CheckCaseTrigger.py #5

# Version 2.1.6

# tag::includeManual[]
"""
CaseCheckTrigger.py

Example 1: Typical usage from the Helix Core server triggers table:

    Triggers:
        CheckCase change-submit //... "/usr/bin/python3 /p4/common/bin/triggers/CheckCaseTrigger.py %changelist% myuser=%user%"

SAMPLE OUTPUT:
    Submit validation failed -- fix problems then use 'p4 submit -c 1234'.
    'CheckCaseTrigger' validation failed:

    Your submission has been rejected because the following files
    are inconsistent in their use of case with respect to existing
    directories:

    Your file:         '//depot/dir/test'
    existing file/dir: '//depot/DIR'

BYPASS LOGIC:
    By default, this trigger can be bypassed by any user by adding the
    token BYPASS_CASE_CHECK to the changelist description.  Specify
    'allowbypass=no' on the command line to disable the ability to
    bypass this trigger

    As an exception, if the user is 'git-fusion-user', the case check
    is always bypassed if 'myuser' is defined.

DEPENDENCIES:
    This trigger requires P4Triggers.py and the P4Python API.

SEE ALSO:
    See 'p4 help triggers'.

"""
# end::includeManual[]

from __future__ import print_function

import P4
import P4Triggers
import os
import re
import subprocess
import sys

class CheckCaseTrigger(P4Triggers.P4Trigger):
# tag::includeManual[]
    """CaseCheckTrigger is a subclass of P4Trigger. Use this trigger to ensure
       that your depot does not contain two filenames or directories that only
       differ in case.
       Having files with different case spelling will cause problems in mixed
       environments where both case-insensitive clients like Windows and case-
       sensitive clients like UNIX access the same server.
    """
# end::includeManual[]

    def __init__(self, *args, maxErrors=10, **kwargs):
        kwargs['charset'] = 'none'
        kwargs['api_level'] = 71

        self.allowBypass=AllowBypass
 
        
        fileFilter = None
        if 'filefilter' in kwargs:
            fileFilter = kwargs['filefilter']
            del kwargs['filefilter']
        
        P4Triggers.P4Trigger.__init__(self, **kwargs)
        self.parse_args(__doc__, args)
        
        # Ensure that -ztag global option is used.
        self.p4.tagged = True

        # need to reset the args in case a p4config file overwrote them
        for (k, v) in kwargs.items():
            if k != "log":
                try:
                    setattr(self.p4, k, v)
                except:
                    self.logger.error("error setting p4 property: '%s' to '%v'" % (k, v))

        self.map = None
        if fileFilter:
            try:
                with open(fileFilter) as f:
                    self.map = P4.Map()
                    for line in f:
                        self.map.insert(line.strip())
            except IOError:
                self.logger.error("Could not open filter file %s" % fileFilter)
        
        self.maxErrors = maxErrors
        self.depotCache = {}
        self.masterCache = {}
        self.maxWildcards = 9

    def add_parse_args(self, parser):
        """Specific args for this trigger - also calls super class to add common trigger args"""
        parser.add_argument('change', help="Change to validate - %%change%% argument from triggers entry.")
        super(CheckCaseTrigger, self).add_parse_args(parser)

    def setUp(self):
        info = self.p4.run_info()[0]
        if "unicode" in info and info["unicode"] == "enabled":
            self.p4.charset = "utf8"
        self.p4.exception_level = 1 # ignore WARNINGS like "no such file"
        self.p4.prog = "CheckCaseTrigger"

        if self.allowBypass:
            self.USER_MESSAGE="""

    Your changelist submit attempt has been rejected because
    one or more file paths opened for add vary only by case
    from existing files/directory paths.  Creating file/folders
    that vary only in case from existing paths causes inconsistent
    behavior across platforms with different case handling behaviors
    (e.g. Windows, Linux/UNIX, Mac OSX). Thus, adding case-only
    variations of existing paths is strongly discouraged.

    If you are certain the files to be added will only be accessed
    from workspaces on case-sensitive platforms (like UNIX/Linux),
    then this trigger can be bypassed by adding the token
    BYPASS_CASE_CHECK to the changelist description and attempting
    the submit again.

    Alternately, you can revert any files opened for add in your
    changelists that vary only in case from existing files, or
    move them to new names that don't conflict with existing files.

    Offending files:
    """
        else:
            self.USER_MESSAGE="""

    Your changelist submit attempt has been rejected because
    one or more file paths opened for add vary only by case
    from existing files/directory paths.  Creating file/folders
    that vary only in case from existing paths causes inconsistent
    behavior across platforms with different case handling behaviors
    (e.g. Windows, Linux/UNIX, Mac OSX). Thus, adding case-only
    variations of existing paths is disallowed.

    To move forward, you can revert any files opened for add in your
    changelists that vary only in case from existing files, or
    move them to new names that don't conflict with existing files.

    Offending files:
    """

        self.BADFILE_FORMAT="""
      Your file:         '%s'
      existing file/dir: '%s'
      """

    def validate(self):
        """Here the fun begins. This method overrides P4Trigger.validate()"""
        badlist = {}
        files = self.change.files
        #self.logger.debug("validate: Files to submit: %s", files)

        self.filterRenames(files)
        
        # Determine valid file list and depots to cache,
        # including the max directory depth for each cache
        
        depth = 0
        validFiles = []
        uniqueDepots = {}
        
        for file in files:
            action = file.revisions[0].action
            if self.map and self.map.includes(file.depotFile):
                continue
            if action not in ("add", "branch", "move/add"):
                continue
            
            mismatch = ""
            
            path = file.depotFile[2:]
            #self.logger.debug("validate: path = %s", path)
            
            parts = path.split('/')            
            #self.logger.debug("validate: parts = %s", parts)
            
            depot = parts[0]
            
            #Determine if we have a valid depot.
            mismatch = self.findDepotMismatch(depot)
            if mismatch:
                badlist[file.depotFile] = mismatch
                continue
            
            if depot in uniqueDepots :
                if len(parts) > uniqueDepots[depot] :
                    uniqueDepots[depot] = len(parts)
            else :
                uniqueDepots[depot] = len(parts)
                        
            validFiles.append(file.depotFile)
            #self.logger.debug("validate: file.depotFile = %s", file.depotFile)
                
        #self.logger.debug("validate: uniqueDepots = %s", uniqueDepots)
        #self.logger.debug("validate: validFiles = %s",validFiles)
        
        # Build cache for each unique depot.
        self.buildCache(uniqueDepots, validFiles)
        #self.logger.debug("validate: masterCache_1 = %s", self.masterCache)
        
        # Look for files in cache. This includes looking for directories in file path.
        
        self.searchCache(validFiles, badlist)
 
        
        self.logger.debug("validate: badlist = %s", badlist)
        #self.logger.debug("validate: masterCache_2 = %s", self.masterCache)

        if len(badlist) > 0:
            self.report(badlist)
        
        return len(badlist) == 0
    
    #end of validate
  
    # This method builds a global cache to use for mismatch searches.
    # IN: depots to use, including depth of cache for each depot
    # OUT: None
    
    def buildCache(self, depots, fileList):
        
        #self.logger.debug("buildCache: depots = %s", depots)
        #self.logger.debug("buildCache: fileList = %s", fileList)
        
        for u in depots :
 
            #Optimize by looking for files from shared root in current depot.
            
            commonList = []
                
            for f in fileList:                
                if f.find('//' + u) != -1:
                    commonList.append(f)
            
            if len(commonList) == 1:
                fileStr = '//' + u + '/...'               
            else:                
                fileStr = '/' + os.path.commonpath( commonList) + '/...'            
            
            #self.logger.debug("buildCache: fileStr = %s", fileStr)
            
            # Ignore case when we get files.
                    
            for f in self.p4.run_files( "-i",fileStr) :
                #self.logger.debug("validate: f = %s", f)
                if "delete" not in f["action"] :
                    f = f["depotFile"]
                    cf = self.canonical(f)
                    
                    if not cf in self.masterCache :
                        self.masterCache[cf] = f
                        
                    #Need to add components of file's path.
                    for i in range(2, len(f)):
                        
                        if f[i] == '/':
                            substr = f[:i]
                            csubstr = self.canonical(substr)
                            if not csubstr in self.masterCache :
                                self.masterCache[csubstr] = substr  
                            #self.logger.debug("buildCache: substr = %s", substr)                          
        
    #end buildCache
    
    # Method canonical 
    # IN:  string, utf8 compatible
    # OUT: unicode string, all lower case
    def canonical(self, aString):
        return aString.lower()

    # Method filterRenames:
    # Removes pairs of files that only differ in case
    # where one action is branch, and the other delete
    def filterRenames(self, files):
        branches = [x for x in files if x.revisions[0].action == 'branch']
        deletes = [x.depotFile.lower() for x in files if x.revisions[0].action == 'delete']
        
        for f in branches:
            if f.depotFile.lower() in deletes:
                files.remove(f)

    # This method returns either a depot or directory that differs from 
    # the supplied path only in case, or None if no mismatch is found
    
    # This method looks for a depot mismatch
    # It caches the depots as it found them (Global cache)
    # IN: depot name (as provided by the change list
    # OUT: None if (no mismatch found) else (stored depot name)

    def findDepotMismatch(self, depot):
        canonicalDepot = self.canonical(depot)

        # if the depot is in the cache, the cached version is authoritative
    
        if canonicalDepot in self.depotCache:
            if self.depotCache[canonicalDepot] == depot:
                return None
            else:
                return '//' + self.depotCache[canonicalDepot]

        # depot not found in cache. Populate the cache
    
        mismatch = None
        for d in self.p4.run_depots():
            dname = d["name"]
            cd = self.canonical(dname)
            self.depotCache[cd] = dname
            if canonicalDepot == cd and depot != dname:
                mismatch = '//' + dname

        return mismatch
    # end findDepotMismatch
    
    def report(self, badfiles):
        msg = self.USER_MESSAGE
        for (n, (file, mismatch)) in enumerate(badfiles.items()):
            if n >= self.maxErrors:
                break
            msg += self.BADFILE_FORMAT % (file, mismatch)

        self.message(msg)

    def run(self):
        """Runs trigger"""
        try:
            self.logger.debug("CheckCaseTrigger firing")
            self.setupP4()
            return self.parseChange(self.options.change)
        except Exception:
            return self.reportException()
        return 0

    # This method searches a global cache to find case mismatches for changelist files.
    # File subdirectories and file itself are added to cache if no mismatches are found.
    # IN: List of changelist files for which we want verify there are no case mismatches.
    #     Mismatch dictionary that records mismatches.
    # OUT: May modifiy mismatches parameter
    
    def searchCache(self, cfiles, mismatches):
        
        #self.logger.debug("searchCache: changelist files = %s, mismatches = %s", cfiles, mismatches)
        
        for f in cfiles :
            
            #self.logger.debug("searchCache: f = %s", f)
            
            mismatch = ""
            
            # Look for file and continue if it's in the cache. It's already been added.
            
            cf = self.canonical(f)
            
            if cf in self.masterCache :
                #self.logger.debug("searchCache: found %s in %s", f, self.masterCache)
                if f != self.masterCache[cf] :
                    mismatch = self.masterCache[cf]
                    mismatches[f] = mismatch
                continue
                
            # Need to check for mismatch of file path components.
            # If none, add components to cache.
            
            for i in range(2, len(f)) :
                
                mismatch =  ""
                 
                if f[i] == '/':
                    substr = f[:i]
                    csubstr = self.canonical(substr)
                    #self.logger.debug("searchCache: csubstr = %s, substr = %s", csubstr, substr)          
                 
                    # Break if we find a mismatch. Otherwise, add subName to cache.
                
                    if csubstr in self.masterCache :
                        if self.masterCache[csubstr] != substr :
                            mismatch = self.masterCache[csubstr]
                            break
                    else :  
                        self.masterCache[csubstr] = substr           
            
            # If no mismatch, add the file.
            
            if mismatch :
                mismatches[f] = mismatch
            else :
                self.masterCache[cf] = f     
    
    #end searchCache
    
# If called from the command line, go in here

if __name__ == "__main__":
    # Generate new args - parsing out port=123 style way of specifying
    # parameters intended for p4 properties
    kwargs = {}
    args = []
    for arg in sys.argv[1:]:
        p = arg.split("=", 1)
        if len(p) == 1:
            args.append(arg)
        else:
            kwargs[p[0]] = p[1]

    # Example of how to exclude the 'git-fusion-user'
    # Note: Need to remove 'myuser' after test as it's not a valid P4 argument.
    if 'myuser' in kwargs: 
        if kwargs['myuser'] == 'git-fusion-user':
            sys.exit(0)
        else:
            del kwargs['myuser']

    AllowBypass = 1

    if 'allowbypass' in kwargs: 
        if kwargs['allowbypass'] == 'no':
            AllowBypass = 0

        # Remove 'allowbypass' after test as it's not a valid P4 argument.
        del kwargs['allowbypass']

    if AllowBypass:
        # Grab the changelist description, and scan for the bypass token string.
        # If the token is detected, silently and immediately exit with a happy 0
        # exit code.
        changelist = sys.argv[1]
        cmd = "%s -ztag -F %%desc%% describe -f -s %s" % (os.getenv('P4BIN','p4'), changelist)
        p = subprocess.Popen(cmd, stdout=subprocess.PIPE, shell=True)
        (changeDesc, err) = p.communicate()
        p_status = p.wait()

        # If the changelist description contains the text BYPASS_CASE_CHECK,
        # bypass the case check logic.
        if (re.search (b'BYPASS_CASE_CHECK', changeDesc, re.MULTILINE)):
            sys.exit(0)

    ct = CheckCaseTrigger(*args, maxErrors=10, **kwargs)
    sys.exit(ct.run())