def dt2latex(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
result << latex.Cmd("item", [(latex.Group(tolatex(tag.content, result)),)], inline=True)
result << latex.Cmd("hfill", inline=True)
result << latex.Cmd("\\", inline=True)
return result
def code2latex(tag, parent):
lang = get_lang(tag, parent)
if "pre" in parent.usr:
page = latex.Env("minipage", args=[latex.Group(["0.95", latex.Cmd("linewidth", inline=True)])], inline=False)
process_usr(tag, page, parent)
result = latex.Env("lstlisting", args=[("mathescape",)])
process_usr(tag, result, parent)
result << convert_code(tag.content, lang, "arg" in parent.usr)
page << result
pre = latex.Group()
pre << latex.Cmd("newbox", ["\\mybox"])
box = latex.Env("lrbox", ["\\mybox"])
box << latex.Cmd("noindent")
box << page
pre << box
pre << latex.Cmd("colorbox", ["code_bg", latex.Cmd("usebox", [latex.Cmd("mybox")])])
return pre
else:
#color = latex.Cmd("colorbox", ["code_bg"], inline=True)
result = latex.Cmd("protect\\lstinline", args=[("mathescape, columns=fixed",)], inline=True, d_open=u'!', d_close=u'!')
process_usr(tag, result, parent)
group = latex.Group()
process_usr(tag, group, result)
group << convert_code(tag.content, lang, "arg" in parent.usr).replace('\n', ' ')
result.args.append(group)
#color.args.append(result)
return result #color
def em2latex(tag, parent):
if "abstract" in parent.usr:
result = latex.Group()
process_usr(tag, result, parent)
result << tolatex(tag.content, result)
return result
else:
result = latex.Cmd("textit", inline=True)
process_usr(tag, result, parent, True)
result.args = [latex.Group()]
process_usr(tag, result.args[0], result)
result.args[0] << tolatex(tag.content, result.args[0])
return result
def i2latex(tag, parent):
result = latex.Cmd("emph", inline=True)
process_usr(tag, result, parent, True)
result.args = [latex.Group()]
process_usr(tag, result.args[0], result)
result.args[0] << tolatex(tag.content, result.args[0])
return result
def _conclusion(tag, parent):
result = latex.Group()
result << latex.Cmd("conclusion")
process_usr(tag, result, parent)
result.usr['skip'] = ['h1']
result << tolatex(tag.content, result)
return result
def thead2latex(tag, parent):
result = latex.Group([], "\n")
process_usr(tag, result, parent)
result.usr["thead"] = True
result << tolatex(tag.content, result)
result << latex.Cmd("hline")
return result
def _title(tag, parent):
result = latex.Cmd("title")
process_usr(tag, result, parent)
result.args = [latex.Group()]
process_usr(tag, result.args[0], result)
result.args[0] << tolatex(tag.content, result.args[0])
return result
def default2latex(tag, parent): if tag.name not in ["section", "subsection", "subsubsection", "div", "document", "html", "body", "span"]: print(tag.name) result = latex.Group() process_usr(tag, result, parent) result << tolatex(tag.content, result) return result
def _appendix(tag, parent):
result = latex.Group()
if not hasattr(_appendix, 'first'):
result << latex.Cmd("appendix")
_appendix.first = True
process_usr(tag, result, parent)
result << tolatex(tag.content, result)
return result
def _listOfAbbreviations(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
result << "{"
result << latex.Cmd("listspacing", ["abbr"])
result << latex.Cmd("printglossary", [("style=list,title=List Of Abbreviations",)])
result << "}"
return result
def abbr2latex(tag, parent):
result = latex.Group()
aid = ''.join(str(x) for x in tag.content)
if "title" in tag.attrs and tag.attrs["title"]:
latex_abbr[aid] = remove_linebreaks(tag.attrs["title"])
result << latex.Cmd("gls", [aid.lower()])
else:
result << aid
return result
def cite2latex(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
if "id" in tag.attrs:
latex_citations.append(tag.attrs["id"])
result << latex.Cmd("bibitem", [tag.attrs["id"]])
result << tolatex(tag.content, result)
result << "\n"
return result
def _appendix(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
if 'skip' in result.usr and result.usr['skip']:
result.usr['skip'].append('h1')
else:
result.usr['skip'] = ['h1']
result << latex.Cmd("appendix")
result << tolatex(tag.content, result)
return result
def _header(tag, parent):
result = latex.Group([], "\n")
process_usr(tag, result, parent)
if 'skip' in result.usr and result.usr['skip']:
result.usr['skip'].append('p')
else:
result.usr['skip'] = ['p']
result << tolatex(tag.content, result)
result << latex.Cmd("maketitle")
return result
def _copyright(tag, parent):
copy = tag.content[0].split(' ')
year = copy[1]
name = ' '.join(copy[2:])
result = latex.Group()
process_usr(tag, result, parent)
result << latex.Cmd("copyrightholder", [name])
result << latex.Cmd("copyrightyear", [year])
result << latex.Cmd("makecopyright")
return result
def hN2latex(tag, parent):
global cfg
index = int(tag.name[1:])
result = latex.Group()
section = latex.Section("", index+cfg['section'])
process_usr(tag, section, parent)
if 'pass' in section.usr:
section.usr['pass'].append('a')
else:
section.usr['pass'] = ['a']
section.args = [latex.Group()]
process_usr(tag, section.args[0], section)
section.args[0] << tolatex([tag.content[0]], section.args[0])
result << section
if isinstance(tag.content[0], html.Tag) and tag.content[0].name == "a" and "name" in tag.content[0].attrs:
result << latex.Cmd("label", [tag.content[0].attrs["name"]])
return result
def a2latex(tag, parent):
if "href" in tag.attrs:
href = tag.attrs["href"]
if len(href) > 0 and href[0] == "#":
if "#" in href[1:]:
result = latex.Cmd("cite", [], inline=True)
process_usr(tag, result, parent, True)
rng = href[1:].split('-#')
if rng[0] in latex_citations and rng[1] in latex_citations:
result.args = [','.join(latex_citations[latex_citations.index(rng[0]):latex_citations.index(rng[1])+1])]
elif href[1:] in latex_citations:
result = latex.Cmd("cite", [escape_ref(href[1:])], inline=True)
process_usr(tag, result, parent, True)
elif href[1:] in latex_figures:
result = latex.Group()
process_usr(tag, result, parent, True)
result << "Fig. "
result << latex.Cmd("ref", [escape_ref(href[1:])], inline=True)
elif href[1:] in latex_tables:
result = latex.Group()
process_usr(tag, result, parent, True)
result << "Table. "
result << latex.Cmd("ref", [escape_ref(href[1:])], inline=True)
elif tag.content:
result = latex.Cmd("hyperref", [(escape_ref(href[1:]),)], inline=True)
process_usr(tag, result, parent, True)
result.args.append(latex.Group(tolatex(tag.content, parent)))
elif href[1:] not in latex_unresolved:
result = latex.Group()
process_usr(tag, result, parent, True)
result << latex.Cmd("ref", [escape_ref(href[1:])], inline=True)
latex_unresolved[href[1:]] = [result]
else:
result = latex.Group()
process_usr(tag, result, parent, True)
result << latex.Cmd("ref", [escape_ref(href[1:])], inline=True)
latex_unresolved[href[1:]].append(result)
return result
else:
result = latex.Cmd("href", [escape_ref(href)], inline=True)
process_usr(tag, result, parent, True)
result.args.append(latex.Group(tolatex(tag.content, parent)))
return result
elif "name" in tag.attrs:
name = tag.attrs["name"]
result = latex.Group()
process_usr(tag, result, parent)
result << tolatex(tag.content, parent)
result << latex.Cmd("label", args=[name], inline=True)
return result
return latex.Group(tolatex(tag.content, parent))
def _preface(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
result << latex.Cmd("preface")
result.usr['skip'] = ['h1']
result << tolatex(tag.content, result)
result << latex.Cmd("normalspacing")
result << latex.Cmd("setcounter", ["page", 1])
result << latex.Cmd("pagenumbering", ["arabic"])
result << latex.Cmd("pagestyle", ["cornell"])
result << latex.Cmd("addtolength", ["\\parskip", "0.5\\baselineskip"])
return result
def td2latex(tag, parent):
result = None
if "colspan" in tag.attrs:
result = latex.Cmd("multicolumn", [int(tag.attrs["colspan"]), "c"], inline=True)
cell = latex.Cmd("makecell", inline=True)
process_usr(tag, cell, parent)
cell.args = [latex.Group(tolatex(tag.content, cell))]
if result is not None:
result.args.append(cell)
return result
else:
return cell
def _abstract(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
p = tag.get('p')
if len(p) > 0:
abst = latex.Env("abstract")
process_usr(p[0], abst, result)
abst << tolatex(p[0].content[1][1:], abst)
result << abst
if len(p) > 1:
keys = latex.Env("IEEEkeywords")
process_usr(p[1], keys, result)
keys << tolatex(p[1].content[2:], result)
result << keys
return result
def _bio(tag, parent):
content = []
img = None
name = None
for c in tag.content:
if isinstance(c, html.STag) and c.name == "img" and not img:
img = c
elif isinstance(c, html.Tag) and c.name == "b":
name = c.content[0]
else:
content.append(c)
result = latex.Env("IEEEbiography")
process_usr(tag, result, parent)
result.args = [(latex.Group([tolatex(img, result)], "", "{", "}"),), tolatex(name, result)]
result << tolatex(content, result)
return result
def tolatex(tag, parent): if isinstance(tag, (list, tuple)): result = [] for elem in tag: item = tolatex(elem, parent) if item: result.append(item) return result if isinstance(tag, html.Tag): if "class" in tag.attrs and "web-only" in tag.attrs["class"]: return "" elif parent and parent.usr and 'skip' in parent.usr and tag.name in parent.usr['skip']: return "" elif parent and parent.usr and 'pass' in parent.usr and tag.name in parent.usr['pass']: result = latex.Group() process_usr(tag, result, parent) result << tolatex(tag.content, result) return result elif latex_module is not None and latex_module.has(tag): return latex_module.set(tag, parent) elif tag.name in handlers: return handlers[tag.name](tag, parent) else: return default2latex(tag, parent) elif isinstance(tag, html.STag): if "class" in tag.attrs and "web-only" in tag.attrs["class"]: return "" elif parent and parent.usr and 'skip' in parent.usr and tag.name in parent.usr['skip']: return "" elif parent and parent.usr and 'pass' in parent.usr and tag.name in parent.usr['pass']: return "" elif latex_module is not None and latex_module.has(tag): return latex_module.set(tag, parent) elif tag.name in handlers: return handlers[tag.name](tag, parent) else: print(tag) return "" else: return escape(tag, "figure" not in parent.usr)
def _title(tag, parent):
title = ""
author = ""
conferral = ""
for elem in tag.content:
if isinstance(elem, html.Tag):
if "class" in elem.attrs:
titlecls = elem.attrs["class"].split()
if "title" in titlecls:
title = elem.content[0].strip()
elif "author" in titlecls:
author = elem.content[2].strip()
conferral = elem.content[4].strip().split()
result = latex.Group()
process_usr(tag, result, parent)
result << latex.Cmd("title", [title])
result << latex.Cmd("author", [author])
result << latex.Cmd("conferraldate", conferral)
result << latex.Cmd("degreefield", ["Ph.D."])
result << latex.Cmd("maketitle")
return result
def img2latex(tag, parent):
if "src" in tag.attrs:
src = tag.attrs["src"]
path = os.path.dirname(src)
outsrc = os.path.join(latex_outpath, src)
if not os.path.exists(os.path.dirname(outsrc)):
try:
os.makedirs(os.path.dirname(outsrc))
except OSError as exc: # Guard against race condition
if exc.errno != errno.EEXIST:
raise
if ".svg" in src:
width = 1.0
if 'width' in parent.usr:
width = parent.usr['width']
if "style" in tag.attrs and tag.attrs["style"]:
attr = tag.attrs["style"].get("width")
if not attr:
attr = tag.attrs["style"].get("max-width")
if "," in attr:
width *= float(attr[attr.index(",")+1:-2])/100.0
elif "%" in attr:
width *= float(attr[0:-1])/100.0
outsrc = outsrc.replace(".svg", ".pdf")
intime = os.path.getmtime(src) if os.path.exists(src) else 0
outtime = os.path.getmtime(outsrc) if os.path.exists(outsrc) else 0
if intime > outtime:
print("inkscape -D -z --file=\"" + src + "\" --export-pdf=\"" + outsrc + "\" --export-latex")
os.system("inkscape -D -z --file=\"" + src + "\" --export-pdf=\"" + outsrc + "\" --export-latex")
src = src.replace(".svg", ".pdf_tex")
result = latex.Group()
result << latex.Cmd("graphicspath", [path+'/'], d_open='{{', d_close='}}')
process_usr(tag, result, parent, True)
if width:
page = latex.Env("minipage", args=[latex.Group([str(width), latex.Cmd("columnwidth", inline=True)])], inline=True)
process_usr(tag, page, parent)
page << latex.Cmd("centering", inline=False)
page << latex.Cmd("def", inline=True)
page << latex.Cmd("svgwidth", args=["\\columnwidth"], inline=False)
page << latex.Cmd("input", args=[src])
result << page
else:
result << latex.Cmd("def", inline=True)
result << latex.Cmd("svgwidth", args=["\\columnwidth"], inline=False)
result << latex.Cmd("input", args=[src], inline=False)
return result
else:
copyfile(src, outsrc)
result = latex.Cmd("includegraphics")
process_usr(tag, result, parent, True)
result.args = [("width=1.0\\columnwidth",), src]
return result
else:
return default2latex(tag, parent)
def dd2latex(tag, parent): result = latex.Group() process_usr(tag, result, parent) result << tolatex(tag.content, result) return result
def li2latex(tag, parent):
result = latex.Group()
process_usr(tag, result, parent)
result << latex.Cmd("item ", inline=True)
result << tolatex(tag.content, result)
return result
def tbody2latex(tag, parent): result = latex.Group([], "\n") process_usr(tag, result, parent) result.usr["tbody"] = True result << tolatex(tag.content, result) return result
def tr2latex(tag, parent): result = latex.Group([], " & ", "", " \\\\\n") process_usr(tag, result, parent) result << tolatex([elem for elem in tag.content if isinstance(elem, html.Tag)], result) return result
def caption2latex(tag, parent):
result = latex.Cmd("caption")
process_usr(tag, result, parent, True)
result.args = [latex.Group(tolatex(tag.content, result))]
return result
def _authors(tag, parent): result = latex.Group([], "\n") process_usr(tag, result, parent) result << tolatex(tag.content, result) return result
