Mercurial > hg > lxmldump
annotate lxmldump.py @ 56:71e66eca1e23
Merge.
author | Matti Hamalainen <ccr@tnsp.org> |
---|---|
date | Thu, 27 May 2021 16:55:22 +0300 |
parents | 301452a71cc7 884770576e74 |
children | 4410e2d1bff3 |
rev | line source |
---|---|
0 | 1 #!/usr/bin/python3 -B |
2 # coding=utf-8 | |
3 ### | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
4 ### lxmldump - Convert and dump ISO/FDIS 1951 XML file data |
4 | 5 ### Programmed and designed by Matti 'ccr' Hämäläinen <ccr@tnsp.org> |
6 ### (C) Copyright 2021 Tecnic Software productions (TNSP) | |
7 ### | |
23 | 8 ### Released / distributed under 3-clause BSD license |
9 ### (see file "COPYING" for more information) | |
10 ### | |
4 | 11 ### Python 3.7+ required! |
0 | 12 ### |
13 import sys | |
14 import signal | |
15 import re | |
16 from pathlib import Path | |
17 import xml.etree.ElementTree as xmlET | |
5
274b2091137c
Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents:
4
diff
changeset
|
18 import unicodedata |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
19 import argparse |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
20 import textwrap |
0 | 21 |
22 assert sys.version_info >= (3, 7) | |
23 | |
24 | |
25 ### | |
26 ### Default settings | |
27 ### | |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
28 # Operation modes |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
29 PKK_MODE_NORMAL = 0 |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
30 PKK_MODE_DUMP = 1 |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
31 PKK_MODE_XML = 2 |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
32 PKK_MODE_ANKI = 3 |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
33 |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
34 |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
35 pkk_modes_list = { |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
36 PKK_MODE_NORMAL: "normal", |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
37 PKK_MODE_DUMP: "dump", |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
38 PKK_MODE_XML: "xml", |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
39 PKK_MODE_ANKI: "anki", |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
40 } |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
41 |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
42 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
43 pkk_mode_defaults = { |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
44 # Default Ptr URL format strings |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
45 "ptr_fmt": { |
52 | 46 PKK_MODE_NORMAL: "<PTR:{href}>{text}</PTR>", |
47 PKK_MODE_ANKI: "<a href='https://kaino.kotus.fi/cgi-bin/kks/karjala.cgi?a={href}'>{text}</a>", | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
48 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
49 |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
50 "word_item": { |
48 | 51 PKK_MODE_NORMAL: "\"{word}\"{search}{attr}\n{hyphenation}{main_sense}{other_senses}\n", |
51
5d3fb2f3aa21
Remove cleanup function as it is unused.
Matti Hamalainen <ccr@tnsp.org>
parents:
50
diff
changeset
|
52 PKK_MODE_ANKI: "\"{word}\"{search}{attr}{hyphenation};{main_sense};{other_senses}\n", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
53 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
54 "word_attr_list": { |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
55 PKK_MODE_NORMAL: " ({alist}) ", |
49
d3d4b547f86c
Some prodding at Anki output.
Matti Hamalainen <ccr@tnsp.org>
parents:
48
diff
changeset
|
56 PKK_MODE_NORMAL: " ({alist})", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
57 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
58 "word_attr_list_empty": { |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
59 PKK_MODE_NORMAL: " ", |
49
d3d4b547f86c
Some prodding at Anki output.
Matti Hamalainen <ccr@tnsp.org>
parents:
48
diff
changeset
|
60 PKK_MODE_ANKI: "", |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
61 }, |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
62 "word_attr_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
63 PKK_MODE_NORMAL: "{text}", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
64 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
65 "word_attr_list_sep": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
66 PKK_MODE_NORMAL: " ; ", |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
67 PKK_MODE_ANKI: " : ", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
68 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
69 |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
70 "search_list": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
71 PKK_MODE_NORMAL: ", {alist}", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
72 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
73 "search_list_empty": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
74 PKK_MODE_NORMAL: "", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
75 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
76 "search_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
77 PKK_MODE_NORMAL: "\"{text}\"", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
78 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
79 "search_list_sep": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
80 PKK_MODE_NORMAL: ", ", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
81 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
82 |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
83 "hyphenation": { |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
84 PKK_MODE_NORMAL: "{indent}hyph \"{text}\"\n", |
50 | 85 PKK_MODE_ANKI: " [hyph: {text}]", |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
86 }, |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
87 "no_hyphenation": { |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
88 PKK_MODE_NORMAL: "", |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
89 }, |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
90 |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
91 "sense_list": { |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
92 PKK_MODE_NORMAL: "{alist}", |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
93 }, |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
94 "sense_list_empty": { |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
95 PKK_MODE_NORMAL: "", |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
96 }, |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
97 "sense_list_item": { |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
98 PKK_MODE_NORMAL: "{indent}sense #{index}\n{text}", |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
99 PKK_MODE_ANKI: "{text}", |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
100 }, |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
101 "sense_list_sep": { |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
102 PKK_MODE_NORMAL: "", |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
103 PKK_MODE_NORMAL: ";", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
104 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
105 |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
106 "definition_item": { |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
107 PKK_MODE_NORMAL: "{indent}defn \"{text}\"\n", |
49
d3d4b547f86c
Some prodding at Anki output.
Matti Hamalainen <ccr@tnsp.org>
parents:
48
diff
changeset
|
108 PKK_MODE_ANKI: " * \"{text}\"", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
109 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
110 |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
111 "example_item": { |
54 | 112 PKK_MODE_NORMAL: "{indent}exmp \"{text}\"{geo_list}\n", |
113 PKK_MODE_ANKI: " ⚫ \"{text}\"{geo_list}", | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
114 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
115 "example_geo_list": { |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
116 PKK_MODE_NORMAL: " ({alist})", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
117 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
118 "example_geo_list_empty": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
119 PKK_MODE_NORMAL: "", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
120 }, |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
121 "example_geo_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
122 PKK_MODE_NORMAL: "{text} [{tclass}]", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
123 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
124 "example_geo_list_sep": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
125 PKK_MODE_NORMAL: ", ", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
126 }, |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
127 } |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
128 |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
129 |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
130 # Element annotation mappings |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
131 pkk_element_annotation_map = { |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
132 "Fragment" : { |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
133 PKK_MODE_NORMAL: ["<", ">"], |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
134 PKK_MODE_ANKI: ["<", ">"], |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
135 }, |
7 | 136 } |
137 | |
138 | |
28 | 139 # List of words in kks1/ useful for debugging, option -p |
7 | 140 pkk_debug_list = [ |
141 "ahas", | |
142 "ahavakkaine", | |
143 "ahavakala", | |
144 "ahavakoittuo", | |
145 "ahvaliha", | |
9 | 146 "aloilleh", |
18 | 147 "hanjahtoakseh", |
19 | 148 "akkalisto", |
33 | 149 "alto-", |
50 | 150 "allot-", |
7 | 151 ] |
152 | |
153 | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
154 pkk_settings = {} |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
155 |
21 | 156 |
0 | 157 ### |
158 ### Misc. helper functions, etc | |
159 ### | |
160 | |
13 | 161 ## Print string to stdout using normalized Unicode if enabled |
5
274b2091137c
Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents:
4
diff
changeset
|
162 def pkk_print(smsg): |
12 | 163 try: |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
164 if pkk_cfg.normalize: |
12 | 165 sys.stdout.write(unicodedata.normalize("NFC", smsg)) |
166 else: | |
167 sys.stdout.write(smsg) | |
168 | |
169 except (BrokenPipeError, IOError) as e: | |
170 sys.stderr.close() | |
171 | |
0 | 172 |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
173 ## Get indentation string |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
174 def pkk_geti(indent): |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
175 return " " * pkk_cfg.indent * indent |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
176 |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
177 |
13 | 178 ## Print string with indentation |
7 | 179 def pkk_printi(indent, smsg): |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
180 pkk_print(pkk_geti(indent) + smsg) |
7 | 181 |
0 | 182 |
183 ## Fatal error handler | |
184 def pkk_fatal(smsg): | |
185 print(u"ERROR: "+ smsg) | |
186 sys.exit(1) | |
187 | |
188 | |
189 ## Handler for SIGINT signals | |
190 def pkk_signal_handler(signal, frame): | |
191 print(u"\nQuitting due to SIGINT / Ctrl+C!") | |
192 sys.exit(1) | |
193 | |
194 | |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
195 class pkk_set_mode(argparse.Action): |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
196 def __call__(self, parser, namespace, value, option_string=None): |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
197 lvalue = value.strip().lower() |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
198 for mode in pkk_modes_list: |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
199 if pkk_modes_list[mode] == lvalue: |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
200 setattr(namespace, self.dest, mode) |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
201 return |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
202 |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
203 pkk_fatal(f"Invalid output mode '{lvalue}'.") |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
204 |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
205 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
206 ## Value handling |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
207 class pkk_set_value(argparse.Action): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
208 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
209 rexpr = re.compile(r'\s*(\w+)\s*=\s*(.*)\s*') |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
210 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
211 def __call__(self, parser, namespace, values, option_string=None): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
212 rmatch = re.match(self.rexpr, values) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
213 if rmatch: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
214 rid = rmatch.group(1).lower().replace("-", "_") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
215 rval = rmatch.group(2) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
216 if rid in pkk_mode_defaults: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
217 pkk_settings[rid] = rval |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
218 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
219 pkk_fatal(f"Invalid option '{option_string} {values}': No such ID '{rid}'.") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
220 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
221 pkk_fatal(f"Invalid option '{option_string} {values}': Expected id=value.") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
222 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
223 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
224 ## Get mode if it exists |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
225 def pkk_test_value(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
226 if mid in pkk_mode_defaults: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
227 if pkk_cfg.mode in pkk_mode_defaults[mid]: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
228 mmode = pkk_cfg.mode |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
229 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
230 mmode = PKK_MODE_NORMAL |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
231 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
232 if mmode in pkk_mode_defaults[mid]: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
233 return mmode |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
234 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
235 return None |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
236 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
237 return None |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
238 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
239 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
240 ## Get default value per mode |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
241 def pkk_get_value(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
242 if mid in pkk_settings and pkk_settings[mid] != None: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
243 return pkk_settings[mid] |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
244 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
245 mmode = pkk_test_value(mid) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
246 if mmode == None: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
247 pkk_fatal(f"Internal error: No mode for ID '{mid}'.") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
248 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
249 return pkk_mode_defaults[mid][mmode] |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
250 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
251 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
252 def pkk_get_fmt(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
253 return pkk_get_value(mid).replace("\\n", "\n") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
254 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
255 |
28 | 256 ## Annotate given string with prefix and suffix based on tag |
26 | 257 def pkk_str_annotate(mtag, mstr): |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
258 if pkk_cfg.annotate and mtag in pkk_element_annotation_map: |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
259 if pkk_cfg.mode in pkk_element_annotation_map[mtag]: |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
260 mmode = pkk_cfg.mode |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
261 else: |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
262 mmode = PKK_MODE_NORMAL |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
263 |
26 | 264 return pkk_element_annotation_map[mtag][mmode][0] + mstr + pkk_element_annotation_map[mtag][mmode][1] |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
265 else: |
26 | 266 return mstr |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
267 |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
268 |
13 | 269 ## Clean string by removing tabs and newlines |
10 | 270 def pkk_str_clean(mstr): |
271 return re.sub(r'[\n\r\t]', '', mstr) | |
272 | |
273 | |
28 | 274 ## Format a "Ptr" node as text |
9 | 275 def pkk_ptr_to_text(pnode): |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
276 pfmt = pkk_get_fmt("ptr_fmt") |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
277 return pfmt.format( |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
278 text=("".join(pnode.itertext())).strip(), |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
279 href=pnode.attrib["{http://www.w3.org/TR/xlink}href"]) |
9 | 280 |
281 | |
13 | 282 ## Get text inside a given node |
16 | 283 def pkk_node_to_text(lnode): |
7 | 284 stmp = "" |
285 for pnode in lnode.iter(): | |
9 | 286 if pnode.tag == "Ptr": |
287 stmp += pkk_ptr_to_text(pnode) | |
288 else: | |
289 if isinstance(pnode.text, str): | |
37
e176fcfc0235
Fix overstripping of concatenated strings.
Matti Hamalainen <ccr@tnsp.org>
parents:
36
diff
changeset
|
290 stmp += pkk_str_annotate(pnode.tag, pkk_str_clean(pnode.text)) |
7 | 291 |
9 | 292 if isinstance(pnode.tail, str): |
10 | 293 stmp += pkk_str_clean(pnode.tail) |
7 | 294 |
295 return stmp.strip() | |
296 | |
297 | |
13 | 298 ## Simple recursive dump starting at given node |
7 | 299 def pkk_dump_recursive(indent, lnode): |
300 if lnode.tag in ["Example"]: | |
16 | 301 stmp = pkk_node_to_text(lnode) |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
302 pkk_printi(indent, f"{lnode.tag} \"{stmp}\"\n") |
6 | 303 else: |
7 | 304 if isinstance(lnode.text, str): |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
305 textstr = pkk_str_clean(lnode.text).strip() |
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
306 if textstr != "": |
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
307 textstr = " \""+ textstr +"\"" |
7 | 308 else: |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
309 textstr = "" |
6 | 310 |
311 if len(lnode.attrib) > 0: | |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
312 attrstr = " "+ str(lnode.attrib) |
6 | 313 else: |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
314 attrstr = "" |
6 | 315 |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
316 pkk_printi(indent, f"{lnode.tag}{attrstr}{textstr}\n") |
6 | 317 for qnode in lnode.findall("./*"): |
7 | 318 pkk_dump_recursive(indent + 1, qnode) |
6 | 319 |
320 | |
17 | 321 ## Output item(s) under given node with given format string |
48 | 322 def pkk_get_subs(indent, dnode, dsub, dfmtname): |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
323 dfmt = pkk_get_fmt(dfmtname) |
48 | 324 ostr = "" |
7 | 325 for qnode in dnode.findall(dsub): |
48 | 326 ostr += dfmt.format( |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
327 text=pkk_node_to_text(qnode), |
48 | 328 indent=pkk_geti(indent)) |
329 return ostr | |
7 | 330 |
331 | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
332 def pkk_get_list_str(dlist, dprefix, dfilter): |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
333 if len(dlist) > 0: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
334 if dfilter: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
335 tfmt = pkk_get_fmt(dprefix + "_list_item") |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
336 tlist = [tfmt.format(text=i) for i in dlist] |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
337 else: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
338 tlist = dlist |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
339 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
340 return pkk_get_fmt(dprefix + "_list").format( |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
341 alist=pkk_get_fmt(dprefix + "_list_sep").join(tlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
342 else: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
343 return pkk_get_fmt(dprefix + "_list_empty") |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
344 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
345 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
346 ## Output a main "Headword" or "Sense" node |
48 | 347 def pkk_get_sense(indent, dnode): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
348 # Definition for this sense |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
349 ostr = pkk_get_subs(indent, dnode, "./Definition", "definition_item") |
7 | 350 |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
351 # Examples for this sense |
7 | 352 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"): |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
353 geolist = [] |
24
8c8e8e4504bb
Remove verbosity option as it is not really used.
Matti Hamalainen <ccr@tnsp.org>
parents:
23
diff
changeset
|
354 for qnode in wnode.findall("./FreeTopic[@type='levikki']/GeographicalUsage"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
355 geolist.append(pkk_get_fmt("example_geo_list_item").format( |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
356 text=pkk_node_to_text(qnode), |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
357 tclass=qnode.attrib["class"])) |
7 | 358 |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
359 ostr += pkk_get_fmt("example_item").format( |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
360 text=pkk_node_to_text(wnode.find("./Example")), |
54 | 361 geo_list=pkk_get_list_str(geolist, "example_geo", False), |
48 | 362 indent=pkk_geti(indent + 1)) |
363 | |
364 return ostr | |
7 | 365 |
366 | |
17 | 367 ## Output one "DictionaryEntry" node |
7 | 368 def pkk_output_node(indent, dnode): |
6 | 369 |
7 | 370 for wnode in dnode.findall("./HeadwordCtn"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
371 # Get head word |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
372 headword = pkk_node_to_text(wnode.find("./Headword")) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
373 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
374 # Collect search forms |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
375 srchlist = [] |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
376 for qnode in wnode.findall("./SearchForm"): |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
377 srchlist.append(pkk_node_to_text(qnode)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
378 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
379 # Remove dupe if headword is also in srchlist |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
380 if headword in srchlist: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
381 srchlist.remove(headword) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
382 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
383 # Remove other duplicates and sort |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
384 srchlist = list(set(srchlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
385 srchlist.sort(reverse=False, key=lambda attr: (attr, len(attr))) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
386 |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
387 # Get hyphenation note, if any |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
388 hnode = wnode.find("./Hyphenation") |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
389 if hnode != None: |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
390 hyphenation = pkk_get_fmt("hyphenation").format( |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
391 text=pkk_node_to_text(hnode), |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
392 indent=pkk_geti(indent + 1)) |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
393 else: |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
394 hyphenation = pkk_get_fmt("no_hyphenation").format( |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
395 indent=pkk_geti(indent + 1)) |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
396 |
18 | 397 # Create list with grammatical attributes (noun, verb, etc.) |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
398 attrlist = [] |
18 | 399 for pnode in wnode.findall("./PartOfSpeechCtn/PartOfSpeech"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
400 attrlist.append(pnode.attrib["freeValue"]) |
18 | 401 |
402 for pnode in wnode.findall("./GrammaticalNote"): | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
403 attrlist.append(pkk_node_to_text(pnode)) |
18 | 404 |
405 # Remove duplicates and sort the list | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
406 attrlist = list(set(attrlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
407 attrlist.sort(reverse=False, key=lambda attr: (attr, len(attr))) |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
408 |
48 | 409 # Get main "sense" |
410 msense = pkk_get_sense(indent + 1, wnode) | |
411 | |
412 # Print any other "senses" | |
413 index = 1 | |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
414 senselist = [] |
48 | 415 for znode in dnode.findall("./SenseGrp"): |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
416 senselist.append(pkk_get_fmt("sense_list_item").format( |
48 | 417 index=index, |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
418 text=pkk_get_sense(indent + 2, znode), |
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
419 indent=pkk_geti(indent + 1))) |
48 | 420 index += 1 |
421 | |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
422 # Print the headword and attributes if any |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
423 pkk_print(pkk_get_fmt("word_item").format( |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
424 word=headword, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
425 attr=pkk_get_list_str(attrlist, "word_attr", True), |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
426 search=pkk_get_list_str(srchlist, "search", True), |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
427 hyphenation=hyphenation, |
48 | 428 main_sense=msense, |
55
301452a71cc7
Make sense output formatting more list-like, rename some items.
Matti Hamalainen <ccr@tnsp.org>
parents:
51
diff
changeset
|
429 other_senses=pkk_get_list_str(senselist, "sense", False), |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
430 indent=pkk_geti(indent))) |
18 | 431 |
6 | 432 |
0 | 433 ### |
434 ### Main program starts | |
435 ### | |
436 signal.signal(signal.SIGINT, pkk_signal_handler) | |
437 | |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
438 optparser = argparse.ArgumentParser( |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
439 description="lxmldump - Convert and dump ISO/FDIS 1951 XML file data", |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
440 usage="%(prog)s [options] <input xml file(s)>", |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
441 add_help=False |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
442 ) |
0 | 443 |
22 | 444 optparser.add_argument("filenames", |
445 type=str, action="extend", nargs="*", | |
446 metavar="filename", | |
447 help="XML filename(s)") | |
0 | 448 |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
449 optparser.add_argument("-h", "--help", |
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
450 dest="show_help", |
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
451 action="store_true", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
452 help="show this help message") |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
453 |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
454 optparser.add_argument("-m", "--mode", |
22 | 455 dest="mode", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
456 action=pkk_set_mode, |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
457 default=PKK_MODE_NORMAL, |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
458 help="set output mode (see below)") |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
459 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
460 optparser.add_argument("-s", "--set", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
461 action=pkk_set_value, |
38
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
462 metavar="ID=STR", |
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
463 help='set format string (see below)') |
32
cfecd039506e
Make word attribute separator string configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
31
diff
changeset
|
464 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
465 optparser.add_argument("-n", "--normalize", |
22 | 466 dest="normalize", |
29
f91ef7d7615b
Use store_true feature of argparse.
Matti Hamalainen <ccr@tnsp.org>
parents:
28
diff
changeset
|
467 action="store_true", |
22 | 468 help="output NFC normalized Unicode") |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
469 |
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
470 optparser.add_argument("-a", "--annotate", |
22 | 471 dest="annotate", |
29
f91ef7d7615b
Use store_true feature of argparse.
Matti Hamalainen <ccr@tnsp.org>
parents:
28
diff
changeset
|
472 action="store_true", |
22 | 473 help="annotate strings") |
0 | 474 |
22 | 475 optparser.add_argument("-i", "--indent", |
476 dest="indent", | |
477 type=int, choices=range(0, 32), default=4, | |
38
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
478 metavar="N", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
479 help='set indentation level (default: %(default)s)') |
0 | 480 |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
481 optparser.add_argument("-p", "--debug", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
482 dest="debug", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
483 action="store_true", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
484 help=argparse.SUPPRESS) |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
485 |
0 | 486 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
487 ### Parse arguments |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
488 pkk_cfg = optparser.parse_args() |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
489 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
490 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
491 ### Show help if needed |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
492 if len(pkk_cfg.filenames) == 0 or pkk_cfg.show_help: |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
493 optparser.print_help() |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
494 print(u"\nAvailable output modes:") |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
495 print(" " + ", ".join(pkk_modes_list.values())) |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
496 |
41 | 497 if pkk_cfg.mode not in [PKK_MODE_NORMAL, PKK_MODE_ANKI]: |
498 pkk_cfg.mode = PKK_MODE_NORMAL | |
499 | |
500 print(u"\nAvailable format strings and values (mode '{}'):".format( | |
501 pkk_modes_list[pkk_cfg.mode])) | |
502 | |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
503 for mid in pkk_mode_defaults: |
41 | 504 stmp = pkk_get_value(mid).replace("\\", "\\\\").replace("\n", "\\n") |
42 | 505 print(u" {:22s} : '{}'".format(mid, stmp)) |
41 | 506 |
0 | 507 sys.exit(0) |
508 | |
509 | |
6 | 510 ### Handle each input file |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
511 for filename in pkk_cfg.filenames: |
0 | 512 # Parse XML file into element tree |
513 try: | |
514 uxml = xmlET.parse(filename) | |
515 except Exception as e: | |
516 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e))) | |
517 | |
518 # Dump output | |
519 try: | |
520 xroot = uxml.getroot() | |
521 for dnode in xroot.findall("./DictionaryEntry"): | |
7 | 522 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
523 if pkk_cfg.debug and dnode.attrib["identifier"] not in pkk_debug_list: |
7 | 524 continue |
525 | |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
526 if pkk_cfg.mode in [PKK_MODE_NORMAL, PKK_MODE_ANKI]: |
19 | 527 try: |
528 pkk_output_node(0, dnode) | |
529 except Exception as e: | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
530 print("") |
19 | 531 pkk_dump_recursive(0, dnode) |
532 print(str(e)) | |
533 sys.exit(0) | |
21 | 534 elif pkk_cfg.mode == PKK_MODE_DUMP: |
7 | 535 pkk_dump_recursive(0, dnode) |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
536 print("") |
21 | 537 elif pkk_cfg.mode == PKK_MODE_XML: |
47 | 538 pkk_print(xmlET.tostring(dnode, encoding="utf8").decode("utf8") + "\n\n") |
0 | 539 else: |
10 | 540 pkk_fatal("Invalid operation mode?") |
7 | 541 |
0 | 542 except (BrokenPipeError, IOError) as e: |
543 sys.stderr.close() | |
544 sys.exit(1) | |
545 | |
546 sys.exit(0) |