Mercurial > hg > lxmldump
annotate lxmldump.py @ 54:884770576e74
Rename a token.
author | Matti Hamalainen <ccr@tnsp.org> |
---|---|
date | Thu, 27 May 2021 14:28:08 +0300 |
parents | 833606b39e35 |
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": { |
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 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
50 "word_fmt": { |
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 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
91 "sense_index": { |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
92 PKK_MODE_NORMAL: "{indent}sense #{index}\n", |
49
d3d4b547f86c
Some prodding at Anki output.
Matti Hamalainen <ccr@tnsp.org>
parents:
48
diff
changeset
|
93 PKK_MODE_ANKI: "#{index}: ", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
94 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
95 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
96 "definition_fmt": { |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
97 PKK_MODE_NORMAL: "{indent}defn \"{text}\"\n", |
49
d3d4b547f86c
Some prodding at Anki output.
Matti Hamalainen <ccr@tnsp.org>
parents:
48
diff
changeset
|
98 PKK_MODE_ANKI: " * \"{text}\"", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
99 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
100 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
101 "example_fmt": { |
54 | 102 PKK_MODE_NORMAL: "{indent}exmp \"{text}\"{geo_list}\n", |
103 PKK_MODE_ANKI: " ⚫ \"{text}\"{geo_list}", | |
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 "example_geo_list": { |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
106 PKK_MODE_NORMAL: " ({alist})", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
107 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
108 "example_geo_list_empty": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
109 PKK_MODE_NORMAL: "", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
110 }, |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
111 "example_geo_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
112 PKK_MODE_NORMAL: "{text} [{tclass}]", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
113 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
114 "example_geo_list_sep": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
115 PKK_MODE_NORMAL: ", ", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
116 }, |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
117 } |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
118 |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
119 |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
120 # Element annotation mappings |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
121 pkk_element_annotation_map = { |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
122 "Fragment" : { |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
123 PKK_MODE_NORMAL: ["<", ">"], |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
124 PKK_MODE_ANKI: ["<", ">"], |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
125 }, |
7 | 126 } |
127 | |
128 | |
28 | 129 # List of words in kks1/ useful for debugging, option -p |
7 | 130 pkk_debug_list = [ |
131 "ahas", | |
132 "ahavakkaine", | |
133 "ahavakala", | |
134 "ahavakoittuo", | |
135 "ahvaliha", | |
9 | 136 "aloilleh", |
18 | 137 "hanjahtoakseh", |
19 | 138 "akkalisto", |
33 | 139 "alto-", |
50 | 140 "allot-", |
7 | 141 ] |
142 | |
143 | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
144 pkk_settings = {} |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
145 |
21 | 146 |
0 | 147 ### |
148 ### Misc. helper functions, etc | |
149 ### | |
150 | |
13 | 151 ## 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
|
152 def pkk_print(smsg): |
12 | 153 try: |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
154 if pkk_cfg.normalize: |
12 | 155 sys.stdout.write(unicodedata.normalize("NFC", smsg)) |
156 else: | |
157 sys.stdout.write(smsg) | |
158 | |
159 except (BrokenPipeError, IOError) as e: | |
160 sys.stderr.close() | |
161 | |
0 | 162 |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
163 ## Get indentation string |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
164 def pkk_geti(indent): |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
165 return " " * pkk_cfg.indent * indent |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
166 |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
167 |
13 | 168 ## Print string with indentation |
7 | 169 def pkk_printi(indent, smsg): |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
170 pkk_print(pkk_geti(indent) + smsg) |
7 | 171 |
0 | 172 |
173 ## Fatal error handler | |
174 def pkk_fatal(smsg): | |
175 print(u"ERROR: "+ smsg) | |
176 sys.exit(1) | |
177 | |
178 | |
179 ## Handler for SIGINT signals | |
180 def pkk_signal_handler(signal, frame): | |
181 print(u"\nQuitting due to SIGINT / Ctrl+C!") | |
182 sys.exit(1) | |
183 | |
184 | |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
185 class pkk_set_mode(argparse.Action): |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
186 def __call__(self, parser, namespace, value, option_string=None): |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
187 lvalue = value.strip().lower() |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
188 for mode in pkk_modes_list: |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
189 if pkk_modes_list[mode] == lvalue: |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
190 setattr(namespace, self.dest, mode) |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
191 return |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
192 |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
193 pkk_fatal(f"Invalid output mode '{lvalue}'.") |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
194 |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
195 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
196 ## Value handling |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
197 class pkk_set_value(argparse.Action): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
198 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
199 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
|
200 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
201 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
|
202 rmatch = re.match(self.rexpr, values) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
203 if rmatch: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
204 rid = rmatch.group(1).lower().replace("-", "_") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
205 rval = rmatch.group(2) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
206 if rid in pkk_mode_defaults: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
207 pkk_settings[rid] = rval |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
208 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
209 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
|
210 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
211 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
|
212 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
213 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
214 ## Get mode if it exists |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
215 def pkk_test_value(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
216 if mid in pkk_mode_defaults: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
217 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
|
218 mmode = pkk_cfg.mode |
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 mmode = PKK_MODE_NORMAL |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
221 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
222 if mmode in pkk_mode_defaults[mid]: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
223 return mmode |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
224 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
225 return None |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
226 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
227 return None |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
228 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
229 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
230 ## Get default value per mode |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
231 def pkk_get_value(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
232 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
|
233 return pkk_settings[mid] |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
234 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
235 mmode = pkk_test_value(mid) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
236 if mmode == None: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
237 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
|
238 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
239 return pkk_mode_defaults[mid][mmode] |
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 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
242 def pkk_get_fmt(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
243 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
|
244 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
245 |
28 | 246 ## Annotate given string with prefix and suffix based on tag |
26 | 247 def pkk_str_annotate(mtag, mstr): |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
248 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
|
249 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
|
250 mmode = pkk_cfg.mode |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
251 else: |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
252 mmode = PKK_MODE_NORMAL |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
253 |
26 | 254 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
|
255 else: |
26 | 256 return mstr |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
257 |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
258 |
13 | 259 ## Clean string by removing tabs and newlines |
10 | 260 def pkk_str_clean(mstr): |
261 return re.sub(r'[\n\r\t]', '', mstr) | |
262 | |
263 | |
28 | 264 ## Format a "Ptr" node as text |
9 | 265 def pkk_ptr_to_text(pnode): |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
266 pfmt = pkk_get_fmt("ptr_fmt") |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
267 return pfmt.format( |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
268 text=("".join(pnode.itertext())).strip(), |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
269 href=pnode.attrib["{http://www.w3.org/TR/xlink}href"]) |
9 | 270 |
271 | |
13 | 272 ## Get text inside a given node |
16 | 273 def pkk_node_to_text(lnode): |
7 | 274 stmp = "" |
275 for pnode in lnode.iter(): | |
9 | 276 if pnode.tag == "Ptr": |
277 stmp += pkk_ptr_to_text(pnode) | |
278 else: | |
279 if isinstance(pnode.text, str): | |
37
e176fcfc0235
Fix overstripping of concatenated strings.
Matti Hamalainen <ccr@tnsp.org>
parents:
36
diff
changeset
|
280 stmp += pkk_str_annotate(pnode.tag, pkk_str_clean(pnode.text)) |
7 | 281 |
9 | 282 if isinstance(pnode.tail, str): |
10 | 283 stmp += pkk_str_clean(pnode.tail) |
7 | 284 |
285 return stmp.strip() | |
286 | |
287 | |
13 | 288 ## Simple recursive dump starting at given node |
7 | 289 def pkk_dump_recursive(indent, lnode): |
290 if lnode.tag in ["Example"]: | |
16 | 291 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
|
292 pkk_printi(indent, f"{lnode.tag} \"{stmp}\"\n") |
6 | 293 else: |
7 | 294 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
|
295 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
|
296 if textstr != "": |
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
297 textstr = " \""+ textstr +"\"" |
7 | 298 else: |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
299 textstr = "" |
6 | 300 |
301 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
|
302 attrstr = " "+ str(lnode.attrib) |
6 | 303 else: |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
304 attrstr = "" |
6 | 305 |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
306 pkk_printi(indent, f"{lnode.tag}{attrstr}{textstr}\n") |
6 | 307 for qnode in lnode.findall("./*"): |
7 | 308 pkk_dump_recursive(indent + 1, qnode) |
6 | 309 |
310 | |
17 | 311 ## Output item(s) under given node with given format string |
48 | 312 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
|
313 dfmt = pkk_get_fmt(dfmtname) |
48 | 314 ostr = "" |
7 | 315 for qnode in dnode.findall(dsub): |
48 | 316 ostr += dfmt.format( |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
317 text=pkk_node_to_text(qnode), |
48 | 318 indent=pkk_geti(indent)) |
319 return ostr | |
7 | 320 |
321 | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
322 def pkk_get_list_str(dlist, dprefix, dfilter): |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
323 if len(dlist) > 0: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
324 if dfilter: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
325 tfmt = pkk_get_fmt(dprefix + "_list_item") |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
326 tlist = [tfmt.format(text=i) for i in dlist] |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
327 else: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
328 tlist = dlist |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
329 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
330 return pkk_get_fmt(dprefix + "_list").format( |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
331 alist=pkk_get_fmt(dprefix + "_list_sep").join(tlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
332 else: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
333 return pkk_get_fmt(dprefix + "_list_empty") |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
334 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
335 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
336 ## Output a main "Headword" or "Sense" node |
48 | 337 def pkk_get_sense(indent, dnode): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
338 # Definition for this sense |
48 | 339 ostr = pkk_get_subs(indent, dnode, "./Definition", "definition_fmt") |
7 | 340 |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
341 # Examples for this sense |
7 | 342 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
|
343 geolist = [] |
24
8c8e8e4504bb
Remove verbosity option as it is not really used.
Matti Hamalainen <ccr@tnsp.org>
parents:
23
diff
changeset
|
344 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
|
345 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
|
346 text=pkk_node_to_text(qnode), |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
347 tclass=qnode.attrib["class"])) |
7 | 348 |
48 | 349 ostr += pkk_get_fmt("example_fmt").format( |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
350 text=pkk_node_to_text(wnode.find("./Example")), |
54 | 351 geo_list=pkk_get_list_str(geolist, "example_geo", False), |
48 | 352 indent=pkk_geti(indent + 1)) |
353 | |
354 return ostr | |
7 | 355 |
356 | |
17 | 357 ## Output one "DictionaryEntry" node |
7 | 358 def pkk_output_node(indent, dnode): |
6 | 359 |
7 | 360 for wnode in dnode.findall("./HeadwordCtn"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
361 # Get head word |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
362 headword = pkk_node_to_text(wnode.find("./Headword")) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
363 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
364 # Collect search forms |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
365 srchlist = [] |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
366 for qnode in wnode.findall("./SearchForm"): |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
367 srchlist.append(pkk_node_to_text(qnode)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
368 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
369 # Remove dupe if headword is also in srchlist |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
370 if headword in srchlist: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
371 srchlist.remove(headword) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
372 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
373 # Remove other duplicates and sort |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
374 srchlist = list(set(srchlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
375 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
|
376 |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
377 # Get hyphenation note, if any |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
378 hnode = wnode.find("./Hyphenation") |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
379 if hnode != None: |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
380 hyphenation = pkk_get_fmt("hyphenation").format( |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
381 text=pkk_node_to_text(hnode), |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
382 indent=pkk_geti(indent + 1)) |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
383 else: |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
384 hyphenation = pkk_get_fmt("no_hyphenation").format( |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
385 indent=pkk_geti(indent + 1)) |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
386 |
18 | 387 # Create list with grammatical attributes (noun, verb, etc.) |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
388 attrlist = [] |
18 | 389 for pnode in wnode.findall("./PartOfSpeechCtn/PartOfSpeech"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
390 attrlist.append(pnode.attrib["freeValue"]) |
18 | 391 |
392 for pnode in wnode.findall("./GrammaticalNote"): | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
393 attrlist.append(pkk_node_to_text(pnode)) |
18 | 394 |
395 # Remove duplicates and sort the list | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
396 attrlist = list(set(attrlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
397 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
|
398 |
48 | 399 # Get main "sense" |
400 msense = pkk_get_sense(indent + 1, wnode) | |
401 | |
402 # Print any other "senses" | |
403 index = 1 | |
404 osenses = "" | |
405 for znode in dnode.findall("./SenseGrp"): | |
406 osenses += pkk_get_fmt("sense_index").format( | |
407 index=index, | |
408 indent=pkk_geti(indent + 1)) | |
409 osenses += pkk_get_sense(indent + 2, znode) | |
410 index += 1 | |
411 | |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
412 # Print the headword and attributes if any |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
413 pkk_print(pkk_get_fmt("word_fmt").format( |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
414 word=headword, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
415 attr=pkk_get_list_str(attrlist, "word_attr", True), |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
416 search=pkk_get_list_str(srchlist, "search", True), |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
417 hyphenation=hyphenation, |
48 | 418 main_sense=msense, |
419 other_senses=osenses, | |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
420 indent=pkk_geti(indent))) |
18 | 421 |
6 | 422 |
0 | 423 ### |
424 ### Main program starts | |
425 ### | |
426 signal.signal(signal.SIGINT, pkk_signal_handler) | |
427 | |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
428 optparser = argparse.ArgumentParser( |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
429 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
|
430 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
|
431 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
|
432 ) |
0 | 433 |
22 | 434 optparser.add_argument("filenames", |
435 type=str, action="extend", nargs="*", | |
436 metavar="filename", | |
437 help="XML filename(s)") | |
0 | 438 |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
439 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
|
440 dest="show_help", |
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
441 action="store_true", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
442 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
|
443 |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
444 optparser.add_argument("-m", "--mode", |
22 | 445 dest="mode", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
446 action=pkk_set_mode, |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
447 default=PKK_MODE_NORMAL, |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
448 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
|
449 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
450 optparser.add_argument("-s", "--set", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
451 action=pkk_set_value, |
38
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
452 metavar="ID=STR", |
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
453 help='set format string (see below)') |
32
cfecd039506e
Make word attribute separator string configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
31
diff
changeset
|
454 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
455 optparser.add_argument("-n", "--normalize", |
22 | 456 dest="normalize", |
29
f91ef7d7615b
Use store_true feature of argparse.
Matti Hamalainen <ccr@tnsp.org>
parents:
28
diff
changeset
|
457 action="store_true", |
22 | 458 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
|
459 |
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
460 optparser.add_argument("-a", "--annotate", |
22 | 461 dest="annotate", |
29
f91ef7d7615b
Use store_true feature of argparse.
Matti Hamalainen <ccr@tnsp.org>
parents:
28
diff
changeset
|
462 action="store_true", |
22 | 463 help="annotate strings") |
0 | 464 |
22 | 465 optparser.add_argument("-i", "--indent", |
466 dest="indent", | |
467 type=int, choices=range(0, 32), default=4, | |
38
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
468 metavar="N", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
469 help='set indentation level (default: %(default)s)') |
0 | 470 |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
471 optparser.add_argument("-p", "--debug", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
472 dest="debug", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
473 action="store_true", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
474 help=argparse.SUPPRESS) |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
475 |
0 | 476 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
477 ### Parse arguments |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
478 pkk_cfg = optparser.parse_args() |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
479 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
480 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
481 ### 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
|
482 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
|
483 optparser.print_help() |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
484 print(u"\nAvailable output modes:") |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
485 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
|
486 |
41 | 487 if pkk_cfg.mode not in [PKK_MODE_NORMAL, PKK_MODE_ANKI]: |
488 pkk_cfg.mode = PKK_MODE_NORMAL | |
489 | |
490 print(u"\nAvailable format strings and values (mode '{}'):".format( | |
491 pkk_modes_list[pkk_cfg.mode])) | |
492 | |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
493 for mid in pkk_mode_defaults: |
41 | 494 stmp = pkk_get_value(mid).replace("\\", "\\\\").replace("\n", "\\n") |
42 | 495 print(u" {:22s} : '{}'".format(mid, stmp)) |
41 | 496 |
0 | 497 sys.exit(0) |
498 | |
499 | |
6 | 500 ### 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
|
501 for filename in pkk_cfg.filenames: |
0 | 502 # Parse XML file into element tree |
503 try: | |
504 uxml = xmlET.parse(filename) | |
505 except Exception as e: | |
506 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e))) | |
507 | |
508 # Dump output | |
509 try: | |
510 xroot = uxml.getroot() | |
511 for dnode in xroot.findall("./DictionaryEntry"): | |
7 | 512 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
513 if pkk_cfg.debug and dnode.attrib["identifier"] not in pkk_debug_list: |
7 | 514 continue |
515 | |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
516 if pkk_cfg.mode in [PKK_MODE_NORMAL, PKK_MODE_ANKI]: |
19 | 517 try: |
518 pkk_output_node(0, dnode) | |
519 except Exception as e: | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
520 print("") |
19 | 521 pkk_dump_recursive(0, dnode) |
522 print(str(e)) | |
523 sys.exit(0) | |
21 | 524 elif pkk_cfg.mode == PKK_MODE_DUMP: |
7 | 525 pkk_dump_recursive(0, dnode) |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
526 print("") |
21 | 527 elif pkk_cfg.mode == PKK_MODE_XML: |
47 | 528 pkk_print(xmlET.tostring(dnode, encoding="utf8").decode("utf8") + "\n\n") |
0 | 529 else: |
10 | 530 pkk_fatal("Invalid operation mode?") |
7 | 531 |
0 | 532 except (BrokenPipeError, IOError) as e: |
533 sys.stderr.close() | |
534 sys.exit(1) | |
535 | |
536 sys.exit(0) |