#!/usr/bin/env python # Creates a man page from a C file. # first argument if present is path to cmark dynamic library # Comments beginning with `/**` are treated as Groff man, except that # 'this' is converted to \fIthis\f[], and ''this'' to \fBthis\f[]. # Non-blank lines immediately following a man page comment are treated # as function signatures or examples and parsed into .Ft, .Fo, .Fa, .Fc. The # immediately preceding man documentation chunk is printed after the example # as a comment on it. # That's about it! import sys, re, os, platform from datetime import date from ctypes import CDLL, c_char_p, c_long, c_void_p sysname = platform.system() if sysname == 'Darwin': cmark = CDLL("build/src/libcmark-gfm.dylib") else: cmark = CDLL("build/src/libcmark-gfm.so") parse_document = cmark.cmark_parse_document parse_document.restype = c_void_p parse_document.argtypes = [c_char_p, c_long] render_man = cmark.cmark_render_man render_man.restype = c_char_p render_man.argtypes = [c_void_p, c_long, c_long] def md2man(text): if sys.version_info >= (3,0): textbytes = text.encode('utf-8') textlen = len(textbytes) return render_man(parse_document(textbytes, textlen), 0, 65).decode('utf-8') else: textbytes = text textlen = len(text) return render_man(parse_document(textbytes, textlen), 0, 72) comment_start_re = re.compile('^\/\*\* ?') comment_delim_re = re.compile('^[/ ]\** ?') comment_end_re = re.compile('^ \**\/') function_re = re.compile('^ *(?:CMARK_GFM_EXPORT\s+)?(?P(?:const\s+)?\w+(?:\s*[*])?)\s*(?P\w+)\s*\((?P[^)]*)\)') blank_re = re.compile('^\s*$') macro_re = re.compile('CMARK_GFM_EXPORT *') typedef_start_re = re.compile('typedef.*{$') typedef_end_re = re.compile('}') single_quote_re = re.compile("(?**', re.sub(single_quote_re, '*\g<1>*', s)) typedef = False mdlines = [] chunk = [] sig = [] if len(sys.argv) > 1: sourcefile = sys.argv[1] else: print("Usage: make_man_page.py sourcefile") exit(1) with open(sourcefile, 'r') as cmarkh: state = 'default' for line in cmarkh: # state transition oldstate = state if comment_start_re.match(line): state = 'man' elif comment_end_re.match(line) and state == 'man': continue elif comment_delim_re.match(line) and state == 'man': state = 'man' elif not typedef and blank_re.match(line): state = 'default' elif typedef and typedef_end_re.match(line): typedef = False elif typedef_start_re.match(line): typedef = True state = 'signature' elif state == 'man': state = 'signature' # handle line if state == 'man': chunk.append(handle_quotes(re.sub(comment_delim_re, '', line))) elif state == 'signature': ln = re.sub(macro_re, '', line) if typedef or not re.match(blank_re, ln): sig.append(ln) elif oldstate == 'signature' and state != 'signature': if len(mdlines) > 0 and mdlines[-1] != '\n': mdlines.append('\n') rawsig = ''.join(sig) m = function_re.match(rawsig) mdlines.append('.PP\n') if m: mdlines.append('\\fI' + m.group('type') + '\\f[]' + ' ') mdlines.append('\\fB' + m.group('name') + '\\f[]' + '(') first = True for argument in re.split(',', m.group('args')): if not first: mdlines.append(', ') first = False mdlines.append('\\fI' + argument.strip() + '\\f[]') mdlines.append(')\n') else: mdlines.append('.nf\n\\fC\n.RS 0n\n') mdlines += sig mdlines.append('.RE\n\\f[]\n.fi\n') if len(mdlines) > 0 and mdlines[-1] != '\n': mdlines.append('\n') mdlines += md2man(''.join(chunk)) mdlines.append('\n') chunk = [] sig = [] elif oldstate == 'man' and state != 'signature': if len(mdlines) > 0 and mdlines[-1] != '\n': mdlines.append('\n') mdlines += md2man(''.join(chunk)) # add man chunk chunk = [] mdlines.append('\n') sys.stdout.write('.TH cmark-gfm 3 "' + date.today().strftime('%B %d, %Y') + '" "LOCAL" "Library Functions Manual"\n') sys.stdout.write(''.join(mdlines))