#!/usr/bin/env python2.2

# get_diff_helper.py, Version 1.0, Raimar Falke

import sys,re,time,pwd,os, string

def log(str):
    sys.stderr.write(str+"\n")    
    
def parse_date(date):
    old=list(time.strptime(date, "%Y/%m/%d %H:%M:%S"))
    old[8]=0
    result=time.mktime(old)
    result=result-time.timezone
    if 0:
        new=list(time.gmtime(result))
        if old!=new:
            print old,new
            assert 0
    return result
        
def parse(branch,name,read_entries):
    lines=open(name).readlines()
    last_line=lines[-1]
    assert last_line=='=============================================================================\n',repr(last_line)
    mo=re.search("(?s)(?m)^symbolic names:(.*)keyword substitution","".join(lines))
    assert mo
    symbolic_names={}
    if mo.group(1).strip():
      for tag_name,rev in map(lambda x:x.strip().split(":"),mo.group(1).strip().split("\n")):
          rev=map(int,rev.strip().split("."))

          # the others are just tags
          if len(rev)>2 and rev[2]==0:
              # 1.19.0.2 -> 1.19.2
              assert len(rev)==4
              rev[2]=2
              del rev[3]
              symbolic_names[tag_name]=rev

    mo=re.search("Working file: (.*)$",lines[2])
    filename=mo.group(1)
    filename=re.sub("^\./","",filename)
    revisions="".join(lines[:-1]).split("\n----------------------------\n")[1:]
    tmp_entries=[]
    for revision in revisions:
        lines=revision.split("\n")
        entry={}
        entry["filename"]=filename
        mo=re.search("^revision (.*)$",lines[0])
        entry["revision"]=mo.group(1)
        entry["num_revision"]=map(int,entry["revision"].split("."))
        for j in lines[1].split("; "):
            key,value=map(lambda x:x.strip(), j.split(": "))
            entry[key]=value
        entry["date"]=parse_date(entry["date"])
        entry["msg"]="\n".join(lines[2:])
        #print entry
        tmp_entries.append(entry)

    tmp_entries.sort(lambda x,y: cmp(x["num_revision"],y["num_revision"]))
    #print "branch=%s name=%s branches=%s"%(branch,name,symbolic_names)
    for entry in tmp_entries:
        add=0
        if branch=="HEAD":
            if len(entry["num_revision"])==2:
                add=1
        elif not symbolic_names.has_key(branch):
            pass # this file was added later
        else:
            if entry["num_revision"] < symbolic_names[branch] or \
               (len(entry["num_revision"])>2 and \
               entry["num_revision"][0:3]==symbolic_names[branch]):
                add=1
        #print "  num_revision=%s add=%d"%(entry["num_revision"],add)
        if add:
            read_entries.append(entry)

def merge(entries):
    result=[]
    for e in entries:
        new_commit=1
        if len(result)>0:
            same_author=e["author"]==result[-1]["author"]
            same_msg=e["msg"].strip()==result[-1]["msg"].strip()
            if same_author and same_msg:
                result[-1]["filenames"].append((e["filename"], e["num_revision"]))
                result[-1]["last_date"]=e["date"]
                new_commit=0
        if new_commit:
            commit={}
            commit["msg"]=e["msg"]
            commit["author"]=e["author"]
            commit["filenames"]=[(e["filename"],e["num_revision"])]
            commit["first_date"]=e["date"]
            commit["last_date"]=e["date"]
            result.append(commit)
    return result

def read_log_files(branch, filenames):
    entries=[]
    log("Reading %d files"%(len(filenames)))
    for i in filenames:
        parse(branch, i,entries)
    log("Read %d revisions"%(len(entries)))
    return entries

def my_wrap(str, max_size):
    all = []
    line = ""
    for l in str.split():
        if len(line+l) >= max_size:
            all = all + [line]
            line = ''
        line = line + (line and ' ') + l
    all = all + [line]
    all=map(lambda x:"\t"+x, all)
    return "\n".join(all)

def print_short_commit(i):
    print "%10s: %s | %s"%(i["author"],
                           i["msg"][0:20].replace("\n"," ").ljust(20),
                           " ".join(i["filenames"])[0:40])

def print_changelog_commit(i):
    i["filenames"].sort()
    first=time.strftime("%a %b %d %H:%M:%S %Y",time.gmtime(i["first_date"]))
    last=time.strftime("%a %b %d %H:%M:%S %Y",time.gmtime(i["last_date"]))
    
    name=i["author"]
    print "%s-%s  %s:"%(first,last,name)
    print
    for x in i["filenames"]:
        print "  "+format(x)
    print
    paragraphs=i["msg"].split("\n\n")
    for i in paragraphs:
        print my_wrap(i, 65)
        print

def format(t):
    name=t[0]
    rev=".".join(map(str,t[1]))
    return "%s:%s"%(name, rev)

def unformat(s):
    name,rev=s.split(":")
    rev=map(int,rev.strip().split("."))
    return (name,rev)
    
def print_cvs_diff_commands(commit):
    cmds=open("get_diff_cvs_commands","w")
    for name,rev in commit["filenames"]:
        prev_rev=rev[:]
        prev_rev[-1]-=1
        rev=".".join(map(str,rev))
        prev_rev=".".join(map(str,prev_rev))
        cmds.write("-N -r %s -r %s %s\n"%(prev_rev, rev, name))

def usage():
    log("get_diff [-r file:revision] | [-d date]")
    sys.exit(1)
    
def main():
    branch="HEAD"

    req_rev=None
    req_date=None

    if len(sys.argv)!=3:
        usage()
    if sys.argv[1] == "-r":
        req_rev=unformat(sys.argv[2])
    elif sys.argv[1] == "-d":
        req_date=parse_date(sys.argv[2])
    else:
        usage()
        
    log_files=sys.stdin.readlines()
    log_files=map(lambda x: string.rstrip(x), log_files)
    read_entries=read_log_files(branch, log_files)
    read_entries.sort(lambda x,y: cmp(x["date"],y["date"]))

    if req_rev:
        l=[]
        for rev in read_entries:
            if rev["filename"] == req_rev[0] and \
                   rev["num_revision"]==req_rev[1]:
                l.append(rev)
        if len(l)==0:
            log("Can't find the given file or revision.")
            sys.exit(1)
        req_date=l[0]["date"]

    from_date=req_date-(12*60*60)
    to_date=req_date+(12*60*60)

    read_entries=filter(lambda x: from_date<=x["date"]<=to_date,read_entries)
    log("%d revisions after date filtering"%len(read_entries))
    if 0:
        for i in read_entries:
            print "%10s: %s | %s"%(i["author"],
                                   i["msg"][0:20].replace("\n"," ").ljust(20),
                                   i["filename"])
    commits=merge(read_entries)
    log("Merged to %d commits"%len(commits))
    if 0:
        for i in commits:            
            print "%10s: %s | %s"%(i["author"],
                                   i["msg"][0:20].replace("\n"," ").ljust(20),
                                   " ".join(map(format,i["filenames"]))[0:80])
    commits=filter(lambda x: x["first_date"]<=req_date<=x["last_date"],commits)
    if len(commits)==0:
        log("No commit at the given time")
        sys.exit(1)
    assert len(commits)==1,repr(commits)
    print_changelog_commit(commits[0])
    print_cvs_diff_commands(commits[0])

if __name__=="__main__":
    main()