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 from __future__ import print_function
80 import getopt, glob, os, re, shutil, sys, tempfile
82 from legacy_lyxpreview2ppm import extract_resolution, legacy_conversion_step1
84 from lyxpreview_tools import bibtex_commands, check_latex_log, copyfileobj, \
85 error, filter_pages, find_exe, find_exe_or_terminate, \
86 join_metrics_and_rename, latex_commands, latex_file_re, make_texcolor, \
87 pdflatex_commands, progress, run_command, run_latex, run_tex, \
88 warning, write_metrics_info
90 PY2 = sys.version_info[0] == 2
94 Usage: %s <options> <input file>
97 --dpi=<res>: Resolution per inch (default: 128)
98 --png, --ppm: Select the output format (default: png)
99 --fg=<color>: Foreground color (default: black, ie '000000')
100 --bg=<color>: Background color (default: white, ie 'ffffff')
101 --latex=<exe>: Specify the executable for latex (default: latex)
102 --bibtex=<exe>: Specify the executable for bibtex (default: bibtex)
103 --lilypond: Preprocess through lilypond-book (default: false)
104 --lilypond-book=<exe>:
105 The executable for lilypond-book (default: lilypond-book)
107 -d, --debug: Show the output from external commands
108 -h, --help: Show this help screen and exit
109 -v, --verbose: Show progress messages
111 The colors are hexadecimal strings, eg 'faf0e6'."""
112 return msg % prog_name
114 # Returns a list of tuples containing page number and ascent fraction
115 # extracted from dvipng output.
116 # Use write_metrics_info to create the .metrics file with this info
117 def extract_metrics_info(dvipng_stdout):
118 # "\[[0-9]+" can match two kinds of numbers: page numbers from dvipng
119 # and glyph numbers from mktexpk. The glyph numbers always match
120 # "\[[0-9]+\]" while the page number never is followed by "\]". Thus:
121 page_re = re.compile("\[([0-9]+)[^]]");
122 metrics_re = re.compile("depth=(-?[0-9]+) height=(-?[0-9]+)")
129 match = page_re.search(dvipng_stdout, pos)
132 page = match.group(1)
134 match = metrics_re.search(dvipng_stdout, pos)
139 # Calculate the 'ascent fraction'.
140 descent = float(match.group(1))
141 ascent = float(match.group(2))
145 # This is an empty image, forbid its display
147 elif ascent >= 0 or descent >= 0:
148 if abs(ascent + descent) > 0.1:
149 frac = ascent / (ascent + descent)
155 results.append((int(page), frac))
156 pos = match.end() + 2
159 error("Failed to extract metrics info from dvipng")
164 def fix_latex_file(latex_file, pdf_output):
165 # python 2 does not allow to declare a string as raw byte so we double
166 # the backslashes and remove the r preffix
167 def_re = re.compile(b"(\\\\newcommandx|\\\\global\\\\long\\\\def)"
170 tmp = tempfile.TemporaryFile()
174 for line in open(latex_file, 'rb').readlines():
175 if not pdf_output and line.startswith(b"\\documentclass"):
177 line += b"\\PassOptionsToPackage{draft}{microtype}\n"
179 match = def_re.match(line)
181 macroname = match.group(2)
182 if macroname in macros:
183 definecmd = match.group(1)
184 if definecmd == b"\\newcommandx":
186 line = line.replace(definecmd, b"\\renewcommandx")
188 macros.append(macroname)
192 copyfileobj(tmp, open(latex_file,"wb"), 1)
197 def convert_to_ppm_format(pngtopnm, basename):
198 png_file_re = re.compile("\.png$")
200 for png_file in glob.glob("%s*.png" % basename):
201 ppm_file = png_file_re.sub(".ppm", png_file)
203 p2p_cmd = '%s "%s"' % (pngtopnm, png_file)
204 p2p_status, p2p_stdout = run_command(p2p_cmd)
206 error("Unable to convert %s to ppm format" % png_file)
208 ppm = open(ppm_file, 'w')
209 ppm.write(p2p_stdout)
212 # Returns a tuple of:
213 # ps_pages: list of page indexes of pages containing PS literals
214 # pdf_pages: list of page indexes of pages requiring running pdflatex
215 # page_count: total number of pages
216 # pages_parameter: parameter for dvipng to exclude pages with PostScript/PDF
217 def find_ps_pages(dvi_file):
219 # FIXME: try with pdflatex
220 if not os.path.isfile(dvi_file):
221 error("No DVI output.")
223 # Check for PostScript specials in the dvi, badly supported by dvipng,
224 # and inclusion of PDF/PNG/JPG files.
225 # This is required for correct rendering of PSTricks and TikZ
226 dv2dt = find_exe_or_terminate(["dv2dt"])
227 dv2dt_call = '%s "%s"' % (dv2dt, dvi_file)
229 # The output from dv2dt goes to stdout
230 dv2dt_status, dv2dt_output = run_command(dv2dt_call)
231 psliteral_re = re.compile("^special[1-4] [0-9]+ '(\"|ps:)")
232 hyperref_re = re.compile("^special[1-4] [0-9]+ 'ps:.*/DEST pdfmark")
233 pdffile_re = re.compile("^special[1-4] [0-9]+ 'PSfile=.*\\.(pdf|png|jpg|jpeg|PDF|PNG|JPG|JPEG)")
235 # Parse the dtl file looking for PostScript specials and pdflatex files.
236 # Pages using PostScript specials or pdflatex files are recorded in
237 # ps_pages or pdf_pages, respectively, and then used to create a
238 # different LaTeX file for processing in legacy mode.
239 # If hyperref is detected, the corresponding page is recorded in pdf_pages.
247 for line in dv2dt_output.split("\n"):
249 if line.startswith("bop"):
255 if line.startswith("eop") and (page_has_ps or page_has_pdf):
256 # We save in a list all the PostScript/PDF pages
258 ps_pages.append(page_index)
260 pdf_pages.append(page_index)
261 ps_or_pdf_pages.append(page_index)
263 if psliteral_re.match(line) != None:
264 # Literal PostScript special detected!
265 # If hyperref is detected, put this page on the pdf pages list
266 if hyperref_re.match(line) != None:
271 elif pdffile_re.match(line) != None:
272 # Inclusion of pdflatex image file detected!
275 # Create the -pp parameter for dvipng
277 if len(ps_or_pdf_pages) > 0 and len(ps_or_pdf_pages) < page_index:
278 # Don't process Postscript/PDF pages with dvipng by selecting the
279 # wanted pages through the -pp parameter. E.g., dvipng -pp 4-12,14,64
280 pages_parameter = " -pp "
284 # Use page ranges, as a list of pages could exceed command line
285 # maximum length (especially under Win32)
286 for index in range(1, page_index + 1):
287 if (not index in ps_or_pdf_pages) and skip:
288 # We were skipping pages but current page shouldn't be skipped.
289 # Add this page to -pp, it could stay alone or become the
291 pages_parameter += str(index)
292 # Save the starting index to avoid things such as "11-11"
294 # We're not skipping anymore
296 elif (index in ps_or_pdf_pages) and (not skip):
297 # We weren't skipping but current page should be skipped
298 if last != index - 1:
299 # If the start index of the range is the previous page
300 # then it's not a range
301 pages_parameter += "-" + str(index - 1)
304 pages_parameter += ","
308 # Remove the trailing separator
309 pages_parameter = pages_parameter.rstrip(",")
310 # We've to manage the case in which the last page is closing a range
311 if (not index in ps_or_pdf_pages) and (not skip) and (last != index):
312 pages_parameter += "-" + str(index)
314 return (ps_pages, pdf_pages, page_index, pages_parameter)
325 output_format = "png"
326 script_name = argv[0]
328 # Parse and manipulate the command line arguments.
330 (opts, args) = getopt.gnu_getopt(argv[1:], "dhv", ["bibtex=", "bg=",
331 "debug", "dpi=", "fg=", "help", "latex=", "lilypond",
332 "lilypond-book=", "png", "ppm", "verbose"])
333 except getopt.GetoptError as err:
334 error("%s\n%s" % (err, usage(script_name)))
337 for opt, val in opts:
338 if opt in ("-h", "--help"):
339 print(usage(script_name))
341 elif opt == "--bibtex":
345 elif opt in ("-d", "--debug"):
346 import lyxpreview_tools
347 lyxpreview_tools.debug = True
352 error("Cannot convert %s to an integer value" % val)
355 elif opt == "--latex":
357 elif opt == "--lilypond":
359 elif opt == "--lilypond-book":
360 lilypond_book = [val]
361 elif opt in ("--png", "--ppm"):
362 output_format = opt[2:]
363 elif opt in ("-v", "--verbose"):
364 import lyxpreview_tools
365 lyxpreview_tools.verbose = True
367 # Determine input file
369 err = "A single input file is required, %s given" % (len(args) or "none")
370 error("%s\n%s" % (err, usage(script_name)))
373 dir, latex_file = os.path.split(input_path)
376 progress("Running Python %s" % str(sys.version_info[:3]))
377 progress("Starting %s..." % script_name)
378 progress("Output format: %s" % output_format)
379 progress("Foreground color: %s" % fg_color)
380 progress("Background color: %s" % bg_color)
381 progress("Resolution (dpi): %s" % dpi)
382 progress("File to process: %s" % input_path)
384 # Check for the input file
385 if not os.path.exists(input_path):
386 error('File "%s" not found.' % input_path)
390 # For python > 2 convert strings to bytes
392 fg_color = bytes(fg_color, 'ascii')
393 bg_color = bytes(bg_color, 'ascii')
395 fg_color_dvipng = make_texcolor(fg_color, False)
396 bg_color_dvipng = make_texcolor(bg_color, False)
398 # For python > 2 convert bytes to string
400 fg_color_dvipng = fg_color_dvipng.decode('ascii')
401 bg_color_dvipng = bg_color_dvipng.decode('ascii')
403 # External programs used by the script.
404 latex = find_exe_or_terminate(latex or latex_commands)
405 bibtex = find_exe(bibtex or bibtex_commands)
407 lilypond_book = find_exe_or_terminate(lilypond_book or
408 ["lilypond-book --safe"])
410 # These flavors of latex are known to produce pdf output
411 pdf_output = latex in pdflatex_commands
413 progress("Latex command: %s" % latex)
414 progress("Latex produces pdf output: %s" % pdf_output)
415 progress("Bibtex command: %s" % bibtex)
416 progress("Lilypond-book command: %s" % lilypond_book)
417 progress("Preprocess through lilypond-book: %s" % lilypond)
418 progress("Altering the latex file for font size and colors")
420 # Make sure that multiple defined macros and the microtype package
421 # don't cause issues in the latex file.
422 fix_latex_file(latex_file, pdf_output)
425 progress("Preprocess the latex file through %s" % lilypond_book)
427 lilypond_book += " --pdf"
428 lilypond_book += " --latex-program=%s" % latex.split()[0]
430 # Make a copy of the latex file
431 lytex_file = latex_file_re.sub(".lytex", latex_file)
432 shutil.copyfile(latex_file, lytex_file)
434 # Preprocess the latex file through lilypond-book.
435 lytex_status, lytex_stdout = run_tex(lilypond_book, lytex_file)
438 progress("Using the legacy conversion method (PDF support)")
439 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
440 bg_color, latex, pdf_output)
442 # This can go once dvipng becomes widespread.
443 dvipng = find_exe(["dvipng"])
445 progress("Using the legacy conversion method (dvipng not found)")
446 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
447 bg_color, latex, pdf_output)
449 dv2dt = find_exe(["dv2dt"])
451 progress("Using the legacy conversion method (dv2dt not found)")
452 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
453 bg_color, latex, pdf_output)
456 if output_format == "ppm":
457 pngtopnm = find_exe(["pngtopnm"])
459 progress("Using the legacy conversion method (pngtopnm not found)")
460 return legacy_conversion_step1(latex_file, dpi, output_format,
461 fg_color, bg_color, latex, pdf_output)
463 # Compile the latex file.
465 latex_status, latex_stdout = run_latex(latex, latex_file, bibtex)
466 latex_log = latex_file_re.sub(".log", latex_file)
468 progress("Will try to recover from %s failure" % latex)
469 error_pages = check_latex_log(latex_log)
471 # The dvi output file name
472 dvi_file = latex_file_re.sub(".dvi", latex_file)
474 # If there's no DVI output, look for PDF and go to legacy or fail
475 if not os.path.isfile(dvi_file):
476 # No DVI, is there a PDF?
477 pdf_file = latex_file_re.sub(".pdf", latex_file)
478 if os.path.isfile(pdf_file):
479 progress("%s produced a PDF output, fallback to legacy." \
480 % (os.path.basename(latex)))
481 progress("Using the legacy conversion method (PDF support)")
482 return legacy_conversion_step1(latex_file, dpi, output_format,
483 fg_color, bg_color, latex, True)
485 error("No DVI or PDF output. %s failed." \
486 % (os.path.basename(latex)))
488 # Look for PS literals or inclusion of pdflatex files in DVI pages
489 # ps_pages: list of indexes of pages containing PS literals
490 # pdf_pages: list of indexes of pages requiring running pdflatex
491 # page_count: total number of pages
492 # pages_parameter: parameter for dvipng to exclude pages with PostScript
493 (ps_pages, pdf_pages, page_count, pages_parameter) = find_ps_pages(dvi_file)
495 # If all pages need PostScript or pdflatex, directly use the legacy method.
496 if len(ps_pages) == page_count:
497 progress("Using the legacy conversion method (PostScript support)")
498 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
499 bg_color, latex, pdf_output)
500 elif len(pdf_pages) == page_count:
501 progress("Using the legacy conversion method (PDF support)")
502 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
503 bg_color, "pdflatex", True)
505 # Retrieve resolution
506 resolution = extract_resolution(latex_log, dpi)
508 # Run the dvi file through dvipng.
509 dvipng_call = '%s -Ttight -depth -height -D %d -fg "%s" -bg "%s" %s "%s"' \
510 % (dvipng, resolution, fg_color_dvipng, bg_color_dvipng, pages_parameter, dvi_file)
511 dvipng_status, dvipng_stdout = run_command(dvipng_call)
514 warning("%s failed to generate images from %s... fallback to legacy method" \
515 % (os.path.basename(dvipng), dvi_file))
516 progress("Using the legacy conversion method (dvipng failed)")
517 return legacy_conversion_step1(latex_file, dpi, output_format, fg_color,
518 bg_color, latex, pdf_output)
520 # Extract metrics info from dvipng_stdout.
521 metrics_file = latex_file_re.sub(".metrics", latex_file)
522 dvipng_metrics = extract_metrics_info(dvipng_stdout)
524 # If some pages require PostScript pass them to legacy method
525 if len(ps_pages) > 0:
526 # Create a new LaTeX file just for the snippets needing
528 legacy_latex_file = latex_file_re.sub("_legacy.tex", latex_file)
529 filter_pages(latex_file, legacy_latex_file, ps_pages)
531 # Pass the new LaTeX file to the legacy method
532 progress("Pages %s include postscript specials" % ps_pages)
533 progress("Using the legacy conversion method (PostScript support)")
534 legacy_status, legacy_metrics = legacy_conversion_step1(legacy_latex_file,
535 dpi, output_format, fg_color, bg_color, latex, pdf_output, True)
537 # Now we need to mix metrics data from dvipng and the legacy method
538 original_bitmap = latex_file_re.sub("%d." + output_format, legacy_latex_file)
539 destination_bitmap = latex_file_re.sub("%d." + output_format, latex_file)
541 # Join metrics from dvipng and legacy, and rename legacy bitmaps
542 join_metrics_and_rename(dvipng_metrics, legacy_metrics, ps_pages,
543 original_bitmap, destination_bitmap)
545 # If some pages require running pdflatex pass them to legacy method
546 if len(pdf_pages) > 0:
547 # Create a new LaTeX file just for the snippets needing
549 legacy_latex_file = latex_file_re.sub("_legacy.tex", latex_file)
550 filter_pages(latex_file, legacy_latex_file, pdf_pages)
552 # Pass the new LaTeX file to the legacy method
553 progress("Pages %s require processing with pdflatex" % pdf_pages)
554 progress("Using the legacy conversion method (PDF support)")
555 legacy_status, legacy_metrics = legacy_conversion_step1(legacy_latex_file,
556 dpi, output_format, fg_color, bg_color, "pdflatex", True, True)
558 # Now we need to mix metrics data from dvipng and the legacy method
559 original_bitmap = latex_file_re.sub("%d." + output_format, legacy_latex_file)
560 destination_bitmap = latex_file_re.sub("%d." + output_format, latex_file)
562 # Join metrics from dvipng and legacy, and rename legacy bitmaps
563 join_metrics_and_rename(dvipng_metrics, legacy_metrics, pdf_pages,
564 original_bitmap, destination_bitmap)
566 # Invalidate metrics for pages that produced errors
567 if len(error_pages) > 0:
569 for index in error_pages:
570 if index not in ps_pages and index not in pdf_pages:
571 dvipng_metrics.pop(index - 1)
572 dvipng_metrics.insert(index - 1, (index, -1.0))
575 warning("Failed to produce %d preview snippet(s)" % error_count)
577 # Convert images to ppm format if necessary.
578 if output_format == "ppm":
579 convert_to_ppm_format(pngtopnm, latex_file_re.sub("", latex_file))
581 # Actually create the .metrics file
582 write_metrics_info(dvipng_metrics, metrics_file)
584 return (0, dvipng_metrics)
586 if __name__ == "__main__":
587 sys.exit(main(sys.argv)[0])