| 1 |
6620
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Translates a data dictionary spreadsheet to Redmine formatting
|
| 3 |
|
|
|
| 4 |
|
|
import csv
|
| 5 |
|
|
import re
|
| 6 |
|
|
import sys
|
| 7 |
|
|
|
| 8 |
|
|
# Spreadsheet format
|
| 9 |
|
|
source_sep = ', '
|
| 10 |
|
|
|
| 11 |
|
|
##### URLs
|
| 12 |
|
|
|
| 13 |
6636
|
aaronmk
|
url_comment_re = r'(?:\([^)]*\))'
|
| 14 |
|
|
|
| 15 |
6620
|
aaronmk
|
def simplify_url(url): return re.sub(r'\(.*?\)', r'', url)
|
| 16 |
|
|
|
| 17 |
6635
|
aaronmk
|
def url_comment_text(comment):
|
| 18 |
|
|
if not comment: return comment
|
| 19 |
|
|
|
| 20 |
6638
|
aaronmk
|
match = re.match(r'^\(\d*:?(.*?)\)$', comment)
|
| 21 |
6635
|
aaronmk
|
assert match
|
| 22 |
|
|
text, = match.groups()
|
| 23 |
|
|
return text
|
| 24 |
|
|
|
| 25 |
6620
|
aaronmk
|
def url_term(url):
|
| 26 |
6637
|
aaronmk
|
'''Prefixes any provider in the URL to the term name, to create a namespace.
|
| 27 |
|
|
Each hierarchical component of the provider is stored in a URL comment.
|
| 28 |
|
|
'''
|
| 29 |
|
|
match = re.match(r'^('+url_comment_re+r'*).*?([\w:-]+)$', url)
|
| 30 |
6620
|
aaronmk
|
assert match
|
| 31 |
6637
|
aaronmk
|
provider_str, term = match.groups()
|
| 32 |
6639
|
aaronmk
|
provider = filter(bool, map(url_comment_text,
|
| 33 |
|
|
re.findall(url_comment_re, provider_str)))
|
| 34 |
6637
|
aaronmk
|
|
| 35 |
|
|
return ':'.join(provider+[term])
|
| 36 |
6620
|
aaronmk
|
|
| 37 |
|
|
##### Redmine
|
| 38 |
|
|
|
| 39 |
6684
|
aaronmk
|
sp = ' ' # Unicode thin space
|
| 40 |
|
|
|
| 41 |
6632
|
aaronmk
|
def redmine_bold(text): return '*'+text+'*'
|
| 42 |
|
|
|
| 43 |
6688
|
aaronmk
|
def redmine_pad(text): return sp+text+sp
|
| 44 |
|
|
|
| 45 |
6620
|
aaronmk
|
def redmine_url(text, url):
|
| 46 |
6686
|
aaronmk
|
if not url: return text
|
| 47 |
|
|
elif url.find('://') >= 0: return '"'+text+'":'+url # external link
|
| 48 |
|
|
else: return '[['+url+'|'+text+']]' # internal link
|
| 49 |
6620
|
aaronmk
|
|
| 50 |
|
|
def source2redmine_url(url):
|
| 51 |
6623
|
aaronmk
|
if url: return redmine_url(url_term(url), simplify_url(url))
|
| 52 |
|
|
else: return ''
|
| 53 |
6620
|
aaronmk
|
|
| 54 |
6684
|
aaronmk
|
def redmine_add_links(page_name, text):
|
| 55 |
|
|
# Link citations to entry in sources list
|
| 56 |
6687
|
aaronmk
|
text = re.sub(r'(?<=\[)(.*?)(?=\])',
|
| 57 |
6689
|
aaronmk
|
redmine_pad(redmine_url('\g<0>', page_name+r'#\g<0>')), text)
|
| 58 |
|
|
# redmine_pad() prevents [[[, which Redmine would parse incorrectly
|
| 59 |
6684
|
aaronmk
|
return text
|
| 60 |
|
|
|
| 61 |
6620
|
aaronmk
|
class RedmineTableWriter:
|
| 62 |
|
|
'''Formats rows as a Redmine table'''
|
| 63 |
|
|
|
| 64 |
|
|
def __init__(self, stream):
|
| 65 |
|
|
self.stream = stream
|
| 66 |
|
|
|
| 67 |
|
|
def writerow(self, row):
|
| 68 |
|
|
self.stream.write(('|'.join(['']+row+['']))+'\n')
|
| 69 |
|
|
|
| 70 |
6673
|
aaronmk
|
class RedmineDictWriter:
|
| 71 |
|
|
'''Formats rows as Redmine sections containing a table'''
|
| 72 |
|
|
|
| 73 |
6676
|
aaronmk
|
def __init__(self, term_col, def_col, header, stream):
|
| 74 |
6673
|
aaronmk
|
self.term_col = term_col
|
| 75 |
6676
|
aaronmk
|
self.def_col = def_col
|
| 76 |
6673
|
aaronmk
|
self.header = header
|
| 77 |
6684
|
aaronmk
|
self.page_name = header[term_col]
|
| 78 |
6673
|
aaronmk
|
self.stream = stream
|
| 79 |
|
|
|
| 80 |
|
|
def writerow(self, row):
|
| 81 |
6684
|
aaronmk
|
row = map(lambda f: redmine_add_links(self.page_name, f), row)
|
| 82 |
|
|
|
| 83 |
6676
|
aaronmk
|
term = row[self.term_col]
|
| 84 |
|
|
def_ = row[self.def_col]
|
| 85 |
|
|
self.stream.write('h3. '+term+'\n\n')
|
| 86 |
|
|
if def_: self.stream.write(def_+'\n\n')
|
| 87 |
6673
|
aaronmk
|
|
| 88 |
|
|
table = RedmineTableWriter(self.stream)
|
| 89 |
|
|
for i, col_name in enumerate(self.header):
|
| 90 |
6677
|
aaronmk
|
if not i in set([self.term_col, self.def_col]):
|
| 91 |
6676
|
aaronmk
|
value = row[i]
|
| 92 |
|
|
if value: table.writerow([redmine_bold(col_name), value])
|
| 93 |
|
|
self.stream.write('\n')
|
| 94 |
6620
|
aaronmk
|
|
| 95 |
6673
|
aaronmk
|
|
| 96 |
6620
|
aaronmk
|
def main():
|
| 97 |
6676
|
aaronmk
|
try: _prog_name, term_col, def_col, sources_col = sys.argv
|
| 98 |
6626
|
aaronmk
|
except ValueError: raise SystemExit('Usage: '+sys.argv[0]
|
| 99 |
6676
|
aaronmk
|
+' <spreadsheet term_col# definition_col# sources_col# >redmine')
|
| 100 |
|
|
term_col, def_col, sources_col = map(int, [term_col, def_col, sources_col])
|
| 101 |
6626
|
aaronmk
|
|
| 102 |
6632
|
aaronmk
|
# Translate input
|
| 103 |
6620
|
aaronmk
|
reader = csv.reader(sys.stdin)
|
| 104 |
6673
|
aaronmk
|
header = reader.next()
|
| 105 |
6676
|
aaronmk
|
writer = RedmineDictWriter(term_col, def_col, header, sys.stdout)
|
| 106 |
6620
|
aaronmk
|
for row in reader:
|
| 107 |
|
|
term = row[term_col]
|
| 108 |
6634
|
aaronmk
|
sources = row[sources_col].split(source_sep)
|
| 109 |
6620
|
aaronmk
|
|
| 110 |
6634
|
aaronmk
|
row[term_col] = redmine_url(term, simplify_url(sources[0]))
|
| 111 |
6632
|
aaronmk
|
row[sources_col] = source_sep.join(map(source2redmine_url, sources))
|
| 112 |
6628
|
aaronmk
|
|
| 113 |
6632
|
aaronmk
|
writer.writerow(row)
|
| 114 |
6620
|
aaronmk
|
|
| 115 |
|
|
main()
|
