icuSources/python/icutools/databuilder/utils.py

   1 # Copyright (C) 2018 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html
   3
   4 # Python 2/3 Compatibility (ICU-20299)
   5 # TODO(ICU-20301): Remove this.
   6 from __future__ import print_function
   7
   8 import sys
   9
  10 from . import *
  11
  12
  13 def dir_for(file):
  14     if isinstance(file, LocalFile):
  15         return get_local_dirname(file.dirname)
  16     if isinstance(file, SrcFile):
  17         return "{SRC_DIR}"
  18     if isinstance(file, InFile):
  19         return "{IN_DIR}"
  20     if isinstance(file, TmpFile):
  21         return "{TMP_DIR}"
  22     if isinstance(file, OutFile):
  23         return "{OUT_DIR}"
  24     if isinstance(file, PkgFile):
  25         return "{PKG_DIR}"
  26     assert False
  27
  28
  29 LOCAL_DIRNAME_SUBSTITUTIONS = {
  30     "SRC": "{SRC_DIR}",
  31     "FILTERS": "{FILTERS_DIR}",
  32     "CWD": "{CWD_DIR}"
  33 }
  34
  35
  36 def get_local_dirname(dirname):
  37     if dirname.startswith("/"):
  38         return dirname
  39     elif dirname.startswith("$"):
  40         # Note: directory separator substitution happens later
  41         sep_idx = dirname.find("/")
  42         if sep_idx == -1:
  43             sep_idx = len(dirname)
  44         variable = dirname[1:sep_idx]
  45         if variable in LOCAL_DIRNAME_SUBSTITUTIONS:
  46             return LOCAL_DIRNAME_SUBSTITUTIONS[variable] + dirname[sep_idx:]
  47     print(
  48         "Error: Local directory must be absolute, or relative to one of: " +
  49         (", ".join("$%s" % v for v in LOCAL_DIRNAME_SUBSTITUTIONS.keys())),
  50         file=sys.stderr
  51     )
  52     exit(1)
  53
  54
  55 ALL_TREES = [
  56     "locales",
  57     "curr",
  58     "lang",
  59     "region",
  60     "zone",
  61     "unit",
  62     "coll",
  63     "brkitr",
  64     "rbnf",
  65 ]
  66
  67
  68 def concat_dicts(*dicts):
  69     # There is not a super great way to do this in Python:
  70     new_dict = {}
  71     for dict in dicts:
  72         new_dict.update(dict)
  73     return new_dict
  74
  75
  76 def repeated_execution_request_looper(request):
  77     # dictionary of lists to list of dictionaries:
  78     ld = [
  79         dict(zip(request.repeat_with, t))
  80         for t in zip(*request.repeat_with.values())
  81     ]
  82     if not ld:
  83         # No special options given in repeat_with
  84         ld = [{} for _ in range(len(request.input_files))]
  85     return zip(ld, request.specific_dep_files, request.input_files, request.output_files)
  86
  87
  88 def format_single_request_command(request, cmd_template, common_vars):
  89     return cmd_template.format(
  90         ARGS = request.args.format(
  91             INPUT_FILES = [file.filename for file in request.input_files],
  92             OUTPUT_FILES = [file.filename for file in request.output_files],
  93             **concat_dicts(common_vars, request.format_with)
  94         )
  95     )
  96
  97
  98 def format_repeated_request_command(request, cmd_template, loop_vars, common_vars):
  99     (iter_vars, _, input_file, output_file) = loop_vars
 100     return cmd_template.format(
 101         ARGS = request.args.format(
 102             INPUT_FILE = input_file.filename,
 103             OUTPUT_FILE = output_file.filename,
 104             **concat_dicts(common_vars, request.format_with, iter_vars)
 105         )
 106     )
 107
 108
 109 def flatten_requests(requests, config, common_vars):
 110     result = []
 111     for request in requests:
 112         result += request.flatten(config, requests, common_vars)
 113     return result
 114
 115
 116 def get_all_output_files(requests, include_tmp=False):
 117     files = []
 118     for request in requests:
 119         files += request.all_output_files()
 120
 121     # Filter out all files but those in OUT_DIR if necessary.
 122     # It is also easy to filter for uniqueness; do it right now and return.
 123     if not include_tmp:
 124         files = (file for file in files if isinstance(file, OutFile))
 125         return list(set(files))
 126
 127     # Filter for unique values.  NOTE: Cannot use set() because we need to accept same filename as
 128     # OutFile and TmpFile as different, and by default they evaluate as equal.
 129     return [f for _, f in set((type(f), f) for f in files)]
 130
 131
 132 def compute_directories(requests):
 133     dirs = set()
 134     for file in get_all_output_files(requests, include_tmp=True):
 135         path = "%s/%s" % (dir_for(file), file.filename)
 136         dirs.add(path[:path.rfind("/")])
 137     return list(sorted(dirs))
 138
 139
 140 class SpaceSeparatedList(list):
 141     """A list that joins itself with spaces when converted to a string."""
 142     def __str__(self):
 143         return " ".join(self)