changeset 136:91e4ece4b9cd

commented out duplicate, handle comments better
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Fri, 29 Sep 2023 15:59:34 +0100
parents df89f0052f3d
children 09019621f790
files lib/python/cc/lmh/test/key_tests.tsv lib/python/cc/lmh/test/test_keys.py
diffstat 2 files changed, 6 insertions(+), 3 deletions(-) [+]
line wrap: on
line diff
--- a/lib/python/cc/lmh/test/key_tests.tsv	Fri Sep 29 15:14:29 2023 +0100
+++ b/lib/python/cc/lmh/test/key_tests.tsv	Fri Sep 29 15:59:34 2023 +0100
@@ -4,11 +4,11 @@
 Final PHPSESSIONID in quoted URI	https://www.eu-kommunal-kompass.de/index.php/foerderdatenbank?option=com_redirectpage&view=redirectpage&url=http%3A%2F%2Fwww.umweltbildung.de%2Ffileadmin%2Finhalte-projekte%2FKommune%2FANU-BNE_im_komm._KS_Abschlussbericht_pblc1b.pdf%3FPHPSESSID%3Da23dac7de24bb2f1e53978474b0d9c2d	de,eu-kommunal-kompass)/index.php/foerderdatenbank?option=com_redirectpage&url=http://www.umweltbildung.de/fileadmin/inhalte-projekte/kommune/anu-bne_im_komm._ks_abschlussbericht_pblc1b.pdf?&view=redirectpage
 x7f raw in path	http://www.televida.biz/en/customers/media/177-%7F5th-anniversary-%E2%80%9Cviva-la-ma%C3%B1ana%E2%80%9D-from-guatevision-channel.html	biz,televida)/en/customers/media/177-\x7f5th-anniversary-%e2%80%9cviva-la-ma%c3%b1ana%e2%80%9d-from-guatevision-channel.html
 Not octal IPv4	http://143.107.225.017/	17,225,107,143)/
-Televida	http://www.televida.biz/en/customers/media/177-%7F5th-anniversary-%E2%80%9Cviva-la-ma%C3%B1ana%E2%80%9D-from-guatevision-channel.html	biz,televida)/en/customers/media/177-\x7f5th-anniversary-%e2%80%9cviva-la-ma%c3%b1ana%e2%80%9d-from-guatevision-channel.html
+# [duplicate deleted]
 %25 in query	http://223.223.160.154/ngt/livecam/index.php?login=+%C4%B9%CC%E7%A1%A6%BB%B0%B6%F9%A1%A1%A5%E9%A5%A4%A5%D6%A5%AB%A5%E1%A5%E9	154,160,223,223)/ngt/livecam/index.php?login=+%c4%b9%25cc%e7%a1%a6%25bb%25b0%25b6%25f9%25a1%25a1%25a5%e9%a5%a4%25a5%d6%a5%25ab%25a5%25e1%25a5%25e9
 %25 in path	http://www.55yq.cn/www.dm190.com/list/%25CE%25FC%D1%AA%25B9%25ED-new--.html	cn,55yq)/www.dm190.com/list/%25ce%25fc%d1%aa%25b9%25ed-new--.html
 CFID last	http://173.161.106.217/Accounts/Login.cfm?ts=%7Bts%20'2018-12-07%2001:22:30'%7D&CFID=81256&CFTOKEN=e2e985379a7dd979-4B757A99-E22D-6CFF-6FD928437D3F2BA3	217,106,161,173)/accounts/login.cfm?&ts={ts%20'2018-12-07%2001:22:30'}
 CFID first	http://200.52.131.30/gpocii/index3.html?CFID=54307&CFTOKEN=67112922	30,131,52,200)/gpocii/index3.html?cfid=54307&cftoken=67112922
 Multi PHPSESS	http://www.rudolfreisen.cz/zajezdy/lastminute-zajezdy/?lastfirstform&amp;search=http%3A%2F%2Fcomfort31.traffics-ibe.com%2Ftibet.php%3Fstp%3D4%26mop%3Dyes%26ixp%3Dpauschal%26cfg%3D0080017030000000%26vnd%3D1555063200%26bsd%3D1558864800%26tdm%3D7%26tdx%3D7%26mmc%3D1w%26typ%3DP%26fdir%3D1%26epc%3D2%26ka1%3D%26ka2%3D%26ka3%3D%26kpc%3D%26kpa%3D%26hmi%3DALLE%26stc%3D3%26zac%3D%252A%26vpc%3DAI%26dsi%3DALLE%26dsr%3D100388%26bst%3D%26vnt%3D%26refer%3D2%26tsc%3D1%26out%3Ddetail%26lng%3Dcs%26icd%3D%26enc%3DUTF-8%26tps%3Dt5%26mpo%3Dl%26sub%3Dall%26CurrencySymbol%3DEUR%26hmask%3D1%26rmask%3D1%26ntc%3D1%26region%3D100388%26verschlagw%5B%5D%3Dben%26PHPSESSID%3D98a1981a1b7b2547f61ee39e7b674f3c%26mmc%3D1w%26mpo%3Dl%26tps%3Dt5%26enc%3DUTF-8%26lng%3Dcs%26PHPSESSID%3D98a1981a1b7b2547f61ee39e7b674f3c%26gid%3D265591%26azh%3DAGA%26hsc%3Dc87229ad96ef3ad4b8b2a1b00e75c429%26PHPSESSID%3D98a1981a1b7b2547f61ee39e7b674f3c	cz,rudolfreisen)/zajezdy/lastminute-zajezdy?&amp;search=http://comfort31.traffics-ibe.com/tibet.php?stp=4&azh=aga&bsd=1558864800&bst=&cfg=0080017030000000&currencysymbol=eur&dsi=alle&dsr=100388&enc=utf-8&enc=utf-8&epc=2&fdir=1&hmask=1&hmi=alle&hsc=c87229ad96ef3ad4b8b2a1b00e75c429&icd=&ixp=pauschal&ka1=&ka2=&ka3=&kpa=&kpc=&lastfirstform&lng=cs&lng=cs&mmc=1w&mmc=1w&mop=yes&mpo=l&mpo=l&ntc=1&out=detail&phpsesgid=265591&phpsessid=98a1981a1b7b2547f61ee39e7b674f3c&refer=2&region=100388&rmask=1&stc=3&sub=all&tdm=7&tdx=7&tps=t5&tps=t5&tsc=1&typ=p&verschlagw[]=ben&vnd=1555063200&vnt=&vpc=ai&zac=*
 UTF-8 in host	https://kabarettfrühling.at/	at,xn--kabarettfrhling-8vb)/
-UTF-8 in URI	https://www.entr\u00FCmpelung-wien.at/entr%C3%BCmpelung1/entr%C3%BCmpelung	at,xn--entrmpelung-wien-mzb)/entr%c3%bcmpelung1/entr%c3%bcmpelung
+Escaped UTF-8 in URI	https://www.entr\u00FCmpelung-wien.at/entr%C3%BCmpelung1/entr%C3%BCmpelung	at,xn--entrmpelung-wien-mzb)/entr%c3%bcmpelung1/entr%c3%bcmpelung
--- a/lib/python/cc/lmh/test/test_keys.py	Fri Sep 29 15:14:29 2023 +0100
+++ b/lib/python/cc/lmh/test/test_keys.py	Fri Sep 29 15:59:34 2023 +0100
@@ -9,9 +9,12 @@
                        'key_tests.tsv'),'r') as f:
   i = 0
   for l in f:
-    if l[0] in '#\n':
+    if l[0] == '\n':
       continue
     i+=1
+    if l[0] == '#':
+      # preserve numbering
+      continue
     d, u, k = l.rstrip().split('\t')
     kk = cdx_key(u)
     if kk == k: