converters/latex.py

from converters.base import Converter
from converters.utils import markdown2latex, remove_ansi
import os
import subprocess
import sys

inkscape = 'inkscape'
if sys.platform == 'darwin':
    inkscape = '/Applications/Inkscape.app/Contents/Resources/bin/inkscape'
    if not os.path.exists(inkscape):
        inkscape = None


class ConverterLaTeX(Converter):
    """Converts a notebook to a .tex file suitable for pdflatex.

    Note: this converter *needs*:

    - `pandoc`: for all conversion of markdown cells.  If your notebook only
       has Raw cells, pandoc will not be needed.

    -  `inkscape`: if your notebook has SVG figures.  These need to be
       converted to PDF before inclusion in the TeX file, as LaTeX doesn't
       understand SVG natively.

    You will in general obtain much better final PDF results if you configure
    the matplotlib backend to create SVG output with

    %config InlineBackend.figure_format = 'svg'

    (or set the equivalent flag at startup or in your configuration profile).
    """
    inkscape = inkscape
    extension = 'tex'
    documentclass = 'article'
    documentclass_options = '11pt,english'
    heading_map = {1: r'\section',
                   2: r'\subsection',
                   3: r'\subsubsection',
                   4: r'\paragraph',
                   5: r'\subparagraph',
                   6: r'\subparagraph'}
    user_preamble = None
    exclude_cells = []

    def in_env(self, environment, lines):
        """Return list of environment lines for input lines

        Parameters
        ----------
        env : string
          Name of the environment to bracket with begin/end.

        lines: """
        out = [ur'\begin{%s}' % environment]
        if isinstance(lines, basestring):
            out.append(lines)
        else:  # list
            out.extend(lines)
        out.append(ur'\end{%s}' % environment)
        return out

    def convert(self, *args, **kwargs):
        # The main body is done by the logic in the parent class, and that's
        # all we need if preamble support has been turned off.
        body = super(ConverterLaTeX, self).convert(*args, **kwargs)
        if not self.with_preamble:
            return body
        # But if preamble is on, then we need to construct a proper, standalone
        # tex file.

        # Tag the document at the top and set latex class
        final = [r'%% This file was auto-generated by IPython, do NOT edit',
                 r'%% Conversion from the original notebook file:',
                 r'%% {0}'.format(self.infile),
                 r'%%',
                 r'\documentclass[%s]{%s}' % (self.documentclass_options,
                                              self.documentclass),
                 '',
                 ]
        # Load our own preamble, which is stored next to the main file.  We
        # need to be careful in case the script entry point is a symlink
        myfile = os.path.realpath(__file__)
        preamble = '../preamble.tex'
        with open(os.path.join(os.path.dirname(myfile), preamble)) as f:
            final.append(f.read())

        # Load any additional user-supplied preamble
        if self.user_preamble:
            final.extend(['', '%% Adding user preamble from file:',
                          '%% {0}'.format(self.user_preamble), ''])
            with open(self.user_preamble) as f:
                final.append(f.read())

        # Include document body
        final.extend([r'\begin{document}', '',
                      body,
                      r'\end{document}', ''])
        # Retun value must be a string
        return '\n'.join(final)

    def render_heading(self, cell):
        marker = self.heading_map[cell.level]
        return ['%s{%s}' % (marker, cell.source)]

    def render_code(self, cell):
        if not cell.input:
            return []

        # Cell codes first carry input code, we use lstlisting for that
        lines = [ur'\begin{codecell}']

        if 'source' not in self.exclude_cells:
            lines.extend(self.in_env('codeinput',
                                     self.in_env('lstlisting', cell.input)))
        else:
            # Empty output is still needed for LaTeX formatting
            lines.extend(self.in_env('codeinput', ''))

        outlines = []
        if 'output' not in self.exclude_cells:
            for output in cell.outputs:
                conv_fn = self.dispatch(output.output_type)
                outlines.extend(conv_fn(output))

        # And then output of many possible types; use a frame for all of it.
        if outlines:
            lines.extend(self.in_env('codeoutput', outlines))

        lines.append(ur'\end{codecell}')

        return lines

    def _img_lines(self, img_file):
        return self.in_env('center',
                [r'\includegraphics[width=6in]{%s}' % img_file, r'\par'])

    def _svg_lines(self, img_file):
        base_file = os.path.splitext(img_file)[0]
        pdf_file = base_file + '.pdf'
        subprocess.check_call([self.inkscape, '--export-pdf=%s' % pdf_file,
                               img_file])
        return self._img_lines(pdf_file)

    def render_markdown(self, cell):
        return [markdown2latex(cell.source)]

    def render_pyout(self, output):
        lines = []

        # output is a dictionary like object with type as a key
        if 'latex' in output:
            lines.extend(self.in_env('equation*', output.latex.lstrip('$$').rstrip('$$')))

        if 'text' in output:
            lines.extend(self.in_env('verbatim', output.text))

        return lines

    def render_pyerr(self, output):
        # Note: a traceback is a *list* of frames.
        return self.in_env('traceback',
                        self.in_env('verbatim',
                                 remove_ansi('\n'.join(output.traceback))))

    def render_raw(self, cell):
        if self.raw_as_verbatim:
            return self.in_env('verbatim', cell.source)
        else:
            return [cell.source]

    def _unknown_lines(self, data):
        return [r'{\vspace{5mm}\bf WARNING:: unknown cell:}'] + \
          self.in_env('verbatim', data)

    def render_display_format_text(self, output):
        lines = []

        if 'text' in output:
            lines.extend(self.in_env('verbatim', output.text.strip()))

        return lines

    def render_display_format_html(self, output):
        return []

    def render_display_format_latex(self, output):
        if type(output.latex) == type([]):
            return output.latex
        return [output.latex]

    def render_display_format_json(self, output):
        # latex ignores json
        return []

    def render_display_format_javascript(self, output):
        # latex ignores javascript
        return []