david_walker  committed 1a51338

new regexes for One Acre Fund

  • Participants
  • Parent commits 5190539

Comments (0)

Files changed (1)

         # Character standardization
         (u'“|”', u'"'),
         (u"’", u"'"),
+        (u'½ (an |)acre', u'half an acre'),
+        (u'½', u'one half'),
         # misspellings
         (u'dependants', u'dependents'),
         (ur'therefor\b', u'therefore'),
-        (ur'(?i)(micro) +finance', u'\1finance'),
+        (ur'(?i)(micro) +finance', ur'\1finance'),
+        (u'south sudan', u'South Sudan'),
+        (u'purshace', u'purchase'),
+        (u'purshacing', u'purchasing'),
+        (u'One Ace Fund', u'One Acre Fund'),
         # proper nouns
         (u'congo town', u'Congo Town'),
         (u'KADET LTD', u'KADET Ltd.'),
         # awkward or verbose constructions
+        (u'in the year ([0-9]+)', ur'in \1'),
+        (u'three/four acre(s?)', ur'three quarters of an acre'),
         (u'requested for', u'requested'),
         (u'has given birth to', 'has'),
         (u'requesting to borrow', u'asking to borrow'),
         # Chiefly British
         (u'([iI]n) future', ur'\1 the future'),
-        (u'tyres', u'tires'),
-        (u'neighbour', u'neighbor'),
-        (u'licencing', u'licensing'),
         # non-ISO currency abbreviations
         (u'/=', u' UGX '),
         (ur'[Pp]hp', 'PHP'),
         (ur'(?i)\bLE([0-9]*)\b', ur'SLL \1'),
         (ur'\bRp\.', 'IDR'),
+        (u'JD', u'JOD'),
         # incorrect punctuation
         (ur'e\.t\.c\.?', u'etc.'),
         (ur'never missed any meeting\.', u'never missed any meetings.'),
         # Field partner template cleanup
+        (u"due to One Acre Fund's best prices and in order to ensure food "
+         "security for (his|her) family and later educate them.",
+         ur"to get One Acre Fund's low prices, to ensure food security for \1"
+         " family, and, later, to educate them."),
+         (ur"The distribution of farming inputs is part of One Acre Fund's "
+          "integrated agriculture package, which includes training, reliable "
+          "input supply \(such as fertilizer and seeds\), credit and insurance\.",
+          ur"The distribution of farming inputs such as fertilizer and seed is"
+          " part of One Acre Fund's integrated agriculture package, which also "
+          "includes training, credit, and insurance."),
+        (ur'was only able to harvest (.+) bags of maize on (.+) of land\.',
+         ur'was only able to harvest \1 bags of maize on that land.'),
+        (ur'one bags', ur'one bag'),
+        (ur'With income from maize sales', ur'With the income from maize sales'),
         (ur'To make a living, (?P<name>(\w|\s)+) owns & operates a business'
            'venture in the [a-z]+ sector \w+ (?P<business>[^.]+\.)',
          ur'\g<name> has a \g<business>'),