Mercurial > hg > lxmldump
annotate lxmldump.py @ 48:6932c3f2bdeb
More flexibility.
author | Matti Hamalainen <ccr@tnsp.org> |
---|---|
date | Wed, 26 May 2021 22:41:29 +0300 |
parents | 2e8282f1a837 |
children | d3d4b547f86c |
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 |
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", |
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}) ", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
56 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
57 "word_attr_list_empty": { |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
58 PKK_MODE_NORMAL: " ", |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
59 }, |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
60 "word_attr_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
61 PKK_MODE_NORMAL: "{text}", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
62 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
63 "word_attr_list_sep": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
64 PKK_MODE_NORMAL: " ; ", |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
65 PKK_MODE_ANKI: " : ", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
66 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
67 |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
68 "search_list": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
69 PKK_MODE_NORMAL: ", {alist}", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
70 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
71 "search_list_empty": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
72 PKK_MODE_NORMAL: "", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
73 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
74 "search_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
75 PKK_MODE_NORMAL: "\"{text}\"", |
45 | 76 PKK_MODE_ANKI: "{text}", |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
77 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
78 "search_list_sep": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
79 PKK_MODE_NORMAL: ", ", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
80 }, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
81 |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
82 "hyphenation": { |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
83 PKK_MODE_NORMAL: "{indent}hyph \"{text}\"\n", |
45 | 84 PKK_MODE_ANKI: "<{text}>", |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
85 }, |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
86 "no_hyphenation": { |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
87 PKK_MODE_NORMAL: "", |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
88 }, |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
89 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
90 "sense_index": { |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
91 PKK_MODE_NORMAL: "{indent}sense #{index}\n", |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
92 PKK_MODE_ANKI: "[{index}]:\n", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
93 }, |
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 "definition_fmt": { |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
96 PKK_MODE_NORMAL: "{indent}defn \"{text}\"\n", |
46 | 97 PKK_MODE_ANKI: "? \"{text}\"\n", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
98 }, |
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 "example_fmt": { |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
101 PKK_MODE_NORMAL: "{indent}exmp \"{text}\"{geostr}\n", |
46 | 102 PKK_MODE_ANKI: "- \"{text}\"{geostr}\n", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
103 }, |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
104 "example_geo_list": { |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
105 PKK_MODE_NORMAL: " ({alist})", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
106 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
107 "example_geo_list_empty": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
108 PKK_MODE_NORMAL: "", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
109 }, |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
110 "example_geo_list_item": { |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
111 PKK_MODE_NORMAL: "{text} [{tclass}]", |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
112 }, |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
113 "example_geo_list_sep": { |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
114 PKK_MODE_NORMAL: ", ", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
115 }, |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
116 } |
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 |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
119 # Element annotation mappings |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
120 pkk_element_annotation_map = { |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
121 "Fragment" : { |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
122 PKK_MODE_NORMAL: ["<", ">"], |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
123 PKK_MODE_ANKI: ["<", ">"], |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
124 }, |
7 | 125 } |
126 | |
127 | |
28 | 128 # List of words in kks1/ useful for debugging, option -p |
7 | 129 pkk_debug_list = [ |
130 "ahas", | |
131 "ahavakkaine", | |
132 "ahavakala", | |
133 "ahavakoittuo", | |
134 "ahvaliha", | |
9 | 135 "aloilleh", |
18 | 136 "hanjahtoakseh", |
19 | 137 "akkalisto", |
33 | 138 "alto-", |
7 | 139 ] |
140 | |
141 | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
142 pkk_settings = {} |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
143 |
21 | 144 |
0 | 145 ### |
146 ### Misc. helper functions, etc | |
147 ### | |
148 def pkk_cleanup(): | |
149 return 0 | |
150 | |
151 | |
13 | 152 ## 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
|
153 def pkk_print(smsg): |
12 | 154 try: |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
155 if pkk_cfg.normalize: |
12 | 156 sys.stdout.write(unicodedata.normalize("NFC", smsg)) |
157 else: | |
158 sys.stdout.write(smsg) | |
159 | |
160 except (BrokenPipeError, IOError) as e: | |
161 sys.stderr.close() | |
162 | |
0 | 163 |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
164 ## Get indentation string |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
165 def pkk_geti(indent): |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
166 return " " * pkk_cfg.indent * indent |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
167 |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
168 |
13 | 169 ## Print string with indentation |
7 | 170 def pkk_printi(indent, smsg): |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
171 pkk_print(pkk_geti(indent) + smsg) |
7 | 172 |
0 | 173 |
174 ## Fatal error handler | |
175 def pkk_fatal(smsg): | |
176 print(u"ERROR: "+ smsg) | |
177 sys.exit(1) | |
178 | |
179 | |
180 ## Handler for SIGINT signals | |
181 def pkk_signal_handler(signal, frame): | |
182 pkk_cleanup() | |
183 print(u"\nQuitting due to SIGINT / Ctrl+C!") | |
184 sys.exit(1) | |
185 | |
186 | |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
187 class pkk_set_mode(argparse.Action): |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
188 def __call__(self, parser, namespace, value, option_string=None): |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
189 lvalue = value.strip().lower() |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
190 for mode in pkk_modes_list: |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
191 if pkk_modes_list[mode] == lvalue: |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
192 setattr(namespace, self.dest, mode) |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
193 return |
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 pkk_fatal(f"Invalid output mode '{lvalue}'.") |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
196 |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
197 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
198 ## Value handling |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
199 class pkk_set_value(argparse.Action): |
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 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
|
202 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
203 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
|
204 rmatch = re.match(self.rexpr, values) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
205 if rmatch: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
206 rid = rmatch.group(1).lower().replace("-", "_") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
207 rval = rmatch.group(2) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
208 if rid in pkk_mode_defaults: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
209 pkk_settings[rid] = rval |
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}': No such ID '{rid}'.") |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
212 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
213 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
|
214 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
215 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
216 ## Get mode if it exists |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
217 def pkk_test_value(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
218 if mid in pkk_mode_defaults: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
219 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
|
220 mmode = pkk_cfg.mode |
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 mmode = PKK_MODE_NORMAL |
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 if mmode in pkk_mode_defaults[mid]: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
225 return mmode |
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 else: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
229 return None |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
230 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
231 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
232 ## Get default value per mode |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
233 def pkk_get_value(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
234 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
|
235 return pkk_settings[mid] |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
236 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
237 mmode = pkk_test_value(mid) |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
238 if mmode == None: |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
239 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
|
240 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
241 return pkk_mode_defaults[mid][mmode] |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
242 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
243 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
244 def pkk_get_fmt(mid): |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
245 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
|
246 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
247 |
28 | 248 ## Annotate given string with prefix and suffix based on tag |
26 | 249 def pkk_str_annotate(mtag, mstr): |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
250 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
|
251 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
|
252 mmode = pkk_cfg.mode |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
253 else: |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
254 mmode = PKK_MODE_NORMAL |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
255 |
26 | 256 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
|
257 else: |
26 | 258 return mstr |
25
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
259 |
8a6738f67106
Factor string annotation into separate function.
Matti Hamalainen <ccr@tnsp.org>
parents:
24
diff
changeset
|
260 |
13 | 261 ## Clean string by removing tabs and newlines |
10 | 262 def pkk_str_clean(mstr): |
263 return re.sub(r'[\n\r\t]', '', mstr) | |
264 | |
265 | |
28 | 266 ## Format a "Ptr" node as text |
9 | 267 def pkk_ptr_to_text(pnode): |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
268 pfmt = pkk_get_fmt("ptr_fmt") |
30
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
269 return pfmt.format( |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
270 text=("".join(pnode.itertext())).strip(), |
34755af2ea1f
Make Ptr field URL formatting configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
29
diff
changeset
|
271 href=pnode.attrib["{http://www.w3.org/TR/xlink}href"]) |
9 | 272 |
273 | |
13 | 274 ## Get text inside a given node |
16 | 275 def pkk_node_to_text(lnode): |
7 | 276 stmp = "" |
277 for pnode in lnode.iter(): | |
9 | 278 if pnode.tag == "Ptr": |
279 stmp += pkk_ptr_to_text(pnode) | |
280 else: | |
281 if isinstance(pnode.text, str): | |
37
e176fcfc0235
Fix overstripping of concatenated strings.
Matti Hamalainen <ccr@tnsp.org>
parents:
36
diff
changeset
|
282 stmp += pkk_str_annotate(pnode.tag, pkk_str_clean(pnode.text)) |
7 | 283 |
9 | 284 if isinstance(pnode.tail, str): |
10 | 285 stmp += pkk_str_clean(pnode.tail) |
7 | 286 |
287 return stmp.strip() | |
288 | |
289 | |
13 | 290 ## Simple recursive dump starting at given node |
7 | 291 def pkk_dump_recursive(indent, lnode): |
292 if lnode.tag in ["Example"]: | |
16 | 293 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
|
294 pkk_printi(indent, f"{lnode.tag} \"{stmp}\"\n") |
6 | 295 else: |
7 | 296 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
|
297 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
|
298 if textstr != "": |
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
299 textstr = " \""+ textstr +"\"" |
7 | 300 else: |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
301 textstr = "" |
6 | 302 |
303 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
|
304 attrstr = " "+ str(lnode.attrib) |
6 | 305 else: |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
306 attrstr = "" |
6 | 307 |
27
d77ab8a300b1
Use Python 3.6/3.7 format strings where we can.
Matti Hamalainen <ccr@tnsp.org>
parents:
26
diff
changeset
|
308 pkk_printi(indent, f"{lnode.tag}{attrstr}{textstr}\n") |
6 | 309 for qnode in lnode.findall("./*"): |
7 | 310 pkk_dump_recursive(indent + 1, qnode) |
6 | 311 |
312 | |
17 | 313 ## Output item(s) under given node with given format string |
48 | 314 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
|
315 dfmt = pkk_get_fmt(dfmtname) |
48 | 316 ostr = "" |
7 | 317 for qnode in dnode.findall(dsub): |
48 | 318 ostr += dfmt.format( |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
319 text=pkk_node_to_text(qnode), |
48 | 320 indent=pkk_geti(indent)) |
321 return ostr | |
7 | 322 |
323 | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
324 def pkk_get_list_str(dlist, dprefix, dfilter): |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
325 if len(dlist) > 0: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
326 if dfilter: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
327 tfmt = pkk_get_fmt(dprefix + "_list_item") |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
328 tlist = [tfmt.format(text=i) for i in dlist] |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
329 else: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
330 tlist = dlist |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
331 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
332 return pkk_get_fmt(dprefix + "_list").format( |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
333 alist=pkk_get_fmt(dprefix + "_list_sep").join(tlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
334 else: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
335 return pkk_get_fmt(dprefix + "_list_empty") |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
336 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
337 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
338 ## Output a main "Headword" or "Sense" node |
48 | 339 def pkk_get_sense(indent, dnode): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
340 # Definition for this sense |
48 | 341 ostr = pkk_get_subs(indent, dnode, "./Definition", "definition_fmt") |
7 | 342 |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
343 # Examples for this sense |
7 | 344 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
|
345 geolist = [] |
24
8c8e8e4504bb
Remove verbosity option as it is not really used.
Matti Hamalainen <ccr@tnsp.org>
parents:
23
diff
changeset
|
346 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
|
347 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
|
348 text=pkk_node_to_text(qnode), |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
349 tclass=qnode.attrib["class"])) |
7 | 350 |
48 | 351 ostr += pkk_get_fmt("example_fmt").format( |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
352 text=pkk_node_to_text(wnode.find("./Example")), |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
353 geostr=pkk_get_list_str(geolist, "example_geo", False), |
48 | 354 indent=pkk_geti(indent + 1)) |
355 | |
356 return ostr | |
7 | 357 |
358 | |
17 | 359 ## Output one "DictionaryEntry" node |
7 | 360 def pkk_output_node(indent, dnode): |
6 | 361 |
7 | 362 for wnode in dnode.findall("./HeadwordCtn"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
363 # Get head word |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
364 headword = pkk_node_to_text(wnode.find("./Headword")) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
365 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
366 # Collect search forms |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
367 srchlist = [] |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
368 for qnode in wnode.findall("./SearchForm"): |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
369 srchlist.append(pkk_node_to_text(qnode)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
370 |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
371 # Remove dupe if headword is also in srchlist |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
372 if headword in srchlist: |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
373 srchlist.remove(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 # Remove other duplicates and sort |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
376 srchlist = list(set(srchlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
377 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
|
378 |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
379 # Get hyphenation note, if any |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
380 hnode = wnode.find("./Hyphenation") |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
381 if hnode != None: |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
382 hyphenation = pkk_get_fmt("hyphenation").format( |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
383 text=pkk_node_to_text(hnode), |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
384 indent=pkk_geti(indent + 1)) |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
385 else: |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
386 hyphenation = pkk_get_fmt("no_hyphenation").format( |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
387 indent=pkk_geti(indent + 1)) |
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
388 |
18 | 389 # Create list with grammatical attributes (noun, verb, etc.) |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
390 attrlist = [] |
18 | 391 for pnode in wnode.findall("./PartOfSpeechCtn/PartOfSpeech"): |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
392 attrlist.append(pnode.attrib["freeValue"]) |
18 | 393 |
394 for pnode in wnode.findall("./GrammaticalNote"): | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
395 attrlist.append(pkk_node_to_text(pnode)) |
18 | 396 |
397 # Remove duplicates and sort the list | |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
398 attrlist = list(set(attrlist)) |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
399 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
|
400 |
48 | 401 # Get main "sense" |
402 msense = pkk_get_sense(indent + 1, wnode) | |
403 | |
404 # Print any other "senses" | |
405 index = 1 | |
406 osenses = "" | |
407 for znode in dnode.findall("./SenseGrp"): | |
408 osenses += pkk_get_fmt("sense_index").format( | |
409 index=index, | |
410 indent=pkk_geti(indent + 1)) | |
411 osenses += pkk_get_sense(indent + 2, znode) | |
412 index += 1 | |
413 | |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
414 # Print the headword and attributes if any |
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
415 pkk_print(pkk_get_fmt("word_fmt").format( |
43
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
416 word=headword, |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
417 attr=pkk_get_list_str(attrlist, "word_attr", True), |
8ed576574712
More improvements to output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
42
diff
changeset
|
418 search=pkk_get_list_str(srchlist, "search", True), |
44
d7b4b2fb0214
Add support for hyphenation data.
Matti Hamalainen <ccr@tnsp.org>
parents:
43
diff
changeset
|
419 hyphenation=hyphenation, |
48 | 420 main_sense=msense, |
421 other_senses=osenses, | |
40
bc8d8ef4a248
More work on output flexibility.
Matti Hamalainen <ccr@tnsp.org>
parents:
39
diff
changeset
|
422 indent=pkk_geti(indent))) |
18 | 423 |
6 | 424 |
0 | 425 ### |
426 ### Main program starts | |
427 ### | |
428 signal.signal(signal.SIGINT, pkk_signal_handler) | |
429 | |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
430 optparser = argparse.ArgumentParser( |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
431 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
|
432 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
|
433 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
|
434 ) |
0 | 435 |
22 | 436 optparser.add_argument("filenames", |
437 type=str, action="extend", nargs="*", | |
438 metavar="filename", | |
439 help="XML filename(s)") | |
0 | 440 |
31
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
441 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
|
442 dest="show_help", |
4cbefe4c6f53
Improve help, list the default Ptr format strings for each mode.
Matti Hamalainen <ccr@tnsp.org>
parents:
30
diff
changeset
|
443 action="store_true", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
444 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
|
445 |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
446 optparser.add_argument("-m", "--mode", |
22 | 447 dest="mode", |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
448 action=pkk_set_mode, |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
449 default=PKK_MODE_NORMAL, |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
450 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
|
451 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
452 optparser.add_argument("-s", "--set", |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
453 action=pkk_set_value, |
38
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
454 metavar="ID=STR", |
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
455 help='set format string (see below)') |
32
cfecd039506e
Make word attribute separator string configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
31
diff
changeset
|
456 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
457 optparser.add_argument("-n", "--normalize", |
22 | 458 dest="normalize", |
29
f91ef7d7615b
Use store_true feature of argparse.
Matti Hamalainen <ccr@tnsp.org>
parents:
28
diff
changeset
|
459 action="store_true", |
22 | 460 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
|
461 |
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
462 optparser.add_argument("-a", "--annotate", |
22 | 463 dest="annotate", |
29
f91ef7d7615b
Use store_true feature of argparse.
Matti Hamalainen <ccr@tnsp.org>
parents:
28
diff
changeset
|
464 action="store_true", |
22 | 465 help="annotate strings") |
0 | 466 |
22 | 467 optparser.add_argument("-i", "--indent", |
468 dest="indent", | |
469 type=int, choices=range(0, 32), default=4, | |
38
0e586b4ab62c
Cosmetic adjustments to help.
Matti Hamalainen <ccr@tnsp.org>
parents:
37
diff
changeset
|
470 metavar="N", |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
471 help='set indentation level (default: %(default)s)') |
0 | 472 |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
473 optparser.add_argument("-p", "--debug", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
474 dest="debug", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
475 action="store_true", |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
476 help=argparse.SUPPRESS) |
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
477 |
0 | 478 |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
479 ### Parse arguments |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
480 pkk_cfg = optparser.parse_args() |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
481 |
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
482 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
483 ### 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
|
484 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
|
485 optparser.print_help() |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
486 print(u"\nAvailable output modes:") |
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
487 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
|
488 |
41 | 489 if pkk_cfg.mode not in [PKK_MODE_NORMAL, PKK_MODE_ANKI]: |
490 pkk_cfg.mode = PKK_MODE_NORMAL | |
491 | |
492 print(u"\nAvailable format strings and values (mode '{}'):".format( | |
493 pkk_modes_list[pkk_cfg.mode])) | |
494 | |
39
f53ea742b57f
Refactor output mode handling.
Matti Hamalainen <ccr@tnsp.org>
parents:
38
diff
changeset
|
495 for mid in pkk_mode_defaults: |
41 | 496 stmp = pkk_get_value(mid).replace("\\", "\\\\").replace("\n", "\\n") |
42 | 497 print(u" {:22s} : '{}'".format(mid, stmp)) |
41 | 498 |
0 | 499 sys.exit(0) |
500 | |
501 | |
6 | 502 ### 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
|
503 for filename in pkk_cfg.filenames: |
0 | 504 # Parse XML file into element tree |
505 try: | |
506 uxml = xmlET.parse(filename) | |
507 except Exception as e: | |
508 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e))) | |
509 | |
510 # Dump output | |
511 try: | |
512 xroot = uxml.getroot() | |
513 for dnode in xroot.findall("./DictionaryEntry"): | |
7 | 514 |
20
f274504eafd0
Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
19
diff
changeset
|
515 if pkk_cfg.debug and dnode.attrib["identifier"] not in pkk_debug_list: |
7 | 516 continue |
517 | |
35
5aafa87dbec2
Some Anki-related stuff, not finalized and thus hidden.
Matti Hamalainen <ccr@tnsp.org>
parents:
34
diff
changeset
|
518 if pkk_cfg.mode in [PKK_MODE_NORMAL, PKK_MODE_ANKI]: |
19 | 519 try: |
520 pkk_output_node(0, dnode) | |
521 except Exception as e: | |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
522 print("") |
19 | 523 pkk_dump_recursive(0, dnode) |
524 print(str(e)) | |
525 sys.exit(0) | |
21 | 526 elif pkk_cfg.mode == PKK_MODE_DUMP: |
7 | 527 pkk_dump_recursive(0, dnode) |
36
4c8aafff8c5f
Refactor output handling to be (mostly) configurable.
Matti Hamalainen <ccr@tnsp.org>
parents:
35
diff
changeset
|
528 print("") |
21 | 529 elif pkk_cfg.mode == PKK_MODE_XML: |
47 | 530 pkk_print(xmlET.tostring(dnode, encoding="utf8").decode("utf8") + "\n\n") |
0 | 531 else: |
10 | 532 pkk_fatal("Invalid operation mode?") |
7 | 533 |
0 | 534 except (BrokenPipeError, IOError) as e: |
535 sys.stderr.close() | |
536 sys.exit(1) | |
537 | |
538 pkk_cleanup() | |
539 sys.exit(0) |