# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
-import sys,string
+import sys,string,re
from parser_tools import *
floats = {
- "footnote": ["\\begin_inset Foot\n",
- "collapsed true\n"],
- "margin": ["\\begin_inset Marginal\n",
- "collapsed true\n"],
- "fig": ["\\begin_inset Float figure\n",
- "placement htbp\n",
- "wide false\n",
- "collapsed false\n"],
- "tab": ["\\begin_inset Float table\n",
- "placement htbp\n",
- "wide false\n",
- "collapsed false\n"],
- "alg": ["\\begin_inset Float algorithm\n",
- "placement htbp\n",
- "wide false\n",
- "collapsed false\n"],
- "wide-fig": ["\\begin_inset Float figure\n",
- "placement htbp\n",
- "wide true\n",
- "collapsed false\n"],
- "wide-tab": ["\\begin_inset Float table\n",
- "placement htbp\n",
- "wide true\n",
- "collapsed false\n"]
+ "footnote": ["\\begin_inset Foot",
+ "collapsed true"],
+ "margin": ["\\begin_inset Marginal",
+ "collapsed true"],
+ "fig": ["\\begin_inset Float figure",
+ "wide false",
+ "collapsed false"],
+ "tab": ["\\begin_inset Float table",
+ "wide false",
+ "collapsed false"],
+ "alg": ["\\begin_inset Float algorithm",
+ "wide false",
+ "collapsed false"],
+ "wide-fig": ["\\begin_inset Float figure",
+ "wide true",
+ "collapsed false"],
+ "wide-tab": ["\\begin_inset Float table",
+ "wide true",
+ "collapsed false"]
}
-def remove_oldfloat(lines):
+font_tokens = ["\\family", "\\series", "\\shape", "\\size", "\\emph",
+ "\\bar", "\\noun", "\\color", "\\lang", "\\latex"]
+
+pextra_type3_rexp = re.compile(r".*\\pextra_type\s+3")
+pextra_rexp = re.compile(r"\\pextra_type\s+(\S+)"+\
+ r"(\s+\\pextra_alignment\s+(\S+))?"+\
+ r"(\s+\\pextra_hfill\s+(\S+))?"+\
+ r"(\s+\\pextra_start_minipage\s+(\S+))?"+\
+ r"(\s+(\\pextra_widthp?)\s+(\S*))?")
+
+def get_width(mo):
+ if mo.group(10):
+ if mo.group(9) == "\\pextra_widthp":
+ return mo.group(10)+"col%"
+ else:
+ return mo.group(10)
+ else:
+ return "100col%"
+
+#
+# Change \begin_float .. \end_float into \begin_inset Float .. \end_inset
+#
+
+def remove_oldfloat(lines, language):
i = 0
while 1:
i = find_token(lines, "\\begin_float", i)
if i == -1:
break
+ # There are no nested floats, so finding the end of the float is simple
j = find_token(lines, "\\end_float", i+1)
+
floattype = string.split(lines[i])[1]
- #print floattype
- start = floats[floattype]+["\n"]
- mid = lines[i+1:j]
- lines[i:j+1]= start+mid+["\\end_inset \n"]
+ if not floats.has_key(floattype):
+ sys.stderr.write("Error! Unknown float type "+floattype+"\n")
+ floattype = "fig"
+
+ # skip \end_deeper tokens
+ i2 = i+1
+ while check_token(lines[i2], "\\end_deeper"):
+ i2 = i2+1
+ if i2 > i+1:
+ j2 = get_next_paragraph(lines, j+1)
+ lines[j2:j2] = ["\\end_deeper "]*(i2-(i+1))
+
+ new = floats[floattype]+[""]
+
+ # Check if the float is floatingfigure
+ k = find_re(lines, pextra_type3_rexp, i, j)
+ if k != -1:
+ mo = pextra_rexp.search(lines[k])
+ width = get_width(mo)
+ lines[k] = re.sub(pextra_rexp, "", lines[k])
+ new = ["\\begin_inset Wrap figure",
+ 'width "%s"' % width,
+ "collapsed false",
+ ""]
+
+ new = new+lines[i2:j]+["\\end_inset ", ""]
+
+ # After a float, all font attributes are reseted.
+ # We need to output '\foo default' for every attribute foo
+ # whose value is not default before the float.
+ # The check here is not accurate, but it doesn't matter
+ # as extra '\foo default' commands are ignored.
+ # In fact, it might be safer to output '\foo default' for all
+ # font attributes.
+ k = get_paragraph(lines, i)
+ flag = 0
+ for token in font_tokens:
+ if find_token(lines, token, k, i) != -1:
+ if not flag:
+ # This is not necessary, but we want the output to be
+ # as similar as posible to the lyx format
+ flag = 1
+ new.append("")
+ if token == "\\lang":
+ new.append(token+" "+language)
+ else:
+ new.append(token+" default ")
+
+ lines[i:j+1] = new
i = i+1
-def remove_oldminipage(lines):
+pextra_type2_rexp = re.compile(r".*\\pextra_type\s+[12]")
+pextra_type2_rexp2 = re.compile(r".*(\\layout|\\pextra_type\s+2)")
+
+def remove_pextra(lines):
i = 0
flag = 0
while 1:
- i = find_token(lines, "\\pextra_type 2", i)
+ i = find_re(lines, pextra_type2_rexp, i)
if i == -1:
break
- hfill = 0
- line = string.split(lines[i])
- if line[4] == "\\pextra_hfill":
- line[4:6] = []
- hfill = 1
- if line[4] == "\\pextra_start_minipage":
- line[4:6] = []
- width = line[5]
- if line[4] == "\\pextra_widthp":
- width = line[5]+"col%"
-
- start = ["\\begin_inset Minipage\n",
- "position %s\n" % line[3],
- "inner_position 0\n",
- 'height "0pt"\n',
- 'width "%s"\n' % width,
- "collapsed false\n"
+
+ mo = pextra_rexp.search(lines[i])
+ width = get_width(mo)
+
+ if mo.group(1) == "1":
+ # handle \pextra_type 1 (indented paragraph)
+ lines[i] = re.sub(pextra_rexp, "\\leftindent "+width+" ", lines[i])
+ i = i+1
+ continue
+
+ # handle \pextra_type 2 (minipage)
+ position = mo.group(3)
+ hfill = mo.group(5)
+ lines[i] = re.sub(pextra_rexp, "", lines[i])
+
+ start = ["\\begin_inset Minipage",
+ "position " + position,
+ "inner_position 0",
+ 'height "0pt"',
+ 'width "%s"' % width,
+ "collapsed false"
]
if flag:
flag = 0
if hfill:
- start = ["\n","\hfill\n","\n"]+start
+ start = ["","\hfill",""]+start
else:
- start = ["\\layout Standard\n"] + start
-
- j = find_token_backwards(lines,"\\layout", i-1)
- j0 = j
- mid = lines[j:i]
+ start = ["\\layout Standard"] + start
- j = find_token2(lines,"\\layout", "\\end_float", i+1)
- mid = mid+lines[i+1:j]
+ j0 = find_token_backwards(lines,"\\layout", i-1)
+ j = get_next_paragraph(lines, i)
+ count = 0
while 1:
- i = find_token2(lines,"\\layout", "\\pextra_type", j+1)
- if i == -1 or not check_token(lines, "\\pextra_type", i):
+ # collect more paragraphs to the minipage
+ count = count+1
+ if j == -1 or not check_token(lines[j], "\\layout"):
+ break
+ i = find_re(lines, pextra_type2_rexp2, j+1)
+ if i == -1:
break
- line = string.split(lines[i])
- if line[4] == "\\pextra_hfill":
- line[4:6] = []
- if line[4] == "\\pextra_start_minipage" and line[5] == "1":
+ mo = pextra_rexp.search(lines[i])
+ if not mo:
+ break
+ if mo.group(7) == "1":
flag = 1
break
- j = find_token_backwards(lines,"\\layout", i-1)
- mid = mid+lines[j:i]
- j = find_token2(lines,"\\layout", "\\end_float", i+1)
- mid = mid+lines[i+1:j]
+ lines[i] = re.sub(pextra_rexp, "", lines[i])
+ j = find_tokens(lines, ["\\layout", "\\end_float"], i+1)
- end = ["\\end_inset \n"]
+ mid = lines[j0:j]
+ end = ["\\end_inset "]
lines[j0:j] = start+mid+end
i = i+1
-def convert(header,body):
- remove_oldminipage(body)
- remove_oldfloat(body)
+def is_empty(lines):
+ return filter(is_nonempty_line, lines) == []
+
+move_rexp = re.compile(r"\\(family|series|shape|size|emph|numeric|bar|noun|end_deeper)")
+ert_rexp = re.compile(r"\\begin_inset|\\hfill|.*\\SpecialChar")
+spchar_rexp = re.compile(r"(.*)(\\SpecialChar.*)")
+ert_begin = ["\\begin_inset ERT",
+ "status Collapsed",
+ "",
+ "\\layout Standard"]
+
+def remove_oldert(lines):
+ i = 0
+ while 1:
+ i = find_tokens(lines, ["\\latex latex", "\\layout LaTeX"], i)
+ if i == -1:
+ break
+ j = i+1
+ while 1:
+ # \end_inset is for ert inside a tabular cell. The other tokens
+ # are obvious.
+ j = find_tokens(lines, ["\\latex default", "\\layout", "\\begin_inset", "\\end_inset", "\\end_float", "\\the_end"],
+ j)
+ if check_token(lines[j], "\\begin_inset"):
+ j = find_end_of_inset(lines, j)+1
+ else:
+ break
+
+ if check_token(lines[j], "\\layout"):
+ while j-1 >= 0 and check_token(lines[j-1], "\\begin_deeper"):
+ j = j-1
+
+ # We need to remove insets, special chars & font commands from ERT text
+ new = []
+ new2 = []
+ if check_token(lines[i], "\\layout LaTeX"):
+ new = ["\layout Standard", "", ""]
+ # We have a problem with classes in which Standard is not the default layout!
+
+ k = i+1
+ while 1:
+ k2 = find_re(lines, ert_rexp, k, j)
+ inset = hfill = specialchar = 0
+ if k2 == -1:
+ k2 = j
+ elif check_token(lines[k2], "\\begin_inset"):
+ inset = 1
+ elif check_token(lines[k2], "\\hfill"):
+ hfill = 1
+ del lines[k2]
+ j = j-1
+ else:
+ specialchar = 1
+ mo = spchar_rexp.match(lines[k2])
+ lines[k2] = mo.group(1)
+ specialchar_str = mo.group(2)
+ k2 = k2+1
+
+ tmp = []
+ for line in lines[k:k2]:
+ # Move some lines outside the ERT inset:
+ if move_rexp.match(line):
+ if new2 == []:
+ # This is not necessary, but we want the output to be
+ # as similar as posible to the lyx format
+ new2 = [""]
+ new2.append(line)
+ elif not check_token(line, "\\latex"):
+ tmp.append(line)
+
+ if is_empty(tmp):
+ if filter(lambda x:x != "", tmp) != []:
+ if new == []:
+ # This is not necessary, but we want the output to be
+ # as similar as posible to the lyx format
+ lines[i-1] = lines[i-1]+" "
+ else:
+ new = new+[" "]
+ else:
+ new = new+ert_begin+tmp+["\\end_inset ", ""]
+
+ if inset:
+ k3 = find_end_of_inset(lines, k2)
+ new = new+[""]+lines[k2:k3+1]+[""] # Put an empty line after \end_inset
+ k = k3+1
+ # Skip the empty line after \end_inset
+ if not is_nonempty_line(lines[k]):
+ k = k+1
+ new.append("")
+ elif hfill:
+ new = new+["\hfill", ""]
+ k = k2
+ elif specialchar:
+ if new == []:
+ # This is not necessary, but we want the output to be
+ # as similar as posible to the lyx format
+ lines[i-1] = lines[i-1]+specialchar_str
+ new = [""]
+ else:
+ new = new+[specialchar_str, ""]
+ k = k2
+ else:
+ break
+
+ new = new+new2
+ if not check_token(lines[j], "\\latex "):
+ new = new+[""]+[lines[j]]
+ lines[i:j+1] = new
+ i = i+1
+
+ # Delete remaining "\latex xxx" tokens
+ i = 0
+ while 1:
+ i = find_token(lines, "\\latex ", i)
+ if i == -1:
+ break
+ del lines[i]
+
+# ERT insert are hidden feature of lyx 1.1.6. This might be removed in the future.
+def remove_oldertinset(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, "\\begin_inset ERT", i)
+ if i == -1:
+ break
+ j = find_end_of_inset(lines, i)
+ k = find_token(lines, "\\layout", i+1)
+ l = get_paragraph(lines, i)
+ if lines[k] == lines[l]: # same layout
+ k = k+1
+ new = lines[k:j]
+ lines[i:j+1] = new
+ i = i+1
+
+def is_ert_paragraph(lines, i):
+ if not check_token(lines[i], "\\layout Standard"):
+ return 0
+
+ i = find_nonempty_line(lines, i+1)
+ if not check_token(lines[i], "\\begin_inset ERT"):
+ return 0
+
+ j = find_end_of_inset(lines, i)
+ k = find_nonempty_line(lines, j+1)
+ return check_token(lines[k], "\\layout")
+
+def combine_ert(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, "\\begin_inset ERT", i)
+ if i == -1:
+ break
+ j = get_paragraph(lines, i)
+ count = 0
+ text = []
+ while is_ert_paragraph(lines, j):
+
+ count = count+1
+ i2 = find_token(lines, "\\layout", j+1)
+ k = find_token(lines, "\\end_inset", i2+1)
+ text = text+lines[i2:k]
+ j = find_token(lines, "\\layout", k+1)
+ if j == -1:
+ break
+
+ if count >= 2:
+ j = find_token(lines, "\\layout", i+1)
+ lines[j:k] = text
+
+ i = i+1
+
+oldunits = ["pt", "cm", "in", "text%", "col%"]
+
+def get_length(lines, name, start, end):
+ i = find_token(lines, name, start, end)
+ if i == -1:
+ return ""
+ x = string.split(lines[i])
+ return x[2]+oldunits[int(x[1])]
+
+def write_attribute(x, token, value):
+ if value != "":
+ x.append("\t"+token+" "+value)
+
+def remove_figinset(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, "\\begin_inset Figure", i)
+ if i == -1:
+ break
+ j = find_end_of_inset(lines, i)
+
+ if ( len(string.split(lines[i])) > 2 ):
+ lyxwidth = string.split(lines[i])[3]+"pt"
+ lyxheight = string.split(lines[i])[4]+"pt"
+ else:
+ lyxwidth = ""
+ lyxheight = ""
+
+ filename = get_value(lines, "file", i+1, j)
+
+ width = get_length(lines, "width", i+1, j)
+ # what does width=5 mean ?
+ height = get_length(lines, "height", i+1, j)
+ rotateAngle = get_value(lines, "angle", i+1, j)
+ if width == "" and height == "":
+ size_type = "0"
+ else:
+ size_type = "1"
+
+ flags = get_value(lines, "flags", i+1, j)
+ x = int(flags)%4
+ if x == 1:
+ display = "monochrome"
+ elif x == 2:
+ display = "gray"
+ else:
+ display = "color"
+
+ subcaptionText = ""
+ subcaptionLine = find_token(lines, "subcaption", i+1, j)
+ if subcaptionLine != -1:
+ subcaptionText = lines[subcaptionLine][11:]
+ if subcaptionText != "":
+ subcaptionText = '"'+subcaptionText+'"'
+
+ k = find_token(lines, "subfigure", i+1,j)
+ if k == -1:
+ subcaption = 0
+ else:
+ subcaption = 1
+
+ new = ["\\begin_inset Graphics FormatVersion 1"]
+ write_attribute(new, "filename", filename)
+ write_attribute(new, "display", display)
+ if subcaption:
+ new.append("\tsubcaption")
+ write_attribute(new, "subcaptionText", subcaptionText)
+ write_attribute(new, "size_type", size_type)
+ write_attribute(new, "width", width)
+ write_attribute(new, "height", height)
+ if rotateAngle != "":
+ new.append("\trotate")
+ write_attribute(new, "rotateAngle", rotateAngle)
+ write_attribute(new, "rotateOrigin", "leftBaseline")
+ write_attribute(new, "lyxsize_type", "1")
+ write_attribute(new, "lyxwidth", lyxwidth)
+ write_attribute(new, "lyxheight", lyxheight)
+ new = new + ["\end_inset"]
+ lines[i:j+1] = new
+
+attr_re = re.compile(r' \w*="(false|0|)"')
+line_re = re.compile(r'<(features|column|row|cell)')
+
+def update_tabular(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, '\\begin_inset Tabular', i)
+ if i == -1:
+ break
+
+ for k in get_tabular_lines(lines, i):
+ if check_token(lines[k], "<lyxtabular"):
+ lines[k] = string.replace(lines[k], 'version="2"', 'version="3"')
+ elif check_token(lines[k], "<column"):
+ lines[k] = string.replace(lines[k], 'width=""', 'width="0pt"')
+
+ if line_re.match(lines[k]):
+ lines[k] = re.sub(attr_re, "", lines[k])
+
+ i = i+1
+
+# Figure insert are hidden feature of lyx 1.1.6. This might be removed in the future.
+def fix_oldfloatinset(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, "\\begin_inset Float ", i)
+ if i == -1:
+ break
+ j = find_token(lines, "collapsed", i)
+ if j != -1:
+ lines[j:j] = ["wide false"]
+ i = i+1
+
+def change_listof(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, "\\begin_inset LatexCommand \\listof", i)
+ if i == -1:
+ break
+ type = re.search(r"listof(\w*)", lines[i]).group(1)[:-1]
+ lines[i] = "\\begin_inset FloatList "+type
+ i = i+1
+
+def change_infoinset(lines):
+ i = 0
+ while 1:
+ i = find_token(lines, "\\begin_inset Info", i)
+ if i == -1:
+ break
+ txt = string.lstrip(lines[i][18:])
+ new = ["\\begin_inset Note", "collapsed true", ""]
+ j = find_token(lines, "\\end_inset", i)
+ if j == -1:
+ break
+
+ note_lines = lines[i+1:j]
+ if len(txt) > 0:
+ note_lines = [txt]+note_lines
+
+ for line in note_lines:
+ new = new + ["\layout Standard", ""]
+ tmp = string.split(line, '\\')
+ new = new + [tmp[0]]
+ for x in tmp[1:]:
+ new = new + ["\\backslash ", x]
+ lines[i:j] = new
+ i = i+5
+
+def change_preamble(lines):
+ i = find_token(lines, "\\use_amsmath", 0)
+ if i == -1:
+ return
+ lines[i+1:i+1] = ["\\use_natbib 0",
+ "\use_numerical_citations 0"]
+
+def convert(header, body):
+ language = get_value(header, "\\language", 0)
+ if language == "":
+ language = "english"
+
+ change_preamble(header)
+ change_listof(body)
+ fix_oldfloatinset(body)
+ update_tabular(body)
+ remove_pextra(body)
+ remove_oldfloat(body, language)
+ remove_figinset(body)
+ remove_oldertinset(body)
+ remove_oldert(body)
+ combine_ert(body)
+ change_infoinset(body)
if __name__ == "__main__":
pass
-