lxmldump: lxmldump.py annotate

annotate lxmldump.py @ 35:5aafa87dbec2

Some Anki-related stuff, not finalized and thus hidden.

author	Matti Hamalainen <ccr@tnsp.org>
date	Tue, 25 May 2021 13:28:24 +0300
parents	73f2f98e3eef
children	4c8aafff8c5f

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 60b789dfee32 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 1 diff changeset	4 ### lxmldump - Dump ISO/FDIS 1951 XML file data
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
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	42 # Default Ptr URL format strings
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	43 pkk_ptr_url_fmt = {
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	44 PKK_MODE_NORMAL: u"<PTR:{href}>{text}</PTR>",
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	45 PKK_MODE_ANKI: u"<a href='https://kaino.kotus.fi/cgi-bin/kks/karjala.cgi?a={href}'>{text}</a>",
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	46 }
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	47
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	48
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	49 # Element annotation mappings
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	50 pkk_element_annotation_map = {
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	51 "Fragment" : {
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	52 PKK_MODE_NORMAL: ["<", ">"],
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	53 PKK_MODE_ANKI: ["<", ">"],
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	54 },
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	55 }
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	56
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	57
28 3442b8700da7 Comments. Matti Hamalainen <ccr@tnsp.org> parents: 27 diff changeset	58 # List of words in kks1/ useful for debugging, option -p
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	59 pkk_debug_list = [
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	60 "ahas",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	61 "ahavakkaine",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	62 "ahavakala",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	63 "ahavakoittuo",
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	64 "ahvaliha",
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	65 "aloilleh",
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	66 "hanjahtoakseh",
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	67 "akkalisto",
33 3dcf8ac43bda Add one word to debug list. Matti Hamalainen <ccr@tnsp.org> parents: 32 diff changeset	68 "alto-",
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	69 ]
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	70
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	71
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	72
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	73 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	74 ### Misc. helper functions, etc
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	75 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	76 def pkk_cleanup():
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	77 return 0
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	78
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	79
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	80 ## 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	81 def pkk_print(smsg):
12 d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	82 try:
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	83 if pkk_cfg.normalize:
12 d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	84 sys.stdout.write(unicodedata.normalize("NFC", smsg))
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	85 else:
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	86 sys.stdout.write(smsg)
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	87
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	88 except (BrokenPipeError, IOError) as e:
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	89 sys.stderr.close()
d50e71642be7 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 11 diff changeset	90
0 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 ## Print string with indentation
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	93 def pkk_printi(indent, smsg):
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	94 pkk_print((" " * pkk_cfg.indent * indent) + smsg)
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	95
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	96
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	97 ## Fatal error handler
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	98 def pkk_fatal(smsg):
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	99 print(u"ERROR: "+ smsg)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	100 sys.exit(1)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	101
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	102
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	103 ## Handler for SIGINT signals
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	104 def pkk_signal_handler(signal, frame):
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	105 pkk_cleanup()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	106 print(u"\nQuitting due to SIGINT / Ctrl+C!")
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	107 sys.exit(1)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	108
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	109
28 3442b8700da7 Comments. Matti Hamalainen <ccr@tnsp.org> parents: 27 diff changeset	110 ## Annotate given string with prefix and suffix based on tag
26 420f13925f20 Rename a variable. Matti Hamalainen <ccr@tnsp.org> parents: 25 diff changeset	111 def pkk_str_annotate(mtag, mstr):
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	112 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	113 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	114 mmode = pkk_cfg.mode
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	115 else:
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	116 mmode = PKK_MODE_NORMAL
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	117
26 420f13925f20 Rename a variable. Matti Hamalainen <ccr@tnsp.org> parents: 25 diff changeset	118 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	119 else:
26 420f13925f20 Rename a variable. Matti Hamalainen <ccr@tnsp.org> parents: 25 diff changeset	120 return mstr
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	121
8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	122
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	123 ## Clean string by removing tabs and newlines
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	124 def pkk_str_clean(mstr):
013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	125 return re.sub(r'[\n\r\t]', '', mstr)
013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	126
013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	127
28 3442b8700da7 Comments. Matti Hamalainen <ccr@tnsp.org> parents: 27 diff changeset	128 ## Format a "Ptr" node as text
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	129 def pkk_ptr_to_text(pnode):
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	130 # If custom format set, use it
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	131 if pkk_cfg.ptr_url_fmt != None:
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	132 pfmt = pkk_cfg.ptr_url_fmt
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	133 elif pkk_cfg.mode in pkk_ptr_url_fmt:
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	134 # Else try mode-specific
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	135 pfmt = pkk_ptr_url_fmt[pkk_cfg.mode]
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	136 else:
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	137 # Last resort is normal mode format
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	138 pfmt = pkk_ptr_url_fmt[PKK_MODE_NORMAL]
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	139
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	140 return pfmt.format(
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	141 text=("".join(pnode.itertext())).strip(),
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	142 href=pnode.attrib["{http://www.w3.org/TR/xlink}href"])
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	143
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	144
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	145 ## Get text inside a given node
16 285b0820d2c6 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 15 diff changeset	146 def pkk_node_to_text(lnode):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	147 stmp = ""
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	148 for pnode in lnode.iter():
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	149 if pnode.tag == "Ptr":
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	150 stmp += pkk_ptr_to_text(pnode)
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	151 else:
2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	152 if isinstance(pnode.text, str):
25 8a6738f67106 Factor string annotation into separate function. Matti Hamalainen <ccr@tnsp.org> parents: 24 diff changeset	153 stmp += pkk_str_annotate(pnode.tag, pkk_str_clean(pnode.text).strip())
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	154
9 2a8c65d22f86 Moar. Matti Hamalainen <ccr@tnsp.org> parents: 8 diff changeset	155 if isinstance(pnode.tail, str):
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	156 stmp += pkk_str_clean(pnode.tail)
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	157
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	158 return stmp.strip()
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	159
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	160
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	161 ## Simple recursive dump starting at given node
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	162 def pkk_dump_recursive(indent, lnode):
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	163 if lnode.tag in ["Example"]:
16 285b0820d2c6 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 15 diff changeset	164 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	165 pkk_printi(indent, f"{lnode.tag} \"{stmp}\"\n")
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	166 else:
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	167 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	168 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	169 if textstr != "":
d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	170 textstr = " \""+ textstr +"\""
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	171 else:
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	172 textstr = ""
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	173
34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	174 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	175 attrstr = " "+ str(lnode.attrib)
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	176 else:
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	177 attrstr = ""
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	178
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	179 pkk_printi(indent, f"{lnode.tag}{attrstr}{textstr}\n")
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	180 for qnode in lnode.findall("./*"):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	181 pkk_dump_recursive(indent + 1, qnode)
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	182
34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	183
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	184 ## Output item(s) under given node with given format string
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	185 def pkk_output_subs_fmt(indent, dnode, dsub, dname, dfmt):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	186 for qnode in dnode.findall(dsub):
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	187 pkk_printi(indent, dfmt.format(nname=dname, ntext=pkk_node_to_text(qnode)))
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	188
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	189 ## Output item(s) under given node with a prefixed name string
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	190 def pkk_output_subs_prefix(indent, dnode, dsub, dname):
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	191 pkk_output_subs_fmt(indent, dnode, dsub, dname, "{nname} \"{ntext}\"\n")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	192
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	193
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	194 ## Output a main "Headword" or "Sense" node under it
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	195 def pkk_output_sense(indent, dnode):
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	196 # Search form and definition
13 3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	197 pkk_output_subs_prefix(indent, dnode, "./SearchForm", "srch")
3bd772fd6a50 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 12 diff changeset	198 pkk_output_subs_prefix(indent, dnode, "./Definition", "defn")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	199
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	200 # Examples
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	201 for wnode in dnode.findall("./ExampleBlock/ExampleCtn"):
16 285b0820d2c6 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 15 diff changeset	202 sstr = pkk_node_to_text(wnode.find("./Example"))
8 ce07bb2a247b More work. Matti Hamalainen <ccr@tnsp.org> parents: 7 diff changeset	203 lstr = ""
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	204
24 8c8e8e4504bb Remove verbosity option as it is not really used. Matti Hamalainen <ccr@tnsp.org> parents: 23 diff changeset	205 ltmp = []
8c8e8e4504bb Remove verbosity option as it is not really used. Matti Hamalainen <ccr@tnsp.org> parents: 23 diff changeset	206 for qnode in wnode.findall("./FreeTopic[@type='levikki']/GeographicalUsage"):
8c8e8e4504bb Remove verbosity option as it is not really used. Matti Hamalainen <ccr@tnsp.org> parents: 23 diff changeset	207 ltmp.append("{} [{}]".format(pkk_node_to_text(qnode), qnode.attrib["class"]))
8 ce07bb2a247b More work. Matti Hamalainen <ccr@tnsp.org> parents: 7 diff changeset	208
24 8c8e8e4504bb Remove verbosity option as it is not really used. Matti Hamalainen <ccr@tnsp.org> parents: 23 diff changeset	209 if len(ltmp) > 0:
8c8e8e4504bb Remove verbosity option as it is not really used. Matti Hamalainen <ccr@tnsp.org> parents: 23 diff changeset	210 lstr = " ({})".format(", ".join(ltmp))
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	211
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	212 pkk_printi(indent + 1, "{} \"{}\"{}\n".format("exmp", sstr, lstr))
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	213
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	214
17 6fa24c711f86 Add some comments. Matti Hamalainen <ccr@tnsp.org> parents: 16 diff changeset	215 ## Output one "DictionaryEntry" node
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	216 def pkk_output_node(indent, dnode):
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	217
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	218 for wnode in dnode.findall("./HeadwordCtn"):
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	219 # Create list with grammatical attributes (noun, verb, etc.)
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	220 tmpl = []
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	221 for pnode in wnode.findall("./PartOfSpeechCtn/PartOfSpeech"):
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	222 tmpl.append(pnode.attrib["freeValue"])
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	223
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	224 for pnode in wnode.findall("./GrammaticalNote"):
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	225 tmpl.append(pkk_node_to_text(pnode))
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	226
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	227 # Remove duplicates and sort the list
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	228 tmpl = list(set(tmpl))
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	229 tmpl.sort(reverse=False, key=lambda attr: (attr, len(attr)))
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	230
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	231 # Print the headword and attributes if any
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	232 pkk_output_subs_fmt(indent, wnode, "./Headword", "", "\"{ntext}\"")
d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	233
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	234 if len(tmpl) > 0:
32 cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	235 pkk_print(" ({nlist})".format(nlist=pkk_cfg.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	236
d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	237 pkk_print("\n")
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	238
ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	239 # Print main "sense"
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	240 pkk_output_sense(indent + 1, wnode)
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	241
18 ff959de0f6c8 Add grammatical attributes. Matti Hamalainen <ccr@tnsp.org> parents: 17 diff changeset	242 # Print any other "senses"
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	243 index = 1
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	244 for wnode in dnode.findall("./SenseGrp"):
27 d77ab8a300b1 Use Python 3.6/3.7 format strings where we can. Matti Hamalainen <ccr@tnsp.org> parents: 26 diff changeset	245 pkk_printi(indent + 1, f"sense #{index}\n")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	246 pkk_output_sense(indent + 2, wnode)
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	247 index += 1
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	248
34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	249
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	250 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	251 ### Main program starts
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	252 ###
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	253 signal.signal(signal.SIGINT, pkk_signal_handler)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	254
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	255 optparser = argparse.ArgumentParser(
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	256 description="lxmldump - Dump ISO/FDIS 1951 XML file data",
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	257 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	258 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	259 )
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	260
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	261 optparser.add_argument("filenames",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	262 type=str, action="extend", nargs="*",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	263 metavar="filename",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	264 help="XML filename(s)")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	265
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	266 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	267 dest="show_help",
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	268 action="store_true",
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	269 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	270
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	271 optparser.add_argument("-d", "--dump",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	272 dest="mode",
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	273 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	274 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	275
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	276 optparser.add_argument("-x", "--xml",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	277 dest="mode",
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	278 action="store_const", const=PKK_MODE_XML,
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	279 help="output as XML")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	280
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	281 optparser.add_argument("-A", "--anki",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	282 dest="mode",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	283 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	284 help=argparse.SUPPRESS)
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	285 # help="output Anki compatible")
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	286
30 34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	287 optparser.add_argument("--ptr-url-fmt",
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	288 dest="ptr_url_fmt",
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	289 type=str,
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	290 default=None,
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	291 metavar="str",
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	292 help='Ptr URL format string (see below)')
34755af2ea1f Make Ptr field URL formatting configurable. Matti Hamalainen <ccr@tnsp.org> parents: 29 diff changeset	293
32 cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	294 optparser.add_argument("--attr-sep",
cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	295 dest="word_attr_sep",
cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	296 type=str, default=" ; ",
cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	297 metavar="str",
cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	298 help='word attribute separator (default: \"%(default)s\")')
cfecd039506e Make word attribute separator string configurable. Matti Hamalainen <ccr@tnsp.org> parents: 31 diff changeset	299
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	300 optparser.add_argument("-n", "--normalize",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	301 dest="normalize",
29 f91ef7d7615b Use store_true feature of argparse. Matti Hamalainen <ccr@tnsp.org> parents: 28 diff changeset	302 action="store_true",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	303 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	304
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	305 optparser.add_argument("-a", "--annotate",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	306 dest="annotate",
29 f91ef7d7615b Use store_true feature of argparse. Matti Hamalainen <ccr@tnsp.org> parents: 28 diff changeset	307 action="store_true",
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	308 help="annotate strings")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	309
22 76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	310 optparser.add_argument("-i", "--indent",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	311 dest="indent",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	312 type=int, choices=range(0, 32), default=4,
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	313 metavar="n",
76856fc4e5fa Clean up argument handling. Matti Hamalainen <ccr@tnsp.org> parents: 21 diff changeset	314 help='indent output by <n> characters (default: %(default)s)')
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	315
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	316 optparser.add_argument("-p", "--debug",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	317 dest="debug",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	318 action="store_true",
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	319 help=argparse.SUPPRESS)
5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	320
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	321
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	322 ### Show help if needed
f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	323 pkk_cfg = optparser.parse_args()
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	324 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	325 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	326
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	327 print(u"\nDefault Ptr format strings per mode:")
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	328 for pmode in pkk_modes_list:
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	329 if pmode in pkk_ptr_url_fmt:
4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	330 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	331
31 4cbefe4c6f53 Improve help, list the default Ptr format strings for each mode. Matti Hamalainen <ccr@tnsp.org> parents: 30 diff changeset	332 print(u"")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	333 sys.exit(0)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	334
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	335
6 34a89d61dbe7 Merge and cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 5 3 diff changeset	336 ### 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	337 for filename in pkk_cfg.filenames:
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	338 # Parse XML file into element tree
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	339 try:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	340 uxml = xmlET.parse(filename)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	341 except Exception as e:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	342 pkk_fatal(u"SVG/XML parsing failed: {0}".format(str(e)))
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	343
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	344 # Dump output
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	345 try:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	346 xroot = uxml.getroot()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	347 for dnode in xroot.findall("./DictionaryEntry"):
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	348
20 f274504eafd0 Use Python argparse module instead of custom self-rolled argument parser. Matti Hamalainen <ccr@tnsp.org> parents: 19 diff changeset	349 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	350 continue
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	351
35 5aafa87dbec2 Some Anki-related stuff, not finalized and thus hidden. Matti Hamalainen <ccr@tnsp.org> parents: 34 diff changeset	352 if pkk_cfg.mode in [PKK_MODE_NORMAL, PKK_MODE_ANKI]:
19 7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	353 try:
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	354 pkk_output_node(0, dnode)
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	355 except Exception as e:
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	356 pkk_dump_recursive(0, dnode)
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	357 print(str(e))
7c6eb57798bd Ja niin. Matti Hamalainen <ccr@tnsp.org> parents: 18 diff changeset	358 sys.exit(0)
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	359 elif pkk_cfg.mode == PKK_MODE_DUMP:
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	360 pkk_dump_recursive(0, dnode)
21 7ef08e05a5bf Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 20 diff changeset	361 elif pkk_cfg.mode == PKK_MODE_XML:
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	362 pkk_print(str(xmlET.tostring(dnode, encoding="utf8")) + "\n")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	363 else:
10 013f0cd9e5b3 More work. Matti Hamalainen <ccr@tnsp.org> parents: 9 diff changeset	364 pkk_fatal("Invalid operation mode?")
7 4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	365
4b4299b62f7f Moar work. Matti Hamalainen <ccr@tnsp.org> parents: 6 diff changeset	366 print("\n")
0 bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	367
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	368 except (BrokenPipeError, IOError) as e:
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	369 sys.stderr.close()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	370 sys.exit(1)
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	371
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	372 pkk_cleanup()
bddf1c283e51 Initial import. Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	373 sys.exit(0)

Mercurial > hg > lxmldump

annotate lxmldump.py @ 35:5aafa87dbec2