annotate lxmldump.py @ 13:3bd772fd6a50

Cleanups.
author Matti Hamalainen <ccr@tnsp.org>
date Tue, 11 May 2021 12:49:15 +0300
parents d50e71642be7
children 7498bda8b4a2
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
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
26 "annotate": False,
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
27 "mode": 0,
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
28 "normalize": False,
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
29 "xml": False,
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
30
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
31 "debug": False,
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
32 }
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
33
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
34
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
35 pkk_str_fmap = {
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
36 "Fragment" : ["<", ">"],
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
37 }
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 pkk_debug_list = [
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
41 "ahas",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
42 "ahavakkaine",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
43 "ahavakala",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
44 "ahavakoittuo",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
45 "ahvaliha",
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
46 "aloilleh",
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
47 ]
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
48
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
49
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 ### Misc. helper functions, etc
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
52 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53 def pkk_cleanup():
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
54 return 0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
55
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
56
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
57 ## 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
58 def pkk_print(smsg):
12
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
59 try:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
60 if pkk_cfg["normalize"]:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
61 sys.stdout.write(unicodedata.normalize("NFC", smsg))
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
62 else:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
63 sys.stdout.write(smsg)
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
64
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
65 except (BrokenPipeError, IOError) as e:
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
66 sys.stderr.close()
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
67
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
68
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
69 ## Print string with indentation
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
70 def pkk_printi(indent, smsg):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
71 pkk_print((" " * indent) + smsg)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
72
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
73
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
74 ## Check value against current verbosity level
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
75 def pkk_verbosity(lvl):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
76 return pkk_cfg["verbosity"] >= lvl
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
77
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
78
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 ## Fatal error handler
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
80 def pkk_fatal(smsg):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
81 print(u"ERROR: "+ smsg)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
83
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
84
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
85 ## Handler for SIGINT signals
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
86 def pkk_signal_handler(signal, frame):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
87 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
88 print(u"\nQuitting due to SIGINT / Ctrl+C!")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
89 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
90
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
91
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
92 ## Clean string by removing tabs and newlines
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
93 def pkk_str_clean(mstr):
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
94 return re.sub(r'[\n\r\t]', '', mstr)
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
95
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
96
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
97 ## Format "Ptr" node as text
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
98 def pkk_ptr_to_text(pnode):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
99 return "PTR: <{}>{}</>".format(
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
100 pnode.attrib["{http://www.w3.org/TR/xlink}href"],
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
101 ("".join(pnode.itertext())).strip())
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
102
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
103
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
104 ## Get text inside a given node
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
105 def pkk_get_text(lnode):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
106 stmp = ""
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
107 for pnode in lnode.iter():
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
108 if pnode.tag == "Ptr":
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
109 stmp += pkk_ptr_to_text(pnode)
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
110 else:
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
111 if isinstance(pnode.text, str):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
112 ptext = pkk_str_clean(pnode.text).strip()
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
113 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
114 stmp += pkk_str_fmap[pnode.tag][0] + ptext + pkk_str_fmap[pnode.tag][1]
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
115 else:
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
116 stmp += ptext
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
117
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
118 if isinstance(pnode.tail, str):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
119 stmp += pkk_str_clean(pnode.tail)
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
120
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
121 return stmp.strip()
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
122
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
123
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
124 ## Simple recursive dump starting at given node
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
125 def pkk_dump_recursive(indent, lnode):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
126 if lnode.tag in ["Example"]:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
127 stmp = pkk_get_text(lnode)
9
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
128 pkk_printi(indent, "{} \"{}\"\n".format(lnode.tag, stmp))
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
129 else:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
130 if isinstance(lnode.text, str):
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
131 stmp = pkk_str_clean(lnode.text).strip()
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
132 if stmp != "":
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
133 stmp = " \""+ stmp +"\""
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
134 else:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
135 stmp = ""
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
136
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
137 if len(lnode.attrib) > 0:
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
138 atmp = " "+ str(lnode.attrib)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
139 else:
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
140 atmp = ""
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
141
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
142 pkk_printi(indent, "{}{}{}\n".format(lnode.tag, atmp, stmp))
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
143 for qnode in lnode.findall("./*"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
144 pkk_dump_recursive(indent + 1, qnode)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
145
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
146
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
147 ## Output item under given node
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
148 def pkk_output_subs_fmt(indent, dnode, dsub, dname, dfmt):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
149 for qnode in dnode.findall(dsub):
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
150 pkk_printi(indent, dfmt.format(dname, pkk_get_text(qnode)))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
151
12
Matti Hamalainen <ccr@tnsp.org>
parents: 11
diff changeset
152
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
153 def pkk_output_subs_prefix(indent, dnode, dsub, dname):
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
154 pkk_output_subs_fmt(indent, dnode, dsub, dname, "{0} \"{1}\"\n")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
155
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
156
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
157 def pkk_output_sense(indent, dnode):
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
158 pkk_output_subs_prefix(indent, dnode, "./SearchForm", "srch")
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
159 pkk_output_subs_prefix(indent, dnode, "./Definition", "defn")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
160
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
161 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
162 sstr = pkk_get_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"):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
168 ltmp.append("{} [{}]".format(pkk_get_text(qnode), qnode.attrib["class"]))
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
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
176 def pkk_output_node(indent, dnode):
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
177
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
178 for wnode in dnode.findall("./HeadwordCtn"):
13
3bd772fd6a50 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 12
diff changeset
179 pkk_output_subs_fmt(indent, wnode, "./Headword", "", "\"{1}\":\n")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
180 pkk_output_sense(indent + 1, wnode)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
181
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
182 index = 1
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
183 for wnode in dnode.findall("./SenseGrp"):
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
184 pkk_printi(indent + 1, "sense #{}\n".format(index))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
185 pkk_output_sense(indent + 2, wnode)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
186 index += 1
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
187
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
188
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
189 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
190 ### Main program starts
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
191 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
192 signal.signal(signal.SIGINT, pkk_signal_handler)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
193
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
194
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
195 ### Check if we have arguments
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
196 pkk_show_help = False
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
197 pkk_filenames = []
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
198 argc = 1
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
199 while argc < len(sys.argv):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
200 arg = sys.argv[argc]
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
201
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
202 needs_param = False
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
203 if argc + 1 < len(sys.argv):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
204 param = sys.argv[argc + 1]
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
205 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
206 param = None
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
207
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
208 # Check for option type arg
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
209 if arg[0:1] == "-":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 oarg = arg
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
211 arg = arg.lstrip("-")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
212
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
213 if arg == "help" or arg == "h":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
214 pkk_show_help = True
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
215 elif arg == "dump" or arg == "d":
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
216 pkk_cfg["mode"] = 1
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
217 elif arg == "xml" or arg == "x":
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
218 pkk_cfg["mode"] = 2
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
219 elif arg == "normalize" or arg == "n":
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
220 pkk_cfg["normalize"] = True
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
221 elif arg == "annotate" or arg == "a":
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
222 pkk_cfg["annotate"] = True
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
223 elif arg == "p":
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
224 pkk_cfg["debug"] = True
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
225 elif arg == "verbosity" or arg == "v":
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
226 needs_param = True
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
227 pkk_cfg["verbosity"] = param
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
228 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
229 pkk_fatal(u"Invalid option argument '{0}'.".format(oarg))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
230
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
231 if needs_param and param == None:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
232 pkk_fatal(u"Option '{0}' requires an argument.".format(oarg))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
233 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
234 # Non-option argument
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
235 pkk_filenames.append(arg)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
236
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
237 if needs_param:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
238 argc += 2
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
239 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
240 argc += 1
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
241
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
242
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
243 ### Show help if requested
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
244 if pkk_show_help or len(pkk_filenames) == 0:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
245 print(u"lxmldump - Dump ISO/FDIS 1951 XML file data")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
246 print(u"Usage: {0} <options> <input xml file(s)>".
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
247 format(str(Path(sys.argv[0]).name)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
248 print(u"")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
249 print(u" --help Show this help")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
250 print(u" -d, --dump Dump mode")
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
251 print(u" -n, --normalize Output NFC normalized Unicode")
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
252 print(u" -a, --annotate Annotate strings")
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
253 print(u" -v, --verbosity <n> Set verbosity level (0 - 3)")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
254 print(u"")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
255 sys.exit(0)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
256
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
257
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
258 ### Validate settings
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
259 try:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
260 pkk_cfg["verbosity"] = int(pkk_cfg["verbosity"])
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
261 except Exception as e:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
262 pkk_fatal(u"Verbosity level is not a valid integer.")
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
263 if pkk_cfg["verbosity"] < 0 or pkk_cfg["verbosity"] > 3:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
264 pkk_fatal(u"Invalid verbosity level value {0}.".format(pkk_cfg["verbosity"]))
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
265
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
266
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
267 ### Handle each input file
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
268 for filename in pkk_filenames:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
269 # Parse XML file into element tree
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
270 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
271 uxml = xmlET.parse(filename)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
272 except Exception as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
273 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
274
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
275 # Dump output
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
276 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
277 xroot = uxml.getroot()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
278 for dnode in xroot.findall("./DictionaryEntry"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
279
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
280 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
281 continue
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
282
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
283 if pkk_cfg["mode"] == 0:
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
284 pkk_output_node(0, dnode)
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
285 elif pkk_cfg["mode"] == 1:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
286 pkk_dump_recursive(0, dnode)
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
287 elif pkk_cfg["mode"] == 2:
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
288 pkk_print(str(xmlET.tostring(dnode, encoding="utf8")) + "\n")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
289 else:
10
013f0cd9e5b3 More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 9
diff changeset
290 pkk_fatal("Invalid operation mode?")
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
291
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
292 print("\n")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
293
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
294 except (BrokenPipeError, IOError) as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
295 sys.stderr.close()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
296 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
297
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
298 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
299 sys.exit(0)