blobhash recursive

hash.py

#!/usr/bin/env python
def blobhash(str1):
    import hashlib
    hash1=hashlib.sha1("blob %d\0" % len(str1))
    hash1.update(str1)
    return hash1.hexdigest()

def walk(dir1,extension=".py"):
    import os
    for root, dirs, files in os.walk(dir1):
        for name in files:
            if name.lower().endswith(extension):
                filepath=os.path.join(root,name)
                h1=blobhash(open(filepath,'rb').read())
                print "%s %s" % (h1, filepath.replace('\\','/'))

def findblobs(hashes,ref='HEAD',stopat=None):
    firsts={}
    lasts={}
    import os
    if ref is not None:
        logcmd="git log %s --pretty=format:\"%%T %%H %%s\"" % ref
    else:
        logcmd="git reflog --pretty=format:\"%T %H %s\""
    mbfh=os.popen(logcmd)
    commit =None
    for cline in mbfh.read().split('\n'):
        if cline.strip():
            try:
                tree,commit,subject = cline.split(' ',2)
                lstr=os.popen("git ls-tree -r %s" % tree)
                for bline in lstr.read().split('\n'):
                    if bline.strip():
                        odata,opath = bline.split('\t',1)
                        oflags,otype,ohash = odata.split(' ',2)
                        for h1 in hashes:
                            if ohash.startswith(h1):
                                firsts[h1] = commit
                                if lasts.get(h1) is None:
                                    lasts[h1] = commit
            except ValueError:
                print cline
                            #print commit, subject, bline
                    
                lastcommit = commit
                lastsubject = subject
                lastbline = bline
            if all ((lasts.get(h1) is not None and firsts.get(h1) != commit) for h1 in hashes):
                #print 'finished'
                #print firsts,lasts
                break
            if stopat is True and all ((lasts.get(h1) is not None and \
                    firsts.get(h1) == commit) for h1 in hashes):
                break

            if stopat is not None and stopat is not True and \
                    commit.startswith(stopat):
                break
            lstr.close
    mbfh.close()
    return (dict((h1,(firsts.get(h1),lasts.get(h1))) for h1 in hashes),commit)

def findblob(h1,ref='HEAD'):
    """search through all commits and trees
returns the last commit in which the blob is present and the
previous commit in which the the blob was introduced.
This does all apearanches since the search is stopped afterwards"""
    first = None
    last = None
    import os
    mbfh=os.popen("git log %s --pretty=format:\"%%T %%h %%s\"" % ref)
    for cline in mbfh.read().split('\n'):
        tree,commit,subject = cline.split(' ',2)
        lstr=os.popen("git ls-tree -r %s" % tree)
        for bline in lstr.read().split('\n'):
            if bline.strip():
                odata,opath = bline.split('\t',1)
                oflags,otype,ohash = odata.split(' ',2)
                if ohash.startswith(h1):
                    first = commit
                    if last is None:
                        last = commit
                        print commit, subject, bline
                    
            lastcommit = commit
            lastsubject = subject
            lastbline = bline
        if last is not None and first != commit:
            print 'first appeard in '
            print lastcommit, lastsubject, lastbline
            return (first,last)
        lstr.close
    mbfh.close()

def refcount(ref):
    import os
    mbfh=os.popen("git --no-replace-objects rev-list --count %s" % ref)
    try:
        return int(mbfh.read().strip())
    except ValueError:
        pass

def refparseverify(ref):
    import os
    rpv=os.popen("git rev-parse --verify %s" % ref)
    ref=rpv.read().strip()
    if rpv.close() is None:
        return ref

def gitfiletype(ref):
    import os
    rpv=os.popen("git cat-file -t %s" % ref)
    ref=rpv.read().strip()
    if rpv.close() is None:
        return ref

def searchblob(str1,ref='HEAD',stopat=None):
    hashes=[]
    nonpresentblobs=[]
    names={}
    for line in str1.split('\n'):
        if line.strip():
            h1,name = line.strip().split(' ',1)
            if gitfiletype(h1) != 'blob':
                print '%s %s not in repo' % (h1,name)
            else:
                hashes.append(h1)
                if h1 not in names:
                    names[h1] = name
                else:
                    names[h1] +=','+name
    ref = refparseverify(ref)
    if ref is None:
        raise ValueError('invalid ref')
    d1,stoppedat=findblobs(hashes,ref=ref,stopat=stopat)
    notfound=[]
    for h1 in hashes:
        value=d1.get(h1,(None,None))
        if value == (None,None):
            notfound.append(h1)
        print '%s %s..%s %s' % (h1[:8],\
            refcount(value[0]),refcount(value[1]),names.get(h1,'?'))
    print 'searched %s..%s starting with %s' % \
            (refcount(stoppedat),refcount(ref),ref)
    #if len(notfound) > 0:
    #    print 'notfound ',notfound
    #    d2,stoppedat=findblobs(notfound,ref='--all',stopat=stopat)
    #    print '--all ',d2
    #    d3,stoppedat=findblobs(notfound,ref=None)
    #    print 'reflog ',d3

def findbaseforpath(patch):
    shorthashes=set()
    fullhashes=set()
    "index fec529f..4b20705 100644"
    for line in patch.split('\n'):
        sline=line.strip()
        if sline.startswith('index '):
            shorthash=sline[6:13]
            if shorthash != '0000000':
                shorthashes.add(shorthash)
                fullhash=refparseverify(shorthash)
                if fullhash is not None:
                    fullhashes.add(fullhash)
                else:
                    raise ValueError('Blob %s not present' % shorthash)
    print fullhashes
    print shorthashes
    fbresults,lastcommit= findblobs(tuple(fullhashes),ref='HEAD',\
            stopat=True)#'aaeba38e781103663e772e8757f0203af38498b9')
    print lastcommit

testdata="""cee38a2c1830bfde8b8d4dd46aeb41a64a71303f Draft.py
c9272acba2a0243a5f8eda6f57c04908a20de462 importDWG.py
00f845039587cd0ce43be796b610b0263186bc57 importDXF.py
f9f48bd8ad766841cb90049ab5c4b7c257909378 importSVG.py
6c2f613bcfc0379ac1f9bb90a01efd24df3f33c1 TestDraft.py
"""
if __name__ == '__main__':
    import sys
    if len(sys.argv):
        findbaseforpath(sys.stdin.read())

    #for dir1 in (sys.argv[1:] or (".")):
    #    walk(dir1)
    #print findblob(*sys.argv[1:])
    #for key,value in findblobs(sys.argv[1].split(','),sys.argv[2]).iteritems():
    #    print '%s %s..%s' % (key,refcount(value[0]),refcount(value[1]))

    #searchblob(testdata,ref='origin/master',\
    #    stopat='691fd1128672c8bd472cece87c9e9d07b71d6fee')