From d94bf72dd7269e126990bb28518c16f68cc5d6cb Mon Sep 17 00:00:00 2001
From: Tony Duckles <tony@nynim.org>
Date: Sat, 17 Mar 2012 00:40:18 -0500
Subject: [PATCH] WIP on diff-summarize

---
 svn2svn/run/svn2svn.py |  3 +++
 svn2svn/svnclient.py   | 30 ++++++++++++++++++++++++++++++
 2 files changed, 33 insertions(+)

diff --git a/svn2svn/run/svn2svn.py b/svn2svn/run/svn2svn.py
index 54e7368..1c88433 100644
--- a/svn2svn/run/svn2svn.py
+++ b/svn2svn/run/svn2svn.py
@@ -533,6 +533,9 @@ def process_svn_log_entry(log_entry, ancestors, commit_paths, prefix = ""):
     source_url = log_entry['url']
     source_base = source_url[len(source_repos_url):]  # e.g. '/trunk'
     ui.status(prefix + ">> process_svn_log_entry: %s", source_url+"@"+str(source_rev), level=ui.DEBUG, color='GREEN')
+    diff_paths = svnclient.get_diff_summarize(source_repos_url, source_url, source_rev)
+    for d in diff_paths:
+        print "%s,%s %s" % (d['item'], d['props'], d['path'])
     for d in log_entry['changed_paths']:
         # Get the full path for this changed_path
         # e.g. '/branches/bug123/projectA/file1.txt'
diff --git a/svn2svn/svnclient.py b/svn2svn/svnclient.py
index efb3212..bc4e15a 100644
--- a/svn2svn/svnclient.py
+++ b/svn2svn/svnclient.py
@@ -453,3 +453,33 @@ def get_all_props(svn_url_or_wc, rev_number=None):
         d = get_prop_value(svn_url_or_wc, prop_name, rev_number)
         l[d['name']] = d['value']
     return l
+
+def parse_diff_summarize(svn_repos_url, svn_url_or_wc, xml_string):
+    """
+    Parse the XML output from an "svn diff -c REV --xml --summarize" command
+    and extract useful info as list of dicts (one per path entry).
+    """
+    l = []
+    xml_string = strip_forbidden_xml_chars(xml_string)
+    tree = ET.fromstring(xml_string)
+    paths = []
+    for path in tree.findall('.//paths/path'):
+        paths.append({
+            'path': path.text[len(svn_repos_url):],
+            'kind': path.get('kind'),
+            'props': path.get('props'),
+            'item': path.get('item'),
+        })
+    # Sort paths (i.e. into hierarchical order), so that process_svn_log_entry()
+    # can process actions in depth-first order.
+    return sorted(paths, key=operator.itemgetter('path'))
+
+def get_diff_summarize(svn_repos_url, svn_url_or_wc, rev_number):
+    args = ['diff', '--xml', '--summarize', '--notice-ancestry']
+    url = str(svn_url_or_wc)
+    args += ['-c', rev_number]
+    if not "@" in svn_url_or_wc:
+        url = "%s@%s" % (svn_url_or_wc, str(rev_number))
+    args += [url]
+    xml_string = run_svn(args)
+    return parse_diff_summarize(svn_repos_url, svn_url_or_wc, xml_string)
-- 
2.49.0