Resolve queries by rowid (nrow)

[datasette-pytables.git] / datasette_pytables / __init__.py

diff --git a/datasette_pytables/__init__.py b/datasette_pytables/__init__.py
index dc6823472e8abdf6190009648a83d8a6f79122f6..50180fdc5c35c1d6ae93d817080dabd6d824ed9d 100644 (file)
--- a/datasette_pytables/__init__.py
+++ b/datasette_pytables/__init__.py
@@ -1,4 +1,6 @@
 import tables
+import sqlparse
+from collections import OrderedDict
 
 _connector_type = 'pytables'
 
@@ -27,10 +29,96 @@ def inspect(path):
     h5file.close()
     return h5tables, views, _connector_type
 
+def _parse_sql(sql):
+    parsed = sqlparse.parse(sql)
+    stmt = parsed[0]
+    parsed_sql = {}
+    current_keyword = ""
+    for token in stmt.tokens:
+        if token.is_keyword:
+            if current_keyword in parsed_sql and parsed_sql[current_keyword] == '':
+                # Check composed keywords like 'order by'
+                del parsed_sql[current_keyword]
+                current_keyword += " " + str(token)
+            else:
+                current_keyword = str(token)
+            parsed_sql[current_keyword] = ""
+        elif type(token) is sqlparse.sql.Where:
+            parsed_sql['where'] = token
+        else:
+            if not token.is_whitespace:
+                parsed_sql[current_keyword] += str(token)
+    return parsed_sql
+
 class Connection:
     def __init__(self, path):
         self.path = path
         self.h5file = tables.open_file(path)
 
-    def execute(self, sql, params):
-        return [], False, []
+    def execute(self, sql, params=None, truncate=False):
+        rows = []
+        truncated = False
+        description = []
+
+        parsed_sql = _parse_sql(sql)
+        table = self.h5file.get_node(parsed_sql['from'][1:-1])
+        table_rows = []
+        fields = parsed_sql['select'].split(',')
+
+        # Use 'where' statement or get all the rows
+        if 'where' in parsed_sql:
+            query = ''
+            start = 0
+            end = table.nrows
+            for condition in parsed_sql['where'].get_sublists():
+                if str(condition) == '"rowid"=:p0':
+                    start = int(params['p0'])
+                    end = start + 1
+            if query:
+                table_rows = table.where(query, start, end)
+            else:
+                table_rows = table.iterrows(start, end)
+        else:
+            table_rows = table.iterrows()
+
+        # Prepare rows
+        if len(fields) == 1 and fields[0] == 'count(*)':
+            rows.append(Row({fields[0]: table.nrows}))
+        else:
+            for table_row in table_rows:
+                row = Row()
+                for field in fields:
+                    if field == 'rowid':
+                        row[field] = table_row.nrow
+                    elif field == '*':
+                        for col in table.colnames:
+                            value = table_row[col]
+                            if type(value) is bytes:
+                                value = value.decode('utf-8')
+                            row[col] = value
+                    else:
+                        row[field] = table_row[field]
+                rows.append(row)
+
+        # Prepare query description
+        for field in fields:
+            if field == '*':
+                for col in table.colnames:
+                    description.append((col,))
+            else:
+                description.append((field,))
+
+        if truncate:
+            return rows, truncated, tuple(description)
+        else:
+            return rows
+
+class Row(OrderedDict):
+    def __getitem__(self, label):
+        if type(label) is int:
+            return super(OrderedDict, self).__getitem__(list(self.keys())[label])
+        else:
+            return super(OrderedDict, self).__getitem__(label)
+
+    def __iter__(self):
+        return self.values().__iter__()