Define & use a Conversion object. It's still really ugly, but at

least there's a token object in here now!  ;-)
This commit is contained in:
Fred Drake 1999-05-07 19:59:02 +00:00
parent b0bc7f2d6c
commit 96c00b0b5e
1 changed files with 258 additions and 230 deletions

View File

@ -27,7 +27,7 @@ class LaTeXFormatError(Error):
_begin_env_rx = re.compile(r"[\\]begin{([^}]*)}") _begin_env_rx = re.compile(r"[\\]begin{([^}]*)}")
_end_env_rx = re.compile(r"[\\]end{([^}]*)}") _end_env_rx = re.compile(r"[\\]end{([^}]*)}")
_begin_macro_rx = re.compile("[\\\\]([a-zA-Z]+[*]?)({|\\s*\n?)") _begin_macro_rx = re.compile("[\\\\]([a-zA-Z]+[*]?)({|\\s*\n?)")
_comment_rx = re.compile("%+ ?(.*)\n *") _comment_rx = re.compile("%+ ?(.*)\n[ \t]*")
_text_rx = re.compile(r"[^]%\\{}]+") _text_rx = re.compile(r"[^]%\\{}]+")
_optional_rx = re.compile(r"\s*[[]([^]]*)[]]") _optional_rx = re.compile(r"\s*[[]([^]]*)[]]")
# _parameter_rx is this complicated to allow {...} inside a parameter; # _parameter_rx is this complicated to allow {...} inside a parameter;
@ -50,248 +50,276 @@ def popping(name, point, depth):
sys.stderr.write("%s</%s> at %s\n" % (" "*depth, name, point)) sys.stderr.write("%s</%s> at %s\n" % (" "*depth, name, point))
def subconvert(line, ofp, table, discards, autoclosing, endchar=None, depth=0): class Conversion:
if DEBUG and endchar: def __init__(self, ifp, ofp, table=None, discards=(), autoclosing=()):
sys.stderr.write("subconvert(%s, ..., endchar=%s)\n" self.ofp_stack = [ofp]
% (`line[:20]`, `endchar`)) self.pop_output()
stack = [] self.table = table
while line: self.discards = discards
if line[0] == endchar and not stack: self.autoclosing = autoclosing
if DEBUG: self.line = string.join(map(string.rstrip, ifp.readlines()), "\n")
sys.stderr.write("subconvert() --> %s\n" % `line[1:21]`) self.err_write = sys.stderr.write
return line[1:] self.preamble = 1
m = _comment_rx.match(line)
if m: def push_output(self, ofp):
text = m.group(1) self.ofp_stack.append(self.ofp)
if text: self.ofp = ofp
ofp.write("(COMMENT\n- %s \n)COMMENT\n-\\n\n" % encode(text)) self.write = ofp.write
line = line[m.end():]
continue def pop_output(self):
m = _begin_env_rx.match(line) self.ofp = self.ofp_stack.pop()
if m: self.write = self.ofp.write
# re-write to use the macro handler
line = r"\%s %s" % (m.group(1), line[m.end():]) def subconvert(self, endchar=None, depth=0):
continue if DEBUG and endchar:
m = _end_env_rx.match(line) self.err_write(
if m: "subconvert(%s)\n line = %s\n" % (`endchar`, `line[:20]`))
# end of environment stack = []
envname = m.group(1) line = self.line
if envname == "document": while line:
# special magic if line[0] == endchar and not stack:
for n in stack[1:]: if DEBUG:
if n not in autoclosing: self.err_write("subconvert() --> %s\n" % `line[1:21]`)
raise LaTeXFormatError("open element on stack: " + `n`) self.line = line
# should be more careful, but this is easier to code: return line
stack = [] m = _comment_rx.match(line)
ofp.write(")document\n") if m:
elif envname == stack[-1]: text = m.group(1)
ofp.write(")%s\n" % envname) if text:
del stack[-1] self.write("(COMMENT\n- %s \n)COMMENT\n-\\n\n"
popping(envname, "a", len(stack) + depth) % encode(text))
else:
sys.stderr.write("stack: %s\n" % `stack`)
raise LaTeXFormatError(
"environment close for %s doesn't match" % envname)
line = line[m.end():]
continue
m = _begin_macro_rx.match(line)
if m:
# start of macro
macroname = m.group(1)
if macroname == "verbatim":
# really magic case!
pos = string.find(line, "\\end{verbatim}")
text = line[m.end(1):pos]
ofp.write("(verbatim\n")
ofp.write("-%s\n" % encode(text))
ofp.write(")verbatim\n")
line = line[pos + len("\\end{verbatim}"):]
continue
numbered = 1
if macroname[-1] == "*":
macroname = macroname[:-1]
numbered = 0
if macroname in autoclosing and macroname in stack:
while stack[-1] != macroname:
if stack[-1] and stack[-1] not in discards:
ofp.write(")%s\n-\\n\n" % stack[-1])
popping(stack[-1], "b", len(stack) + depth - 1)
del stack[-1]
if macroname not in discards:
ofp.write("-\\n\n)%s\n-\\n\n" % macroname)
popping(macroname, "c", len(stack) + depth - 1)
del stack[-1]
real_ofp = ofp
if macroname in discards:
ofp = StringIO.StringIO()
#
conversion = table.get(macroname, ([], 0, 0, 0, 0))
params, optional, empty, environ, nocontent = conversion
if empty:
ofp.write("e\n")
elif nocontent:
empty = 1
if not numbered:
ofp.write("Anumbered TOKEN no\n")
opened = 0
# rip off the macroname
if params:
if optional and len(params) == 1:
line = line = line[m.end():]
else:
line = line[m.end(1):]
elif empty:
line = line[m.end(1):]
else:
line = line[m.end():] line = line[m.end():]
#
# Very ugly special case to deal with \item[]. The catch is that
# this needs to occur outside the for loop that handles attribute
# parsing so we can 'continue' the outer loop.
#
if optional and type(params[0]) is type(()):
# the attribute name isn't used in this special case
pushing(macroname, "a", depth + len(stack))
stack.append(macroname)
ofp.write("(%s\n" % macroname)
m = _start_optional_rx.match(line)
if m:
line = line[m.end():]
line = subconvert(line, ofp, table, discards,
autoclosing, endchar="]",
depth=depth + len(stack))
line = "}" + line
continue continue
# handle attribute mappings here: m = _begin_env_rx.match(line)
for attrname in params: if m:
if optional: # re-write to use the macro handler
optional = 0 line = r"\%s %s" % (m.group(1), line[m.end():])
if type(attrname) is type(""): continue
m = _optional_rx.match(line) m = _end_env_rx.match(line)
if m:
# end of environment
envname = m.group(1)
if envname == "document":
# special magic
for n in stack[1:]:
if n not in self.autoclosing:
raise LaTeXFormatError(
"open element on stack: " + `n`)
# should be more careful, but this is easier to code:
stack = []
self.write(")document\n")
elif envname == stack[-1]:
self.write(")%s\n" % envname)
del stack[-1]
popping(envname, "a", len(stack) + depth)
else:
self.err_write("stack: %s\n" % `stack`)
raise LaTeXFormatError(
"environment close for %s doesn't match" % envname)
line = line[m.end():]
continue
m = _begin_macro_rx.match(line)
if m:
# start of macro
macroname = m.group(1)
if macroname == "verbatim":
# really magic case!
pos = string.find(line, "\\end{verbatim}")
text = line[m.end(1):pos]
self.write("(verbatim\n")
self.write("-%s\n" % encode(text))
self.write(")verbatim\n")
line = line[pos + len("\\end{verbatim}"):]
continue
numbered = 1
opened = 0
if macroname[-1] == "*":
macroname = macroname[:-1]
numbered = 0
if macroname in self.autoclosing and macroname in stack:
while stack[-1] != macroname:
top = stack.pop()
if top and top not in self.discards:
self.write(")%s\n-\\n\n" % top)
popping(top, "b", len(stack) + depth)
if macroname not in self.discards:
self.write("-\\n\n)%s\n-\\n\n" % macroname)
popping(macroname, "c", len(stack) + depth - 1)
del stack[-1]
#
if macroname in self.discards:
self.push_output(StringIO.StringIO())
else:
self.push_output(self.ofp)
#
params, optional, empty, environ = self.start_macro(macroname)
if not numbered:
self.write("Anumbered TOKEN no\n")
# rip off the macroname
if params:
if optional and len(params) == 1:
line = line[m.end():]
else:
line = line[m.end(1):]
elif empty:
line = line[m.end(1):]
else:
line = line[m.end():]
#
# Very ugly special case to deal with \item[]. The catch
# is that this needs to occur outside the for loop that
# handles attribute parsing so we can 'continue' the outer
# loop.
#
if optional and type(params[0]) is type(()):
# the attribute name isn't used in this special case
pushing(macroname, "a", depth + len(stack))
stack.append(macroname)
self.write("(%s\n" % macroname)
m = _start_optional_rx.match(line)
if m:
self.line = line[m.end():]
line = self.subconvert("]", depth + len(stack))
line = "}" + line
continue
# handle attribute mappings here:
for attrname in params:
if optional:
optional = 0
if type(attrname) is type(""):
m = _optional_rx.match(line)
if m:
line = line[m.end():]
self.write("A%s TOKEN %s\n"
% (attrname, encode(m.group(1))))
elif type(attrname) is type(()):
# This is a sub-element; but don't place the
# element we found on the stack (\section-like)
pushing(macroname, "b", len(stack) + depth)
stack.append(macroname)
self.write("(%s\n" % macroname)
macroname = attrname[0]
m = _start_group_rx.match(line)
if m: if m:
line = line[m.end():] line = line[m.end():]
ofp.write("A%s TOKEN %s\n" elif type(attrname) is type([]):
% (attrname, encode(m.group(1)))) # A normal subelement.
elif type(attrname) is type(()): attrname = attrname[0]
# This is a sub-element; but don't place the if not opened:
# element we found on the stack (\section-like) opened = 1
pushing(macroname, "b", len(stack) + depth) self.write("(%s\n" % macroname)
stack.append(macroname) pushing(macroname, "c", len(stack) + depth)
ofp.write("(%s\n" % macroname) self.write("(%s\n" % attrname)
macroname = attrname[0] pushing(attrname, "sub-elem", len(stack) + depth + 1)
m = _start_group_rx.match(line) self.line = skip_white(line)[1:]
if m: line = subconvert("}", depth + len(stack) + 2)
popping(attrname, "sub-elem", len(stack) + depth + 1)
self.write(")%s\n" % attrname)
else:
m = _parameter_rx.match(line)
if not m:
raise LaTeXFormatError(
"could not extract parameter %s for %s: %s"
% (attrname, macroname, `line[:100]`))
value = m.group(1)
if _token_rx.match(value):
dtype = "TOKEN"
else:
dtype = "CDATA"
self.write("A%s %s %s\n"
% (attrname, dtype, encode(value)))
line = line[m.end():] line = line[m.end():]
elif type(attrname) is type([]): if params and type(params[-1]) is type('') \
# A normal subelement. and (not empty) and not environ:
attrname = attrname[0] # attempt to strip off next '{'
if not opened: m = _start_group_rx.match(line)
opened = 1
ofp.write("(%s\n" % macroname)
pushing(macroname, "c", len(stack) + depth)
ofp.write("(%s\n" % attrname)
pushing(attrname, "sub-elem", len(stack) + depth + 1)
line = subconvert(skip_white(line)[1:], ofp, table,
discards, autoclosing, endchar="}",
depth=depth + len(stack) + 2)
popping(attrname, "sub-elem", len(stack) + depth + 1)
ofp.write(")%s\n" % attrname)
else:
m = _parameter_rx.match(line)
if not m: if not m:
raise LaTeXFormatError( raise LaTeXFormatError(
"could not extract parameter %s for %s: %s" "non-empty element '%s' has no content: %s"
% (attrname, macroname, `line[:100]`)) % (macroname, line[:12]))
value = m.group(1)
if _token_rx.match(value):
dtype = "TOKEN"
else:
dtype = "CDATA"
ofp.write("A%s %s %s\n"
% (attrname, dtype, encode(value)))
line = line[m.end():] line = line[m.end():]
if params and type(params[-1]) is type('') \ if not opened:
and (not empty) and not environ: self.write("(%s\n" % macroname)
# attempt to strip off next '{' pushing(macroname, "d", len(stack) + depth)
m = _start_group_rx.match(line) if empty:
if not m: line = "}" + line
raise LaTeXFormatError( stack.append(macroname)
"non-empty element '%s' has no content: %s" self.pop_output()
% (macroname, line[:12])) continue
if line[0] == endchar and not stack:
if DEBUG:
self.err_write("subconvert() --> %s\n" % `line[1:21]`)
self.line = line[1:]
return self.line
if line[0] == "}":
# end of macro or group
macroname = stack[-1]
conversion = self.table.get(macroname)
if macroname \
and macroname not in self.discards \
and type(conversion) is not type(""):
# otherwise, it was just a bare group
self.write(")%s\n" % stack[-1])
popping(macroname, "d", len(stack) + depth - 1)
del stack[-1]
line = line[1:]
continue
if line[0] == "{":
pushing("", "e", len(stack) + depth)
stack.append("")
line = line[1:]
continue
if line[0] == "\\" and line[1] in ESCAPED_CHARS:
self.write("-%s\n" % encode(line[1]))
line = line[2:]
continue
if line[:2] == r"\\":
self.write("(BREAK\n)BREAK\n")
line = line[2:]
continue
m = _text_rx.match(line)
if m:
text = encode(m.group())
self.write("-%s\n" % text)
line = line[m.end():] line = line[m.end():]
if not opened: continue
ofp.write("(%s\n" % macroname) # special case because of \item[]
pushing(macroname, "d", len(stack) + depth) if line[0] == "]":
if empty: self.write("-]\n")
line = "}" + line line = line[1:]
stack.append(macroname) continue
ofp = real_ofp # avoid infinite loops
continue extra = ""
if line[0] == endchar and not stack: if len(line) > 100:
if DEBUG: extra = "..."
sys.stderr.write("subconvert() --> %s\n" % `line[1:21]`) raise LaTeXFormatError("could not identify markup: %s%s"
return line[1:] % (`line[:100]`, extra))
if line[0] == "}": while stack and stack[-1] in self.autoclosing:
# end of macro self.write("-\\n\n")
macroname = stack[-1] self.write(")%s\n" % stack[-1])
conversion = table.get(macroname) popping(stack.pop(), "e", len(stack) + depth - 1)
if macroname \ if stack:
and macroname not in discards \ raise LaTeXFormatError("elements remain on stack: "
and type(conversion) is not type(""): + string.join(stack, ", "))
# otherwise, it was just a bare group # otherwise we just ran out of input here...
ofp.write(")%s\n" % stack[-1])
popping(macroname, "d", len(stack) + depth - 1) def convert(self):
del stack[-1] self.subconvert()
line = line[1:]
continue def start_macro(self, name):
if line[0] == "{": conversion = self.table.get(name, ([], 0, 0, 0, 0))
pushing("", "e", len(stack) + depth) params, optional, empty, environ, nocontent = conversion
stack.append("") if empty:
line = line[1:] self.write("e\n")
continue elif nocontent:
if line[0] == "\\" and line[1] in ESCAPED_CHARS: empty = 1
ofp.write("-%s\n" % encode(line[1])) return params, optional, empty, environ
line = line[2:]
continue
if line[:2] == r"\\":
ofp.write("(BREAK\n)BREAK\n")
line = line[2:]
continue
m = _text_rx.match(line)
if m:
text = encode(m.group())
ofp.write("-%s\n" % text)
line = line[m.end():]
continue
# special case because of \item[]
if line[0] == "]":
ofp.write("-]\n")
line = line[1:]
continue
# avoid infinite loops
extra = ""
if len(line) > 100:
extra = "..."
raise LaTeXFormatError("could not identify markup: %s%s"
% (`line[:100]`, extra))
while stack and stack[-1] in autoclosing:
ofp.write("-\\n\n")
ofp.write(")%s\n" % stack[-1])
popping(stack[-1], "e", len(stack) + depth - 1)
del stack[-1]
if stack:
raise LaTeXFormatError("elements remain on stack: "
+ string.join(stack))
# otherwise we just ran out of input here...
def convert(ifp, ofp, table={}, discards=(), autoclosing=()): def convert(ifp, ofp, table={}, discards=(), autoclosing=()):
lines = string.split(ifp.read(), "\n") c = Conversion(ifp, ofp, table, discards, autoclosing)
for i in range(len(lines)):
lines[i] = string.rstrip(lines[i])
data = string.join(lines, "\n")
try: try:
subconvert(data, ofp, table, discards, autoclosing) c.convert()
except IOError, (err, msg): except IOError, (err, msg):
if err != errno.EPIPE: if err != errno.EPIPE:
raise raise