Mercurial > hg > cc > azure

comparison master/bin/fixDates.py @ 34:ad6eff2bc6f9

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
fixes to logging and efficiency, see also notes.txt wrt patches to dateparser
author Henry S. Thompson <ht@markup.co.uk>
date Wed, 21 Nov 2018 18:42:56 +0000
parents 4c117ee8ed75
children bb09db2afe6b
comparison
equal deleted inserted replaced
33:4c117ee8ed75 34:ad6eff2bc6f9
12 HTTPS=1 12 HTTPS=1
13 sn=['http','https'] 13 sn=['http','https']
14 http_ytab=list(201*[None]) # 1900--2100 14 http_ytab=list(201*[None]) # 1900--2100
15 https_ytab=list(201*[None]) 15 https_ytab=list(201*[None])
16 tab=[http_ytab,https_ytab] 16 tab=[http_ytab,https_ytab]
17 nd=[0,0] 17 nd=[0,0] # no date
18 ed=[0,0] # date < 1900
19 ld=[0,0] # date > 2100
18 mn=[None,'Jan','Feb','Mar','Apr','May','Jun', 20 mn=[None,'Jan','Feb','Mar','Apr','May','Jun',
19 'Jul','Aug','Sep','Oct','Nov','Dec'] 21 'Jul','Aug','Sep','Oct','Nov','Dec']
20 months=dict(zip(mn[1:],range(1,13))) 22 months=dict(zip(mn[1:],range(1,13)))
21 for l in sys.stdin: 23 for l in sys.stdin:
22 if l[0]=='#': 24 if l[0]=='#':
23 print('# %s'%strftime('%Y-%m-%d %H:%M:%S'),file=sys.stderr) 25 print('#1 %s'%strftime('%Y-%m-%d %H:%M:%S'),file=sys.stderr)
24 continue 26 continue
25 n+=1 27 n+=1
26 ff=ok.match(l) 28 ff=ok.match(l)
27 if ff is not None: 29 if ff is not None:
28 #print(l,end='') 30 #print(l,end='')
37 year=int(ff.group(3)) 39 year=int(ff.group(3))
38 count=int(ff.group(4)) 40 count=int(ff.group(4))
39 except: 41 except:
40 # Unusual month or year field 42 # Unusual month or year field
41 try: 43 try:
42 d=parse("%s %s"%(ff.group(2),ff.group(3))) 44 d=parse("%s %s"%(ff.group(2),ff.group(3)))#,languages=['en'])
43 if d is None or count is None: 45 if d is None or count is None:
44 print(5,ff.group(1),ff.group(2),ff.group(3),ff.group(4), 46 print(5,ff.group(1),ff.group(2),ff.group(3),ff.group(4),
45 file=sys.stderr) 47 file=sys.stderr)
46 bogons+=1 48 bogons+=1
47 continue 49 continue
81 nd[scheme]+=count 83 nd[scheme]+=count
82 continue 84 continue
83 else: 85 else:
84 l=' '.join(cols) 86 l=' '.join(cols)
85 try: 87 try:
86 d=parse(l) 88 d=parse(l)#,languages=['en']))
87 if d is None: 89 if d is None:
88 print(3,d,l,count,file=sys.stderr) 90 print(3,d,l,count,file=sys.stderr)
89 bogons+=1 91 bogons+=1
90 continue 92 continue
91 elif d.year<1900 or d.year>2100: 93 elif d.year<1900 or d.year>2100:
92 # Shouldn't happen 94 # Jan 0001 does show up, so log these as early / late
93 print(8,sn[scheme],d.month,d.year,count, 95 (ed if d.year<1900 else ld)[scheme]+=count
94 file=sys.stderr)
95 bogons+=1
96 continue 96 continue
97 else: 97 else:
98 year=d.year 98 year=d.year
99 month=d.month 99 month=d.month
100 except Exception as e: 100 except Exception as e:
110 mm=yy[y] 110 mm=yy[y]
111 mm[month]+=count 111 mm[month]+=count
112 if n==0: 112 if n==0:
113 # ssh screwed up 113 # ssh screwed up
114 exit(1) 114 exit(1)
115 print('#2 %s'%strftime('%Y-%m-%d %H:%M:%S'),file=sys.stderr)
115 for s in (HTTP,HTTPS): 116 for s in (HTTP,HTTPS):
116 if nd[s]!=0: 117 if nd[s]!=0:
117 print(sn[s],0,0,nd[s],sep='\t') 118 print(sn[s],0,0,nd[s],sep='\t')
119 if ed[s]!=0:
120 print(sn[s],0,1,ed[s],sep='\t')
121 if ld[s]!=0:
122 print(sn[s],0,2,ld[s],sep='\t')
118 yy=tab[s] 123 yy=tab[s]
119 for y in range(201): 124 for y in range(201):
120 mm=yy[y] 125 mm=yy[y]
121 if mm is not None: 126 if mm is not None:
122 for m in range(1,13): 127 for m in range(1,13):
123 if mm[m]!=0: 128 if mm[m]!=0:
124 print(sn[s],mn[m],y+1900,mm[m],sep='\t') 129 print(sn[s],mn[m],y+1900,mm[m],sep='\t')
125 print(n,bogons,file=sys.stderr) 130 print(n,bogons,file=sys.stderr)
131 print('#3 %s'%strftime('%Y-%m-%d %H:%M:%S'),file=sys.stderr)
126 132
127 133
128 134