Unescape strings

diff --git a/ChangeLog b/ChangeLog
index 6d0e1944899aff915987f7beeb58c3f95c1f1db2..ec6b82f7953467d0bf1a43b54def6b6177b0d18d 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -4,6 +4,8 @@ Version 0.0.6 (2016-09-??)
 
     Rename remove_directives -> remove_directive_tokens.
 
+    Unescape strings.
+
 Version 0.0.5 (2016-09-07)
 
     Remove /*! directives */; and newlines after them.

diff --git a/README.txt b/README.txt
index 86b1b0513690252b34a648759d32039955bd8109..2e444746fcdebf7a61844fa24ed2e606970e4bba 100644 (file)
--- a/README.txt
+++ b/README.txt
@@ -5,7 +5,8 @@ License: GPL
 This is sqlconvert, a a library to perform SQL convertions. It uses
 sqlparse to parse SQL.
 
-The library is in its initial phase and currently cannot do much.
+The library is in the early stage of development and currently cannot do
+much.
 
 The first goal is to implemet mysq2sql, a script intended primarily to
 convert mysqldump output (especially with extended INSERT syntax) to

diff --git a/docs/mysql2sql.rst b/docs/mysql2sql.rst
index 9c1892d62a7054d88bf410624e5fb5144f25b1e9..12e574b3676702d30b42ac3f37c18b5bd43c5049 100644 (file)
--- a/docs/mysql2sql.rst
+++ b/docs/mysql2sql.rst
@@ -5,8 +5,9 @@ This is mysql2sql, a mysql to sql converter. It is primary intended to
 convert mysqldump (especially with extended INSERT syntax) to standard
 SQL to load at least to PostgreSQL or SQLite.
 
-The program is in its initial phase and currently cannot do much. It only
-removes /\*! directives \*/ and passes everything else unmodified.
+The program is in the early stage of development and currently cannot do much.
+It removes /\*! directives \*/, unescapes strings and passes everything else
+unmodified.
 
 
 .. highlight:: none

diff --git a/sqlconvert/process_mysql.py b/sqlconvert/process_mysql.py
index f91f516d2d9236f53dd320f5eb2a958545d54c9d..d72479e7111bfc0db29b12c7f6bdb1d8aa0561ed 100644 (file)
--- a/sqlconvert/process_mysql.py
+++ b/sqlconvert/process_mysql.py
@@ -49,6 +49,21 @@ def requote_names(token_list):
                 token.normalized = token.value = '"%s"' % value
 
 
+def unescape_strings(token_list):
+    """Unescape strings"""
+    for token in token_list.flatten():
+        if token.ttype is T.String.Single:
+            value = token.value
+            for orig, repl in (
+                ('\\"', '"'),
+                ("\\'", "''"),
+                ('\\\032', '\032'),
+            ):
+                value = value.replace(orig, repl)
+            token.normalized = token.value = value
+
+
 def process_statement(statement):
     remove_directive_tokens(statement)
     requote_names(statement)
+    unescape_strings(statement)

diff --git a/tests/test_tokens.py b/tests/test_tokens.py
index 24866548cfa3b9abbbbd77d2ef35803badacb5b8..de11485d1f70bcb7bbc042c52829cbc4d27f9c13 100755 (executable)
--- a/tests/test_tokens.py
+++ b/tests/test_tokens.py
@@ -5,8 +5,9 @@ import unittest
 from sqlparse import parse
 
 from sqlconvert.print_tokens import tlist2str
-from sqlconvert.process_mysql import remove_directive_tokens, requote_names, \
-        is_directive_statement, process_statement
+from sqlconvert.process_mysql import remove_directive_tokens, \
+        is_directive_statement, requote_names, unescape_strings, \
+        process_statement
 from tests import main
 
 
@@ -22,17 +23,11 @@ class TestTokens(unittest.TestCase):
         query = tlist2str(parsed)
         self.assertEqual(query, u"INSERT INTO test (1, 'тест')")
 
-    def test_requote(self):
-        parsed = parse("select * from `T`")[0]
-        requote_names(parsed)
-        query = tlist2str(parsed)
-        self.assertEqual(query, 'SELECT * FROM "T"')
-
     def test_directive(self):
         parsed = parse("select /*! test */ * from /* test */ `T`")[0]
         remove_directive_tokens(parsed)
         query = tlist2str(parsed)
-        self.assertEqual(query, 'SELECT * FROM /* test */ `T`')
+        self.assertEqual(query, u'SELECT * FROM /* test */ `T`')
 
     def test_directive_statement(self):
         parsed = parse("/*! test */ test ;")[0]
@@ -40,11 +35,23 @@ class TestTokens(unittest.TestCase):
         parsed = parse("/*! test */ ;")[0]
         self.assertTrue(is_directive_statement(parsed))
 
+    def test_requote(self):
+        parsed = parse("select * from `T`")[0]
+        requote_names(parsed)
+        query = tlist2str(parsed)
+        self.assertEqual(query, u'SELECT * FROM "T"')
+
+    def test_string(self):
+        parsed = parse("insert into test values ('\"test\\\"')")[0]
+        unescape_strings(parsed)
+        query = tlist2str(parsed)
+        self.assertEqual(query, u"INSERT INTO test VALUES ('\"test\"')")
+
     def test_process(self):
         parsed = parse("select /*! test */ * from /* test */ `T`")[0]
         process_statement(parsed)
         query = tlist2str(parsed)
-        self.assertEqual(query, 'SELECT * FROM /* test */ "T"')
+        self.assertEqual(query, u'SELECT * FROM /* test */ "T"')
 
 
 if __name__ == "__main__":