lxmldump: lxmldump.py annotate

annotate lxmldump.py @ 38:0e586b4ab62c

Cosmetic adjustments to help.

author	Matti Hamalainen <ccr@tnsp.org>
date	Tue, 25 May 2021 23:51:56 +0300
parents	e176fcfc0235
children	f53ea742b57f

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 ###
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	4 ### lxmldump - Convert and dump ISO/FDIS 1951 XML file data
4 60b789dfee32 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 1 diff changeset	5 ### Programmed and designed by Matti 'ccr' Hämäläinen <ccr@tnsp.org>
60b789dfee32 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 1 diff changeset	6 ### (C) Copyright 2021 Tecnic Software productions (TNSP)
60b789dfee32 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 1 diff changeset	7 ###
23 3ef1c5463b8f Add license. Matti Hamalainen <ccr@tnsp.org> parents: 22 diff changeset	8 ### Released / distributed under 3-clause BSD license
3ef1c5463b8f Add license. Matti Hamalainen <ccr@tnsp.org> parents: 22 diff changeset	9 ### (see file "COPYING" for more information)
3ef1c5463b8f Add license. Matti Hamalainen <ccr@tnsp.org> parents: 22 diff changeset	10 ###
4 60b789dfee32 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 1 diff changeset	11 ### Python 3.7+ required!
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	12 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	13 import sys
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	14 import signal
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	15 import re
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	16 from pathlib import Path
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	17 import xml.etree.ElementTree as xmlET
5 274b2091137c Some more work on cleaning this up. Matti Hamalainen <ccr@tnsp.org> parents: 4 diff changeset	18 import unicodedata
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	19 import argparse
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	20
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	21 assert sys.version_info >= (3, 7)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	22
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	23
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	24 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	25 ### Default settings
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	26 ###
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	27 # Operation modes
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	28 PKK_MODE_NORMAL = 0
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	29 PKK_MODE_DUMP = 1
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	30 PKK_MODE_XML = 2
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	31 PKK_MODE_ANKI = 3
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	32
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	33
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	34 pkk_modes_list = {
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	35 PKK_MODE_NORMAL: "normal",
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	36 PKK_MODE_DUMP: "dump",
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	37 PKK_MODE_XML: "xml",
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	38 # PKK_MODE_ANKI: "anki",
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	39 }
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	40
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	41
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	42 pkk_mode_defaults = {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	43 # Default Ptr URL format strings
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	44 "ptr_url_fmt": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	45 PKK_MODE_NORMAL: u"<PTR:{href}>{text}</PTR>",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	46 PKK_MODE_ANKI: u"<a href='https://kaino.kotus.fi/cgi-bin/kks/karjala.cgi?a={href}'>{text}</a>",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	47 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	48
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	49 "word_fmt": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	50 PKK_MODE_NORMAL: "\"{text}\"",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	51 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	52 "word_attr_sep": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	53 PKK_MODE_NORMAL: " ; ",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	54 PKK_MODE_ANKI: ":",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	55 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	56 "word_attr_fmt": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	57 PKK_MODE_NORMAL: " ({alist})",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	58 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	59
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	60 "word_eol": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	61 PKK_MODE_NORMAL: "\n",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	62 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	63
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	64 "sense_index": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	65 PKK_MODE_NORMAL: "sense #{index}\n",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	66 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	67
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	68 "search_fmt": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	69 PKK_MODE_NORMAL: "srch \"{text}\"\n",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	70 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	71
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	72 "definition_fmt": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	73 PKK_MODE_NORMAL: "defn \"{text}\"\n",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	74 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	75
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	76 "example_fmt": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	77 PKK_MODE_NORMAL: "exmp \"{text}\"{geostr}\n",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	78 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	79 "example_geo_list": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	80 PKK_MODE_NORMAL: " ({glist})",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	81 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	82 "example_geo_empty": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	83 PKK_MODE_NORMAL: "",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	84 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	85 "example_geo_sep": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	86 PKK_MODE_NORMAL: ", ",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	87 },
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	88
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	89 "word_end": {
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	90 PKK_MODE_NORMAL: "\n",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	91 },
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	92 }
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	93
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	94
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	95 # Element annotation mappings
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	96 pkk_element_annotation_map = {
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	97 "Fragment" : {
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	98 PKK_MODE_NORMAL: ["<", ">"],
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	99 PKK_MODE_ANKI: ["<", ">"],
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	100 },
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	101 }
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	102
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	103
28 3442b8700da7 Comments. Matti Hamalainen <ccr@tnsp.org> parents: 27 diff changeset	104 # List of words in kks1/ useful for debugging, option -p
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	105 pkk_debug_list = [
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	106 "ahas",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	107 "ahavakkaine",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	108 "ahavakala",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	109 "ahavakoittuo",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	110 "ahvaliha",
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	111 "aloilleh",
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	112 "hanjahtoakseh",
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	113 "akkalisto",
33 3dcf8ac43bda Add one word to debug list. Matti Hamalainen <ccr@tnsp.org> parents: 32 diff changeset	114 "alto-",
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	115 ]
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	116
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	117
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	118 pkk_settings = {}
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	119
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	120
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	121 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	122 ### Misc. helper functions, etc
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	123 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	124 def pkk_cleanup():
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	125 return 0
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	126
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	127
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	128 ## 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	129 def pkk_print(smsg):
12 d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	130 try:
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	131 if pkk_cfg.normalize:
12 d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	132 sys.stdout.write(unicodedata.normalize("NFC", smsg))
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	133 else:
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	134 sys.stdout.write(smsg)
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	135
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	136 except (BrokenPipeError, IOError) as e:
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	137 sys.stderr.close()
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	138
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	139
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	140 ## Print string with indentation
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	141 def pkk_printi(indent, smsg):
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	142 pkk_print((" " * pkk_cfg.indent * indent) + smsg)
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	143
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	144
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	145 ## Fatal error handler
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	146 def pkk_fatal(smsg):
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	147 print(u"ERROR: "+ smsg)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	148 sys.exit(1)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	149
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	150
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	151 ## Handler for SIGINT signals
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152 def pkk_signal_handler(signal, frame):
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	153 pkk_cleanup()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	154 print(u"\nQuitting due to SIGINT / Ctrl+C!")
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	155 sys.exit(1)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	156
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	157
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	158 ## Value handling
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	159 class pkk_set_value(argparse.Action):
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	160
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	161 rexpr = re.compile(r'\s(\w+)\s=\s(.)\s*')
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	162
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	163 def __call__(self, parser, namespace, values, option_string=None):
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	164 rmatch = re.match(self.rexpr, values)
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	165 if rmatch:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	166 rid = rmatch.group(1).lower().replace("-", "_")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	167 rval = rmatch.group(2)
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	168 if rid in pkk_mode_defaults:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	169 pkk_settings[rid] = rval
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	170 else:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	171 pkk_fatal(f"Invalid option '{option_string} {values}': No such ID '{rid}'.")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	172 else:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	173 pkk_fatal(f"Invalid option '{option_string} {values}': Expected id=value.")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	174
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	175
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	176 ## Get mode if it exists
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	177 def pkk_test_value(mid):
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	178 if mid in pkk_mode_defaults:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	179 if pkk_cfg.mode in pkk_mode_defaults[mid]:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	180 mmode = pkk_cfg.mode
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	181 else:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	182 mmode = PKK_MODE_NORMAL
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	183
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	184 if mmode in pkk_mode_defaults[mid]:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	185 return mmode
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	186 else:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	187 return None
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	188 else:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	189 return None
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	190
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	191
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	192 ## Get default value per mode
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	193 def pkk_get_value(mid):
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	194 if mid in pkk_settings and pkk_settings[mid] != None:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	195 return pkk_settings[mid]
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	196
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	197 mmode = pkk_test_value(mid)
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	198 if mmode == None:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	199 pkk_fatal(f"Internal error: No mode for ID '{mid}'.")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	200
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	201 return pkk_mode_defaults[mid][mmode]
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	202
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	203
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	204 def pkk_get_fmt(mid):
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	205 return pkk_get_value(mid).replace("\\n", "\n")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	206
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	207
28 3442b8700da7 Comments. Matti Hamalainen <ccr@tnsp.org> parents: 27 diff changeset	208 ## Annotate given string with prefix and suffix based on tag
26 420f13925f20 Rename a variable. Matti Hamalainen <ccr@tnsp.org> parents: 25 diff changeset	209 def pkk_str_annotate(mtag, mstr):
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	210 if pkk_cfg.annotate and mtag in pkk_element_annotation_map:
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	211 if pkk_cfg.mode in pkk_element_annotation_map[mtag]:
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	212 mmode = pkk_cfg.mode
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	213 else:
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	214 mmode = PKK_MODE_NORMAL
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	215
26 420f13925f20 Rename a variable. Matti Hamalainen <ccr@tnsp.org> parents: 25 diff changeset	216 return pkk_element_annotation_map[mtag][mmode][0] + mstr + pkk_element_annotation_map[mtag][mmode][1]
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	217 else:
26 420f13925f20 Rename a variable. Matti Hamalainen <ccr@tnsp.org> parents: 25 diff changeset	218 return mstr
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	219
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	220
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	221 ## Clean string by removing tabs and newlines
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	222 def pkk_str_clean(mstr):
013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	223 return re.sub(r'[\n\r\t]', '', mstr)
013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	224
013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	225
28 3442b8700da7 Comments. Matti Hamalainen <ccr@tnsp.org> parents: 27 diff changeset	226 ## Format a "Ptr" node as text
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	227 def pkk_ptr_to_text(pnode):
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	228 pfmt = pkk_get_fmt("ptr_url_fmt")
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	229 return pfmt.format(
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	230 text=("".join(pnode.itertext())).strip(),
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	231 href=pnode.attrib["{http://www.w3.org/TR/xlink}href"])
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	232
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	233
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	234 ## Get text inside a given node
16 285b0820d2c6 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 15 diff changeset	235 def pkk_node_to_text(lnode):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	236 stmp = ""
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	237 for pnode in lnode.iter():
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	238 if pnode.tag == "Ptr":
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	239 stmp += pkk_ptr_to_text(pnode)
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	240 else:
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	241 if isinstance(pnode.text, str):
37 e176fcfc0235 Fix overstripping of concatenated strings. Matti Hamalainen <ccr@tnsp.org> parents: 36 diff changeset	242 stmp += pkk_str_annotate(pnode.tag, pkk_str_clean(pnode.text))
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	243
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	244 if isinstance(pnode.tail, str):
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	245 stmp += pkk_str_clean(pnode.tail)
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	246
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	247 return stmp.strip()
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	248
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	249
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	250 ## Simple recursive dump starting at given node
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	251 def pkk_dump_recursive(indent, lnode):
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	252 if lnode.tag in ["Example"]:
16 285b0820d2c6 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 15 diff changeset	253 stmp = pkk_node_to_text(lnode)
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	254 pkk_printi(indent, f"{lnode.tag} \"{stmp}\"\n")
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	255 else:
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	256 if isinstance(lnode.text, str):
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	257 textstr = pkk_str_clean(lnode.text).strip()
d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	258 if textstr != "":
d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	259 textstr = " \""+ textstr +"\""
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	260 else:
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	261 textstr = ""
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	262
34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	263 if len(lnode.attrib) > 0:
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	264 attrstr = " "+ str(lnode.attrib)
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	265 else:
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	266 attrstr = ""
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	267
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	268 pkk_printi(indent, f"{lnode.tag}{attrstr}{textstr}\n")
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	269 for qnode in lnode.findall("./*"):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	270 pkk_dump_recursive(indent + 1, qnode)
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	271
34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	272
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	273 ## Output item(s) under given node with given format string
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	274 def pkk_output_subs(indent, dnode, dsub, dfmtname):
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	275 dfmt = pkk_get_fmt(dfmtname)
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	276 for qnode in dnode.findall(dsub):
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	277 pkk_printi(indent, dfmt.format(text=pkk_node_to_text(qnode)))
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	278
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	279
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	280 ## Output a main "Headword" or "Sense" node under it
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	281 def pkk_output_sense(indent, dnode):
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	282 # Search form and definition
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	283 pkk_output_subs(indent, dnode, "./SearchForm", "search_fmt")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	284 pkk_output_subs(indent, dnode, "./Definition", "definition_fmt")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	285
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	286 # Examples
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	287 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"):
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	288 geolist = []
24 8c8e8e4504bb Remove verbosity option as it is not really used. Matti Hamalainen <ccr@tnsp.org> parents: 23 diff changeset	289 for qnode in wnode.findall("./FreeTopic[@type='levikki']/GeographicalUsage"):
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	290 geolist.append("{} [{}]".format(pkk_node_to_text(qnode), qnode.attrib["class"]))
8 ce07bb2a247b More work. Matti Hamalainen <ccr@tnsp.org> parents: 7 diff changeset	291
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	292 if len(geolist) > 0:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	293 geostr = pkk_get_fmt("example_geo_list").format(glist=pkk_get_fmt("example_geo_sep").join(geolist))
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	294 else:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	295 geostr = pkk_get_fmt("example_geo_empty")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	296
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	297 pkk_printi(indent + 1, pkk_get_fmt("example_fmt").format(
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	298 text=pkk_node_to_text(wnode.find("./Example")), geostr=geostr))
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	299
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	300
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	301 ## Output one "DictionaryEntry" node
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	302 def pkk_output_node(indent, dnode):
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	303
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	304 for wnode in dnode.findall("./HeadwordCtn"):
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	305 # Create list with grammatical attributes (noun, verb, etc.)
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	306 tmpl = []
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	307 for pnode in wnode.findall("./PartOfSpeechCtn/PartOfSpeech"):
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	308 tmpl.append(pnode.attrib["freeValue"])
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	309
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	310 for pnode in wnode.findall("./GrammaticalNote"):
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	311 tmpl.append(pkk_node_to_text(pnode))
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	312
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	313 # Remove duplicates and sort the list
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	314 tmpl = list(set(tmpl))
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	315 tmpl.sort(reverse=False, key=lambda attr: (attr, len(attr)))
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	316
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	317 # Print the headword and attributes if any
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	318 pkk_output_subs(indent, wnode, "./Headword", "word_fmt")
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	319
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	320 if len(tmpl) > 0:
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	321 pkk_print(pkk_get_fmt("word_attr_fmt").format(
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	322 alist=pkk_get_fmt("word_attr_sep").join(tmpl)))
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	323
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	324 pkk_print(pkk_get_fmt("word_eol"))
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	325
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	326 # Print main "sense"
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	327 pkk_output_sense(indent + 1, wnode)
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	328
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	329 # Print any other "senses"
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	330 index = 1
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	331 for wnode in dnode.findall("./SenseGrp"):
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	332 pkk_printi(indent + 1, pkk_get_fmt("sense_index").format(index=index))
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	333 pkk_output_sense(indent + 2, wnode)
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	334 index += 1
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	335
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	336 pkk_print(pkk_get_fmt("word_end"))
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	337
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	338
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	339 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	340 ### Main program starts
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	341 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	342 signal.signal(signal.SIGINT, pkk_signal_handler)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	343
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	344 optparser = argparse.ArgumentParser(
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	345 description="lxmldump - Convert and dump ISO/FDIS 1951 XML file data",
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	346 usage="%(prog)s [options] <input xml file(s)>",
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	347 add_help=False
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	348 )
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	349
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	350 optparser.add_argument("filenames",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	351 type=str, action="extend", nargs="*",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	352 metavar="filename",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	353 help="XML filename(s)")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	354
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	355 optparser.add_argument("-h", "--help",
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	356 dest="show_help",
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	357 action="store_true",
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	358 help="show this help message and exit")
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	359
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	360 optparser.add_argument("-d", "--dump",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	361 dest="mode",
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	362 action="store_const", const=PKK_MODE_DUMP, default=PKK_MODE_NORMAL,
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	363 help="output as simple dump")
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	364
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	365 optparser.add_argument("-x", "--xml",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	366 dest="mode",
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	367 action="store_const", const=PKK_MODE_XML,
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	368 help="output as XML")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	369
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	370 optparser.add_argument("-A", "--anki",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	371 dest="mode",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	372 action="store_const", const=PKK_MODE_ANKI,
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	373 help=argparse.SUPPRESS)
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	374 # help="output Anki compatible")
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	375
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	376 optparser.add_argument("-s", "--set",
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	377 action=pkk_set_value,
38 0e586b4ab62c Cosmetic adjustments to help. Matti Hamalainen <ccr@tnsp.org> parents: 37 diff changeset	378 metavar="ID=STR",
0e586b4ab62c Cosmetic adjustments to help. Matti Hamalainen <ccr@tnsp.org> parents: 37 diff changeset	379 help='set format string (see below)')
32 cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	380
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	381 optparser.add_argument("-n", "--normalize",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	382 dest="normalize",
29 f91ef7d7615b Use store_true feature of argparse. Matti Hamalainen <ccr@tnsp.org> parents: 28 diff changeset	383 action="store_true",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	384 help="output NFC normalized Unicode")
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	385
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	386 optparser.add_argument("-a", "--annotate",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	387 dest="annotate",
29 f91ef7d7615b Use store_true feature of argparse. Matti Hamalainen <ccr@tnsp.org> parents: 28 diff changeset	388 action="store_true",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	389 help="annotate strings")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	390
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	391 optparser.add_argument("-i", "--indent",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	392 dest="indent",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	393 type=int, choices=range(0, 32), default=4,
38 0e586b4ab62c Cosmetic adjustments to help. Matti Hamalainen <ccr@tnsp.org> parents: 37 diff changeset	394 metavar="N",
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	395 help='set indentation level (default: %(default)s)')
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	396
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	397 optparser.add_argument("-p", "--debug",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	398 dest="debug",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	399 action="store_true",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	400 help=argparse.SUPPRESS)
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	401
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	402
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	403 ### Parse arguments
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	404 pkk_cfg = optparser.parse_args()
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	405
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	406
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	407 ### Show help if needed
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	408 if len(pkk_cfg.filenames) == 0 or pkk_cfg.show_help:
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	409 optparser.print_help()
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	410
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	411 # print(u"\nDefault Ptr format strings per mode:")
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	412 # for pmode in pkk_modes_list:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	413 # if pmode in pkk_ptr_url_fmt:
4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	414 # print(u" {:6s} : \"{}\"".format(pkk_modes_list[pmode], pkk_ptr_url_fmt[pmode]))
34 73f2f98e3eef Cosmetic. Matti Hamalainen <ccr@tnsp.org> parents: 33 diff changeset	415
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	416 print(u"")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	417 sys.exit(0)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	418
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	419
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	420 ### Handle each input file
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	421 for filename in pkk_cfg.filenames:
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	422 # Parse XML file into element tree
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	423 try:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	424 uxml = xmlET.parse(filename)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	425 except Exception as e:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	426 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e)))
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	427
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	428 # Dump output
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	429 try:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	430 xroot = uxml.getroot()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	431 for dnode in xroot.findall("./DictionaryEntry"):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	432
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	433 if pkk_cfg.debug and dnode.attrib["identifier"] not in pkk_debug_list:
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	434 continue
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	435
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	436 if pkk_cfg.mode in [PKK_MODE_NORMAL, PKK_MODE_ANKI]:
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	437 try:
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	438 pkk_output_node(0, dnode)
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	439 except Exception as e:
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	440 print("")
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	441 pkk_dump_recursive(0, dnode)
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	442 print(str(e))
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	443 sys.exit(0)
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	444 elif pkk_cfg.mode == PKK_MODE_DUMP:
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	445 pkk_dump_recursive(0, dnode)
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	446 print("")
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	447 elif pkk_cfg.mode == PKK_MODE_XML:
36 4c8aafff8c5f Refactor output handling to be (mostly) configurable. Matti Hamalainen <ccr@tnsp.org> parents: 35 diff changeset	448 pkk_print(str(xmlET.tostring(dnode, encoding="utf8")) + "\n\n")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	449 else:
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	450 pkk_fatal("Invalid operation mode?")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	451
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	452 except (BrokenPipeError, IOError) as e:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	453 sys.stderr.close()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	454 sys.exit(1)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	455
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	456 pkk_cleanup()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	457 sys.exit(0)

Mercurial > hg > lxmldump

annotate lxmldump.py @ 38:0e586b4ab62c