From eb058b6eb526d1b1f486c45cad9858f382200757 Mon Sep 17 00:00:00 2001
From: Nick Pellegrino <npellegrino@dimagi.com>
Date: Mon, 24 Dec 2018 11:33:28 -0500
Subject: [PATCH 1/4] use properties to determine database type

---
 commcare_export/writers.py | 18 +++++++++++++++---
 1 file changed, 15 insertions(+), 3 deletions(-)

diff --git a/commcare_export/writers.py b/commcare_export/writers.py
index c90ddc75..73ce6ed2 100644
--- a/commcare_export/writers.py
+++ b/commcare_export/writers.py
@@ -269,15 +269,27 @@ def __enter__(self):
     def __exit__(self, exc_type, exc_val, exc_tb):
         self.connection.close()
 
+    @property
+    def is_postgres(self):
+        return 'postgres' in self.db_url
+
+    @property
+    def is_mysql(self):
+        return 'mysql' in self.db_url
+
+    @property
+    def is_mssql(self):
+        return 'mssql' in self.db_url
+
     @property
     def max_column_length(self):
-        if 'postgres' in self.db_url:
+        if self.is_postgres:
             # https://www.postgresql.org/docs/current/static/sql-syntax-lexical.html#SQL-SYNTAX-IDENTIFIERS
             return 63
-        if 'mysql' in self.db_url:
+        if self.is_mysql:
             # https://dev.mysql.com/doc/refman/8.0/en/identifiers.html
             return 64
-        if 'mssql' in self.db_url:
+        if self.is_mssql:
             # https://docs.microsoft.com/en-us/sql/relational-databases/databases/database-identifiers?view=sql-server-2017
             return 128
         raise Exception("Unknown database dialect: {}".format(self.db_url))

From a5a422136b39563f602d7bace23496d0b3885ce5 Mon Sep 17 00:00:00 2001
From: Nick Pellegrino <npellegrino@dimagi.com>
Date: Mon, 24 Dec 2018 11:34:13 -0500
Subject: [PATCH 2/4] never change primary key column type

---
 commcare_export/writers.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/commcare_export/writers.py b/commcare_export/writers.py
index 73ce6ed2..30bd989b 100644
--- a/commcare_export/writers.py
+++ b/commcare_export/writers.py
@@ -455,7 +455,7 @@ def get_cols():
                 self.metadata.clear()
                 self.metadata.reflect()
                 columns = get_cols()
-            else:
+            elif not columns[column].primary_key:
                 current_ty = columns[column].type
                 new_type = None
                 if self.strict_types:

From c19d9812a22d9c92625f66ea800f2f865c91efaf Mon Sep 17 00:00:00 2001
From: Nick Pellegrino <npellegrino@dimagi.com>
Date: Mon, 24 Dec 2018 11:35:36 -0500
Subject: [PATCH 3/4] for string data, pick a column type appropriate for that
 database

---
 commcare_export/writers.py | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/commcare_export/writers.py b/commcare_export/writers.py
index 30bd989b..07478ac4 100644
--- a/commcare_export/writers.py
+++ b/commcare_export/writers.py
@@ -340,14 +340,20 @@ def best_type_for(self, val):
         if isinstance(val, int):
             return sqlalchemy.Integer()
         elif isinstance(val, six.string_types):
-            # Notes on the conversions between various string types:
-            # 1. PostgreSQL is the best; you can use TEXT everywhere and it works like a charm.
-            # 2. MySQL cannot build an index on TEXT due to the lack of a field length, so we
-            #    try to use VARCHAR when possible.
-            if len(val) < self.MAX_VARCHAR_LEN: # FIXME: Is 255 an interesting cutoff?
-                return sqlalchemy.Unicode( max(len(val), self.MIN_VARCHAR_LEN), collation=self.collation)
-            else:
+            if self.is_postgres:
+                # PostgreSQL is the best; you can use TEXT everywhere and it works like a charm.
                 return sqlalchemy.UnicodeText(collation=self.collation)
+            elif self.is_mysql:
+                # MySQL cannot build an index on TEXT due to the lack of a field length, so we
+                # try to use VARCHAR when possible.
+                if len(val) < self.MAX_VARCHAR_LEN: # FIXME: Is 255 an interesting cutoff?
+                    return sqlalchemy.Unicode( max(len(val), self.MIN_VARCHAR_LEN), collation=self.collation)
+                else:
+                    return sqlalchemy.UnicodeText(collation=self.collation)
+            elif self.is_mssql:
+                return sqlalchemy.NVARCHAR(collation=self.collation)
+            else:
+                raise Exception("Unknown database dialect: {}".format(self.db_url))
         else:
             # We do not have a name for "bottom" in SQL aka the type whose least upper bound
             # with any other type is the other type.

From 876c46d69fc3aab849260950cfc03020175da0d5 Mon Sep 17 00:00:00 2001
From: Nick Pellegrino <npellegrino@dimagi.com>
Date: Mon, 24 Dec 2018 11:35:59 -0500
Subject: [PATCH 4/4] lint

---
 commcare_export/writers.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/commcare_export/writers.py b/commcare_export/writers.py
index 07478ac4..b026431b 100644
--- a/commcare_export/writers.py
+++ b/commcare_export/writers.py
@@ -346,8 +346,8 @@ def best_type_for(self, val):
             elif self.is_mysql:
                 # MySQL cannot build an index on TEXT due to the lack of a field length, so we
                 # try to use VARCHAR when possible.
-                if len(val) < self.MAX_VARCHAR_LEN: # FIXME: Is 255 an interesting cutoff?
-                    return sqlalchemy.Unicode( max(len(val), self.MIN_VARCHAR_LEN), collation=self.collation)
+                if len(val) < self.MAX_VARCHAR_LEN:  # FIXME: Is 255 an interesting cutoff?
+                    return sqlalchemy.Unicode(max(len(val), self.MIN_VARCHAR_LEN), collation=self.collation)
                 else:
                     return sqlalchemy.UnicodeText(collation=self.collation)
             elif self.is_mssql: