontBBox.p.x, font->FontBBox.p.y, &m, &p0);
    gs_point_transform(font->FontBBox.p.x, font->FontBBox.q.y, &m, &p1);
    gs_point_transform(font->FontBBox.q.x, font->FontBBox.p.y, &m, &p2);
    gs_point_transform(font->FontBBox.q.x, font->FontBBox.q.y, &m, &p3);
    text_bbox->p.x = min(min(p0.x, p1.x), min(p1.x, p2.x));
    text_bbox->p.y = min(min(p0.y, p1.y), min(p1.y, p2.y));
    text_bbox->q.x = max(max(p0.x, p1.x), max(p1.x, p2.x));
    text_bbox->q.y = max(max(p0.y, p1.y), max(p1.y, p2.y));

    return 0;
}

/* Record widths and CID => GID mappings. */
static int
scan_cmap_text(pdf_text_enum_t *pte, void *vbuf)
{
    gx_device_pdf *pdev = (gx_device_pdf *)pte->dev;
    /* gs_font_type0 *const font = (gs_font_type0 *)pte->current_font;*/ /* Type 0, fmap_CMap */
    gs_font_type0 *const font = (gs_font_type0 *)pte->orig_font; /* Type 0, fmap_CMap */
    /* Not sure. Changed for CDevProc callout. Was pte->current_font */
    gs_text_enum_t scan = *(gs_text_enum_t *)pte;
    int wmode = font->WMode, code, rcode = 0;
    pdf_font_resource_t *pdsubf0 = NULL;
    gs_font *subfont0 = NULL, *saved_subfont = NULL;
    uint index = scan.index, xy_index = scan.xy_index, start_index = index;
    uint font_index0 = 0x7badf00d;
    bool done = false;
    pdf_char_glyph_pairs_t p;
    gs_glyph *type1_glyphs = (gs_glyph *)vbuf;
    int num_type1_glyphs = 0;

    p.num_all_chars = 1;
    p.num_unused_chars = 1;
    p.unused_offset = 0;
    pte->returned.total_width.x = pte->returned.total_width.y = 0;;
    for (;;) {
        uint break_index, break_xy_index;
        uint font_index = 0x7badf00d;
        gs_const_string str;
        pdf_text_process_state_t text_state;
        pdf_font_resource_t *pdsubf;
        gs_font *subfont = NULL;
        gs_point wxy;
        bool font_change = 0;

        code = gx_path_current_point(pte->path, &pte->origin);
        if (code < 0)
            return code;
        do {
            gs_char chr;
            gs_glyph glyph;
            pdf_font_descriptor_t *pfd;
            byte *glyph_usage;
            double *real_widths, *w, *v, *w0;
            int char_cache_size, width_cache_size;
            gs_char cid;

            break_index = scan.index;
            break_xy_index = scan.xy_index;
            code = font->procs.next_char_glyph(&scan, &chr, &glyph);
            if (code == 2) {		/* end of string */
                done = true;
                break;
            }
            if (code < 0)
                return code;
            subfont = scan.fstack.items[scan.fstack.depth].font;
            font_index = scan.fstack.items[scan.fstack.depth - 1].index;
            scan.xy_index++;
            if (glyph == GS_NO_GLYPH)
                glyph = GS_MIN_CID_GLYPH;
            cid = glyph - GS_MIN_CID_GLYPH;
            switch (subfont->FontType) {
                case ft_encrypted:
                case ft_encrypted2:{
                    if (glyph == GS_MIN_CID_GLYPH) {
                        glyph = subfont->procs.encode_char(subfont, chr, GLYPH_SPACE_NAME);
                    }
                    type1_glyphs[num_type1_glyphs] = glyph;
                    num_type1_glyphs++;
                    break;
                }
                case ft_CID_encrypted:
                case ft_CID_TrueType: {
                    p.s[0].glyph = glyph;
                    p.s[0].chr = cid;
                    code = pdf_obtain_cidfont_resource(pdev, subfont, &pdsubf, &p);
                    if (code < 0)
                        return code;
                    break;
                }
                case ft_user_defined:
                case ft_PDF_user_defined:
                {
                    gs_string str1;

                    str1.data = NULL;
                    str1.size = 0;
                    pte->current_font = subfont;
                    code = pdf_obtain_font_resource(pte, &str1, &pdsubf);
                    if (code < 0)
                        return code;
                    cid = pdf_find_glyph(pdsubf, glyph);
                    if (cid == GS_NO_CHAR) {
                        code = pdf_make_font3_resource(pdev, subfont, &pdsubf);
                        if (code < 0)
                            return code;
                        code = pdf_attach_font_resource(pdev, subfont, pdsubf);
                        if (code < 0)
                            return code;
                        cid = 0;
                    }
                    break;
                }
                default:
                    /* An unsupported case, fall back to default implementation. */
                    return_error(gs_error_rangecheck);
            }
            code = pdf_attached_font_resource(pdev, (gs_font *)subfont, &pdsubf,
                                       &glyph_usage, &real_widths, &char_cache_size, &width_cache_size);
            if (code < 0)
                return code;
            if (break_index > start_index && pdev->charproc_just_accumulated)
                break;
            if ((subfont->FontType == ft_user_defined || subfont->FontType == ft_PDF_user_defined )&&
                (break_index > start_index || !pdev->charproc_just_accumulated) &&
                !(pdsubf->u.simple.s.type3.cached[cid >> 3] & (0x80 >> (cid & 7)))) {
                if (subfont0 && subfont0->FontType != ft_user_defined && subfont0->FontType != ft_PDF_user_defined)
                    /* This is hacky. By pretending to be in a type 3 font doing a charpath we force
                     * text handling to fall right back to bitmap glyphs. This is because we can't handle
                     * CIDFonts with mixed type 1/3 descendants. Ugly but it produces correct output for
                     * what is after all a dumb setup.
                     */
                    pdev->type3charpath = 1;
                pte->current_font = subfont;
                return_error(gs_error_undefined);
            }
            if (subfont->FontType == ft_encrypted || subfont->FontType == ft_encrypted2) {
                font_change = (subfont != subfont0 && subfont0 != NULL);
                if (font_change) {
                    saved_subfont = subfont;
                    subfont = subfont0;
                    num_type1_glyphs--;
                }
            } else
                font_change = (pdsubf != pdsubf0 && pdsubf0 != NULL);
            if (!font_change) {
                pdsubf0 = pdsubf;
                font_index0 = font_index;
                subfont0 = subfont;
            }
            if (subfont->FontType != ft_encrypted && subfont->FontType != ft_encrypted2) {
                pfd = pdsubf->FontDescriptor;
                code = pdf_resize_resource_arrays(pdev, pdsubf, cid + 1);
                if (code < 0)
                    return code;
                if (subfont->FontType == ft_CID_encrypted || subfont->FontType == ft_CID_TrueType) {
                    if (cid >=width_cache_size) {
                        /* fixme: we add the CID=0 glyph as CID=cid glyph to the output font.
                           Really it must not add and leave the CID undefined. */
                        cid = 0; /* notdef. */
                    }
                }
                if (cid >= char_cache_size || cid >= width_cache_size)
                    return_error(gs_error_unregistered); /* Must not happen */
                if (pdsubf->FontType == ft_user_defined  || pdsubf->FontType == ft_PDF_user_defined  || pdsubf->FontType == ft_encrypted ||
                                pdsubf->FontType == ft_encrypted2) {
                } else {
                    pdf_font_resource_t *pdfont;
                    bool notdef_subst = false;

                    code = pdf_obtain_cidfont_widths_arrays(pdev, pdsubf, wmode, &w, &w0, &v);
                    if (code < 0)
                        return code;
                    code = pdf_obtain_parent_type0_font_resource(pdev, pdsubf, font_index,
                        &font->data.CMap->CMapName, &pdfont);
                    if (code < 0)
                        return code;
                    if (pdf_is_CID_font(subfont)) {
                        /* Some Pscript5 output has non-identity mappings between character code and CID
                         * and the GlyphNames2Unicode dictionary uses character codes, not glyph names. So
                         * if we detect ths condition we cheat and claim not to be a CIDFont, so that the
                         * decode_glyph procedure can use the character code to look up the GlyphNames2Unicode
                         * dictionary. See bugs #696021, #688768 and #687954 for examples of the various ways
                         * this code can be exercised.
                         */
                        if (chr == glyph - GS_MIN_CID_GLYPH)
                            code = subfont->procs.decode_glyph((gs_font *)subfont, glyph, -1, NULL, 0);
                        else
                            code = subfont->procs.decode_glyph((gs_font *)subfont, glyph, chr, NULL, 0);
                        if (code != 0)
                            /* Since PScript5.dll creates GlyphNames2Unicode with character codes
                               instead CIDs, and with the WinCharSetFFFF-H2 CMap
                               character codes appears different than CIDs (Bug 687954),
                               pass the character code intead the CID. */
                            code = pdf_add_ToUnicode(pdev, subfont, pdfont,
                                chr + GS_MIN_CID_GLYPH, chr, NULL);
                        else {
                            /* If we interpret a PDF document, ToUnicode
                               CMap may be attached to the Type 0 font. */
                            code = pdf_add_ToUnicode(pdev, pte->orig_font, pdfont,
                                chr + GS_MIN_CID_GLYPH, chr, NULL);
                        }
                    }
                    else
                        code = pdf_add_ToUnicode(pdev, subfont, pdfont, glyph, cid, NULL);
                    if (code < 0)
                        return code;
                    /*  We can't check pdsubf->used[cid >> 3] here,
                        because it mixed data for different values of WMode.
                        Perhaps pdf_font_used_glyph returns fast with reused glyphs.
                     */
                    code = pdf_font_used_glyph(pfd, glyph, (gs_font_base *)subfont);
                    if (code == gs_error_rangecheck) {
                        if (!(pdsubf->used[cid >> 3] & (0x80 >> (cid & 7)))) {
                            char buf[gs_font_name_max + 1];
                            int l = min(sizeof(buf) - 1, subfont->font_name.size);

                            memcpy(buf, subfont->font_name.chars, l);
                            buf[l] = 0;
                            emprintf3(pdev->memory,
                                      "Missing glyph CID=%d, glyph=%04x in the font %s . The output PDF may fail with some viewers.\n",
                                      (int)cid,
                                      (unsigned int)(glyph - GS_MIN_CID_GLYPH),
                                      buf);
                            pdsubf->used[cid >> 3] |= 0x80 >> (cid & 7);
                            if (pdev->PDFA != 0) {
                                switch (pdev->PDFACompatibilityPolicy) {
                                    /* Default behaviour matches Adobe Acrobat, warn and continue,
                                     * output file will not be PDF/A compliant
                                     */
                                    case 0:
                                    case 1:
                                    case 3:
                                        emprintf(pdev->memory,
                                             "All used glyphs mst be present in fonts for PDF/A, reverting to normal PDF output.\n");
                                        pdev->AbortPDFAX = true;
                                        pdev->PDFA = 0;
                                        break;
                                    case 2:
                                        emprintf(pdev->memory,
                                             "All used glyphs mst be present in fonts for PDF/A, aborting conversion.\n");
                                        return_error(gs_error_invalidfont);
                                        break;
                                    default:
                                        emprintf(pdev->memory,
                                             "All used glyphs mst be present in fonts for PDF/A, unrecognised PDFACompatibilityLevel,\nreverting to normal PDF output\n");
                                        pdev->AbortPDFAX = true;
                                        pdev->PDFA = 0;
                                        break;
                                }
                            }
                        }
                        cid = 0, code = 1;  /* undefined glyph. */
                        notdef_subst = true;
                        /* If this is the first use of CID=0, get its width */
                        if (pdsubf->Widths[cid] == 0) {
                            pdf_glyph_widths_t widths;

                            code = pdf_glyph_widths(pdsubf, wmode, glyph, (gs_font *)subfont, &widths,
                                pte->cdevproc_callout ? pte->cdevproc_result : NULL);
                        }
                    } else if (code < 0)
                        return code;
                    if (glyph == GS_MIN_CID_GLYPH && pdev->PDFA != 0) {
                        switch (pdev->PDFACompatibilityPolicy) {
                            case 0:
                            case 1:
                            case 3:
                                emprintf(pdev->memory,
                                     "A CIDFont uses CID 0, which is not legal for PDF/A, reverting to normal PDF output.\n");
                                pdev->AbortPDFAX = true;
                                pdev->PDFA = 0;
                                break;
                            case 2:
                                emprintf(pdev->memory,
                                     "A CIDFont uses CID 0, which is not legal for PDF/A, aborting conversion.\n");
                                return_error(gs_error_invalidfont);
                                break;
                            default:
                                emprintf(pdev->memory,
                                     "A CIDFont uses CID 0, which is not legal for PDF/A, unrecognised PDFACompatibilityLevel,\nreverting to normal PDF output\n");
                                pdev->AbortPDFAX = true;
                                pdev->PDFA = 0;
                                break;
                        }
                    }
                    if ((code == 0 /* just copied */ || pdsubf->Widths[cid] == 0) && !notdef_subst) {
                        pdf_glyph_widths_t widths;

                    code = pdf_glyph_widths(pdsubf, wmode, glyph, (gs_font *)subfont, &widths,
                        pte->cdevproc_callout ? pte->cdevproc_result : NULL);
                    if (code < 0)
                        return code;
                    if (code == TEXT_PROCESS_CDEVPROC) {
                        pte->returned.current_glyph = glyph;
                        pte->current_font = subfont;
                        rcode = TEXT_PROCESS_CDEVPROC;
                        break;
                    }
                    if (code >= 0) {
                        if (cid > pdsubf->count)
                            return_error(gs_error_unregistered); /* Must not happen. */
                        w[cid] = widths.Width.w;
                        if (v != NULL) {
                            v[cid * 2 + 0] = widths.Width.v.x;
                            v[cid * 2 + 1] = widths.Width.v.y;
                        }
                        real_widths[cid] = widths.real_width.w;
                    }
                    if (wmode) {
                        /* Since AR5 use W or DW to compute the x-coordinate of
                           v-vector, comupte and store the glyph width for WMode 0. */
                        /* fixme : skip computing real_width here. */
                        code = pdf_glyph_widths(pdsubf, 0, glyph, (gs_font *)subfont, &widths,
                            pte->cdevproc_callout ? pte->cdevproc_result : NULL);
                        if (code < 0)
                            return code;
                        w0[cid] = widths.Width.w;
                    }
                    if (pdsubf->u.cidfont.CIDToGIDMap != 0) {
                        uint gid = 0;
                        gs_font_cid2 *subfont2 = (gs_font_cid2 *)subfont;

                        gid = subfont2->cidata.CIDMap_proc(subfont2, glyph);

                        /* If this is a TrueType CIDFont, check the GSUB table to see if there's
                         * a suitable substitute glyph.
                         */
                        if (subfont2->FontType == ft_CID_TrueType)
                            gid = subfont2->data.substitute_glyph_index_vertical((gs_font_type42 *)subfont, gid, subfont2->WMode, glyph);
                        pdsubf->u.cidfont.CIDToGIDMap[cid] = gid;
                    }
                }
                if (wmode)
                    pdsubf->u.cidfont.used2[cid >> 3] |= 0x80 >> (cid & 7);
                }
                pdsubf->used[cid >> 3] |= 0x80 >> (cid & 7);
            }
            if (pte->cdevproc_callout) {
                /* Only handle a single character because its width is stored
                  into pte->cdevproc_result, and process_text_modify_width neds it.
                  fixme: next time take from w, v, real_widths. */
                break_index = scan.index;
                break_xy_index = scan.xy_index;
                break;
            }
        } while (!font_change);
        if (break_index > index) {
            pdf_font_resource_t *pdfont;
            gs_matrix m3;
            int xy_index_step = (!(pte->text.operation & TEXT_REPLACE_WIDTHS) ? 0 :
                                 pte->text.x_widths == pte->text.y_widths ? 2 : 1);
            gs_text_params_t save_text;

            if (!subfont && num_type1_glyphs != 0)
                subfont = subfont0;
            if (subfont && (subfont->FontType == ft_encrypted || subfont->FontType == ft_encrypted2)) {
                int save_op = pte->text.operation;
                gs_font *save_font = pte->current_font;
                const gs_glyph *save_data = pte->text.data.glyphs;

                pte->current_font = subfont;
                pte->text.operation |= TEXT_FROM_GLYPHS;
                pte->text.data.glyphs = type1_glyphs;
                str.data = ((const byte *)vbuf) + ((pte->text.size - pte->index) * sizeof(gs_glyph));
                str.size = num_type1_glyphs;
                code = pdf_obtain_font_resource_unencoded(pte, (const gs_string *)&str, &pdsubf0,
                    type1_glyphs);
                if (code < 0) {
                    /* Replace the modified values, fall back to default implementation
                     * (type 3 bitmap image font)
                     */
                    pte->current_font = save_font;
                    pte->text.operation |= save_op;
                    pte->text.data.glyphs = save_data;
                    return(code);
                }
                memcpy((void *)scan.text.data.bytes, (void *)str.data, str.size);
                str.data = scan.text.data.bytes;
                pdsubf = pdsubf0;
                pte->text.operation = save_op;
            }
            pte->current_font = subfont0;
            if (!subfont0 || !pdsubf0)
                /* This should be impossible */
                return_error(gs_error_invalidfont);

            code = gs_matrix_multiply(&subfont0->FontMatrix, &font->FontMatrix, &m3);
            /* We thought that it should be gs_matrix_multiply(&font->FontMatrix, &subfont0->FontMatrix, &m3); */
            if (code < 0)
                return code;
            if (pdsubf0->FontType == ft_user_defined  || pdsubf0->FontType == ft_PDF_user_defined  || pdsubf->FontType == ft_encrypted ||
                pdsubf->FontType == ft_encrypted2)
                    pdfont = pdsubf0;
            else {
                code = pdf_obtain_parent_type0_font_resource(pdev, pdsubf0, font_index0,
                            &font->data.CMap->CMapName, &pdfont);
                if (code < 0)
                    return code;
                if (!pdfont->u.type0.Encoding_name[0]) {
                    /*
                    * If pdfont->u.type0.Encoding_name is set,
                    * a CMap resource is already attached.
                    * See attach_cmap_resource.
                    */
                    code = attach_cmap_resource(pdev, pdfont, font->data.CMap, font_index0);
                    if (code < 0)
                        return code;
                }
            }
            pdf_set_text_wmode(pdev, font->WMode);
            code = pdf_update_text_state(&text_state, (pdf_text_enum_t *)pte, pdfont, &m3);
            if (code < 0)
                return code;
            /* process_text_modify_width breaks text parameters.
               We would like to improve it someday.
               Now save them locally and restore after the call. */
            save_text = pte->text;
            if (subfont && (subfont->FontType != ft_encrypted &&
                subfont->FontType != ft_encrypted2)) {
                /* If we are a type 1 descendant, we already sorted this out above */
                str.data = scan.text.data.bytes + index;
                str.size = break_index - index;
            }
            if (pte->text.operation & TEXT_REPLACE_WIDTHS) {
                if (pte->text.x_widths != NULL)
                    pte->text.x_widths += xy_index * xy_index_step;
                if (pte->text.y_widths != NULL)
                    pte->text.y_widths += xy_index * xy_index_step;
            }
            pte->xy_index = 0;
            if (subfont && (subfont->FontType == ft_encrypted ||
                subfont->FontType == ft_encrypted2)) {
                gs_font *f = pte->orig_font;

                adjust_first_last_char(pdfont, (byte *)str.data, str.size);

                /* Make sure we use the descendant font, not the original type 0 ! */
                pte->orig_font = subfont;
                code = process_text_modify_width((pdf_text_enum_t *)pte,
                    (gs_font *)subfont, &text_state, &str, &wxy, type1_glyphs, false, scan.index - index);
                if (code < 0)
                    return(code);
                if(font_change) {
                    type1_glyphs[0] = type1_glyphs[num_type1_glyphs];
                    num_type1_glyphs = 1;
                    subfont = saved_subfont;
                } else {
                    num_type1_glyphs = 0;
                }
                pte->orig_font = f;
            } else {
                code = process_text_modify_width((pdf_text_enum_t *)pte, (gs_font *)font,
                    &text_state, &str, &wxy, NULL, true, scan.index - index);
            }
            if (pte->text.operation & TEXT_REPLACE_WIDTHS) {
                if (pte->text.x_widths != NULL)
                    pte->text.x_widths -= xy_index * xy_index_step;
                if (pte->text.y_widths != NULL)
                    pte->text.y_widths -= xy_index * xy_index_step;
            }
            pte->text = save_text;
            pte->cdevproc_callout = false;
            if (code < 0) {
                pte->index = index;
                pte->xy_index = xy_index;
                return code;
            }
            pte->index = break_index;
            pte->xy_index = break_xy_index;
            if (pdev->Eps2Write) {
                gs_rect text_bbox;
                gx_device_clip cdev;
                gx_drawing_color devc;
                fixed x0, y0, bx2, by2;

                text_bbox.q.x = text_bbox.p.y = text_bbox.q.y = 0;
                estimate_fontbbox(pte, (gs_font_base *)font, NULL, &text_bbox);
                text_bbox.p.x = fixed2float(pte->origin.x);
                text_bbox.q.x = text_bbox.p.x + wxy.x;

                x0 = float2fixed(text_bbox.p.x);
                y0 = float2fixed(text_bbox.p.y);
                bx2 = float2fixed(text_bbox.q.x) - x0;
                by2 = float2fixed(text_bbox.q.y) - y0;

                pdev->AccumulatingBBox++;
                gx_make_clip_device_on_stack(&cdev, pte->pcpath, (gx_device *)pdev);
                set_nonclient_dev_color(&devc, gx_device_black((gx_device *)pdev));  /* any non-white color will do */
                gx_default_fill_triangle((gx_device *) pdev, x0, y0,
                                         float2fixed(text_bbox.p.x) - x0,
                                         float2fixed(text_bbox.q.y) - y0,
                                         bx2, by2, &devc, lop_default);
                gx_default_fill_triangle((gx_device *) & cdev, x0, y0,
                                         float2fixed(text_bbox.q.x) - x0,
                                         float2fixed(text_bbox.p.y) - y0,
                                         bx2, by2, &devc, lop_default);
                pdev->AccumulatingBBox--;
            }
            code = pdf_shift_text_currentpoint(pte, &wxy);
            if (code < 0)
                return code;
        }
        pdf_text_release_cgp(pte);
        index = break_index;
        xy_index = break_xy_index;
        if (done || rcode != 0)
            break;
        pdsubf0 = pdsubf;
        font_index0 = font_index;
        subfont0 = subfont;
    }
    pte->index = index;
    pte->xy_index = xy_index;
    return rcode;
}

int
process_cmap_text(gs_text_enum_t *penum, void *vbuf, uint bsize)
{
    int code;
    pdf_text_enum_t *pte = (pdf_text_enum_t *)penum;
    byte *save;
    uint start = pte->index;

    if (pte->text.operation &
        (TEXT_FROM_ANY - (TEXT_FROM_STRING | TEXT_FROM_BYTES))
        )
        return_error(gs_error_rangecheck);
    if (pte->text.operation & TEXT_INTERVENE) {
        /* Not implemented.  (PostScript doesn't allow TEXT_INTERVENE.) */
        return_error(gs_error_rangecheck);
    }
    /* scan_cmap_text has the unfortunate side effect of meddling with the
     * text data in the enumerator. In general that's OK but in the case where
     * the string is (eg) in a bound procedure, and we run that procedure more
     * than once, the string is corrupted on the first use and then produces
     * incorrect output for the subsequent use(s).
     * The routine is, sadly, extremely convoluted so instead of trying to fix
     * it so that it doesn't corrupt the string (which looks likely to be impossible
     * without copying the string at some point) I've chosen to take a copy of the
     * string here, and restore it after the call to scan_cmap_text.
     * See bug #695322 and test file Bug691680.ps
     */
    save = (byte *)pte->text.data.bytes;
    pte->text.data.bytes = gs_alloc_string(pte->memory, pte->text.size, "pdf_text_process");
    memcpy((byte *)pte->text.data.bytes, save, pte->text.size);
    code = scan_cmap_text(pte, vbuf);
    gs_free_string(pte->memory, (byte *)pte->text.data.bytes,  pte->text.size, "pdf_text_process");
    pte->text.data.bytes = save;
    pte->bytes_decoded = pte->index - start;

    if (code == TEXT_PROCESS_CDEVPROC)
        pte->cdevproc_callout = true;
    else
        pte->cdevproc_callout = false;
    return code;
}

/* ---------------- CIDFont ---------------- */

/*
 * Process a text string in a CIDFont.  (Only glyphshow is supported.)
 */
int
process_cid_text(gs_text_enum_t *pte, void *vbuf, uint bsize)
{
    pdf_text_enum_t *penum = (pdf_text_enum_t *)pte;
    uint operation = pte->text.operation;
    gs_text_enum_t save;
    gs_font *scaled_font = pte->current_font; /* CIDFont */
    gs_font *font;		/* unscaled font (CIDFont) */
    const gs_glyph *glyphs;
    gs_matrix scale_matrix;
    pdf_font_resource_t *pdsubf; /* CIDFont */
    gs_font_type0 *font0 = NULL;
    uint size;
    int code;

    if (operation & TEXT_FROM_GLYPHS) {
        glyphs = pte->text.data.glyphs;
        size = pte->text.size - pte->index;
    } else if (operation & TEXT_FROM_SINGLE_GLYPH) {
        glyphs = &pte->text.data.d_glyph;
        size = 1;
    } else if (operation & TEXT_FROM_STRING) {
        glyphs = &pte->outer_CID;
        size = 1;
    } else
        return_error(gs_error_rangecheck);

    /*
     * PDF doesn't support glyphshow directly: we need to create a Type 0
     * font with an Identity CMap.  Make sure all the glyph numbers fit
     * into 16 bits.  (Eventually we should support wider glyphs too,
     * but this would require a different CMap.)
     */
    if (bsize < size * 2)
        return_error(gs_error_unregistered); /* Must not happen. */
    {
        int i;
        byte *pchars = vbuf;

        for (i = 0; i < size; ++i) {
            ulong gnum = glyphs[i] - GS_MIN_CID_GLYPH;

            if (gnum & ~0xffffL)
                return_error(gs_error_rangecheck);
            *pchars++ = (byte)(gnum >> 8);
            *pchars++ = (byte)gnum;
        }
    }

    /* Find the original (unscaled) version of this font. */

    for (font = scaled_font; font->base != font; )
        font = font->base;
    /* Compute the scaling matrix. */
    code = gs_matrix_invert(&font->FontMatrix, &scale_matrix);
    if (code < 0)
        return code;
    gs_matrix_multiply(&scale_matrix, &scaled_font->FontMatrix, &scale_matrix);

    /* Find or create the CIDFont resource. */

    code = pdf_obtain_font_resource(penum, NULL, &pdsubf);
    if (code < 0)
        return code;

    /* Create the CMap and Type 0 font if they don't exist already. */

    if (pdsubf->u.cidfont.glyphshow_font_id != 0)
        font0 = (gs_font_type0 *)gs_find_font_by_id(font->dir,
                    pdsubf->u.cidfont.glyphshow_font_id, &scaled_font->FontMatrix);
    if (font0 == NULL) {
        code = gs_font_type0_from_cidfont(&font0, font, font->WMode,
                                          &scale_matrix, font->memory);
        if (code < 0)
            return code;
        pdsubf->u.cidfont.glyphshow_font_id = font0->id;
    }

    /* Now handle the glyphshow as a show in the Type 0 font. */

    save = *pte;
    pte->current_font = pte->orig_font = (gs_font *)font0;
    /* Patch the operation temporarily for init_fstack. */
    pte->text.operation = (operation & ~TEXT_FROM_ANY) | TEXT_FROM_BYTES;
    /* Patch the data for process_cmap_text. */
    pte->text.data.bytes = vbuf;
    pte->text.size = size * 2;
    pte->index = 0;
    gs_type0_init_fstack(pte, pte->current_font);
    code = process_cmap_text(pte, vbuf, bsize);
    pte->current_font = scaled_font;
    pte->orig_font = save.orig_font;
    pte->text = save.text;
    pte->index = save.index + pte->index / 2;
    pte->fstack = save.fstack;
    return code;
}