annotate lxmldump.py @ 16:285b0820d2c6

Cleanup.
author Matti Hamalainen <ccr@tnsp.org>
date Tue, 11 May 2021 14:47:01 +0300
parents 2f86537ff1f7
children 6fa24c711f86
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
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
16
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
17 assert sys.version_info >= (3, 7)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
18
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 ### Default settings
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
22 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
23 pkk_cfg = {
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
24 "verbosity": 3,
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
25 "annotate": False,
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
26 "mode": 0,
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
27 "normalize": False,
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
28 "debug": False,
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
29 }
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
30
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
31
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
32 pkk_str_fmap = {
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
33 "Fragment" : ["<", ">"],
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
34 }
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
35
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
36
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
37 pkk_debug_list = [
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
38 "ahas",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
39 "ahavakkaine",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
40 "ahavakala",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
41 "ahavakoittuo",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
42 "ahvaliha",
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
43 "aloilleh",
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
44 ]
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
45
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
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:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
57 if pkk_cfg["normalize"]:
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):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
68 pkk_print((" " * indent) + smsg)
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):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
73 return pkk_cfg["verbosity"] >= lvl
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()
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
110 if pkk_cfg["annotate"] and isinstance(pnode.tag, str) and pnode.tag in pkk_str_fmap:
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
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
144 ## Output item under given node
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
12
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
149
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
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
154 def pkk_output_sense(indent, dnode):
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
155 pkk_output_subs_prefix(indent, dnode, "./SearchForm", "srch")
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
156 pkk_output_subs_prefix(indent, dnode, "./Definition", "defn")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
157
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
158 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"):
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
159 sstr = pkk_node_to_text(wnode.find("./Example"))
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
160 lstr = ""
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
161
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
162 if pkk_verbosity(1):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
163 ltmp = []
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
164 for qnode in wnode.findall("./FreeTopic[@type='levikki']/GeographicalUsage"):
16
285b0820d2c6 Cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 15
diff changeset
165 ltmp.append("{} [{}]".format(pkk_node_to_text(qnode), qnode.attrib["class"]))
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
166
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
167 if len(ltmp) > 0:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
168 lstr = " ({})".format(", ".join(ltmp))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
169
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
170 pkk_printi(indent + 1, "{} \"{}\"{}\n".format("exmp", sstr, lstr))
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
171
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 def pkk_output_node(indent, dnode):
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
174
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
175 for wnode in dnode.findall("./HeadwordCtn"):
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
176 pkk_output_subs_fmt(indent, wnode, "./Headword", "", "\"{1}\":\n")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
177 pkk_output_sense(indent + 1, wnode)
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 index = 1
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
180 for wnode in dnode.findall("./SenseGrp"):
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
181 pkk_printi(indent + 1, "sense #{}\n".format(index))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
182 pkk_output_sense(indent + 2, wnode)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
183 index += 1
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
184
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
185
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
186 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
187 ### Main program starts
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
188 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
189 signal.signal(signal.SIGINT, pkk_signal_handler)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
190
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
191
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
192 ### Check if we have arguments
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
193 pkk_show_help = False
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
194 pkk_filenames = []
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
195 argc = 1
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
196 while argc < len(sys.argv):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
197 arg = sys.argv[argc]
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
198
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
199 needs_param = False
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
200 if argc + 1 < len(sys.argv):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
201 param = sys.argv[argc + 1]
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
202 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
203 param = None
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
204
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
205 # Check for option type arg
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
206 if arg[0:1] == "-":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
207 oarg = arg
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
208 arg = arg.lstrip("-")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
209
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 if arg == "help" or arg == "h":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
211 pkk_show_help = True
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
212 elif arg == "dump" or arg == "d":
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
213 pkk_cfg["mode"] = 1
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
214 elif arg == "xml" or arg == "x":
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
215 pkk_cfg["mode"] = 2
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
216 elif arg == "normalize" or arg == "n":
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
217 pkk_cfg["normalize"] = True
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
218 elif arg == "annotate" or arg == "a":
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
219 pkk_cfg["annotate"] = True
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
220 elif arg == "p":
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
221 pkk_cfg["debug"] = True
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
222 elif arg == "verbosity" or arg == "v":
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
223 needs_param = True
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
224 pkk_cfg["verbosity"] = param
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
225 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
226 pkk_fatal(u"Invalid option argument '{0}'.".format(oarg))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
227
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
228 if needs_param and param == None:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
229 pkk_fatal(u"Option '{0}' requires an argument.".format(oarg))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
230 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
231 # Non-option argument
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
232 pkk_filenames.append(arg)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
233
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
234 if needs_param:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
235 argc += 2
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
236 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
237 argc += 1
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
238
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
239
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
240 ### Show help if requested
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
241 if pkk_show_help or len(pkk_filenames) == 0:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
242 print(u"lxmldump - Dump ISO/FDIS 1951 XML file data")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
243 print(u"Usage: {0} <options> <input xml file(s)>".
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
244 format(str(Path(sys.argv[0]).name)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
245 print(u"")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
246 print(u" --help Show this help")
15
2f86537ff1f7 Improve help.
Matti Hamalainen <ccr@tnsp.org>
parents: 14
diff changeset
247 print(u" -d, --dump Output as simple dump")
2f86537ff1f7 Improve help.
Matti Hamalainen <ccr@tnsp.org>
parents: 14
diff changeset
248 print(u" -x, --xml Output as XML")
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
249 print(u" -n, --normalize Output NFC normalized Unicode")
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
250 print(u" -a, --annotate Annotate strings")
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
251 print(u" -v, --verbosity <n> Set verbosity level (0 - 3)")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
252 print(u"")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
253 sys.exit(0)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
254
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
255
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
256 ### Validate settings
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
257 try:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
258 pkk_cfg["verbosity"] = int(pkk_cfg["verbosity"])
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
259 except Exception as e:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
260 pkk_fatal(u"Verbosity level is not a valid integer.")
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
261 if pkk_cfg["verbosity"] < 0 or pkk_cfg["verbosity"] > 3:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
262 pkk_fatal(u"Invalid verbosity level value {0}.".format(pkk_cfg["verbosity"]))
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
263
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
264
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
265 ### Handle each input file
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
266 for filename in pkk_filenames:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
267 # Parse XML file into element tree
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
268 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
269 uxml = xmlET.parse(filename)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
270 except Exception as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
271 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
272
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
273 # Dump output
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
274 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
275 xroot = uxml.getroot()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
276 for dnode in xroot.findall("./DictionaryEntry"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
277
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
278 if pkk_cfg["debug"] and dnode.attrib["identifier"] not in pkk_debug_list:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
279 continue
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
280
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
281 if pkk_cfg["mode"] == 0:
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
282 pkk_output_node(0, dnode)
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
283 elif pkk_cfg["mode"] == 1:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
284 pkk_dump_recursive(0, dnode)
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
285 elif pkk_cfg["mode"] == 2:
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
286 pkk_print(str(xmlET.tostring(dnode, encoding="utf8")) + "\n")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
287 else:
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
288 pkk_fatal("Invalid operation mode?")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
289
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
290 print("\n")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
291
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
292 except (BrokenPipeError, IOError) as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
293 sys.stderr.close()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
294 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
295
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
296 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
297 sys.exit(0)