X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=sqlconvert%2Fprocess_mysql.py;h=e99ea79edc2816c57787f0578589a38bacea6173;hb=843981ca5ac41be09212f3aa427500580a6e9f5c;hp=b11eccb38f4376f526a7b8b3130e618f7ff58c8f;hpb=dbc9220a2b29725f94637607f8d8b00c762deb67;p=sqlconvert.git

diff --git a/sqlconvert/process_mysql.py b/sqlconvert/process_mysql.py
index b11eccb..e99ea79 100644
--- a/sqlconvert/process_mysql.py
+++ b/sqlconvert/process_mysql.py
@@ -1,11 +1,47 @@
 
-from sqlparse.tokens import Name
+from sqlparse.sql import Comment, Function, Identifier, Parenthesis, \
+    Statement, Token
+from sqlparse import tokens as T
+from .process_tokens import escape_strings, is_comment_or_space
+
+
+def _is_directive_token(token):
+    if isinstance(token, Comment):
+        subtokens = token.tokens
+        if subtokens:
+            comment = subtokens[0]
+            if comment.ttype is T.Comment.Multiline and \
+                    comment.value.startswith('/*!'):
+                return True
+    return False
+
+
+def is_directive_statement(statement):
+    tokens = statement.tokens
+    if not _is_directive_token(tokens[0]):
+        return False
+    if tokens[-1].ttype is not T.Punctuation or tokens[-1].value != ';':
+        return False
+    for token in tokens[1:-1]:
+        if token.ttype not in (T.Newline, T.Whitespace):
+            return False
+    return True
+
+
+def remove_directive_tokens(statement):
+    """Remove /\*! directives \*/ from the first-level"""
+    new_tokens = []
+    for token in statement.tokens:
+        if _is_directive_token(token):
+            continue
+        new_tokens.append(token)
+    statement.tokens = new_tokens
 
 
 def requote_names(token_list):
     """Remove backticks, quote non-lowercase identifiers"""
     for token in token_list.flatten():
-        if token.ttype is Name:
+        if token.ttype is T.Name:
             value = token.value
             if (value[0] == "`") and (value[-1] == "`"):
                 value = value[1:-1]
@@ -13,3 +49,109 @@ def requote_names(token_list):
                 token.normalized = token.value = value
             else:
                 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 (
+                ('\\"', '"'),
+                ("\\'", "'"),
+                ("''", "'"),
+                ('\\b', '\b'),
+                ('\\n', '\n'),
+                ('\\r', '\r'),
+                ('\\t', '\t'),
+                ('\\\032', '\032'),
+                ('\\\\', '\\'),
+            ):
+                value = value.replace(orig, repl)
+            token.normalized = token.value = value
+
+
+def get_DML_type(statement):
+    for token in statement.tokens:
+        if is_comment_or_space(token):
+            continue
+        if (token.ttype is T.DML):
+            return token.normalized
+        break
+    raise ValueError("Not a DML statement")
+
+
+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 (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, dbname)
+    try:
+        is_insert = get_DML_type(statement) == 'INSERT'
+    except ValueError:
+        is_insert = False
+    if is_insert:
+        for statement in split_ext_insert(statement):
+            yield statement
+    else:
+        yield statement