drime

commit 6559338d7e22464f7eafdd42b56fe2a74c6eda76
parent 56a35afb5d21d5f51c62f6f685fff85859937970
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Wed,  9 Nov 2011 12:06:08 +0100

reorder code

Diffstat:
lexique2sql.py
 | 
67
++++++++++++++++++++++++++++++++++++++-----------------------------
lexique2sql.sh
 | 
2
+-

2 files changed, 39 insertions(+), 30 deletions(-)
diff --git a/lexique2sql.py b/lexique2sql.py
@@ -3,7 +3,7 @@
 """Prepare the rhyme database
 
 Input should have tab-separated fields: word, pronunciation, base word,
-grammatical category, frequency. Output is a """
+grammatical category, frequency. Output is a sequence of SQL statements"""
 
 # TODO frequencies are off
 # TODO "bibliographe" number of syllables?!
@@ -20,20 +20,14 @@ phon_vowels = "()$#289aeEioOuy@"
 # not a feminine ending, independently of spelling
 phon_non_end_fem = ['#', ')']
 
-# 
-print ("""CREATE TABLE words(
-  word varchar(100), -- word
-  phon varchar(100), -- pronunciation
-  base varchar(100), -- base word
-  kind varchar(10), -- grammatical category
-  freq float, -- frequency
-  min_nsyl int, -- lower bound on the number of syllabes
-  max_nsyl int, -- upper bound on the number of syllabes
-  word_end varchar(10), -- minimal word-level rhyme
-  phon_end varchar(10), -- minimal phon-level rhyme
-  elidable bool, -- can cause elision
-  feminine bool -- genre of the rhyme
-);""")
+def escape(x):
+  """Escape for SQL"""
+  s = []
+  for a in x:
+    if a in ["\\", "'"]:
+      s.append('\\')
+    s.append(a)
+  return ''.join(s)
 
 class Word:
   @property
@@ -139,18 +133,33 @@ class Word:
     else:
       self.nsyl = [min(self.nsyl[0], item), max(self.nsyl[1], item)]
 
-while True:
-  line = sys.stdin.readline()
-  if not line:
-    break
-  l = line.rstrip().split("\t")
-  word = l.pop(0)
-  phon = l.pop(0)
-  base = l.pop(0)
-  kind = l.pop(0)
-  freq = float(l.pop(0))
-  assert(len(l) == 0)
-  w = Word(word, phon, base, kind, freq)
-  if w.ok:
-    print(w.render_sql)
+if __name__ == '__main__':
+  print ("""CREATE TABLE words(
+    word varchar(100), -- word
+    phon varchar(100), -- pronunciation
+    base varchar(100), -- base word
+    kind varchar(10), -- grammatical category
+    freq float, -- frequency
+    min_nsyl int, -- lower bound on the number of syllabes
+    max_nsyl int, -- upper bound on the number of syllabes
+    word_end varchar(10), -- minimal word-level rhyme
+    phon_end varchar(10), -- minimal phon-level rhyme
+    elidable bool, -- can cause elision
+    feminine bool -- genre of the rhyme
+  );""")
+
+  while True:
+    line = sys.stdin.readline()
+    if not line:
+      break
+    l = line.rstrip().split("\t")
+    word = l.pop(0)
+    phon = l.pop(0)
+    base = l.pop(0)
+    kind = l.pop(0)
+    freq = float(l.pop(0))
+    assert(len(l) == 0)
+    w = Word(word, phon, base, kind, freq)
+    if w.ok:
+      print(w.render_sql)
 
diff --git a/lexique2sql.sh b/lexique2sql.sh
@@ -5,4 +5,4 @@ cd "$( dirname "$0" )"
 cat - additions | # add custom exceptions
   cut -f 1,2,3,4,7,8,9,10,24,28 | # select relevant fields
   awk '{FS="	"; OFS="	"; print $1, $2, $3, $4, ($5+$6)/2 + 100*($7+$8)/2}' | # aggregate frequencies
-  ./make_db.py
+  ./lexique2sql.py