annotate curlyq @ 23:dc30266d4d5b

Add --backtick mode.
author David Barts <n5jrn@me.com>
date Wed, 15 Jan 2020 09:06:55 -0800
parents a771878f6cf4
children f4cc6d8cafe8
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
1 #!/usr/bin/env python3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
2 # -*- coding: utf-8 -*-
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
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
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
6 import os, sys
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
7 import argparse
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
8 import codecs
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
9
4
7a83e82e65a6 Remove some deadwood.
David Barts <n5jrn@me.com>
parents: 3
diff changeset
10 from curlers import TextCurler, HtmlCurler, uncurl
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
11 from runes import Workspace
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
12
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
13 # V a r i a b l e s
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
14
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
15 # Name invoked by
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
16 MYNAME = os.path.basename(sys.argv[0])
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
17
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
18 # Streams
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
19 input_fp = None
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
20 output_fp = None
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
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
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
28 # For feet/inches/min/sec
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
29 BACKT = "`"
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
30 FTMIN = "'"
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
31 INSEC = '"'
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
32
22
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
33 # C l a s s e s
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
34
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
35 class SafeWorkspace(Workspace):
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
36 def __getitem__(self, key):
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
37 try:
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
38 return super().__getitem__(key)
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
39 except IndexError:
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
40 return ""
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
41
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
42 # F u n c t i o n s
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
43
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
44 def normal():
4
7a83e82e65a6 Remove some deadwood.
David Barts <n5jrn@me.com>
parents: 3
diff changeset
45 global input_fp, output_fp, args
22
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
46 ws = SafeWorkspace()
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
47 curler = TextCurler(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
48 while True:
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
49 line = input_fp.readline()
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
50 ws.append(line)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
51 if line == "" or line == "\n":
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
52 if args.force: uncurl(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
53 curler.feed()
23
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
54 if args.backtick: fims(ws)
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
55 output_fp.write(str(ws))
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
56 ws.clear()
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
57 if line == "":
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
58 break
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
59
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
60 def flowed():
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
61 global input_fp, output_fp, args
22
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
62 ws = SafeWorkspace()
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
63 curler = TextCurler(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
64 while True:
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
65 line = input_fp.readline()
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
66 if line == "":
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
67 break
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
68 ws.append(line)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
69 if args.force: uncurl(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
70 curler.feed()
23
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
71 if args.backtick: fims(ws)
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
72 output_fp.write(str(ws))
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
73 ws.clear()
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
74
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
75 def html():
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
76 global input_fp, output_fp
22
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
77 ws = SafeWorkspace(input_fp.read())
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
78 curler = HtmlCurler(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
79 if args.force: uncurl(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
80 curler.feed()
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
81 output_fp.write(str(ws))
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
82
5
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
83 def do_uncurl():
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
84 global input_fp, output_fp
22
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
85 ws = SafeWorkspace(input_fp.read())
10
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
86 uncurl(ws)
397c178c5b98 Make it array-based.
David Barts <n5jrn@me.com>
parents: 8
diff changeset
87 output_fp.write(str(ws))
5
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
88
23
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
89 def fims(buf):
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
90 pos = 0
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
91 while True:
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
92 pos = buf.find(BACKT)
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
93 if pos < 0:
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
94 break
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
95 if buf[pos+1] == BACKT:
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
96 buf[pos:pos+2] = INSEC
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
97 else:
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
98 buf[pos] = FTMIN
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
99 pos += 1
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
100
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
101 # M a i n P r o g r a m
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
102
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
103 # Parse arguments
8
05363e803272 Improve help text.
David Barts <n5jrn@me.com>
parents: 5
diff changeset
104 parser = argparse.ArgumentParser(
05363e803272 Improve help text.
David Barts <n5jrn@me.com>
parents: 5
diff changeset
105 description='Make straight quotes curly.', prog=MYNAME)
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
106 group = parser.add_mutually_exclusive_group()
23
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
107 parser.add_argument("--backtick", action="store_true", help="Use backticks on input for ft/in/min/sec.")
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
108 group.add_argument("--flowed", action="store_true", help="Input is flowed text.")
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
109 group.add_argument("--html", action="store_true", help="Input is HTML.")
5
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
110 group.add_argument("--uncurl", action="store_true", help="Uncurl quotes instead of curling them.")
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
111 parser.add_argument("--force", action="store_true", help="Force all quotes to straight ones first.")
8
05363e803272 Improve help text.
David Barts <n5jrn@me.com>
parents: 5
diff changeset
112 parser.add_argument("--icoding", default="UTF-8", help="Input encoding (default UTF-8).")
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
113 parser.add_argument("--inplace", action="store_true", help="Edit file in-place.")
8
05363e803272 Improve help text.
David Barts <n5jrn@me.com>
parents: 5
diff changeset
114 parser.add_argument("--ocoding", default="UTF-8", help="Output encoding (default UTF-8).")
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
115 parser.add_argument("input", nargs="?", help="Input file.")
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
116 parser.add_argument("output", nargs="?", help="Output file.")
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
117 try:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
118 args = parser.parse_args()
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
119 except SystemExit:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
120 sys.exit(2)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
121
23
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
122 # Sanity check
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
123 if args.html and args.backtick:
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
124 sys.stderr.write(MYNAME + ": --backtick not supported in --html mode\n")
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
125 sys.exit(2)
dc30266d4d5b Add --backtick mode.
David Barts <n5jrn@me.com>
parents: 22
diff changeset
126
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
127 # Sanity-check codings
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
128 try:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
129 codec = codecs.lookup(args.icoding)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
130 codec = codecs.lookup(args.ocoding)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
131 except LookupError as e:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
132 sys.stderr.write("{0}: {1!s}\n".format(MYNAME, e))
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
133 sys.exit(2)
22
a771878f6cf4 Remove deadwood, update runes.py.
David Barts <n5jrn@me.com>
parents: 10
diff changeset
134 if codec not in CODECS_TO_NAME:
8
05363e803272 Improve help text.
David Barts <n5jrn@me.com>
parents: 5
diff changeset
135 sys.stderr.write("{0}: {1!r} output coding does not support Unicode\n".format(MYNAME, args.ocoding))
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
136 sys.exit(1)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
137 del codec
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
138
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
139 # Get streams
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
140 try:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
141 if args.input and (not args.output) and args.inplace:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
142 args.output = args.input
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
143 args.input += "~"
5
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
144 os.rename(args.output, args.input)
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
145 if args.input:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
146 input_fp = open(args.input, "r", encoding=args.icoding)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
147 else:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
148 input_fp = open(0, "r", encoding=args.icoding)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
149 if args.output:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
150 output_fp = open(args.output, "w", encoding=args.ocoding)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
151 else:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
152 output_fp = open(1, "w", encoding=args.ocoding)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
153 except (OSError, LookupError) as e:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
154 sys.stderr.write("{0}: {1!s}\n".format(MYNAME, e))
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
155 sys.exit(1)
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
156
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
157 # Choose our mode
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
158 if args.flowed:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
159 flowed()
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
160 elif args.html:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
161 html()
5
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
162 elif args.uncurl:
d5198c7ec54d Add --uncurl option.
David Barts <n5jrn@me.com>
parents: 4
diff changeset
163 do_uncurl()
3
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
164 else:
091c03f1b2e8 Getting it working...
David Barts <n5jrn@me.com>
parents:
diff changeset
165 normal()