Skip to content

Commit 3801718

Browse files
andiwandandiwand
committed
wvware figure name handling
1 parent fa9a87e commit 3801718

File tree

2 files changed

+55
-73
lines changed

2 files changed

+55
-73
lines changed

‎src/odr/internal/html/pdf2htmlex_wrapper.cpp‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -119,6 +119,8 @@ Html html::translate_poppler_pdf_file(const PopplerPdfFile &pdf_file,
119119
!param.poppler_data_dir.empty() ? param.poppler_data_dir.c_str()
120120
: nullptr);
121121

122+
// TODO not sure what the `progPath` is used for. it cannot be `nullptr`
123+
// TODO potentially just a cache dir?
122124
pdf2htmlEX::HTMLRenderer(fontconfig_path, param).process(&pdf_doc);
123125

124126
globalParams.reset();

‎src/odr/internal/html/wvware_wrapper.cpp‎

Lines changed: 53 additions & 73 deletions
Original file line numberDiff line numberDiff line change
@@ -48,6 +48,8 @@ struct TranslationState : public expand_data {
4848
int message = 0;
4949
} special_char_handler_state = {};
5050

51+
std::size_t figure_number = 0;
52+
5153
html::HtmlWriter out;
5254
};
5355

@@ -168,7 +170,6 @@ thing and just put ... instead of this
168170
return 1;
169171

170172
/* Mac specials (MV): */
171-
172173
case 0xf020:
173174
out.out() << " ";
174175
return 1;
@@ -218,10 +219,8 @@ void output_from_unicode(wvParseStruct *ps, std::uint16_t eachchar,
218219
auto *data = (TranslationState *)ps->userData;
219220
auto &out = data->out;
220221

221-
// TODO static
222-
static char cached_outputtype[33]; /* Last outputtype */
223-
static GIConv g_iconv_handle = (GIConv)-1; /* Cached iconv descriptor */
224-
static int need_swapping;
222+
GIConv g_iconv_handle = (GIConv)-1;
223+
int need_swapping;
225224
gchar *ibuf, *obuf;
226225
std::size_t ibuflen, obuflen, len, count, i;
227226
std::uint8_t buffer[2], buffer2[5];
@@ -230,12 +229,7 @@ void output_from_unicode(wvParseStruct *ps, std::uint16_t eachchar,
230229
return;
231230
}
232231

233-
if ((g_iconv_handle == (GIConv)-1) ||
234-
strcmp(cached_outputtype, outputtype) != 0) {
235-
if ((g_iconv_handle != (GIConv)-1)) {
236-
g_iconv_close(g_iconv_handle);
237-
}
238-
232+
{
239233
g_iconv_handle = g_iconv_open(outputtype, "UCS-2");
240234
if (g_iconv_handle == (GIConv)-1) {
241235
std::cerr << "g_iconv_open fail: " << errno
@@ -244,9 +238,6 @@ void output_from_unicode(wvParseStruct *ps, std::uint16_t eachchar,
244238
return;
245239
}
246240

247-
/* safe to cache the output type here */
248-
strcpy(cached_outputtype, outputtype);
249-
250241
/* Determining if unicode biteorder is swapped (glibc < 2.2) */
251242
need_swapping = 1;
252243

@@ -295,6 +286,9 @@ void output_from_unicode(wvParseStruct *ps, std::uint16_t eachchar,
295286
out.out() << buffer2[i];
296287
}
297288
}
289+
290+
// TODO iconv could be cached
291+
{ g_iconv_close(g_iconv_handle); }
298292
}
299293

300294
/// Originally from `wvWare.c` `wvStrangeNoGraphicData`
@@ -315,7 +309,7 @@ void strange_no_graphic_data(wvParseStruct *ps, int graphicstype) {
315309
/// https://github.com/opendocument-app/wvWare/blob/c015326b001f1ad6dfb1f5e718461c16c56cca5f/wvWare.c#L1239-L1287
316310
/// simplified to HTML output
317311
void print_graphics(wvParseStruct *ps, int graphicstype, int width, int height,
318-
char *source) {
312+
const std::string &source) {
319313
// upstream converts to PNG, we just use the original format as the browser
320314
// should support them
321315

@@ -329,15 +323,15 @@ void print_graphics(wvParseStruct *ps, int graphicstype, int width, int height,
329323
<< source << R"("/><br/>)";
330324
}
331325

332-
int handle_bitmap(wvParseStruct * /*ps*/, char *name, BitmapBlip *bitmap) {
326+
void handle_bitmap(wvParseStruct * /*ps*/, const std::string &name,
327+
BitmapBlip *bitmap) {
333328
wvStream *pwv = bitmap->m_pvBits;
334329
FILE *fd = nullptr;
335330
std::size_t size = 0, i;
336331

337-
fd = fopen(name, "wb");
332+
fd = fopen(name.c_str(), "wb");
338333
if (fd == nullptr) {
339-
fprintf(stderr, "\nCannot open %s for writing\n", name);
340-
exit(1);
334+
throw std::runtime_error("Cannot open " + name + " file for writing");
341335
}
342336
size = wvStream_size(pwv);
343337
wvStream_rewind(pwv);
@@ -346,11 +340,10 @@ int handle_bitmap(wvParseStruct * /*ps*/, char *name, BitmapBlip *bitmap) {
346340
fputc(read_8ubit(pwv), fd);
347341
}
348342
fclose(fd);
349-
wvTrace(("Name is %s\n", name));
350-
return 0;
351343
}
352344

353-
int handle_metafile(wvParseStruct * /*ps*/, char *name, MetaFileBlip *bitmap) {
345+
int handle_metafile(wvParseStruct * /*ps*/, const char *name,
346+
MetaFileBlip *bitmap) {
354347
wvStream *pwv = bitmap->m_pvBits;
355348
FILE *fd = nullptr;
356349
std::size_t size = 0, i;
@@ -364,12 +357,14 @@ int handle_metafile(wvParseStruct * /*ps*/, char *name, MetaFileBlip *bitmap) {
364357
size = wvStream_size(pwv);
365358
wvStream_rewind(pwv);
366359

367-
if (bitmap->m_fCompression == msocompressionDeflate)
360+
if (bitmap->m_fCompression == msocompressionDeflate) {
368361
decompressf = setdecom();
362+
}
369363

370364
if (!decompressf) {
371-
for (i = 0; i < size; i++)
365+
for (i = 0; i < size; i++) {
372366
fputc(read_8ubit(pwv), fd);
367+
}
373368
} else /* decompress here */
374369
{
375370
FILE *tmp = tmpfile();
@@ -385,49 +380,51 @@ int handle_metafile(wvParseStruct * /*ps*/, char *name, MetaFileBlip *bitmap) {
385380

386381
rewind(out);
387382

388-
for (i = 0; i < bitmap->m_cb; i++)
383+
for (i = 0; i < bitmap->m_cb; i++) {
389384
fputc(fgetc(out), fd);
385+
}
390386

391387
fclose(out);
392388
}
393389

394390
fclose(fd);
395-
wvTrace(("Name is %s\n", name));
396391
return 0;
397392
}
398393

399-
char *html_graphic(wvParseStruct *ps, Blip *blip) {
400-
char *name;
394+
std::string figure_name(wvParseStruct *ps) {
395+
auto *data = (TranslationState *)ps->userData;
396+
397+
std::size_t number = data->figure_number++;
398+
std::string name = "figure" + std::to_string(number);
399+
400+
return name;
401+
}
402+
403+
std::string html_graphic(wvParseStruct *ps, Blip *blip) {
404+
std::string name;
401405
wvStream *fd;
402406
char test[3];
403407

404-
// TODO handle figure name
405-
name = nullptr;
406-
if (name == nullptr) {
407-
return nullptr;
408-
}
408+
name = figure_name(ps);
409409

410410
/*
411411
temp hack to test older included bmps in word 6 and 7,
412412
should be wrapped in a modern escher strucure before getting
413413
to here, and then handled as normal
414414
*/
415-
wvTrace(("type is %d\n", blip->type));
416415
switch (blip->type) {
417416
case msoblipJPEG:
418417
case msoblipDIB:
419418
case msoblipPNG:
420419
fd = (blip->blip.bitmap.m_pvBits);
421420
test[2] = '\0';
422-
test[0] = read_8ubit(fd);
421+
test[0] = (char)read_8ubit(fd);
423422

424-
test[1] = read_8ubit(fd);
423+
test[1] = (char)read_8ubit(fd);
425424
wvStream_rewind(fd);
426425
if (!(strcmp(test, "BM"))) {
427-
wvAppendStr(&name, ".bmp");
428-
if (0 != handle_bitmap(ps, name, &blip->blip.bitmap)) {
429-
return nullptr;
430-
}
426+
name += ".bmp";
427+
handle_bitmap(ps, name, &blip->blip.bitmap);
431428
return name;
432429
}
433430
default:
@@ -436,40 +433,28 @@ char *html_graphic(wvParseStruct *ps, Blip *blip) {
436433

437434
switch (blip->type) {
438435
case msoblipWMF:
439-
wvAppendStr(&name, ".wmf");
440-
if (0 != handle_metafile(ps, name, &blip->blip.metafile)) {
441-
return nullptr;
442-
}
436+
name += ".wmf";
437+
handle_metafile(ps, name.c_str(), &blip->blip.metafile);
443438
break;
444439
case msoblipEMF:
445-
wvAppendStr(&name, ".emf");
446-
if (0 != handle_metafile(ps, name, &blip->blip.metafile)) {
447-
return nullptr;
448-
}
440+
name += ".emf";
441+
handle_metafile(ps, name.c_str(), &blip->blip.metafile);
449442
break;
450443
case msoblipPICT:
451-
wvAppendStr(&name, ".pict");
452-
if (0 != handle_metafile(ps, name, &blip->blip.metafile)) {
453-
return nullptr;
454-
}
444+
name += ".pict";
445+
handle_metafile(ps, name.c_str(), &blip->blip.metafile);
455446
break;
456447
case msoblipJPEG:
457-
wvAppendStr(&name, ".jpg");
458-
if (0 != handle_bitmap(ps, name, &blip->blip.bitmap)) {
459-
return nullptr;
460-
}
448+
name += ".jpg";
449+
handle_bitmap(ps, name.c_str(), &blip->blip.bitmap);
461450
break;
462451
case msoblipDIB:
463-
wvAppendStr(&name, ".dib");
464-
if (0 != handle_bitmap(ps, name, &blip->blip.bitmap)) {
465-
return nullptr;
466-
}
452+
name += ".dib";
453+
handle_bitmap(ps, name.c_str(), &blip->blip.bitmap);
467454
break;
468455
case msoblipPNG:
469-
wvAppendStr(&name, ".png");
470-
if (0 != handle_bitmap(ps, name, &blip->blip.bitmap)) {
471-
return nullptr;
472-
}
456+
name += ".png";
457+
handle_bitmap(ps, name.c_str(), &blip->blip.bitmap);
473458
break;
474459
}
475460
return name;
@@ -702,7 +687,6 @@ int special_char_handler(wvParseStruct *ps, std::uint16_t eachchar, CHP *achp) {
702687
case 0x01: {
703688
wvStream *f;
704689
Blip blip;
705-
char *name;
706690
long p = wvStream_tell(ps->data);
707691

708692
if (achp->fOle2 != 0) {
@@ -713,10 +697,9 @@ int special_char_handler(wvParseStruct *ps, std::uint16_t eachchar, CHP *achp) {
713697
wvGetPICF(wvQuerySupported(&ps->fib, nullptr), &picf, ps->data);
714698
f = picf.rgb;
715699
if (wv0x01(&blip, f, picf.lcb - picf.cbHeader) != 0) {
716-
name = html_graphic(ps, &blip);
700+
std::string name = html_graphic(ps, &blip);
717701
print_graphics(ps, 0x01, (int)wvTwipsToHPixels(picf.dxaGoal),
718702
(int)wvTwipsToVPixels(picf.dyaGoal), name);
719-
wvFree(name);
720703
} else {
721704
strange_no_graphic_data(ps, 0x01);
722705
}
@@ -726,7 +709,6 @@ int special_char_handler(wvParseStruct *ps, std::uint16_t eachchar, CHP *achp) {
726709
}
727710
case 0x08: {
728711
Blip blip;
729-
char *name;
730712
if (wvQuerySupported(&ps->fib, nullptr) == WORD8) {
731713
if (ps->nooffspa > 0) {
732714
fspa =
@@ -739,23 +721,21 @@ int special_char_handler(wvParseStruct *ps, std::uint16_t eachchar, CHP *achp) {
739721

740722
data->props = fspa;
741723
if (wv0x08(&blip, (int)fspa->spid, ps) != 0) {
742-
name = html_graphic(ps, &blip);
724+
std::string name = html_graphic(ps, &blip);
743725
print_graphics(
744726
ps, 0x08,
745727
(int)wvTwipsToHPixels((short)(fspa->xaRight - fspa->xaLeft)),
746728
(int)wvTwipsToVPixels((short)(fspa->yaBottom - fspa->yaTop)),
747729
name);
748-
wvFree(name);
749730
} else {
750731
strange_no_graphic_data(ps, 0x08);
751732
}
752733
} else {
753734
std::cerr << "nooffspa was <=0! Ignoring.\n";
754735
}
755736
} else {
756-
FDOA *fdoa;
757737
std::cerr << "pre word8 0x08 graphic, unsupported at the moment\n";
758-
fdoa =
738+
FDOA *fdoa =
759739
wvGetFDOAFromCP(ps->currentcp, ps->fdoa, ps->fdoapos, ps->nooffdoa);
760740
data->props = fdoa;
761741
}
@@ -823,7 +803,7 @@ Html html::translate_wvware_oldms_file(
823803
HtmlResourceLocator resourceLocator =
824804
local_resource_locator(output_path, config);
825805

826-
auto output_file_path = output_path + "/document.html";
806+
std::string output_file_path = output_path + "/document.html";
827807

828808
std::ofstream ostream(output_file_path, std::ios::out);
829809
if (!ostream.is_open()) {

0 commit comments

Comments
 (0)