def __init__(self,
in_file,
bug_handler,
headings_to_sections,
list_of_lists,
copy = None,
run_level = 1,
no_headings_as_list = 1,
write_list_info = 0,
):
"""
Required:
'file'
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__run_level = run_level
self.__no_headings_as_list = no_headings_as_list
self.__headings_to_sections = headings_to_sections
self.__copy = copy
self.__write_to = better_mktemp()
self.__list_of_lists = list_of_lists
self.__write_list_info = write_list_info
def __init__(self,
in_file,
bug_handler,
dtd_path,
no_dtd,
encoding,
indent=None,
copy=None,
run_level=1,
):
"""
Required:
'file'
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__dtd_path = dtd_path
self.__no_dtd = no_dtd
self.__encoding = 'cp' + encoding
# if encoding == 'mac_roman':
# self.__encoding = 'mac_roman'
self.__indent = indent
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__convert_utf = False
self.__bad_encoding = False
def __init__(
self,
in_file,
bug_handler,
default_font,
copy=None,
run_level=1,
):
"""
Required:
'file'--file to parse
'default_font' --document default font
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__default_font = default_font
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
def __init__(self,
in_file,
bug_handler,
headings_to_sections,
list_of_lists,
copy=None,
run_level=1,
no_headings_as_list=1,
write_list_info=0,
):
"""
Required:
'file'
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__run_level = run_level
self.__no_headings_as_list = no_headings_as_list
self.__headings_to_sections = headings_to_sections
self.__copy = copy
self.__write_to = better_mktemp()
self.__list_of_lists = list_of_lists
self.__write_list_info = write_list_info
def __init__(
self,
in_file,
bug_handler,
table_data,
copy=None,
run_level=1,
):
"""
Required:
'file'--file to parse
'table_data' -- a dictionary for each table.
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__table_data = table_data
self.__run_level = run_level
self.__write_to = better_mktemp()
def __init__(
self,
in_file,
bug_handler,
dtd_path,
no_dtd,
encoding,
indent=None,
copy=None,
run_level=1,
):
"""
Required:
'file'
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__dtd_path = dtd_path
self.__no_dtd = no_dtd
self.__encoding = 'cp' + encoding
# if encoding == 'mac_roman':
# self.__encoding = 'mac_roman'
self.__indent = indent
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__convert_utf = False
self.__bad_encoding = False
def __init__(self,
in_file,
area_to_convert,
char_file,
default_char_map,
bug_handler,
invalid_rtf_handler,
copy=None,
temp_dir=None,
symbol = None,
wingdings = None,
caps = None,
convert_caps = None,
dingbats = None,
run_level = 1,
):
"""
Required:
'file'
'area_to_convert'--the area of file to convert
'char_file'--the file containing the character mappings
'default_char_map'--name of default character map
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
'symbol'--whether to load the symbol character map
'winddings'--whether to load the wingdings character map
'caps'--whether to load the caps characer map
'convert_to_caps'--wether to convert caps to utf-8
Returns:
nothing
"""
self.__file = in_file
self.__copy = copy
if area_to_convert not in ('preamble', 'body'):
msg = (
'Developer error! Wrong flag.\n'
'in module "hex_2_utf8.py\n'
'"area_to_convert" must be "body" or "preamble"\n'
)
raise self.__bug_handler, msg
self.__char_file = char_file
self.__area_to_convert = area_to_convert
self.__default_char_map = default_char_map
self.__symbol = symbol
self.__wingdings = wingdings
self.__dingbats = dingbats
self.__caps = caps
self.__convert_caps = 0
self.__convert_symbol = 0
self.__convert_wingdings = 0
self.__convert_zapf = 0
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__bug_handler = bug_handler
self.__invalid_rtf_handler = invalid_rtf_handler
def __init__(self,
in_file,
area_to_convert,
char_file,
default_char_map,
bug_handler,
invalid_rtf_handler,
copy=None,
temp_dir=None,
symbol=None,
wingdings=None,
caps=None,
convert_caps=None,
dingbats=None,
run_level=1,
):
"""
Required:
'file'
'area_to_convert'--the area of file to convert
'char_file'--the file containing the character mappings
'default_char_map'--name of default character map
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
'symbol'--whether to load the symbol character map
'winddings'--whether to load the wingdings character map
'caps'--whether to load the caps characer map
'convert_to_caps'--wether to convert caps to utf-8
Returns:
nothing
"""
self.__file = in_file
self.__copy = copy
if area_to_convert not in ('preamble', 'body'):
msg = (
'Developer error! Wrong flag.\n'
'in module "hex_2_utf8.py\n'
'"area_to_convert" must be "body" or "preamble"\n'
)
raise self.__bug_handler(msg)
self.__char_file = char_file
self.__area_to_convert = area_to_convert
self.__default_char_map = default_char_map
self.__symbol = symbol
self.__wingdings = wingdings
self.__dingbats = dingbats
self.__caps = caps
self.__convert_caps = 0
self.__convert_symbol = 0
self.__convert_wingdings = 0
self.__convert_zapf = 0
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__bug_handler = bug_handler
self.__invalid_rtf_handler = invalid_rtf_handler
def __init__(
self,
in_file,
bug_handler,
copy=None,
run_level=1,
):
"""
Required:
'file'--file to parse
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__run_level = run_level
self.__state_dict = {
'before_body': self.__before_body_func,
'in_body': self.__in_body_func,
'after_control_word': self.__after_control_word_func,
'in_ignore': self.__ignore_func,
}
self.__accept = [
'cw<ci<bold______',
'cw<ci<annotation',
'cw<ci<blue______',
# 'cw<ci<bold______' ,
'cw<ci<caps______',
'cw<ci<char-style',
'cw<ci<dbl-strike',
'cw<ci<emboss____',
'cw<ci<engrave___',
'cw<ci<font-color',
'cw<ci<font-down_',
'cw<ci<font-size_',
'cw<ci<font-style',
'cw<ci<font-up___',
'cw<ci<footnot-mk',
'cw<ci<green_____',
'cw<ci<hidden____',
'cw<ci<italics___',
'cw<ci<outline___',
'cw<ci<red_______',
'cw<ci<shadow____',
'cw<ci<small-caps',
'cw<ci<strike-thr',
'cw<ci<subscript_',
'cw<ci<superscrip',
'cw<ci<underlined',
# 'cw<ul<underlined' ,
]
def __init__(self,
in_file,
copy=None,
run_level=1,
):
self.__file = in_file
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
def __init__(
self,
in_file,
copy=None,
run_level=1,
):
self.__file = in_file
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
def __init__(self, in_file,
bug_handler,
copy=None,
run_level=1,
):
"""
Required:
'file'--file to parse
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__run_level = run_level
self.__state_dict = {
'before_body' : self.__before_body_func,
'in_body' : self.__in_body_func,
'after_control_word' : self.__after_control_word_func,
'in_ignore' : self.__ignore_func,
}
self.__accept = [
'cw<ci<bold______' ,
'cw<ci<annotation' ,
'cw<ci<blue______' ,
# 'cw<ci<bold______' ,
'cw<ci<caps______' ,
'cw<ci<char-style' ,
'cw<ci<dbl-strike' ,
'cw<ci<emboss____' ,
'cw<ci<engrave___' ,
'cw<ci<font-color' ,
'cw<ci<font-down_' ,
'cw<ci<font-size_' ,
'cw<ci<font-style' ,
'cw<ci<font-up___' ,
'cw<ci<footnot-mk' ,
'cw<ci<green_____' ,
'cw<ci<hidden____' ,
'cw<ci<italics___' ,
'cw<ci<outline___' ,
'cw<ci<red_______' ,
'cw<ci<shadow____' ,
'cw<ci<small-caps' ,
'cw<ci<strike-thr' ,
'cw<ci<subscript_' ,
'cw<ci<superscrip' ,
'cw<ci<underlined' ,
# 'cw<ul<underlined' ,
]
def __init__(self,
in_file ,
bug_handler,
copy=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__found_a_footnote = 0
def __init__(self,
in_file ,
bug_handler,
copy=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__found_a_header = False
def __init__(self,
in_file ,
bug_handler,
copy=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__state = 'default'
self.__bord_pos = 'default'
self.__bord_att = []
def __init__(self,
bug_handler,
in_file=None,
copy=None,
run_level=1,
replace_illegals=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__replace_illegals = replace_illegals
def __init__(self,
in_file ,
bug_handler,
copy=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__state = 'default'
self.__bord_pos = 'default'
self.__bord_att = []
def __init__(self, in_file, bug_handler, copy=None, run_level=1):
"""
Required:
'file'--file to parse
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
def __init__(self,
in_file,
bug_handler,
copy=None,
run_level=1,
# out_file = None,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
# self.__write_to = out_file
self.__compile_expressions()
# variables
self.__uc_char = 0
self.__uc_bin = False
self.__uc_value = [1]
def separate_footnotes(self):
"""
Separate all the footnotes in an RTF file and put them at the bottom,
where they are easier to process. Each time a footnote is found,
print all of its contents to a temporary file. Close both the main and
temporary file. Print the footnotes from the temporary file to the
bottom of the main file.
"""
self.__initiate_sep_values()
self.__footnote_holder = better_mktemp()
with open(self.__file) as read_obj:
with open(self.__write_to, 'w') as self.__write_obj:
with open(self.__footnote_holder, 'w') as self.__write_to_foot_obj:
for line in read_obj:
self.__token_info = line[:16]
# keep track of opening and closing brackets
if self.__token_info == 'ob<nu<open-brack':
self.__ob_count = line[-5:-1]
if self.__token_info == 'cb<nu<clos-brack':
self.__cb_count = line[-5:-1]
# In the middle of footnote text
if self.__in_footnote:
self.__in_footnote_func(line)
# not in the middle of footnote text
else:
self.__default_sep(line)
with open(self.__footnote_holder, 'r') as read_obj:
with open(self.__write_to, 'a') as write_obj:
write_obj.write(
'mi<mk<sect-close\n'
'mi<mk<body-close\n'
'mi<tg<close_____<section\n'
'mi<tg<close_____<body\n'
'mi<tg<close_____<doc\n'
'mi<mk<footnt-beg\n')
for line in read_obj:
write_obj.write(line)
write_obj.write(
'mi<mk<footnt-end\n')
os.remove(self.__footnote_holder)
copy_obj = copy.Copy(bug_handler=self.__bug_handler)
if self.__copy:
copy_obj.copy_file(self.__write_to, "footnote_separate.data")
copy_obj.rename(self.__write_to, self.__file)
os.remove(self.__write_to)
def separate_footnotes(self):
"""
Separate all the footnotes in an RTF file and put them at the bottom,
where they are easier to process. Each time a footnote is found,
print all of its contents to a temporary file. Close both the main and
temporary file. Print the footnotes from the temporary file to the
bottom of the main file.
"""
self.__initiate_sep_values()
self.__footnote_holder = better_mktemp()
with open(self.__file) as read_obj:
with open(self.__write_to, 'w') as self.__write_obj:
with open(self.__footnote_holder, 'w') as self.__write_to_foot_obj:
for line in read_obj:
self.__token_info = line[:16]
# keep track of opening and closing brackets
if self.__token_info == 'ob<nu<open-brack':
self.__ob_count = line[-5:-1]
if self.__token_info == 'cb<nu<clos-brack':
self.__cb_count = line[-5:-1]
# In the middle of footnote text
if self.__in_footnote:
self.__in_footnote_func(line)
# not in the middle of footnote text
else:
self.__default_sep(line)
with open(self.__footnote_holder, 'r') as read_obj:
with open(self.__write_to, 'a') as write_obj:
write_obj.write(
'mi<mk<sect-close\n'
'mi<mk<body-close\n'
'mi<tg<close_____<section\n'
'mi<tg<close_____<body\n'
'mi<tg<close_____<doc\n'
'mi<mk<footnt-beg\n')
for line in read_obj:
write_obj.write(line)
write_obj.write(
'mi<mk<footnt-end\n')
os.remove(self.__footnote_holder)
copy_obj = copy.Copy(bug_handler=self.__bug_handler)
if self.__copy:
copy_obj.copy_file(self.__write_to, "footnote_separate.data")
copy_obj.rename(self.__write_to, self.__file)
os.remove(self.__write_to)
def __init__(self,
in_file,
exception_handler,
bug_handler,
copy=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
self.initiate_token_dict()
# self.initiate_token_actions()
self.compile_expressions()
self.__bracket_count=0
self.__exception_handler = exception_handler
self.__bug_handler = bug_handler
def __init__(self,
in_file ,
bug_handler,
copy=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__write_to = better_mktemp()
self.__run_level = run_level
self.__initiate_allow()
self.__bracket_count= 0
self.__ob_count = 0
self.__cb_count = 0
self.__ob = 0
self.__write_cb = False
self.__found_delete = False
def __init__(self,
in_file,
exception_handler,
bug_handler,
copy = None,
run_level = 1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
self.initiate_token_dict()
##self.initiate_token_actions()
self.compile_expressions()
self.__bracket_count=0
self.__exception_handler = exception_handler
self.__bug_handler = bug_handler
def __init__(self,
in_file,
bug_handler,
out_file,
copy=None,
orig_file=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__bracket_count = 0
self.__ob_count = 0
self.__cb_count = 0
self.__pict_count = 0
self.__in_pict = False
self.__already_found_pict = False
self.__orig_file = orig_file
self.__initiate_pict_dict()
self.__out_file = out_file
def __init__(self,
in_file,
bug_handler,
out_file,
copy=None,
orig_file=None,
run_level=1,
):
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__run_level = run_level
self.__write_to = better_mktemp()
self.__bracket_count = 0
self.__ob_count = 0
self.__cb_count = 0
self.__pict_count = 0
self.__in_pict = False
self.__already_found_pict = False
self.__orig_file = orig_file
self.__initiate_pict_dict()
self.__out_file = out_file
def __init__(self, in_file,
bug_handler,
copy=None,
no_namespace=None,
run_level=1,
):
"""
Required:
'file'
Optional:
'copy'-- whether to make a copy of result for debugging
'temp_dir' --where to output temporary results (default is
directory from which the script is run.)
Returns:
nothing
"""
self.__file = in_file
self.__bug_handler = bug_handler
self.__copy = copy
self.__no_namespace = no_namespace
self.__write_to = better_mktemp()
self.__run_level = run_level
def join_footnotes(self):
"""
Join the footnotes from the bottom of the file and put them in their
former places. First, remove the footnotes from the bottom of the
input file, outputting them to a temporary file. This creates two new
files, one without footnotes, and one of just footnotes. Open both
these files to read. When a marker is found in the main file, find the
corresponding marker in the footnote file. Output the mix of body and
footnotes to a third file.
"""
if not self.__found_a_footnote:
return
self.__write_to2 = better_mktemp()
self.__state = 'body'
self.__get_footnotes()
self.__join_from_temp()
# self.__write_obj.close()
# self.__read_from_foot_obj.close()
copy_obj = copy.Copy(bug_handler=self.__bug_handler)
if self.__copy:
copy_obj.copy_file(self.__write_to2, "footnote_joined.data")
copy_obj.rename(self.__write_to2, self.__file)
os.remove(self.__write_to2)
os.remove(self.__footnote_holder)
def join_headers(self):
"""
Join the footnotes from the bottom of the file and put them in their
former places. First, remove the footnotes from the bottom of the
input file, outputting them to a temporary file. This creates two new
files, one without footnotes, and one of just footnotes. Open both
these files to read. When a marker is found in the main file, find the
corresponding marker in the footnote file. Output the mix of body and
footnotes to a third file.
"""
if not self.__found_a_header:
return
self.__write_to2 = better_mktemp()
self.__state = 'body'
self.__get_headers()
self.__join_from_temp()
self.__write_obj.close()
self.__read_from_head_obj.close()
copy_obj = copy.Copy(bug_handler=self.__bug_handler)
if self.__copy:
copy_obj.copy_file(self.__write_to, "header_join.data")
copy_obj.rename(self.__write_to, self.__file)
os.remove(self.__write_to)
os.remove(self.__header_holder)
