Mercurial > cgi-bin > hgweb.cgi > curlyq
annotate curlyq @ 26:3264788aa0c8
Add --tex option.
| author | David Barts <n5jrn@me.com> |
|---|---|
| date | Sun, 22 Mar 2020 08:32:52 -0700 |
| parents | f4cc6d8cafe8 |
| children | 70e75dd07e03 |
| rev | line source |
|---|---|
| 3 | 1 #!/usr/bin/env python3 |
| 2 # -*- coding: utf-8 -*- | |
| 3 | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
4 # I m p o r t s |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
5 |
| 3 | 6 import os, sys |
| 7 import argparse | |
| 8 import codecs | |
| 9 | |
| 4 | 10 from curlers import TextCurler, HtmlCurler, uncurl |
| 10 | 11 from runes import Workspace |
| 3 | 12 |
| 13 # V a r i a b l e s | |
| 14 | |
| 15 # Name invoked by | |
| 16 MYNAME = os.path.basename(sys.argv[0]) | |
| 17 | |
| 18 # Streams | |
| 19 input_fp = None | |
| 20 output_fp = None | |
| 21 | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
22 # Codecs we support |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
23 CODECS_TO_NAME = {} |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
24 for i in [ "UTF-8", "UTF-16", "UTF-16LE", "UTF-16BE", "UTF-32", "UTF-32LE", "UTF-32BE" ]: |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
25 CODECS_TO_NAME[codecs.lookup(i)] = i |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
26 del i |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
27 |
| 23 | 28 # For feet/inches/min/sec |
| 29 BACKT = "`" | |
| 30 FTMIN = "'" | |
| 31 INSEC = '"' | |
| 32 | |
| 26 | 33 # For --tex option |
| 34 TEX_SUBST = [ ("---", "—"), ("--", "–"), ("...", "…") ] | |
| 35 | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
36 # C l a s s e s |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
37 |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
38 class SafeWorkspace(Workspace): |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
39 def __getitem__(self, key): |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
40 try: |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
41 return super().__getitem__(key) |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
42 except IndexError: |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
43 return "" |
|
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
44 |
| 3 | 45 # F u n c t i o n s |
| 46 | |
| 47 def normal(): | |
| 4 | 48 global input_fp, output_fp, args |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
49 ws = SafeWorkspace() |
| 10 | 50 curler = TextCurler(ws) |
| 51 while True: | |
| 52 line = input_fp.readline() | |
| 53 ws.append(line) | |
| 54 if line == "" or line == "\n": | |
| 55 if args.force: uncurl(ws) | |
| 56 curler.feed() | |
| 23 | 57 if args.backtick: fims(ws) |
| 26 | 58 if args.tex: tex(ws) |
| 10 | 59 output_fp.write(str(ws)) |
| 60 ws.clear() | |
| 61 if line == "": | |
| 62 break | |
| 3 | 63 |
| 64 def flowed(): | |
| 65 global input_fp, output_fp, args | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
66 ws = SafeWorkspace() |
| 10 | 67 curler = TextCurler(ws) |
| 68 while True: | |
| 69 line = input_fp.readline() | |
| 70 if line == "": | |
| 71 break | |
| 72 ws.append(line) | |
| 73 if args.force: uncurl(ws) | |
| 74 curler.feed() | |
| 23 | 75 if args.backtick: fims(ws) |
| 26 | 76 if args.tex: tex(ws) |
| 10 | 77 output_fp.write(str(ws)) |
| 78 ws.clear() | |
| 3 | 79 |
| 80 def html(): | |
| 81 global input_fp, output_fp | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
82 ws = SafeWorkspace(input_fp.read()) |
| 10 | 83 curler = HtmlCurler(ws) |
| 84 if args.force: uncurl(ws) | |
| 85 curler.feed() | |
| 86 output_fp.write(str(ws)) | |
| 3 | 87 |
| 5 | 88 def do_uncurl(): |
| 89 global input_fp, output_fp | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
90 ws = SafeWorkspace(input_fp.read()) |
| 10 | 91 uncurl(ws) |
| 92 output_fp.write(str(ws)) | |
| 5 | 93 |
| 23 | 94 def fims(buf): |
| 95 pos = 0 | |
| 96 while True: | |
| 24 | 97 pos = buf.find(BACKT, pos) |
| 23 | 98 if pos < 0: |
| 99 break | |
| 100 if buf[pos+1] == BACKT: | |
| 101 buf[pos:pos+2] = INSEC | |
| 102 else: | |
| 103 buf[pos] = FTMIN | |
| 104 pos += 1 | |
| 105 | |
| 26 | 106 def gsub(buf, old, repl): |
| 107 old = Workspace(old) | |
| 108 repl = Workspace(repl) | |
| 109 olen = len(old) | |
| 110 delta = len(repl) | |
| 111 pos = 0 | |
| 112 while True: | |
| 113 pos = buf.find(old, pos) | |
| 114 if pos < 0: | |
| 115 break | |
| 116 buf[pos:pos+olen] = repl | |
| 117 pos += delta | |
| 118 | |
| 119 def tex(buf): | |
| 120 global TEX_SUBST | |
| 121 for i in TEX_SUBST: | |
| 122 gsub(buf, i[0], i[1]) | |
| 123 | |
| 3 | 124 # M a i n P r o g r a m |
| 125 | |
| 126 # Parse arguments | |
| 8 | 127 parser = argparse.ArgumentParser( |
| 128 description='Make straight quotes curly.', prog=MYNAME) | |
| 3 | 129 group = parser.add_mutually_exclusive_group() |
| 23 | 130 parser.add_argument("--backtick", action="store_true", help="Use backticks on input for ft/in/min/sec.") |
| 3 | 131 group.add_argument("--flowed", action="store_true", help="Input is flowed text.") |
| 132 group.add_argument("--html", action="store_true", help="Input is HTML.") | |
| 5 | 133 group.add_argument("--uncurl", action="store_true", help="Uncurl quotes instead of curling them.") |
| 3 | 134 parser.add_argument("--force", action="store_true", help="Force all quotes to straight ones first.") |
| 8 | 135 parser.add_argument("--icoding", default="UTF-8", help="Input encoding (default UTF-8).") |
| 3 | 136 parser.add_argument("--inplace", action="store_true", help="Edit file in-place.") |
| 8 | 137 parser.add_argument("--ocoding", default="UTF-8", help="Output encoding (default UTF-8).") |
| 26 | 138 parser.add_argument("--tex", action="store_true", help="TeX/LaTeX style dash and ellipsis substitution.") |
| 3 | 139 parser.add_argument("input", nargs="?", help="Input file.") |
| 140 parser.add_argument("output", nargs="?", help="Output file.") | |
| 141 try: | |
| 142 args = parser.parse_args() | |
| 143 except SystemExit: | |
| 144 sys.exit(2) | |
| 145 | |
| 26 | 146 # Sanity checks |
| 23 | 147 if args.html and args.backtick: |
| 148 sys.stderr.write(MYNAME + ": --backtick not supported in --html mode\n") | |
| 149 sys.exit(2) | |
| 26 | 150 if args.html and args.tex: |
| 151 sys.stderr.write(MYNAME + ": --tex not supported in --html mode\n") | |
| 152 sys.exit(2) | |
| 23 | 153 |
| 3 | 154 # Sanity-check codings |
| 155 try: | |
| 156 codec = codecs.lookup(args.icoding) | |
| 157 codec = codecs.lookup(args.ocoding) | |
| 158 except LookupError as e: | |
| 159 sys.stderr.write("{0}: {1!s}\n".format(MYNAME, e)) | |
| 160 sys.exit(2) | |
|
22
a771878f6cf4
Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents:
10
diff
changeset
|
161 if codec not in CODECS_TO_NAME: |
| 8 | 162 sys.stderr.write("{0}: {1!r} output coding does not support Unicode\n".format(MYNAME, args.ocoding)) |
| 3 | 163 sys.exit(1) |
| 164 del codec | |
| 165 | |
| 166 # Get streams | |
| 167 try: | |
| 168 if args.input and (not args.output) and args.inplace: | |
| 169 args.output = args.input | |
| 170 args.input += "~" | |
| 5 | 171 os.rename(args.output, args.input) |
| 3 | 172 if args.input: |
| 173 input_fp = open(args.input, "r", encoding=args.icoding) | |
| 174 else: | |
| 175 input_fp = open(0, "r", encoding=args.icoding) | |
| 176 if args.output: | |
| 177 output_fp = open(args.output, "w", encoding=args.ocoding) | |
| 178 else: | |
| 179 output_fp = open(1, "w", encoding=args.ocoding) | |
| 180 except (OSError, LookupError) as e: | |
| 181 sys.stderr.write("{0}: {1!s}\n".format(MYNAME, e)) | |
| 182 sys.exit(1) | |
| 183 | |
| 184 # Choose our mode | |
| 185 if args.flowed: | |
| 186 flowed() | |
| 187 elif args.html: | |
| 188 html() | |
| 5 | 189 elif args.uncurl: |
| 190 do_uncurl() | |
| 3 | 191 else: |
| 192 normal() |