drime

commit 77221a978a5e49fe5a16d6606559bc91576423e7
parent c1acbbd1530fd004a8141deea2705d0c0136a653
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Mon, 26 Dec 2011 23:13:54 +0100

lexique2sql.py now aggregates all possible derivations

Diffstat:
lexique2sql.py
 | 
38
+++++++++++++++++++++++++-------------

1 file changed, 25 insertions(+), 13 deletions(-)
diff --git a/lexique2sql.py b/lexique2sql.py
@@ -18,8 +18,7 @@ def string_type(max_len):
 sql_fields = [
   ('word', string_type(100)),
   ('phon', string_type(100)),
-  ('base', string_type(100)),
-  ('kind', string_type(10)),
+  ('orig', ('string2', {'max_len':100})),
   ('freq', ('float', {})),
   ('min_nsyl', ('int', {})),
   ('max_nsyl', ('int', {})),
@@ -110,14 +109,18 @@ class Word:
   @property
   def sql(self):
     render = {
-      'string': lambda s: '"'+escape(s)+'"', # no escaping: use parametrized queries!
-      'float': str,
-      'int': lambda s: str(int(s)),
-      'bool': lambda s: str(int(s)),
+      'string': lambda s, w: '"'+escape(s)+'"',
+      'string2': lambda s, w:
+            '"'+escape(
+              ', '.join([x[0]+(' ('+x[1]+')' if w.word != x[1] else '')
+                for x in s]))+'"',
+      'float': lambda s, w: str(s),
+      'int': lambda s, w: str(int(s)),
+      'bool': lambda s, w: str(int(s)),
     }
     def sql_field(field):
       (name, (ty, _)) = field
-      return render[ty](getattr(self, name))
+      return render[ty](getattr(self, name), self)
     return ('INSERT INTO words VALUES('
             + ', '.join([sql_field(f) for f in sql_fields])
             + ');')
@@ -125,8 +128,7 @@ class Word:
   def __init__(self, word, phon, base, kind, freq):
     self.word = word
     self.phon = phon
-    self.base = base
-    self.kind = kind
+    self.orig = [(kind, base)]
     self.freq = float(freq)
     self.nsyl = None
     self.do_extends()
@@ -153,7 +155,7 @@ class Word:
 def create_table():
   def decl(field):
     (name, (ty, data)) = field
-    if ty == 'string':
+    if ty.startswith('string'):
       ty = 'varchar(' + str(data['max_len']) + ')'
     return name + ' ' + ty
   return ('CREATE TABLE words('
@@ -161,14 +163,24 @@ def create_table():
           + ');')
 
 if __name__ == '__main__':
-  print (create_table())
+  cache = None
 
+  print (create_table())
   while True:
     line = sys.stdin.readline()
     if not line:
       break
     l = line.rstrip().split("\t")
     w = Word(*l)
-    if w.ok:
-      print(w.sql)
+    k = (w.word, w.phon)
+    if not w.ok:
+      continue
+    if cache and cache.word == w.word and cache.phon == w.phon:
+      cache.freq += w.freq
+      cache.orig += [w.orig[0]]
+    else:
+      if cache:
+        print(cache.sql)
+      cache = w
+  print(cache.sql)