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