annotate lxmldump.py @ 22:76856fc4e5fa

Clean up argument handling.
author Matti Hamalainen <ccr@tnsp.org>
date Sun, 16 May 2021 16:10:44 +0300
parents 7ef08e05a5bf
children 3ef1c5463b8f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
1 #!/usr/bin/python3 -B
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
2 # coding=utf-8
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
3 ###
4
Matti Hamalainen <ccr@tnsp.org>
parents: 1
diff changeset
4 ### lxmldump - Dump ISO/FDIS 1951 XML file data
Matti Hamalainen <ccr@tnsp.org>
parents: 1
diff changeset
5 ### Programmed and designed by Matti 'ccr' Hämäläinen <ccr@tnsp.org>
Matti Hamalainen <ccr@tnsp.org>
parents: 1
diff changeset
6 ### (C) Copyright 2021 Tecnic Software productions (TNSP)
Matti Hamalainen <ccr@tnsp.org>
parents: 1
diff changeset
7 ###
Matti Hamalainen <ccr@tnsp.org>
parents: 1
diff changeset
8 ### Python 3.7+ required!
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
9 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
10 import sys
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
11 import signal
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
12 import re
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
13 from pathlib import Path
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
14 import xml.etree.ElementTree as xmlET
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
15 import unicodedata
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
16 import argparse
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
17
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
18 assert sys.version_info >= (3, 7)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
19
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
20
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
21 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
22 ### Default settings
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
23 ###
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
24 pkk_str_fmap = {
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
25 "Fragment" : ["<", ">"],
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
26 }
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
27
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
28
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
29 pkk_debug_list = [
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
30 "ahas",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
31 "ahavakkaine",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
32 "ahavakala",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
33 "ahavakoittuo",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
34 "ahvaliha",
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
35 "aloilleh",
18
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
36 "hanjahtoakseh",
19
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
37 "akkalisto",
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
38 ]
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
39
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
40
21
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
41 # Operation modes
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
42 PKK_MODE_NORMAL = 0
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
43 PKK_MODE_DUMP = 1
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
44 PKK_MODE_XML = 2
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
45
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
46
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
47 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
48 ### Misc. helper functions, etc
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
49 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50 def pkk_cleanup():
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 return 0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
52
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
54 ## 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
55 def pkk_print(smsg):
12
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
56 try:
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
57 if pkk_cfg.normalize:
12
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
58 sys.stdout.write(unicodedata.normalize("NFC", smsg))
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
59 else:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
60 sys.stdout.write(smsg)
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
61
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
62 except (BrokenPipeError, IOError) as e:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
63 sys.stderr.close()
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
64
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
65
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
66 ## Print string with indentation
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
67 def pkk_printi(indent, smsg):
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
68 pkk_print((" " * pkk_cfg.indent * indent) + smsg)
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
69
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
71 ## Check value against current verbosity level
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
72 def pkk_verbosity(lvl):
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
73 return pkk_cfg.verbosity >= lvl
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
74
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
75
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
76 ## Fatal error handler
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
77 def pkk_fatal(smsg):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
78 print(u"ERROR: "+ smsg)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
80
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
81
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82 ## Handler for SIGINT signals
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
83 def pkk_signal_handler(signal, frame):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
84 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
85 print(u"\nQuitting due to SIGINT / Ctrl+C!")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
86 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
87
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
88
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
89 ## Clean string by removing tabs and newlines
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
90 def pkk_str_clean(mstr):
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
91 return re.sub(r'[\n\r\t]', '', mstr)
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
92
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
93
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
94 ## Format "Ptr" node as text
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
95 def pkk_ptr_to_text(pnode):
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
96 return "<PTR:{}>{}</PTR>".format(
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
97 pnode.attrib["{http://www.w3.org/TR/xlink}href"],
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
98 ("".join(pnode.itertext())).strip())
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
99
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
100
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
101 ## Get text inside a given node
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
102 def pkk_node_to_text(lnode):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
103 stmp = ""
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
104 for pnode in lnode.iter():
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
105 if pnode.tag == "Ptr":
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
106 stmp += pkk_ptr_to_text(pnode)
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
107 else:
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
108 if isinstance(pnode.text, str):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
109 ptext = pkk_str_clean(pnode.text).strip()
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
110 if pkk_cfg.annotate and isinstance(pnode.tag, str) and pnode.tag in pkk_str_fmap:
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
111 stmp += pkk_str_fmap[pnode.tag][0] + ptext + pkk_str_fmap[pnode.tag][1]
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
112 else:
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
113 stmp += ptext
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
114
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
115 if isinstance(pnode.tail, str):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
116 stmp += pkk_str_clean(pnode.tail)
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
117
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
118 return stmp.strip()
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
119
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
120
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
121 ## Simple recursive dump starting at given node
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
122 def pkk_dump_recursive(indent, lnode):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
123 if lnode.tag in ["Example"]:
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
124 stmp = pkk_node_to_text(lnode)
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
125 pkk_printi(indent, "{} \"{}\"\n".format(lnode.tag, stmp))
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
126 else:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
127 if isinstance(lnode.text, str):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
128 stmp = pkk_str_clean(lnode.text).strip()
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
129 if stmp != "":
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
130 stmp = " \""+ stmp +"\""
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
131 else:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
132 stmp = ""
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
133
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
134 if len(lnode.attrib) > 0:
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
135 atmp = " "+ str(lnode.attrib)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
136 else:
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
137 atmp = ""
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
138
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
139 pkk_printi(indent, "{}{}{}\n".format(lnode.tag, atmp, stmp))
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
140 for qnode in lnode.findall("./*"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
141 pkk_dump_recursive(indent + 1, qnode)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
142
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
143
17
6fa24c711f86 Add some comments.
Matti Hamalainen <ccr@tnsp.org>
parents: 16
diff changeset
144 ## Output item(s) under given node with given format string
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
145 def pkk_output_subs_fmt(indent, dnode, dsub, dname, dfmt):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
146 for qnode in dnode.findall(dsub):
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
147 pkk_printi(indent, dfmt.format(dname, pkk_node_to_text(qnode)))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
148
17
6fa24c711f86 Add some comments.
Matti Hamalainen <ccr@tnsp.org>
parents: 16
diff changeset
149 ## Output item(s) under given node with a prefixed name string
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
150 def pkk_output_subs_prefix(indent, dnode, dsub, dname):
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
151 pkk_output_subs_fmt(indent, dnode, dsub, dname, "{0} \"{1}\"\n")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
152
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
153
17
6fa24c711f86 Add some comments.
Matti Hamalainen <ccr@tnsp.org>
parents: 16
diff changeset
154 ## Output a main "Headword" or "Sense" node under it
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
155 def pkk_output_sense(indent, dnode):
17
6fa24c711f86 Add some comments.
Matti Hamalainen <ccr@tnsp.org>
parents: 16
diff changeset
156 # Search form and definition
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
157 pkk_output_subs_prefix(indent, dnode, "./SearchForm", "srch")
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
158 pkk_output_subs_prefix(indent, dnode, "./Definition", "defn")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
159
17
6fa24c711f86 Add some comments.
Matti Hamalainen <ccr@tnsp.org>
parents: 16
diff changeset
160 # Examples
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
161 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"):
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
162 sstr = pkk_node_to_text(wnode.find("./Example"))
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
163 lstr = ""
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
164
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
165 if pkk_verbosity(1):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
166 ltmp = []
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
167 for qnode in wnode.findall("./FreeTopic[@type='levikki']/GeographicalUsage"):
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
168 ltmp.append("{} [{}]".format(pkk_node_to_text(qnode), qnode.attrib["class"]))
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
169
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
170 if len(ltmp) > 0:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
171 lstr = " ({})".format(", ".join(ltmp))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
172
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
173 pkk_printi(indent + 1, "{} \"{}\"{}\n".format("exmp", sstr, lstr))
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
174
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
175
17
6fa24c711f86 Add some comments.
Matti Hamalainen <ccr@tnsp.org>
parents: 16
diff changeset
176 ## Output one "DictionaryEntry" node
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
177 def pkk_output_node(indent, dnode):
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
178
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
179 for wnode in dnode.findall("./HeadwordCtn"):
18
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
180 # Create list with grammatical attributes (noun, verb, etc.)
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
181 tmpl = []
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
182 for pnode in wnode.findall("./PartOfSpeechCtn/PartOfSpeech"):
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
183 tmpl.append(pnode.attrib["freeValue"])
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
184
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
185 for pnode in wnode.findall("./GrammaticalNote"):
19
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
186 tmpl.append(pkk_node_to_text(pnode))
18
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
187
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
188 # Remove duplicates and sort the list
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
189 tmpl = list(set(tmpl))
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
190 tmpl.sort(reverse=False, key=lambda attr: (attr, len(attr)))
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
191
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
192 # Print the headword and attributes if any
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
193 pkk_output_subs_fmt(indent, wnode, "./Headword", "", "\"{1}\"")
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
194 if len(tmpl) > 0:
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
195 pkk_print(" ({})\n".format(" ; ".join(tmpl)))
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
196 else:
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
197 pkk_print("\n")
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
198
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
199 # Print main "sense"
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
200 pkk_output_sense(indent + 1, wnode)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
201
18
ff959de0f6c8 Add grammatical attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 17
diff changeset
202 # Print any other "senses"
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
203 index = 1
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
204 for wnode in dnode.findall("./SenseGrp"):
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
205 pkk_printi(indent + 1, "sense #{}\n".format(index))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
206 pkk_output_sense(indent + 2, wnode)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
207 index += 1
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
208
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
209
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
211 ### Main program starts
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
212 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
213 signal.signal(signal.SIGINT, pkk_signal_handler)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
214
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
215 optparser = argparse.ArgumentParser(
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
216 description="lxmldump - Dump ISO/FDIS 1951 XML file data",
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
217 usage="%(prog)s [options] <input xml file(s)>",
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
218 epilog="\n\n"
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
219 )
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
220
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
221 optparser.add_argument("filenames",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
222 type=str, action="extend", nargs="*",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
223 metavar="filename",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
224 help="XML filename(s)")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
225
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
226 optparser.add_argument("-d", "--dump",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
227 dest="mode",
21
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
228 action="store_const", const=PKK_MODE_DUMP, default=PKK_MODE_NORMAL,
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
229 help="output as simple dump")
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
230
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
231 optparser.add_argument("-x", "--xml",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
232 dest="mode",
21
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
233 action="store_const", const=PKK_MODE_XML,
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
234 help="output as XML")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
235
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
236 optparser.add_argument("-n", "--normalize",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
237 dest="normalize",
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
238 action="store_const", const=True, default=False,
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
239 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
240
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
241 optparser.add_argument("-a", "--annotate",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
242 dest="annotate",
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
243 action="store_const", const=True, default=False,
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
244 help="annotate strings")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
245
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
246 optparser.add_argument("-v", "--verbosity",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
247 dest="verbosity",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
248 type=int, choices=range(0, 4), default=3,
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
249 metavar="n",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
250 help='set verbosity level (0-3, default: %(default)s)')
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
251
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
252 optparser.add_argument("-p", "--debug",
22
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
253 dest="debug", action="store_const", const=True, default=False,
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
254 help=argparse.SUPPRESS)
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
255
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
256 optparser.add_argument("-i", "--indent",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
257 dest="indent",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
258 type=int, choices=range(0, 32), default=4,
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
259 metavar="n",
76856fc4e5fa Clean up argument handling.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
260 help='indent output by <n> characters (default: %(default)s)')
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
261
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
262
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
263 ### Show help if needed
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
264 pkk_cfg = optparser.parse_args()
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
265 if len(pkk_cfg.filenames) == 0:
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
266 optparser.print_help()
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
267 sys.exit(0)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
268
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
269
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
270 ### 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
271 for filename in pkk_cfg.filenames:
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
272 # Parse XML file into element tree
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
273 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
274 uxml = xmlET.parse(filename)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
275 except Exception as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
276 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
277
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
278 # Dump output
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
279 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
280 xroot = uxml.getroot()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
281 for dnode in xroot.findall("./DictionaryEntry"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
282
20
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 19
diff changeset
283 if pkk_cfg.debug and dnode.attrib["identifier"] not in pkk_debug_list:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
284 continue
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
285
21
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
286 if pkk_cfg.mode == PKK_MODE_NORMAL:
19
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
287 try:
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
288 pkk_output_node(0, dnode)
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
289 except Exception as e:
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
290 pkk_dump_recursive(0, dnode)
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
291 print(str(e))
7c6eb57798bd Ja niin.
Matti Hamalainen <ccr@tnsp.org>
parents: 18
diff changeset
292 sys.exit(0)
21
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
293 elif pkk_cfg.mode == PKK_MODE_DUMP:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
294 pkk_dump_recursive(0, dnode)
21
7ef08e05a5bf Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 20
diff changeset
295 elif pkk_cfg.mode == PKK_MODE_XML:
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
296 pkk_print(str(xmlET.tostring(dnode, encoding="utf8")) + "\n")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
297 else:
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
298 pkk_fatal("Invalid operation mode?")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
299
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
300 print("\n")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
301
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
302 except (BrokenPipeError, IOError) as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
303 sys.stderr.close()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
304 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
305
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
306 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
307 sys.exit(0)