1 # -*- coding: utf-8 -*-
3 # file lyxpreview2bitmap.py
4 # This file is part of LyX, the document processor.
5 # Licence details can be found in the file COPYING.
8 # with much advice from members of the preview-latex project:
9 # David Kastrup, dak@gnu.org and
10 # Jan-Åke Larsson, jalar@mai.liu.se.
12 # Full author contact details are available in file CREDITS
14 # This script takes a LaTeX file and generates a collection of
15 # png or ppm image files, one per previewed snippet.
18 # * python 2.4 or later (subprocess module);
19 # * A latex executable;
23 # * pngtoppm (if outputing ppm format images).
25 # preview.sty and dvipng are part of the preview-latex project
26 # http://preview-latex.sourceforge.net/
28 # preview.sty can alternatively be obtained from
29 # CTAN/support/preview-latex/
32 # lyxpreview2bitmap.py --bg=faf0e6 0lyxpreview.tex
34 # This script takes one obligatory argument:
36 # <input file>: The name of the .tex file to be converted.
38 # and these optional arguments:
40 # --png, --ppm: The desired output format. Either 'png' or 'ppm'.
41 # --dpi=<res>: A scale factor, used to ascertain the resolution of the
42 # generated image which is then passed to gs.
43 # --fg=<color>: The foreground color as a hexadecimal string, eg '000000'.
44 # --bg=<color>: The background color as a hexadecimal string, eg 'faf0e6'.
45 # --latex=<exe>: The converter for latex files. Default is latex.
46 # --bibtex=<exe>: The converter for bibtex files. Default is bibtex.
47 # --lilypond: Preprocess through lilypond-book. Default is false.
48 # --lilypond-book=<exe>:
49 # The converter for lytex files. Default is lilypond-book.
51 # -d, --debug Show the output from external commands.
52 # -h, --help Show an help screen and exit.
53 # -v, --verbose Show progress messages.
55 # Decomposing TEXFILE's name as DIR/BASE.tex, this script will,
56 # if executed successfully, leave in DIR:
57 # * a (possibly large) number of image files with names
59 # * a file BASE.metrics, containing info needed by LyX to position
60 # the images correctly on the screen.
62 # What does this script do?
63 # 1) Call latex/pdflatex/xelatex/whatever (CONVERTER parameter)
64 # 2) If the output is a PDF fallback to legacy
65 # 3) Otherwise check each page of the DVI (with dv2dt) looking for
66 # PostScript literals, not well supported by dvipng. Pages
67 # containing them are passed to the legacy method in a new LaTeX file.
68 # 4) Call dvipng on the pages without PS literals
69 # 5) Join metrics info coming from both methods (legacy and dvipng)
70 # and write them to file
72 # dvipng is fast but gives problem in several cases, like with
73 # PSTricks, TikZ and other packages using PostScript literals
74 # for all these cases the legacy route is taken (step 3).
75 # Moreover dvipng can't work with PDF files, so, if the CONVERTER
76 # paramter is pdflatex we have to fallback to legacy route (step 2).
78 import getopt, glob, os, re, shutil, string, sys
80 from legacy_lyxpreview2ppm import legacy_conversion_step1
82 from lyxpreview_tools import bibtex_commands, check_latex_log, copyfileobj, \
83 error, filter_pages, find_exe, find_exe_or_terminate, \
84 join_metrics_and_rename, latex_commands, latex_file_re, make_texcolor, \
85 mkstemp, pdflatex_commands, progress, run_command, run_latex, run_tex, \
86 warning, write_metrics_info
91 Usage: %s <options> <input file>
94 --dpi=<res>: Resolution per inch (default: 128)
95 --png, --ppm: Select the output format (default: png)
96 --fg=<color>: Foreground color (default: black, ie '000000')
97 --bg=<color>: Background color (default: white, ie 'ffffff')
98 --latex=<exe>: Specify the executable for latex (default: latex)
99 --bibtex=<exe>: Specify the executable for bibtex (default: bibtex)
100 --lilypond: Preprocess through lilypond-book (default: false)
101 --lilypond-book=<exe>:
102 The executable for lilypond-book (default: lilypond-book)
104 -d, --debug: Show the output from external commands
105 -h, --help: Show this help screen and exit
106 -v, --verbose: Show progress messages
108 The colors are hexadecimal strings, eg 'faf0e6'."""
109 return msg % prog_name
111 # Returns a list of tuples containing page number and ascent fraction
112 # extracted from dvipng output.
113 # Use write_metrics_info to create the .metrics file with this info
114 def extract_metrics_info(dvipng_stdout):
115 # "\[[0-9]+" can match two kinds of numbers: page numbers from dvipng
116 # and glyph numbers from mktexpk. The glyph numbers always match
117 # "\[[0-9]+\]" while the page number never is followed by "\]". Thus:
118 page_re = re.compile("\[([0-9]+)[^]]");
119 metrics_re = re.compile("depth=(-?[0-9]+) height=(-?[0-9]+)")
126 match = page_re.search(dvipng_stdout, pos)
129 page = match.group(1)
131 match = metrics_re.search(dvipng_stdout, pos)
136 # Calculate the 'ascent fraction'.
137 descent = string.atof(match.group(1))
138 ascent = string.atof(match.group(2))
142 # This is an empty image, forbid its display
144 elif ascent >= 0 or descent >= 0:
145 if abs(ascent + descent) > 0.1:
146 frac = ascent / (ascent + descent)
152 results.append((int(page), frac))
153 pos = match.end() + 2
156 error("Failed to extract metrics info from dvipng")
161 def fix_latex_file(latex_file):
162 documentclass_re = re.compile("(\\\\documentclass\[)(1[012]pt,?)(.+)")
167 for line in open(latex_file, 'r').readlines():
168 match = documentclass_re.match(line)
174 tmp.write("%s%s\n" % (match.group(1), match.group(3)))
177 copyfileobj(tmp, open(latex_file,"wb"), 1)
182 def convert_to_ppm_format(pngtopnm, basename):
183 png_file_re = re.compile("\.png$")
185 for png_file in glob.glob("%s*.png" % basename):
186 ppm_file = png_file_re.sub(".ppm", png_file)
188 p2p_cmd = '%s "%s"' % (pngtopnm, png_file)
189 p2p_status, p2p_stdout = run_command(p2p_cmd)
191 error("Unable to convert %s to ppm format" % png_file)
193 ppm = open(ppm_file, 'w')
194 ppm.write(p2p_stdout)
197 # Returns a tuple of:
198 # ps_pages: list of page indexes of pages containing PS literals
199 # pdf_pages: list of page indexes of pages requiring running pdflatex
200 # page_count: total number of pages
201 # pages_parameter: parameter for dvipng to exclude pages with PostScript/PDF
202 def find_ps_pages(dvi_file):
204 # FIXME: try with pdflatex
205 if not os.path.isfile(dvi_file):
206 error("No DVI output.")
208 # Check for PostScript specials in the dvi, badly supported by dvipng,
209 # and inclusion of PDF/PNG/JPG files.
210 # This is required for correct rendering of PSTricks and TikZ
211 dv2dt = find_exe_or_terminate(["dv2dt"])
212 dv2dt_call = '%s "%s"' % (dv2dt, dvi_file)
214 # The output from dv2dt goes to stdout
215 dv2dt_status, dv2dt_output = run_command(dv2dt_call)
216 psliteral_re = re.compile("^special[1-4] [0-9]+ '(\"|ps:)")
217 hyperref_re = re.compile("^special[1-4] [0-9]+ 'ps:.*/DEST pdfmark")
218 pdffile_re = re.compile("^special[1-4] [0-9]+ 'PSfile=.*\\.(pdf|png|jpg|jpeg|PDF|PNG|JPG|JPEG)")
220 # Parse the dtl file looking for PostScript specials and pdflatex files.
221 # Pages using PostScript specials or pdflatex files are recorded in
222 # ps_pages or pdf_pages, respectively, and then used to create a
223 # different LaTeX file for processing in legacy mode.
224 # If hyperref is detected, the corresponding page is recorded in pdf_pages.
232 for line in dv2dt_output.split("\n"):
234 if line.startswith("bop"):
240 if line.startswith("eop") and (page_has_ps or page_has_pdf):
241 # We save in a list all the PostScript/PDF pages
243 ps_pages.append(page_index)
245 pdf_pages.append(page_index)
246 ps_or_pdf_pages.append(page_index)
248 if psliteral_re.match(line) != None:
249 # Literal PostScript special detected!
250 # If hyperref is detected, put this page on the pdf pages list
251 if hyperref_re.match(line) != None:
256 elif pdffile_re.match(line) != None:
257 # Inclusion of pdflatex image file detected!
260 # Create the -pp parameter for dvipng
262 if len(ps_or_pdf_pages) > 0 and len(ps_or_pdf_pages) < page_index:
263 # Don't process Postscript/PDF pages with dvipng by selecting the
264 # wanted pages through the -pp parameter. E.g., dvipng -pp 4-12,14,64
265 pages_parameter = " -pp "
269 # Use page ranges, as a list of pages could exceed command line
270 # maximum length (especially under Win32)
271 for index in xrange(1, page_index + 1):
272 if (not index in ps_or_pdf_pages) and skip:
273 # We were skipping pages but current page shouldn't be skipped.
274 # Add this page to -pp, it could stay alone or become the
276 pages_parameter += str(index)
277 # Save the starting index to avoid things such as "11-11"
279 # We're not skipping anymore
281 elif (index in ps_or_pdf_pages) and (not skip):
282 # We weren't skipping but current page should be skipped
283 if last != index - 1:
284 # If the start index of the range is the previous page
285 # then it's not a range
286 pages_parameter += "-" + str(index - 1)
289 pages_parameter += ","
293 # Remove the trailing separator
294 pages_parameter = pages_parameter.rstrip(",")
295 # We've to manage the case in which the last page is closing a range
296 if (not index in ps_or_pdf_pages) and (not skip) and (last != index):
297 pages_parameter += "-" + str(index)
299 return (ps_pages, pdf_pages, page_index, pages_parameter)
310 output_format = "png"
311 script_name = argv[0]
313 # Parse and manipulate the command line arguments.
315 (opts, args) = getopt.gnu_getopt(argv[1:], "dhv", ["bibtex=", "bg=",
316 "debug", "dpi=", "fg=", "help", "latex=", "lilypond",
317 "lilypond-book=", "png", "ppm", "verbose"])
318 except getopt.GetoptError, err:
319 error("%s\n%s" % (err, usage(script_name)))
322 for opt, val in opts:
323 if opt in ("-h", "--help"):
324 print usage(script_name)
326 elif opt == "--bibtex":
330 elif opt in ("-d", "--debug"):
331 import lyxpreview_tools
332 lyxpreview_tools.debug = True
335 dpi = string.atoi(val)
337 error("Cannot convert %s to an integer value" % val)
340 elif opt == "--latex":
342 elif opt == "--lilypond":
344 elif opt == "--lilypond-book":
345 lilypond_book = [val]
346 elif opt in ("--png", "--ppm"):
347 output_format = opt[2:]
348 elif opt in ("-v", "--verbose"):
349 import lyxpreview_tools
350 lyxpreview_tools.verbose = True
352 # Determine input file
354 err = "A single input file is required, %s given" % (len(args) or "none")
355 error("%s\n%s" % (err, usage(script_name)))
358 dir, latex_file = os.path.split(input_path)
361 progress("Starting %s..." % script_name)
362 progress("Output format: %s" % output_format)
363 progress("Foreground color: %s" % fg_color)
364 progress("Background color: %s" % bg_color)
365 progress("Resolution (dpi): %s" % dpi)
366 progress("File to process: %s" % input_path)
368 # Check for the input file
369 if not os.path.exists(input_path):
370 error('File "%s" not found.' % input_path)
374 fg_color_dvipng = make_texcolor(fg_color, False)
375 bg_color_dvipng = make_texcolor(bg_color, False)
377 # External programs used by the script.
378 latex = find_exe_or_terminate(latex or latex_commands)
379 bibtex = find_exe(bibtex or bibtex_commands)
381 lilypond_book = find_exe_or_terminate(lilypond_book or
382 ["lilypond-book --safe"])
384 # These flavors of latex are known to produce pdf output
385 pdf_output = latex in pdflatex_commands
387 progress("Latex command: %s" % latex)
388 progress("Latex produces pdf output: %s" % pdf_output)
389 progress("Bibtex command: %s" % bibtex)
390 progress("Lilypond-book command: %s" % lilypond_book)
391 progress("Preprocess through lilypond-book: %s" % lilypond)
392 progress("Altering the latex file for font size and colors")
394 # Omit font size specification in latex file and make sure multiple
395 # defined macros are not an issue.
396 fix_latex_file(latex_file)
399 progress("Preprocess the latex file through %s" % lilypond_book)
401 lilypond_book += " --pdf"
402 lilypond_book += " --latex-program=%s" % latex.split()[0]
404 # Make a copy of the latex file
405 lytex_file = latex_file_re.sub(".lytex", latex_file)
406 shutil.copyfile(latex_file, lytex_file)
408 # Preprocess the latex file through lilypond-book.
409 lytex_status, lytex_stdout = run_tex(lilypond_book, lytex_file)
412 progress("Using the legacy conversion method (PDF support)")
413 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
414 bg_color, latex, pdf_output)
416 # This can go once dvipng becomes widespread.
417 dvipng = find_exe(["dvipng"])
419 progress("Using the legacy conversion method (dvipng not found)")
420 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
421 bg_color, latex, pdf_output)
423 dv2dt = find_exe(["dv2dt"])
425 progress("Using the legacy conversion method (dv2dt not found)")
426 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
427 bg_color, latex, pdf_output)
430 if output_format == "ppm":
431 pngtopnm = find_exe(["pngtopnm"])
433 progress("Using the legacy conversion method (pngtopnm not found)")
434 return legacy_conversion_step1(latex_file, dpi, output_format,
435 fg_color, bg_color, latex, pdf_output)
437 # Compile the latex file.
439 latex_status, latex_stdout = run_latex(latex, latex_file, bibtex)
441 warning("trying to recover from failed compilation")
442 error_pages = check_latex_log(latex_file_re.sub(".log", latex_file))
444 # The dvi output file name
445 dvi_file = latex_file_re.sub(".dvi", latex_file)
447 # If there's no DVI output, look for PDF and go to legacy or fail
448 if not os.path.isfile(dvi_file):
449 # No DVI, is there a PDF?
450 pdf_file = latex_file_re.sub(".pdf", latex_file)
451 if os.path.isfile(pdf_file):
452 progress("%s produced a PDF output, fallback to legacy." \
453 % (os.path.basename(latex)))
454 progress("Using the legacy conversion method (PDF support)")
455 return legacy_conversion_step1(latex_file, dpi, output_format,
456 fg_color, bg_color, latex, True)
458 error("No DVI or PDF output. %s failed." \
459 % (os.path.basename(latex)))
461 # Look for PS literals or inclusion of pdflatex files in DVI pages
462 # ps_pages: list of indexes of pages containing PS literals
463 # pdf_pages: list of indexes of pages requiring running pdflatex
464 # page_count: total number of pages
465 # pages_parameter: parameter for dvipng to exclude pages with PostScript
466 (ps_pages, pdf_pages, page_count, pages_parameter) = find_ps_pages(dvi_file)
468 # If all pages need PostScript or pdflatex, directly use the legacy method.
469 if len(ps_pages) == page_count:
470 progress("Using the legacy conversion method (PostScript support)")
471 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
472 bg_color, latex, pdf_output)
473 elif len(pdf_pages) == page_count:
474 progress("Using the legacy conversion method (PDF support)")
475 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
476 bg_color, "pdflatex", True)
478 # Run the dvi file through dvipng.
479 dvipng_call = '%s -Ttight -depth -height -D %d -fg "%s" -bg "%s" %s "%s"' \
480 % (dvipng, dpi, fg_color_dvipng, bg_color_dvipng, pages_parameter, dvi_file)
481 dvipng_status, dvipng_stdout = run_command(dvipng_call)
484 warning("%s failed to generate images from %s... fallback to legacy method" \
485 % (os.path.basename(dvipng), dvi_file))
486 progress("Using the legacy conversion method (dvipng failed)")
487 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
488 bg_color, latex, pdf_output)
490 # Extract metrics info from dvipng_stdout.
491 metrics_file = latex_file_re.sub(".metrics", latex_file)
492 dvipng_metrics = extract_metrics_info(dvipng_stdout)
494 # If some pages require PostScript pass them to legacy method
495 if len(ps_pages) > 0:
496 # Create a new LaTeX file just for the snippets needing
498 legacy_latex_file = latex_file_re.sub("_legacy.tex", latex_file)
499 filter_pages(latex_file, legacy_latex_file, ps_pages)
501 # Pass the new LaTeX file to the legacy method
502 progress("Pages %s include postscript specials" % ps_pages)
503 progress("Using the legacy conversion method (PostScript support)")
504 legacy_status, legacy_metrics = legacy_conversion_step1(legacy_latex_file,
505 dpi, output_format, fg_color, bg_color, latex, pdf_output, True)
507 # Now we need to mix metrics data from dvipng and the legacy method
508 original_bitmap = latex_file_re.sub("%d." + output_format, legacy_latex_file)
509 destination_bitmap = latex_file_re.sub("%d." + output_format, latex_file)
511 # Join metrics from dvipng and legacy, and rename legacy bitmaps
512 join_metrics_and_rename(dvipng_metrics, legacy_metrics, ps_pages,
513 original_bitmap, destination_bitmap)
515 # If some pages require running pdflatex pass them to legacy method
516 if len(pdf_pages) > 0:
517 # Create a new LaTeX file just for the snippets needing
519 legacy_latex_file = latex_file_re.sub("_legacy.tex", latex_file)
520 filter_pages(latex_file, legacy_latex_file, pdf_pages)
522 # Pass the new LaTeX file to the legacy method
523 progress("Pages %s require processing with pdflatex" % pdf_pages)
524 progress("Using the legacy conversion method (PDF support)")
525 legacy_status, legacy_metrics = legacy_conversion_step1(legacy_latex_file,
526 dpi, output_format, fg_color, bg_color, "pdflatex", True, True)
528 # Now we need to mix metrics data from dvipng and the legacy method
529 original_bitmap = latex_file_re.sub("%d." + output_format, legacy_latex_file)
530 destination_bitmap = latex_file_re.sub("%d." + output_format, latex_file)
532 # Join metrics from dvipng and legacy, and rename legacy bitmaps
533 join_metrics_and_rename(dvipng_metrics, legacy_metrics, pdf_pages,
534 original_bitmap, destination_bitmap)
536 # Invalidate metrics for pages that produced errors
537 if len(error_pages) > 0:
538 for index in error_pages:
539 if index not in ps_pages and index not in pdf_pages:
540 dvipng_metrics.pop(index - 1)
541 dvipng_metrics.insert(index - 1, (index, -1.0))
543 # Convert images to ppm format if necessary.
544 if output_format == "ppm":
545 convert_to_ppm_format(pngtopnm, latex_file_re.sub("", latex_file))
547 # Actually create the .metrics file
548 write_metrics_info(dvipng_metrics, metrics_file)
550 return (0, dvipng_metrics)
552 if __name__ == "__main__":
553 sys.exit(main(sys.argv)[0])