Compare commits

...

2 Commits

Author SHA1 Message Date
Michael R Sweet
294f5e07c5
Add support for internal hyperlinks. 2024-12-15 10:23:59 -05:00
Michael R Sweet
4baafde74b
Add support for remote URL links. 2024-12-14 19:56:45 -05:00
3 changed files with 249 additions and 11 deletions

View File

@ -20,6 +20,7 @@
#include <stdio.h> #include <stdio.h>
#include <stdlib.h> #include <stdlib.h>
#include <string.h> #include <string.h>
#include <ctype.h>
#include <math.h> #include <math.h>
#ifdef _WIN32 #ifdef _WIN32
# include <io.h> # include <io.h>
@ -64,11 +65,28 @@ typedef struct docimage_s // Document image info
typedef struct doclink_s // Document link info typedef struct doclink_s // Document link info
{ {
const char *url; // Reference URL const char *url; // Target URL
pdfio_rect_t box; // Link box pdfio_rect_t box; // Link box
} doclink_t; } doclink_t;
#define DOCLINK_MAX 1000 // Maximum number of links/page #define DOCLINK_MAX 1000 // Maximum number of links per page
typedef struct docaction_s // Document action info
{
const char *target; // Target name
pdfio_obj_t *obj; // Link object
} docaction_t;
#define DOCACTION_MAX 10000 // Maximum number of actions per document
typedef struct doctarget_s // Document target info
{
char name[128]; // Target name
size_t page; // Target page
double y; // Target page position
} doctarget_t;
#define DOCTARGET_MAX 1000 // Maximum number of targets per document
typedef struct docdata_s // Document formatting data typedef struct docdata_s // Document formatting data
{ {
@ -86,9 +104,14 @@ typedef struct docdata_s // Document formatting data
docfont_t font; // Current font docfont_t font; // Current font
double fsize; // Current font size double fsize; // Current font size
doccolor_t color; // Current color doccolor_t color; // Current color
pdfio_obj_t *annots; // Annotations object (for links) pdfio_array_t *annots_array; // Annotations array (for links)
pdfio_obj_t *annots_obj; // Annotations object (for links)
size_t num_links; // Number of links for this page size_t num_links; // Number of links for this page
doclink_t links[DOCLINK_MAX]; // Links for this page doclink_t links[DOCLINK_MAX]; // Links for this page
size_t num_actions; // Number of actions for this document
docaction_t actions[DOCACTION_MAX]; // Actions for this document
size_t num_targets; // Number of targets for this document
doctarget_t targets[DOCTARGET_MAX]; // Targets for this document
} docdata_t; } docdata_t;
typedef struct linefrag_s // Line fragment typedef struct linefrag_s // Line fragment
@ -99,6 +122,7 @@ typedef struct linefrag_s // Line fragment
height; // Height of item height; // Height of item
size_t imagenum; // Image number size_t imagenum; // Image number
const char *text; // Text string const char *text; // Text string
const char *url; // Link URL string
bool ws; // Whitespace before text? bool ws; // Whitespace before text?
docfont_t font; // Text font docfont_t font; // Text font
doccolor_t color; // Text color doccolor_t color; // Text color
@ -203,11 +227,13 @@ static const char * const docfont_names[] =
// //
static void add_images(docdata_t *dd, mmd_t *doc); static void add_images(docdata_t *dd, mmd_t *doc);
static void add_links(docdata_t *dd);
static pdfio_obj_t *find_image(docdata_t *dd, const char *url, size_t *imagenum); static pdfio_obj_t *find_image(docdata_t *dd, const char *url, size_t *imagenum);
static void format_block(docdata_t *dd, mmd_t *block, docfont_t deffont, double fsize, double left, double right, const char *leader); static void format_block(docdata_t *dd, mmd_t *block, docfont_t deffont, double fsize, double left, double right, const char *leader);
static void format_code(docdata_t *dd, mmd_t *block, double left, double right); static void format_code(docdata_t *dd, mmd_t *block, double left, double right);
static void format_doc(docdata_t *dd, mmd_t *doc, double left, double right); static void format_doc(docdata_t *dd, mmd_t *doc, double left, double right);
static void format_table(docdata_t *dd, mmd_t *table, double left, double right); static void format_table(docdata_t *dd, mmd_t *table, double left, double right);
static void make_target_name(char *dst, const char *src, size_t dstsize);
static double measure_cell(docdata_t *dd, mmd_t *cell, tablecol_t *col); static double measure_cell(docdata_t *dd, mmd_t *cell, tablecol_t *col);
static mmd_t *mmd_walk_next(mmd_t *top, mmd_t *node); static mmd_t *mmd_walk_next(mmd_t *top, mmd_t *node);
static void new_page(docdata_t *dd); static void new_page(docdata_t *dd);
@ -216,6 +242,7 @@ static void render_line(docdata_t *dd, double margin_left, double margin_top, do
static void render_row(docdata_t *dd, size_t num_cols, tablecol_t *cols, tablerow_t *row); static void render_row(docdata_t *dd, size_t num_cols, tablecol_t *cols, tablerow_t *row);
static void set_color(docdata_t *dd, doccolor_t color); static void set_color(docdata_t *dd, doccolor_t color);
static void set_font(docdata_t *dd, docfont_t font, double fsize); static void set_font(docdata_t *dd, docfont_t font, double fsize);
static void write_actions(docdata_t *dd);
// //
@ -303,7 +330,12 @@ main(int argc, // I - Number of command-line arguments
format_doc(&dd, doc, dd.art_box.x1, dd.art_box.x2); format_doc(&dd, doc, dd.art_box.x1, dd.art_box.x2);
// Close the PDF and return... // Close the PDF and return...
if (dd.st)
{
pdfioStreamClose(dd.st); pdfioStreamClose(dd.st);
add_links(&dd);
}
write_actions(&dd);
pdfioFileClose(dd.pdf); pdfioFileClose(dd.pdf);
mmdFree(doc); mmdFree(doc);
@ -362,6 +394,88 @@ add_images(docdata_t *dd, // I - Document data
} }
//
// 'add_links()' - Add the page links, if any.
//
static void
add_links(docdata_t *dd) // I - Document data
{
size_t i; // Looping var
doclink_t *l; // Current link
pdfio_dict_t *dict; // Object dictionary
pdfio_obj_t *aobj, // Annotation object
*lobj; // Link object
pdfio_array_t *border; // Border values
// Add any links we have...
for (i = 0, l = dd->links; i < dd->num_links; i ++, l ++)
{
if (l->url[0] == '#')
{
// No remote action...
aobj = NULL;
}
else
{
// Create the link action (remote URL)
dict = pdfioDictCreate(dd->pdf);
pdfioDictSetName(dict, "S", "URI");
pdfioDictSetString(dict, "URI", pdfioStringCreate(dd->pdf, l->url));
aobj = pdfioFileCreateObj(dd->pdf, dict);
pdfioObjClose(aobj);
}
// Create the annotation object pointing to the action...
dict = pdfioDictCreate(dd->pdf);
pdfioDictSetName(dict, "Subtype", "Link");
pdfioDictSetRect(dict, "Rect", &l->box);
border = pdfioArrayCreate(dd->pdf);
pdfioArrayAppendNumber(border, 0.0);
pdfioArrayAppendNumber(border, 0.0);
pdfioArrayAppendNumber(border, 0.0);
pdfioDictSetArray(dict, "Border", border);
lobj = pdfioFileCreateObj(dd->pdf, dict);
if (l->url[0] == '#' && dd->num_actions < DOCACTION_MAX)
{
// Save this link action for later...
docaction_t *a = dd->actions + dd->num_actions;
// New action
a->target = l->url + 1;
a->obj = lobj;
dd->num_actions ++;
}
else if (aobj)
{
// Close out this link since we have a remote URL...
pdfioDictSetObj(dict, "A", aobj);
pdfioObjClose(lobj);
}
else
{
// Nothing that can be done for this one...
pdfioObjClose(lobj);
}
pdfioArrayAppendObj(dd->annots_array, lobj);
}
// Close the Annots array object...
pdfioObjClose(dd->annots_obj);
// Reset links...
dd->annots_array = NULL;
dd->annots_obj = NULL;
dd->num_links = 0;
}
// //
// 'find_image()' - Find an image in the document. // 'find_image()' - Find an image in the document.
// //
@ -584,6 +698,7 @@ format_block(docdata_t *dd, // I - Document data
frag->height = text ? fsize : height; frag->height = text ? fsize : height;
frag->imagenum = imagenum; frag->imagenum = imagenum;
frag->text = text; frag->text = text;
frag->url = url;
frag->ws = ws; frag->ws = ws;
frag->font = font; frag->font = font;
frag->color = color; frag->color = color;
@ -743,6 +858,18 @@ format_doc(docdata_t *dd, // I - Document data
dd->heading = mmdCopyAllText(current); dd->heading = mmdCopyAllText(current);
format_block(dd, current, DOCFONT_BOLD, heading_sizes[curtype - MMD_TYPE_HEADING_1], left, right, /*leader*/NULL); format_block(dd, current, DOCFONT_BOLD, heading_sizes[curtype - MMD_TYPE_HEADING_1], left, right, /*leader*/NULL);
if (dd->num_targets < DOCTARGET_MAX)
{
doctarget_t *t = dd->targets + dd->num_targets;
// New target
make_target_name(t->name, dd->heading, sizeof(t->name));
t->page = pdfioFileGetNumPages(dd->pdf) - 1;
t->y = dd->y + heading_sizes[curtype - MMD_TYPE_HEADING_1] * LINE_HEIGHT;
dd->num_targets ++;
}
break; break;
case MMD_TYPE_PARAGRAPH : case MMD_TYPE_PARAGRAPH :
@ -895,6 +1022,33 @@ format_table(docdata_t *dd, // I - Document data
} }
//
// 'make_target_name()' - Convert text to a target name.
//
static void
make_target_name(char *dst, // I - Destination buffer
const char *src, // I - Source text
size_t dstsize) // I - Size of destination buffer
{
char *dstptr = dst, // Pointer into destination
*dstend = dst + dstsize - 1; // End of destination
while (*src && dstptr < dstend)
{
if (isalnum(*src) || *src == '.' || *src == '-')
*dstptr++ = tolower(*src);
else if (*src == ' ')
*dstptr++ = '-';
src ++;
}
*dstptr = '\0';
}
// //
// 'measure_cell()' - Measure the dimensions of a table cell. // 'measure_cell()' - Measure the dimensions of a table cell.
// //
@ -1076,15 +1230,23 @@ new_page(docdata_t *dd) // I - Document data
// Close the current page... // Close the current page...
if (dd->st)
{
pdfioStreamClose(dd->st); pdfioStreamClose(dd->st);
add_links(dd);
}
// Prep the new page... // Prep the new page...
page_dict = pdfioDictCreate(dd->pdf); page_dict = pdfioDictCreate(dd->pdf);
dd->annots_array = pdfioArrayCreate(dd->pdf);
dd->annots_obj = pdfioFileCreateArrayObj(dd->pdf, dd->annots_array);
pdfioDictSetRect(page_dict, "MediaBox", &dd->media_box); pdfioDictSetRect(page_dict, "MediaBox", &dd->media_box);
// pdfioDictSetRect(page_dict, "CropBox", &dd->crop_box); // pdfioDictSetRect(page_dict, "CropBox", &dd->crop_box);
pdfioDictSetRect(page_dict, "ArtBox", &dd->art_box); pdfioDictSetRect(page_dict, "ArtBox", &dd->art_box);
pdfioDictSetObj(page_dict, "Annots", dd->annots_obj);
for (fontface = DOCFONT_REGULAR; fontface < DOCFONT_MAX; fontface ++) for (fontface = DOCFONT_REGULAR; fontface < DOCFONT_MAX; fontface ++)
pdfioPageDictAddFont(page_dict, docfont_names[fontface], dd->fonts[fontface]); pdfioPageDictAddFont(page_dict, docfont_names[fontface], dd->fonts[fontface]);
@ -1192,7 +1354,10 @@ render_line(docdata_t *dd, // I - Document data
if (!dd->st) if (!dd->st)
{
new_page(dd); new_page(dd);
margin_top = 0.0;
}
dd->y -= margin_top + lineheight; dd->y -= margin_top + lineheight;
if (dd->y < dd->art_box.y1) if (dd->y < dd->art_box.y1)
@ -1248,6 +1413,40 @@ render_line(docdata_t *dd, // I - Document data
pdfioContentTextShowf(dd->st, UNICODE_VALUE, " %s", frag->text); pdfioContentTextShowf(dd->st, UNICODE_VALUE, " %s", frag->text);
else else
pdfioContentTextShow(dd->st, UNICODE_VALUE, frag->text); pdfioContentTextShow(dd->st, UNICODE_VALUE, frag->text);
if (frag->type == MMD_TYPE_LINKED_TEXT && frag->url && dd->num_links < DOCLINK_MAX)
{
doclink_t *l = dd->links + dd->num_links;
// Pointer to this link record
if (!strcmp(frag->url, "@"))
{
// Use mapped text as link target...
char targetlink[129]; // Targeted link
targetlink[0] = '#';
make_target_name(targetlink + 1, frag->text, sizeof(targetlink) - 1);
l->url = pdfioStringCreate(dd->pdf, targetlink);
}
else if (!strcmp(frag->url, "@@"))
{
// Use literal text as anchor...
l->url = pdfioStringCreatef(dd->pdf, "#%s", frag->text);
}
else
{
// Use URL as-is...
l->url = frag->url;
}
l->box.x1 = frag->x;
l->box.y1 = dd->y;
l->box.x2 = frag->x + frag->width;
l->box.y2 = dd->y + frag->height;
dd->num_links ++;
}
} }
else else
{ {
@ -1391,3 +1590,42 @@ set_font(docdata_t *dd, // I - Document data
dd->font = font; dd->font = font;
dd->fsize = fsize; dd->fsize = fsize;
} }
//
// 'write_actions()' - Write remaining actions to the PDF file.
//
static void
write_actions(docdata_t *dd) // I - Document data
{
size_t i, j; // Looping vars
docaction_t *a; // Current action
doctarget_t *t; // Current target
for (i = 0, a = dd->actions; i < dd->num_actions; i ++, a ++)
{
for (j = 0, t = dd->targets; j < dd->num_targets; j ++, t ++)
{
if (!strcmp(a->target, t->name))
break;
}
if (j < dd->num_targets)
{
pdfio_array_t *dest = pdfioArrayCreate(dd->pdf);
// Destination array
pdfioArrayAppendObj(dest, pdfioFileGetPage(dd->pdf, t->page));
pdfioArrayAppendName(dest, "XYZ");
pdfioArrayAppendNumber(dest, PAGE_LEFT);
pdfioArrayAppendNumber(dest, t->y);
pdfioArrayAppendNumber(dest, 0.0);
pdfioDictSetArray(pdfioObjGetDict(a->obj), "Dest", dest);
}
pdfioObjClose(a->obj);
}
}

View File

@ -12,16 +12,16 @@ pages in a PDF file. It demonstrates how to:
- Format text, - Format text,
- Embed JPEG and PNG images, - Embed JPEG and PNG images,
- Add headers and footers, and - Add headers and footers, and
- (Future) Add hyperlinks. - Add hyperlinks.
Source Files Source Files
------------ ------------
The `md2pdf` program is organized into three source files: `md2pdf.c` which The `md2pdf` program is organized into three source files: `md2pdf.c` which
contains the code to format the markdown content and `mmd.h` and `mmd.c` which contains the code to format the markdown content and `mmd.h` and `mmd.c` (from
load the markdown content, which come from the [Miniature Markdown Library][MMD] the [Miniature Markdown Library][MMD] project) which load the markdown content.
project which is provided under the terms of the Apache License v2.0.
![Apache License v2.0](apache-badge.png)
[MMD]: https://www.msweet.org/mmd/ [MMD]: https://www.msweet.org/mmd/

View File

@ -544,7 +544,7 @@ mmdLoadIO(mmd_t *root, // I - Root node for document or `NULL` for a new d
block = NULL; block = NULL;
continue; continue;
} }
else if (*lineptr == '>' && (lineptr - linestart) < 4) else if (stackptr->parent->type != MMD_TYPE_CODE_BLOCK && *lineptr == '>' && (lineptr - linestart) < 4)
{ {
// Block quote. See if there is an existing blockquote... // Block quote. See if there is an existing blockquote...
DEBUG_printf(" BLOCKQUOTE (stackptr=%ld)\n", stackptr - stack); DEBUG_printf(" BLOCKQUOTE (stackptr=%ld)\n", stackptr - stack);