Merge branch 'jn/gitweb-patch'
[gitweb.git] / contrib / fast-import / git-p4
index a85a7b2a583ee9270fc2d765ec8c8c6e9d6b5e32..e710219ca52af3c90c28bb90273d062a26543864 100755 (executable)
@@ -8,12 +8,10 @@
 # License: MIT <http://www.opensource.org/licenses/mit-license.php>
 #
 
-import optparse, sys, os, marshal, popen2, subprocess, shelve
-import tempfile, getopt, sha, os.path, time, platform
+import optparse, sys, os, marshal, subprocess, shelve
+import tempfile, getopt, os.path, time, platform
 import re
 
-from sets import Set;
-
 verbose = False
 
 
@@ -201,7 +199,7 @@ def isModeExec(mode):
 def isModeExecChanged(src_mode, dst_mode):
     return isModeExec(src_mode) != isModeExec(dst_mode)
 
-def p4CmdList(cmd, stdin=None, stdin_mode='w+b'):
+def p4CmdList(cmd, stdin=None, stdin_mode='w+b', cb=None):
     cmd = p4_build_cmd("-G %s" % (cmd))
     if verbose:
         sys.stderr.write("Opening pipe: %s\n" % cmd)
@@ -224,7 +222,10 @@ def p4CmdList(cmd, stdin=None, stdin_mode='w+b'):
     try:
         while True:
             entry = marshal.load(p4.stdout)
-            result.append(entry)
+           if cb is not None:
+               cb(entry)
+           else:
+               result.append(entry)
     except EOFError:
         pass
     exitCode = p4.wait()
@@ -442,13 +443,14 @@ def p4ChangesForPaths(depotPaths, changeRange):
     output = p4_read_pipe_lines("changes " + ' '.join (["%s...%s" % (p, changeRange)
                                                         for p in depotPaths]))
 
-    changes = []
+    changes = {}
     for line in output:
-        changeNum = line.split(" ")[1]
-        changes.append(int(changeNum))
+       changeNum = int(line.split(" ")[1])
+       changes[changeNum] = True
 
-    changes.sort()
-    return changes
+    changelist = changes.keys()
+    changelist.sort()
+    return changelist
 
 class Command:
     def __init__(self):
@@ -860,8 +862,8 @@ class P4Sync(Command):
 
         self.usage += " //depot/path[@revRange]"
         self.silent = False
-        self.createdBranches = Set()
-        self.committedChanges = Set()
+        self.createdBranches = set()
+        self.committedChanges = set()
         self.branch = ""
         self.detectBranches = False
         self.detectLabels = False
@@ -949,10 +951,84 @@ class P4Sync(Command):
 
         return branches
 
-    ## Should move this out, doesn't use SELF.
-    def readP4Files(self, files):
+    # output one file from the P4 stream
+    # - helper for streamP4Files
+
+    def streamOneP4File(self, file, contents):
+       if file["type"] == "apple":
+           print "\nfile %s is a strange apple file that forks. Ignoring" % \
+               file['depotFile']
+           return
+
+        relPath = self.stripRepoPath(file['depotFile'], self.branchPrefixes)
+        if verbose:
+            sys.stderr.write("%s\n" % relPath)
+
+        mode = "644"
+        if isP4Exec(file["type"]):
+            mode = "755"
+        elif file["type"] == "symlink":
+            mode = "120000"
+            # p4 print on a symlink contains "target\n", so strip it off
+            last = contents.pop()
+            last = last[:-1]
+            contents.append(last)
+
+        if self.isWindows and file["type"].endswith("text"):
+            mangled = []
+            for data in contents:
+                data = data.replace("\r\n", "\n")
+                mangled.append(data)
+            contents = mangled
+
+        if file['type'] in ('text+ko', 'unicode+ko', 'binary+ko'):
+            contents = map(lambda text: re.sub(r'(?i)\$(Id|Header):[^$]*\$',r'$\1$', text), contents)
+        elif file['type'] in ('text+k', 'ktext', 'kxtext', 'unicode+k', 'binary+k'):
+            contents = map(lambda text: re.sub(r'\$(Id|Header|Author|Date|DateTime|Change|File|Revision):[^$\n]*\$',r'$\1$', text), contents)
+
+        self.gitStream.write("M %s inline %s\n" % (mode, relPath))
+
+        # total length...
+        length = 0
+        for d in contents:
+            length = length + len(d)
+
+        self.gitStream.write("data %d\n" % length)
+        for d in contents:
+            self.gitStream.write(d)
+        self.gitStream.write("\n")
+
+    def streamOneP4Deletion(self, file):
+        relPath = self.stripRepoPath(file['path'], self.branchPrefixes)
+        if verbose:
+            sys.stderr.write("delete %s\n" % relPath)
+        self.gitStream.write("D %s\n" % relPath)
+
+    # handle another chunk of streaming data
+    def streamP4FilesCb(self, marshalled):
+
+       if marshalled.has_key('depotFile') and self.stream_have_file_info:
+           # start of a new file - output the old one first
+           self.streamOneP4File(self.stream_file, self.stream_contents)
+           self.stream_file = {}
+           self.stream_contents = []
+           self.stream_have_file_info = False
+
+       # pick up the new file information... for the
+       # 'data' field we need to append to our array
+       for k in marshalled.keys():
+           if k == 'data':
+               self.stream_contents.append(marshalled['data'])
+           else:
+               self.stream_file[k] = marshalled[k]
+
+       self.stream_have_file_info = True
+
+    # Stream directly from "p4 files" into "git fast-import"
+    def streamP4Files(self, files):
         filesForCommit = []
         filesToRead = []
+        filesToDelete = []
 
         for f in files:
             includeFile = True
@@ -966,50 +1042,35 @@ class P4Sync(Command):
                 filesForCommit.append(f)
                 if f['action'] not in ('delete', 'purge'):
                     filesToRead.append(f)
+                else:
+                    filesToDelete.append(f)
 
-        filedata = []
-        if len(filesToRead) > 0:
-            filedata = p4CmdList('-x - print',
-                                 stdin='\n'.join(['%s#%s' % (f['path'], f['rev'])
-                                                  for f in filesToRead]),
-                                 stdin_mode='w+')
-
-            if "p4ExitCode" in filedata[0]:
-                die("Problems executing p4. Error: [%d]."
-                    % (filedata[0]['p4ExitCode']));
-
-        j = 0;
-        contents = {}
-        while j < len(filedata):
-            stat = filedata[j]
-            j += 1
-            text = ''
-            while j < len(filedata) and filedata[j]['code'] in ('text', 'unicode', 'binary'):
-                text += filedata[j]['data']
-                del filedata[j]['data']
-                j += 1
-
-            if not stat.has_key('depotFile'):
-                sys.stderr.write("p4 print fails with: %s\n" % repr(stat))
-                continue
+        # deleted files...
+        for f in filesToDelete:
+            self.streamOneP4Deletion(f)
 
-            if stat['type'] in ('text+ko', 'unicode+ko', 'binary+ko'):
-                text = re.sub(r'(?i)\$(Id|Header):[^$]*\$',r'$\1$', text)
-            elif stat['type'] in ('text+k', 'ktext', 'kxtext', 'unicode+k', 'binary+k'):
-                text = re.sub(r'\$(Id|Header|Author|Date|DateTime|Change|File|Revision):[^$\n]*\$',r'$\1$', text)
+        if len(filesToRead) > 0:
+            self.stream_file = {}
+            self.stream_contents = []
+            self.stream_have_file_info = False
 
-            contents[stat['depotFile']] = text
+           # curry self argument
+           def streamP4FilesCbSelf(entry):
+               self.streamP4FilesCb(entry)
 
-        for f in filesForCommit:
-            path = f['path']
-            if contents.has_key(path):
-                f['data'] = contents[path]
+           p4CmdList("-x - print",
+               '\n'.join(['%s#%s' % (f['path'], f['rev'])
+                                                  for f in filesToRead]),
+               cb=streamP4FilesCbSelf)
 
-        return filesForCommit
+            # do the last chunk
+            if self.stream_file.has_key('depotFile'):
+                self.streamOneP4File(self.stream_file, self.stream_contents)
 
     def commit(self, details, files, branch, branchPrefixes, parent = ""):
         epoch = details["time"]
         author = details["user"]
+       self.branchPrefixes = branchPrefixes
 
         if self.verbose:
             print "commit into %s" % branch
@@ -1022,7 +1083,6 @@ class P4Sync(Command):
                 new_files.append (f)
             else:
                 sys.stderr.write("Ignoring file outside of prefix: %s\n" % path)
-        files = self.readP4Files(new_files)
 
         self.gitStream.write("commit %s\n" % branch)
 #        gitStream.write("mark :%s\n" % details["change"])
@@ -1050,33 +1110,7 @@ class P4Sync(Command):
                 print "parent %s" % parent
             self.gitStream.write("from %s\n" % parent)
 
-        for file in files:
-            if file["type"] == "apple":
-                print "\nfile %s is a strange apple file that forks. Ignoring!" % file['path']
-                continue
-
-            relPath = self.stripRepoPath(file['path'], branchPrefixes)
-            if file["action"] in ("delete", "purge"):
-                self.gitStream.write("D %s\n" % relPath)
-            else:
-                data = file['data']
-
-                mode = "644"
-                if isP4Exec(file["type"]):
-                    mode = "755"
-                elif file["type"] == "symlink":
-                    mode = "120000"
-                    # p4 print on a symlink contains "target\n", so strip it off
-                    data = data[:-1]
-
-                if self.isWindows and file["type"].endswith("text"):
-                    data = data.replace("\r\n", "\n")
-
-                self.gitStream.write("M %s inline %s\n" % (mode, relPath))
-                self.gitStream.write("data %s\n" % len(data))
-                self.gitStream.write(data)
-                self.gitStream.write("\n")
-
+        self.streamP4Files(new_files)
         self.gitStream.write("\n")
 
         change = int(details["change"])
@@ -1141,7 +1175,7 @@ class P4Sync(Command):
 
         s = ''
         for (key, val) in self.users.items():
-            s += "%s\t%s\n" % (key, val)
+           s += "%s\t%s\n" % (key.expandtabs(1), val.expandtabs(1))
 
         open(self.getUserCacheFilename(), "wb").write(s)
         self.userMapFromPerforceServer = True
@@ -1626,7 +1660,7 @@ class P4Sync(Command):
 
             if len(self.changesFile) > 0:
                 output = open(self.changesFile).readlines()
-                changeSet = Set()
+                changeSet = set()
                 for line in output:
                     changeSet.add(int(line))