Skip to content

Commit

Permalink
update mupdf
Browse files Browse the repository at this point in the history
  • Loading branch information
@kjk
kjk committed Mar 29, 2023
1 parent 2cc4297 commit 987bb6a
Show file tree
Hide file tree
Showing 4 changed files with 135 additions and 121 deletions.
20 changes: 17 additions & 3 deletions mupdf/include/mupdf/fitz/document.h
View file
Expand Up @@ -874,7 +874,7 @@ void *fz_process_opened_pages(fz_context *ctx, fz_document *doc, fz_process_open
struct fz_page
{
int refs;
fz_document *doc; /* reference to parent document */
fz_document *doc; /* kept reference to parent document. Guaranteed non-NULL. */
int chapter; /* chapter number */
int number; /* page number in chapter */
int incomplete; /* incomplete from progressive loading; don't cache! */
Expand All @@ -891,7 +891,13 @@ struct fz_page
fz_page_uses_overprint_fn *overprint;
fz_page_create_link_fn *create_link;
fz_page_delete_link_fn *delete_link;
fz_page **prev, *next; /* linked list of currently open pages */

/* linked list of currently open pages. This list is maintained
* by fz_load_chapter_page and fz_drop_page. All pages hold a
* kept reference to the document, so the document cannot disappear
* while pages exist. 'Incomplete' pages are NOT kept in this
* list. */
fz_page **prev, *next;
};

/**
Expand Down Expand Up @@ -924,7 +930,15 @@ struct fz_document
fz_document_output_accelerator_fn *output_accelerator;
int did_layout;
int is_reflowable;
fz_page *open; /* linked list of currently open pages */

/* Linked list of currently open pages. These are not
* references, but just a linked list of open pages,
* maintained by fz_load_chapter_page, and fz_drop_page.
* Every page holds a kept reference to the document, so
* the document cannot be destroyed while a page exists.
* Incomplete pages are NOT inserted into this list, but
* do still hold a real document reference. */
fz_page *open;
};

struct fz_document_handler
Expand Down
8 changes: 7 additions & 1 deletion mupdf/source/html/html-imp.h
View file
Expand Up @@ -498,14 +498,20 @@ struct fz_html_flow_s
/* Whether this node is currently taken as a line break */
unsigned int breaks_line : 1;

/* Whether this word node can be split or consists of a single glyph cluster */
unsigned int atomic : 1;

/* Whether lines may be broken before this word for overflow-wrap: word-break */
unsigned int overflow_wrap : 1;

/* Direction setting for text - UAX#9 says 125 is the max */
unsigned int bidi_level : 7;

/* The script detected by the bidi code. */
unsigned int script : 8;

/* Whether the markup specifies a given language. */
unsigned int markup_lang : 15;
unsigned short markup_lang;

float x, y, w, h;
fz_html_box *box; /* for style and em */
Expand Down
223 changes: 106 additions & 117 deletions mupdf/source/html/html-layout.c
View file
Expand Up @@ -378,58 +378,6 @@ static void measure_string_h(fz_context *ctx, fz_html_flow *node)
node->h = fz_from_css_number_scale(node->box->style->line_height, em);
}

static unsigned int measure_string_to_fit(fz_context *ctx, const char *s, fz_html_flow *node, hb_buffer_t *hb_buf, float max_w)
{
string_walker walker;
unsigned int i;
float em;
float line_w;
uint32_t min;
int fragment_offset;
float node_w;

node_w = 0;
em = node->box->s.layout.em;

line_w = 0;
fragment_offset = 0;
init_string_walker(ctx, &walker, hb_buf, node->bidi_level & 1, node->box->style->font, node->script, node->markup_lang, node->box->style->small_caps, s);
while (walk_string(&walker))
{
for (i = 0; i < walker.glyph_count; i++)
{
line_w += walker.glyph_pos[i].x_advance * em / walker.scale;
if (line_w > max_w)
goto split;
node_w = line_w;
}
fragment_offset = walker.end - s;
}

/* This indicates that the whole string fitted. That should never be possible
* as we'd never have called this function in that case! */
assert("Spanish Inquisition!" == NULL);

return 0;

split:
/* Nothing fitted. Exit here. Don't update the node width. */
if (i == 0)
return 0;

/* Find min, the byte offset of the smallest cluster seen after
* where we need to split the string to fit. That's the split point. */
min = walker.glyph_info[i].cluster;
for (i++; i < walker.glyph_count; i++)
if (walker.glyph_info[i].cluster < min)
min = walker.glyph_info[i].cluster;

/* Update the width of the node for when we split it.*/
node->w = node_w;

/* So return the offset in bytes at which to split. */
return min + fragment_offset;
}

static float measure_line(fz_html_flow *node, fz_html_flow *end, float *baseline, float *vert_adv)
{
Expand Down Expand Up @@ -658,36 +606,55 @@ static int flush_line(fz_context *ctx, fz_html_box *box, layout_data *ld, float
return 0;
}

static fz_html_flow *
break_node(fz_context *ctx, fz_html_flow *node, layout_data *ld, float w)
static void break_word_for_overflow_wrap(fz_context *ctx, fz_html_flow *node, layout_data *ld)
{
const char *s = get_node_text(ctx, node);
unsigned int split_pos;
fz_html_flow *new_node;
hb_buffer_t *hb_buf = ld->hb_buf;
const char *text = node->content.text;
string_walker walker;

/* Only break nodes if overflow_wrap is set to break-word. */
if (node->box->style->overflow_wrap != OVERFLOW_WRAP_BREAK_WORD)
return NULL;
assert(node->type == FLOW_WORD);
assert(node->atomic == 0);

/* Split a word node after the first cluster (usually a character), and
* flag the second half as a valid node to break before if in desperate
* need. This may break earlier than necessary, but in that case we'll
* break the second half again when we come to it, until we find a
* suitable breaking point.
*
* We split after each clusters here so we can flag each fragment as
* "atomic" so we don't try breaking it again, and also to flag the
* following word fragment as a possible break point. Breaking at the
* exact desired point would make this more complicated than necessary.
*
* Desperately breaking in the middle of a word like this should should
* rarely (if ever) come up.
*
* TODO: Split at all the clusters in the word at once.
*/

split_pos = measure_string_to_fit(ctx, s, node, ld->hb_buf, w);
if (split_pos == 0)
/* Walk string and split at the first cluster. */
init_string_walker(ctx, &walker, hb_buf, node->bidi_level & 1, node->box->style->font, node->script, node->markup_lang, node->box->style->small_caps, text);
while (walk_string(&walker))
{
/* No sensible chunk fitted - we can't split, but this should
* be a candidate for breaking. */
return node;
unsigned int i, a, b;
a = walker.glyph_info[0].cluster;
for (i = 0; i < walker.glyph_count; ++i)
{
b = walker.glyph_info[i].cluster;
if (b != a)
{
fz_html_split_flow(ctx, ld->pool, node, fz_runeidx(text, text + b));
node->atomic = 1;
node->next->overflow_wrap = 1;
measure_string_w(ctx, node, ld->hb_buf);
measure_string_w(ctx, node->next, ld->hb_buf);
return;
}
}
}

new_node = fz_html_split_flow(ctx, ld->pool, node, split_pos);
new_node->type = FLOW_WORD;
new_node->h = node->h;
new_node->expand = node->expand;
new_node->script = node->script;
new_node->markup_lang = node->markup_lang;
new_node->bidi_level = node->bidi_level;
new_node->breaks_line = node->breaks_line;
measure_string_w(ctx, new_node, ld->hb_buf);

return new_node;
/* Word is already only one cluster. Don't try breaking here again! */
node->atomic = 1;
}

/*
Expand All @@ -700,10 +667,10 @@ break_node(fz_context *ctx, fz_html_flow *node, layout_data *ld, float w)
*/
static void layout_flow(fz_context *ctx, layout_data *ld, fz_html_box *box, fz_html_box *top)
{
fz_html_flow *node, *line, *candidate;
fz_html_flow *node, *line, *candidate, *desperate;
fz_html_flow *start_flow = NULL;
float line_w, candidate_w, indent, break_w, nonbreak_w;
int line_align, align;
float line_w, candidate_w, desperate_w, indent;
int align;
fz_html_restarter *restart = ld->restart;

float em = box->s.layout.em = fz_from_css_number(box->style->font_size, top->s.layout.em, top->s.layout.em, top->s.layout.em);
Expand Down Expand Up @@ -738,6 +705,7 @@ static void layout_flow(fz_context *ctx, layout_data *ld, fz_html_box *box, fz_h
if (!box->u.flow.head)
return;

/* Measure the size of all the words and images in the flow. */
for (node = box->u.flow.head; node; node = node->next)
{
if (restart && restart->start_flow)
Expand Down Expand Up @@ -793,9 +761,6 @@ static void layout_flow(fz_context *ctx, layout_data *ld, fz_html_box *box, fz_h
}

node = box->u.flow.head;

candidate = NULL;
candidate_w = 0;
if (start_flow)
{
line = start_flow;
Expand All @@ -807,8 +772,14 @@ static void layout_flow(fz_context *ctx, layout_data *ld, fz_html_box *box, fz_h
line_w = indent;
}

candidate = NULL;
candidate_w = 0;
desperate = NULL;
desperate_w = 0;

while (node)
{
/* Fast-forward to the restart point. */
if (start_flow)
{
if (start_flow != node)
Expand All @@ -818,66 +789,84 @@ static void layout_flow(fz_context *ctx, layout_data *ld, fz_html_box *box, fz_h
}
start_flow = NULL;
}

switch (node->type)
{
default:
// always same width
line_w += node->w;
break;

case FLOW_WORD:
if (node->w > box->s.layout.w - line_w && !candidate)
/* Allow desperate breaking in middle of word if overflow-wrap: break-word. */
if (line_w + node->w > box->s.layout.w && !candidate)
{
candidate = break_node(ctx, node, ld, box->s.layout.w - line_w);
if (!node->atomic && node->box->style->overflow_wrap == OVERFLOW_WRAP_BREAK_WORD)
{
break_word_for_overflow_wrap(ctx, node, ld);
}
}
nonbreak_w = break_w = node->w;
break;
case FLOW_IMAGE:
nonbreak_w = break_w = node->w;
break;

case FLOW_SHYPHEN:
case FLOW_SBREAK:
case FLOW_SPACE:
nonbreak_w = break_w = 0;

/* Determine broken and unbroken widths of this node. */
if (node->type == FLOW_SPACE)
nonbreak_w = node->w;
else if (node->type == FLOW_SHYPHEN)
break_w = node->w;

/* If the broken node fits, remember it. */
/* Also remember it if we have no other candidate and need to break in desperation. */
if (line_w + break_w <= box->s.layout.w || !candidate)
/* Remember overflow-wrap word fragments, unless at the beginning of a line. */
if (node->overflow_wrap && node != line)
{
candidate = node;
candidate_w = line_w + break_w;
desperate = node;
desperate_w = line_w;
}
line_w += node->w;
break;

case FLOW_BREAK:
nonbreak_w = break_w = 0;
case FLOW_SBREAK:
// always zero width
candidate = node;
candidate_w = line_w;
break;

case FLOW_SPACE:
// zero width if broken, default width if unbroken
candidate = node;
candidate_w = line_w;
line_w += node->w;
break;

case FLOW_SHYPHEN:
// default width if broken, zero width if unbroken
candidate = node;
candidate_w = line_w + node->w;
break;
}

/* The current node either does not fit or we saw a hard break. */
/* Break the line if we have a candidate break point. */
line_w += nonbreak_w;
if (node->type == FLOW_BREAK || (line_w > box->s.layout.w && candidate))
if (node->type == FLOW_BREAK || (line_w > box->s.layout.w && (candidate || desperate)))
{
fz_html_flow *break_at = (candidate->type == FLOW_WORD ? candidate : candidate->next);
int line_align = align;
fz_html_flow *break_at;
float break_w;

candidate->breaks_line = 1;
if (candidate->type == FLOW_BREAK)
line_align = (align == TA_JUSTIFY) ? TA_LEFT : align;
if (candidate)
{
if (candidate->type == FLOW_BREAK)
line_align = (align == TA_JUSTIFY) ? TA_LEFT : align;
candidate->breaks_line = 1;
break_at = candidate->next;
break_w = candidate_w;
}
else
line_align = align;
if (flush_line(ctx, box, ld, box->s.layout.w, candidate_w, line_align, indent, line, break_at, restart))
{
break_at = desperate;
break_w = desperate_w;
}

if (flush_line(ctx, box, ld, box->s.layout.w, break_w, line_align, indent, line, break_at, restart))
return;

line = break_at;
node = break_at;
candidate = NULL;
candidate_w = 0;
desperate = NULL;
desperate_w = 0;
indent = 0;
line_w = 0;
}
Expand All @@ -889,7 +878,7 @@ static void layout_flow(fz_context *ctx, layout_data *ld, fz_html_box *box, fz_h

if (line)
{
line_align = (align == TA_JUSTIFY) ? TA_LEFT : align;
int line_align = (align == TA_JUSTIFY) ? TA_LEFT : align;
flush_line(ctx, box, ld, box->s.layout.w, line_w, line_align, indent, line, NULL, restart);
}
}
Expand Down Expand Up @@ -1512,7 +1501,7 @@ static void layout_update_widths(fz_context *ctx, fz_html_box *box, fz_html_box
if (node->type == FLOW_IMAGE)
/* start with "native" size (only used for table width calculations) */
node->w = node->content.image->w * 72.0f / 96.0f;
else
else if (node->type == FLOW_WORD || node->type == FLOW_SPACE || node->type == FLOW_SHYPHEN)
measure_string_w(ctx, node, hb_buf);
}
}
Expand Down
5 changes: 5 additions & 0 deletions mupdf/source/pdf/pdf-op-run.c
View file
Expand Up @@ -1006,6 +1006,11 @@ pdf_show_char(fz_context *ctx, pdf_run_processor *pr, int cid, fz_text_language
ucslen = 0;
if (fontdesc->to_unicode)
ucslen = pdf_lookup_cmap_full(fontdesc->to_unicode, cid, ucsbuf);

/* ignore obviously bad values in ToUnicode, fall back to the cid_to_ucs table */
if (ucslen == 1 && (ucsbuf[0] < 32 || (ucsbuf[0] >= 127 && ucsbuf[0] < 160)))
ucslen = 0;

if (ucslen == 0 && (size_t)cid < fontdesc->cid_to_ucs_len)
{
ucsbuf[0] = fontdesc->cid_to_ucs[cid];
Expand Down

0 comments on commit 987bb6a

Please sign in to comment.