issue mhagger#18: Use internal lines instead of plain text in checkou…

…t_db.

As the return value of RCSStream.invert_diff() is not to be applied
for flat text content but for internal logical lines in RCSStream,
which may contain some unterminated logical lines at any position,
so we should use the content of internal logical lines in RCSStream
as base text which is used to get the content of the next newer revision.
With this commit, it is implemented by splitting the checkout() method
in TextRecord, checkout() method for external use for as in the past
and checkout_as_lines() method for internal use.

* cvs2svn_lib/checkout_internal.py
  (TextRecord.checout_as_lines): New method. Replacement of checkout()
    method but returns internal lines in RCSStream instead of a plain
    text.
  (TextRecord.checkout):
    Use checkout_as_lines() for default implementation.
  (FullTextRecord.checkout, DeltaTextRecord.checkout):
    Removed to use default implementation.
  (FullTextRecord.checkout_as_lines):
    New method. Just same logic as the past checkout() method.
  (DeltaTextRecord.checkout_as_lines):
    New method. Just same logic as the past checkout() method but
    uses internal lines in rcs_stream instead of its text.
  (_Sink.set_revision_info):
    Record the internal lines in rcs_stream instead of its text
    at revision 1.1.

* cvs2svn_lib/rcs_stream.py
  (RCSStream.__init__):
    Allow to set lines directly in addition to a text.
  (RCSStream.get_lines):
    New method.

cvs2svn_lib/checkout_internal.py

@@ -134,15 +134,22 @@ def decrement_refcount(self, text_record_db):
     if self.refcount == 0:
       text_record_db.discard(self.id)
 
-  def checkout(self, text_record_db):
+  def checkout_as_lines(self, text_record_db):
     """Workhorse of the checkout process.
 
-    Return the text for this revision, decrement our reference count,
-    and update the databases depending on whether there will be future
-    checkouts."""
+    Return the text for this revision as list of logical lines,
+    decrement our reference count, and update the databases depending
+    on whether there will be future checkouts."""
 
     raise NotImplementedError()
 
+  def checkout(self, text_record_db):
+    """Return the text for this revision.
+
+     Just same as checkout_as_lines() but returns text as flat text string."""
+
+    return "".join(self.checkout_as_lines(text_record_db))
+
   def free(self, text_record_db):
     """This instance will never again be checked out; free it.
 
@@ -168,10 +175,10 @@ def __getstate__(self):
   def __setstate__(self, state):
     (self.id, self.refcount,) = state
 
-  def checkout(self, text_record_db):
-    text = text_record_db.delta_db[self.id]
+  def checkout_as_lines(self, text_record_db):
+    lines = text_record_db.delta_db[self.id]
     self.decrement_refcount(text_record_db)
-    return text
+    return lines
 
   def free(self, text_record_db):
     del text_record_db.delta_db[self.id]
@@ -205,12 +212,12 @@ def __setstate__(self, state):
   def increment_dependency_refcounts(self, text_record_db):
     text_record_db[self.pred_id].refcount += 1
 
-  def checkout(self, text_record_db):
-    base_text = text_record_db[self.pred_id].checkout(text_record_db)
-    rcs_stream = RCSStream(base_text)
+  def checkout_as_lines(self, text_record_db):
+    base_lines = text_record_db[self.pred_id].checkout_as_lines(text_record_db)
+    rcs_stream = RCSStream(base_lines)
     delta_text = text_record_db.delta_db[self.id]
     rcs_stream.apply_diff(delta_text)
-    text = rcs_stream.get_text()
+    lines = rcs_stream.get_lines()
     del rcs_stream
     self.refcount -= 1
     if self.refcount == 0:
@@ -220,11 +227,11 @@ def checkout(self, text_record_db):
       del text_record_db[self.id]
     else:
       # Store a new CheckedOutTextRecord in place of ourselves:
-      text_record_db.checkout_db['%x' % self.id] = text
+      text_record_db.checkout_db['%x' % self.id] = lines
       new_text_record = CheckedOutTextRecord(self.id)
       new_text_record.refcount = self.refcount
       text_record_db.replace(new_text_record)
-    return text
+    return lines
 
   def free(self, text_record_db):
     del text_record_db.delta_db[self.id]
@@ -251,10 +258,10 @@ def __getstate__(self):
   def __setstate__(self, state):
     (self.id, self.refcount,) = state
 
-  def checkout(self, text_record_db):
-    text = text_record_db.checkout_db['%x' % self.id]
+  def checkout_as_lines(self, text_record_db):
+    lines = text_record_db.checkout_db['%x' % self.id]
     self.decrement_refcount(text_record_db)
-    return text
+    return lines
 
   def free(self, text_record_db):
     del text_record_db.checkout_db['%x' % self.id]
@@ -533,7 +540,7 @@ def set_revision_info(self, revision, log, text):
         # This is revision 1.1.  Write its fulltext:
         text_record = FullTextRecord(cvs_rev_id)
         self.revision_collector._writeout(
-            text_record, self._rcs_stream.get_text()
+            text_record, self._rcs_stream.get_lines()
             )
 
         # There will be no more trunk revisions delivered, so free the

cvs2svn_lib/rcs_stream.py

@@ -174,13 +174,21 @@ class RCSStream:
   def __init__(self, text):
     """Instantiate and initialize the file content with TEXT."""
 
-    self.set_text(text)
+    if isinstance(text, bytes):
+      self.set_text(text)
+    else:
+      self.set_lines(text)
 
   def get_text(self):
     """Return the current file content."""
 
     return "".join(self._lines)
 
+  def get_lines(self):
+    """Return the current file content as list of logical lines."""
+
+    return self._lines
+
   def set_lines(self, lines):
     """Set the current contents to the specified LINES.