X-Git-Url: https://git.phdru.name/?p=sqlconvert.git;a=blobdiff_plain;f=sqlconvert%2Fprocess_mysql.py;h=ad4d422b4c13fa0ca2f1821009038b0e512c9214;hp=9e342c2e04ec7c5ad4ba1796731932320762e5f3;hb=HEAD;hpb=d33677c7a57a955f59e46f2abe03e9f5a192d883

diff --git a/sqlconvert/process_mysql.py b/sqlconvert/process_mysql.py
index 9e342c2..c89d6d1 100644
--- a/sqlconvert/process_mysql.py
+++ b/sqlconvert/process_mysql.py
@@ -1,5 +1,6 @@
 
-from sqlparse.sql import Comment
+from sqlparse.sql import Comment, Function, Identifier, Parenthesis, \
+    Statement, Token, Values
 from sqlparse import tokens as T
 from .process_tokens import escape_strings, is_comment_or_space
 
@@ -70,17 +71,95 @@ def unescape_strings(token_list):
             token.normalized = token.value = value
 
 
-def is_insert(statement):
+def get_DML_type(statement):
     for token in statement.tokens:
         if is_comment_or_space(token):
             continue
-        return (token.ttype is T.DML) and (token.normalized == 'INSERT')
+        if (token.ttype is T.DML):
+            return token.normalized
+        break
+    raise ValueError("Not a DML statement")
 
 
-def process_statement(statement, quoting_style='sqlite'):
+def split_ext_insert(statement):
+    """Split extended INSERT into multiple standard INSERTs"""
+    insert_tokens = []
+    values_tokens = []
+    end_tokens = []
+    expected = 'INSERT'
+    for token in statement.tokens:
+        if is_comment_or_space(token):
+            if expected == 'END':
+                end_tokens.append(token)
+            else:
+                insert_tokens.append(token)
+            continue
+        elif expected == 'INSERT':
+            if (token.ttype is T.DML) and (token.normalized == 'INSERT'):
+                insert_tokens.append(token)
+                expected = 'INTO'
+                continue
+        elif expected == 'INTO':
+            if (token.ttype is T.Keyword) and (token.normalized == 'INTO'):
+                insert_tokens.append(token)
+                expected = 'TABLE_NAME'
+                continue
+        elif expected == 'TABLE_NAME':
+            if isinstance(token, (Function, Identifier)):
+                insert_tokens.append(token)
+                expected = 'VALUES'
+                continue
+        elif expected == 'VALUES':
+            if isinstance(token, Values):
+                for subtoken in token.tokens:
+                    if isinstance(subtoken, Parenthesis):
+                        values_tokens.append(subtoken)
+                insert_tokens.append(Token(T.Keyword, 'VALUES'))
+                insert_tokens.append(Token(T.Whitespace, ' '))
+                expected = 'VALUES_OR_SEMICOLON'
+                continue
+            if (token.ttype is T.Keyword) and (token.normalized == 'VALUES'):
+                insert_tokens.append(token)
+                expected = 'VALUES_OR_SEMICOLON'
+                continue
+        elif expected == 'VALUES_OR_SEMICOLON':
+            if isinstance(token, Parenthesis):
+                values_tokens.append(token)
+                continue
+            elif token.ttype is T.Punctuation:
+                if token.value == ',':
+                    continue
+                elif token.value == ';':
+                    end_tokens.append(token)
+                    expected = 'END'
+                    continue
+        raise ValueError(
+            'SQL syntax error: expected "%s", got %s "%s"' % (
+                expected, token.ttype, token.normalized))
+    new_line = Token(T.Newline, '\n')
+    new_lines = [new_line]  # Insert newlines between split statements
+    for i, values in enumerate(values_tokens):
+        if i == len(values_tokens) - 1:  # Last but one statement
+            # Insert newlines only between split statements but not after
+            new_lines = []
+        # The statement sets `parent` attribute of the every token to self
+        # but we don't care.
+        statement = Statement(insert_tokens + [values] +
+                              end_tokens + new_lines)
+        yield statement
+
+
+def process_statement(statement, dbname='sqlite'):
     requote_names(statement)
     unescape_strings(statement)
     remove_directive_tokens(statement)
-    escape_strings(statement, quoting_style)
-    yield statement
-    return
+    escape_strings(statement, dbname)
+    try:
+        dml_type = get_DML_type(statement)
+    except ValueError:
+        dml_type = 'UNKNOWN'
+    if dml_type == 'INSERT':
+        for statement in split_ext_insert(statement):
+            yield statement
+    else:
+        yield statement