annotate lxmldump.py @ 8:ce07bb2a247b

More work.
author Matti Hamalainen <ccr@tnsp.org>
date Tue, 11 May 2021 02:08:30 +0300
parents 4b4299b62f7f
children 2a8c65d22f86
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,
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
27 "dump": False,
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
28 "normalize": False,
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
29
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
30 "debug": False,
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
31 }
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
32
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
33
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
34 pkk_str_fmap = {
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
35 "Fragment" : ["<", ">"],
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
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 pkk_debug_list = [
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
40 "ahas",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
41 "ahavakkaine",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
42 "ahavakala",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
43 "ahavakoittuo",
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
44 "ahvaliha",
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
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
47
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
48 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
49 ### Misc. helper functions, etc
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 def pkk_cleanup():
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
52 return 0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
54
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
55 ## Wrapper for print()
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
56 def pkk_print(smsg):
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
57 if pkk_cfg["normalize"]:
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
58 sys.stdout.write(unicodedata.normalize("NFC", smsg))
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
59 else:
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
60 sys.stdout.write(smsg)
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
61
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
62 def pkk_printi(indent, smsg):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
63 pkk_print((" " * indent) + smsg)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
64
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
65
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
66 def pkk_verbosity(lvl):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
67 return pkk_cfg["verbosity"] >= lvl
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
68
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
69
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70 ## Fatal error handler
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
71 def pkk_fatal(smsg):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
72 print(u"ERROR: "+ smsg)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
73 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
74
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
75
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
76 ## Handler for SIGINT signals
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
77 def pkk_signal_handler(signal, frame):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
78 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 print(u"\nQuitting due to SIGINT / Ctrl+C!")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
80 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
81
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
83 def pkk_get_text(lnode):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
84 stmp = ""
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
85 for pnode in lnode.iter():
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
86 if isinstance(pnode.text, str):
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
87 if pkk_cfg["annotate"] and isinstance(pnode.tag, str) and pnode.tag in pkk_str_fmap:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
88 stmp += pkk_str_fmap[pnode.tag][0] + pnode.text + pkk_str_fmap[pnode.tag][1]
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
89 else:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
90 stmp += pnode.text
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
91
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
92 if isinstance(pnode.tail, str):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
93 stmp += pnode.tail
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
94
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
95 return stmp.strip()
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
96
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
97
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
98 ##
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
99 def pkk_dump_recursive(indent, lnode):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
100 if lnode.tag in ["Example"]:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
101 stmp = pkk_get_text(lnode)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
102 pkk_printi(indent, "{} \"{}\"".format(lnode.tag, stmp))
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
103 else:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
104 if isinstance(lnode.text, str):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
105 stmp = lnode.text.strip()
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
106 if stmp != "":
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
107 stmp = " \""+ stmp +"\""
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
108 else:
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
109 stmp = ""
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
110
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
111 if len(lnode.attrib) > 0:
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
112 atmp = " "+str(lnode.attrib)
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
113 else:
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
114 atmp = ""
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
115
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
116 pkk_printi(indent, "{}{}{}\n".format(lnode.tag, atmp, stmp))
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
117 for qnode in lnode.findall("./*"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
118 pkk_dump_recursive(indent + 1, qnode)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
119
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
120
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
121 ##
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
122 def pkk_output_one(indent, dnode, dsub):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
123 for qnode in dnode.findall(dsub):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
124 pkk_printi(indent, "{}\n".format(pkk_get_text(qnode)))
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
125
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
126 def pkk_output_subs(indent, dnode, dsub, dname):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
127 for qnode in dnode.findall(dsub):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
128 pkk_printi(indent, "{} \"{}\"\n".format(dname, pkk_get_text(qnode)))
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
129
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
130
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
131 def pkk_output_sense(indent, dnode):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
132 pkk_output_subs(indent, dnode, "./SearchForm", "srch")
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
133 pkk_output_subs(indent, dnode, "./Definition", "defn")
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
134
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
135 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
136 sstr = pkk_get_text(wnode.find("./Example"))
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
137 lstr = ""
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
138
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
139 if pkk_verbosity(1):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
140 ltmp = []
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
141 for qnode in wnode.findall("./FreeTopic[@type='levikki']/GeographicalUsage"):
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
142 ltmp.append("{} [{}]".format(pkk_get_text(qnode), qnode.attrib["class"]))
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
143
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
144 if len(ltmp) > 0:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
145 lstr = " ({})".format(", ".join(ltmp))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
146
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
147 pkk_printi(indent + 1, "{} \"{}\"{}\n".format("exmp", sstr, lstr))
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
148
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
149
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
150 def pkk_output_node(indent, dnode):
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
151
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
152 for wnode in dnode.findall("./HeadwordCtn"):
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
153 pkk_output_one (indent, wnode, "./Headword")
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
154 pkk_output_sense(indent + 1, wnode)
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
155
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
156 index = 1
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
157 for wnode in dnode.findall("./SenseGrp"):
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
158 pkk_printi(indent + 1, "sense #{}\n".format(index))
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
159 pkk_output_sense(indent + 2, wnode)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
160 index += 1
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
161
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
162
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
163 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
164 ### Main program starts
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
165 ###
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
166 signal.signal(signal.SIGINT, pkk_signal_handler)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
167
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
168
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
169 ### Check if we have arguments
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
170 pkk_show_help = False
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
171 pkk_filenames = []
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
172 argc = 1
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
173 while argc < len(sys.argv):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
174 arg = sys.argv[argc]
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
175
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
176 needs_param = False
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
177 if argc + 1 < len(sys.argv):
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
178 param = sys.argv[argc + 1]
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
179 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
180 param = None
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
181
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
182 # Check for option type arg
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
183 if arg[0:1] == "-":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
184 oarg = arg
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
185 arg = arg.lstrip("-")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
186
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
187 if arg == "help" or arg == "h":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
188 pkk_show_help = True
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
189 elif arg == "dump" or arg == "d":
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
190 pkk_cfg["dump"] = True
5
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
191 elif arg == "normalize" or arg == "n":
274b2091137c Some more work on cleaning this up.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
192 pkk_cfg["normalize"] = True
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
193 elif arg == "annotate" or arg == "a":
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
194 pkk_cfg["annotate"] = True
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
195 elif arg == "p":
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
196 pkk_cfg["debug"] = True
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
197 elif arg == "verbosity" or arg == "v":
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
198 needs_param = True
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
199 pkk_cfg["verbosity"] = param
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
200 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
201 pkk_fatal(u"Invalid option argument '{0}'.".format(oarg))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
202
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
203 if needs_param and param == None:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
204 pkk_fatal(u"Option '{0}' requires an argument.".format(oarg))
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 # Non-option argument
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
207 pkk_filenames.append(arg)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
208
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
209 if needs_param:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 argc += 2
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
211 else:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
212 argc += 1
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
213
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
214
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
215 ### Show help if requested
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
216 if pkk_show_help or len(pkk_filenames) == 0:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
217 print(u"lxmldump - Dump ISO/FDIS 1951 XML file data")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
218 print(u"Usage: {0} <options> <input xml file(s)>".
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
219 format(str(Path(sys.argv[0]).name)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
220 print(u"")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
221 print(u" --help Show this help")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
222 print(u" -d, --dump Dump mode")
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
223 print(u" -n, --normalize Output NFC normalized Unicode")
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
224 print(u" -a, --annotate Annotate strings")
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
225 print(u" -v, --verbosity <n> Set verbosity level (0 - 3)")
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
226 print(u"")
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
227 sys.exit(0)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
228
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
229
8
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
230 ### Validate settings
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
231 try:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
232 pkk_cfg["verbosity"] = int(pkk_cfg["verbosity"])
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
233 except Exception as e:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
234 pkk_fatal(u"Verbosity level is not a valid integer.")
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
235 if pkk_cfg["verbosity"] < 0 or pkk_cfg["verbosity"] > 3:
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
236 pkk_fatal(u"Invalid verbosity level value {0}.".format(pkk_cfg["verbosity"]))
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
237
ce07bb2a247b More work.
Matti Hamalainen <ccr@tnsp.org>
parents: 7
diff changeset
238
6
34a89d61dbe7 Merge and cleanup.
Matti Hamalainen <ccr@tnsp.org>
parents: 5 3
diff changeset
239 ### Handle each input file
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
240 for filename in pkk_filenames:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
241 # Parse XML file into element tree
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
242 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
243 uxml = xmlET.parse(filename)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
244 except Exception as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
245 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e)))
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
246
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
247 # Dump output
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
248 try:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
249 xroot = uxml.getroot()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
250 for dnode in xroot.findall("./DictionaryEntry"):
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
251
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
252 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
253 continue
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
254
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
255 if pkk_cfg["dump"]:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
256 pkk_dump_recursive(0, dnode)
0
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
257 else:
7
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
258 pkk_output_node(0, dnode)
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
259
4b4299b62f7f Moar work.
Matti Hamalainen <ccr@tnsp.org>
parents: 6
diff changeset
260 print("\n")
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 except (BrokenPipeError, IOError) as e:
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
263 sys.stderr.close()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
264 sys.exit(1)
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
265
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
266 pkk_cleanup()
bddf1c283e51 Initial import.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
267 sys.exit(0)