| 1 | 1 |  #
 | 
| 2 | 2 |  #  Copyright (C) 2016 Codethink Limited
 | 
|  | 3 | +#  Copyright (C) 2019 Bloomberg L.P.
 | 
| 3 | 4 |  #
 | 
| 4 | 5 |  #  This program is free software; you can redistribute it and/or
 | 
| 5 | 6 |  #  modify it under the terms of the GNU Lesser General Public
 | 
| ... | ... | @@ -16,6 +17,7 @@ | 
| 16 | 17 |  #
 | 
| 17 | 18 |  #  Authors:
 | 
| 18 | 19 |  #        Tristan Van Berkom <tristan vanberkom codethink co uk>
 | 
|  | 20 | +#        Daniel Silverstone <daniel silverstone codethink co uk>
 | 
| 19 | 21 |  
 | 
| 20 | 22 |  import re
 | 
| 21 | 23 |  import sys
 | 
| ... | ... | @@ -25,7 +27,7 @@ from . import _yaml | 
| 25 | 27 |  
 | 
| 26 | 28 |  # Variables are allowed to have dashes here
 | 
| 27 | 29 |  #
 | 
| 28 |  | -_VARIABLE_MATCH = r'\%\{([a-zA-Z][a-zA-Z0-9_-]*)\}'
 | 
|  | 30 | +PARSE_EXPANSION = re.compile(r"\%\{([a-zA-Z][a-zA-Z0-9_-]*)\}")
 | 
| 29 | 31 |  
 | 
| 30 | 32 |  
 | 
| 31 | 33 |  # The Variables helper object will resolve the variable references in
 | 
| ... | ... | @@ -46,8 +48,14 @@ class Variables(): | 
| 46 | 48 |      def __init__(self, node):
 | 
| 47 | 49 |  
 | 
| 48 | 50 |          self.original = node
 | 
| 49 |  | -#        self.variables = self._resolve(node)
 | 
| 50 |  | -        self.newexp = self._resolve_exp(node)
 | 
|  | 51 | +        self.newexp = self._resolve(node)
 | 
|  | 52 | +        self._sanity_check()
 | 
|  | 53 | +
 | 
|  | 54 | +    def get(self, var):
 | 
|  | 55 | +        if var in self.newexp:
 | 
|  | 56 | +            return _expand_expstr(self.newexp, self.newexp[var])
 | 
|  | 57 | +        else:
 | 
|  | 58 | +            return None
 | 
| 51 | 59 |  
 | 
| 52 | 60 |      # subst():
 | 
| 53 | 61 |      #
 | 
| ... | ... | @@ -63,203 +71,138 @@ class Variables(): | 
| 63 | 71 |      #    LoadError, if the string contains unresolved variable references.
 | 
| 64 | 72 |      #
 | 
| 65 | 73 |      def subst(self, string):
 | 
| 66 |  | -#        real = self.real_subst(string)
 | 
| 67 |  | -        exp = self.subst_exp(string)
 | 
| 68 |  | -#        if real != exp:
 | 
| 69 |  | -#            raise LoadError(LoadErrorReason.UNRESOLVED_VARIABLE, "Oh dear!")
 | 
| 70 |  | -#        return real
 | 
| 71 |  | -        return exp
 | 
| 72 |  | -    
 | 
| 73 |  | -    def real_subst(self, string):
 | 
| 74 |  | -        substitute, unmatched, _ = self._subst(string, self.variables)
 | 
| 75 |  | -        unmatched = list(set(unmatched))
 | 
| 76 |  | -        if unmatched:
 | 
| 77 |  | -            if len(unmatched) == 1:
 | 
| 78 |  | -                message = "Unresolved variable '{var}'".format(var=unmatched[0])
 | 
| 79 |  | -            else:
 | 
| 80 |  | -                message = "Unresolved variables: "
 | 
| 81 |  | -                for unmatch in unmatched:
 | 
| 82 |  | -                    if unmatched.index(unmatch) > 0:
 | 
| 83 |  | -                        message += ', '
 | 
| 84 |  | -                    message += unmatch
 | 
| 85 |  | -
 | 
| 86 |  | -            raise LoadError(LoadErrorReason.UNRESOLVED_VARIABLE, message)
 | 
| 87 |  | -
 | 
| 88 |  | -        return substitute
 | 
| 89 |  | -
 | 
| 90 |  | -    def _subst(self, string, variables):
 | 
| 91 |  | -
 | 
| 92 |  | -        def subst_callback(match):
 | 
| 93 |  | -            nonlocal variables
 | 
| 94 |  | -            nonlocal unmatched
 | 
| 95 |  | -            nonlocal matched
 | 
| 96 |  | -
 | 
| 97 |  | -            token = match.group(0)
 | 
| 98 |  | -            varname = match.group(1)
 | 
|  | 74 | +        exp = _parse_expstr(string)
 | 
| 99 | 75 |  
 | 
| 100 |  | -            value = _yaml.node_get(variables, str, varname, default_value=None)
 | 
| 101 |  | -            if value is not None:
 | 
| 102 |  | -                # We have to check if the inner string has variables
 | 
| 103 |  | -                # and return unmatches for those
 | 
| 104 |  | -                unmatched += re.findall(_VARIABLE_MATCH, value)
 | 
| 105 |  | -                matched += [varname]
 | 
| 106 |  | -            else:
 | 
| 107 |  | -                # Return unmodified token
 | 
| 108 |  | -                unmatched += [varname]
 | 
| 109 |  | -                value = token
 | 
|  | 76 | +        try:
 | 
|  | 77 | +            return _expand_expstr(self.newexp, exp)
 | 
|  | 78 | +        except KeyError:
 | 
|  | 79 | +            unmatched = []
 | 
| 110 | 80 |  
 | 
| 111 |  | -            return value
 | 
|  | 81 | +            for v in exp[1][1::2]:
 | 
|  | 82 | +                if v not in self.newexp:
 | 
|  | 83 | +                    unmatched.append(v)
 | 
| 112 | 84 |  
 | 
| 113 |  | -        matched = []
 | 
| 114 |  | -        unmatched = []
 | 
| 115 |  | -        replacement = re.sub(_VARIABLE_MATCH, subst_callback, string)
 | 
|  | 85 | +            if unmatched:
 | 
|  | 86 | +                if len(unmatched) == 1:
 | 
|  | 87 | +                    message = "Unresolved variable '{var}'".format(var=unmatched[0])
 | 
|  | 88 | +                else:
 | 
|  | 89 | +                    message = "Unresolved variables: "
 | 
|  | 90 | +                    for unmatch in unmatched:
 | 
|  | 91 | +                        if unmatched.index(unmatch) > 0:
 | 
|  | 92 | +                            message += ', '
 | 
|  | 93 | +                            message += unmatch
 | 
| 116 | 94 |  
 | 
| 117 |  | -        return (replacement, unmatched, matched)
 | 
|  | 95 | +                raise LoadError(LoadErrorReason.UNRESOLVED_VARIABLE, message)
 | 
|  | 96 | +            raise
 | 
| 118 | 97 |  
 | 
| 119 | 98 |      # Variable resolving code
 | 
| 120 | 99 |      #
 | 
| 121 |  | -    # Here we substitute variables for values (resolve variables) repeatedly
 | 
| 122 |  | -    # in a dictionary, each time creating a new dictionary until there is no
 | 
| 123 |  | -    # more unresolved variables to resolve, or, until resolving further no
 | 
| 124 |  | -    # longer resolves anything, in which case we throw an exception.
 | 
|  | 100 | +    # Here we resolve all of our inputs into a dictionary, ready for use
 | 
|  | 101 | +    # in subst()
 | 
| 125 | 102 |      def _resolve(self, node):
 | 
| 126 |  | -        variables = node
 | 
| 127 |  | -
 | 
| 128 | 103 |          # Special case, if notparallel is specified in the variables for this
 | 
| 129 | 104 |          # element, then override max-jobs to be 1.
 | 
| 130 | 105 |          # Initialize it as a string as all variables are processed as strings.
 | 
| 131 | 106 |          #
 | 
| 132 |  | -        if _yaml.node_get(variables, bool, 'notparallel', default_value=False):
 | 
| 133 |  | -            variables['max-jobs'] = str(1)
 | 
| 134 |  | -
 | 
| 135 |  | -        # Resolve the dictionary once, reporting the new dictionary with things
 | 
| 136 |  | -        # substituted in it, and reporting unmatched tokens.
 | 
| 137 |  | -        #
 | 
| 138 |  | -        def resolve_one(variables):
 | 
| 139 |  | -            unmatched = []
 | 
| 140 |  | -            resolved = {}
 | 
| 141 |  | -
 | 
| 142 |  | -            for key, value in _yaml.node_items(variables):
 | 
| 143 |  | -
 | 
| 144 |  | -                # Ensure stringness of the value before substitution
 | 
| 145 |  | -                value = _yaml.node_get(variables, str, key)
 | 
| 146 |  | -
 | 
| 147 |  | -                resolved_var, item_unmatched, matched = self._subst(value, variables)
 | 
| 148 |  | -
 | 
| 149 |  | -                if _wrap_variable(key) in resolved_var:
 | 
| 150 |  | -                    referenced_through = find_recursive_variable(key, matched, variables)
 | 
| 151 |  | -                    raise LoadError(LoadErrorReason.RECURSIVE_VARIABLE,
 | 
| 152 |  | -                                    "{}: ".format(_yaml.node_get_provenance(variables, key)) +
 | 
| 153 |  | -                                    ("Variable '{}' expands to contain a reference to itself. " +
 | 
| 154 |  | -                                     "Perhaps '{}' contains '{}").format(key, referenced_through, _wrap_variable(key)))
 | 
| 155 |  | -
 | 
| 156 |  | -                resolved[key] = resolved_var
 | 
| 157 |  | -                unmatched += item_unmatched
 | 
| 158 |  | -
 | 
| 159 |  | -            # Carry over provenance
 | 
| 160 |  | -            resolved[_yaml.PROVENANCE_KEY] = variables[_yaml.PROVENANCE_KEY]
 | 
| 161 |  | -            return (resolved, unmatched)
 | 
| 162 |  | -
 | 
| 163 |  | -        # Resolve it until it's resolved or broken
 | 
| 164 |  | -        #
 | 
| 165 |  | -        resolved = variables
 | 
| 166 |  | -        unmatched = ['dummy']
 | 
| 167 |  | -        last_unmatched = ['dummy']
 | 
| 168 |  | -        while unmatched:
 | 
| 169 |  | -            resolved, unmatched = resolve_one(resolved)
 | 
|  | 107 | +        if _yaml.node_get(node, bool, 'notparallel', default_value=False):
 | 
|  | 108 | +            node['max-jobs'] = str(1)
 | 
| 170 | 109 |  
 | 
| 171 |  | -            # Lists of strings can be compared like this
 | 
| 172 |  | -            if unmatched == last_unmatched:
 | 
| 173 |  | -                # We've got the same result twice without matching everything,
 | 
| 174 |  | -                # something is undeclared or cyclic, compose a summary.
 | 
| 175 |  | -                #
 | 
| 176 |  | -                summary = ''
 | 
| 177 |  | -                for unmatch in set(unmatched):
 | 
| 178 |  | -                    for var, provenance in self._find_references(unmatch):
 | 
| 179 |  | -                        line = "  unresolved variable '{unmatched}' in declaration of '{variable}' at: {provenance}\n"
 | 
| 180 |  | -                        summary += line.format(unmatched=unmatch, variable=var, provenance=provenance)
 | 
| 181 |  | -
 | 
| 182 |  | -                raise LoadError(LoadErrorReason.UNRESOLVED_VARIABLE,
 | 
| 183 |  | -                                "Failed to resolve one or more variable:\n{}".format(summary))
 | 
| 184 |  | -
 | 
| 185 |  | -            last_unmatched = unmatched
 | 
| 186 |  | -
 | 
| 187 |  | -        return resolved
 | 
| 188 |  | -
 | 
| 189 |  | -    def _resolve_exp(self, node):
 | 
| 190 | 110 |          ret = {}
 | 
| 191 | 111 |          for key, value in _yaml.node_items(node):
 | 
| 192 | 112 |              value = _yaml.node_get(node, str, key)
 | 
| 193 | 113 |              ret[sys.intern(key)] = _parse_expstr(value)
 | 
| 194 | 114 |          return ret
 | 
| 195 | 115 |  
 | 
| 196 |  | -    def subst_exp(self, string):
 | 
| 197 |  | -        exp = _parse_expstr(string)
 | 
| 198 |  | -        return _expand_expstr(self.newexp, exp)
 | 
| 199 |  | -    
 | 
| 200 |  | -    # Helper function to fetch information about the node referring to a variable
 | 
|  | 116 | +    # Sanity checks on the resolved variables dictionary
 | 
| 201 | 117 |      #
 | 
| 202 |  | -    def _find_references(self, varname):
 | 
| 203 |  | -        fullname = _wrap_variable(varname)
 | 
| 204 |  | -        for key, value in _yaml.node_items(self.original):
 | 
| 205 |  | -            if fullname in value:
 | 
| 206 |  | -                provenance = _yaml.node_get_provenance(self.original, key)
 | 
| 207 |  | -                yield (key, provenance)
 | 
| 208 |  | -
 | 
| 209 |  | -
 | 
| 210 |  | -def find_recursive_variable(variable, matched_variables, all_vars):
 | 
| 211 |  | -    matched_values = (_yaml.node_get(all_vars, str, key) for key in matched_variables)
 | 
| 212 |  | -    for key, value in zip(matched_variables, matched_values):
 | 
| 213 |  | -        if _wrap_variable(variable) in value:
 | 
| 214 |  | -            return key
 | 
| 215 |  | -    # We failed to find a recursive variable
 | 
| 216 |  | -    return None
 | 
| 217 |  | -
 | 
| 218 |  | -
 | 
| 219 |  | -def _wrap_variable(var):
 | 
| 220 |  | -    return "%{" + var + "}"
 | 
| 221 |  | -
 | 
| 222 |  | -## Stuff for new style expansion strings
 | 
| 223 |  | -
 | 
| 224 |  | -PARSE_EXPANSION = re.compile(r"\%\{([a-zA-Z][a-zA-Z0-9_-]*)\}")
 | 
|  | 118 | +    # Here we check for loops and for missing inputs which might cause the
 | 
|  | 119 | +    # resolved dictionary of inputs to not be sufficient.
 | 
|  | 120 | +    def _sanity_check(self):
 | 
|  | 121 | +        # First the check for anything unresolvable
 | 
|  | 122 | +        summary = []
 | 
|  | 123 | +        wants = {}
 | 
|  | 124 | +        for k, es in self.newexp.items():
 | 
|  | 125 | +            wants[k] = wants.get(k, set())
 | 
|  | 126 | +            for var in es[1][1::2]:
 | 
|  | 127 | +                wants[k].add(var)
 | 
|  | 128 | +                if var not in self.newexp:
 | 
|  | 129 | +                    line = "  unresolved variable '{unmatched}' in declaration of '{variable}' at: {provenance}"
 | 
|  | 130 | +                    provenance = _yaml.node_get_provenance(self.original, k)
 | 
|  | 131 | +                    summary.append(line.format(unmatched=var, variable=k, provenance=provenance))
 | 
|  | 132 | +        if summary:
 | 
|  | 133 | +            raise LoadError(LoadErrorReason.UNRESOLVED_VARIABLE,
 | 
|  | 134 | +                            "Failed to resolve one or more variable:\n{}\n".format("\n".join(summary)))
 | 
|  | 135 | +
 | 
|  | 136 | +        # And now the cycle checks
 | 
|  | 137 | +        def cycle_check(exp, visited, cleared):
 | 
|  | 138 | +            for var in exp[1][1::2]:
 | 
|  | 139 | +                if var in cleared:
 | 
|  | 140 | +                    continue
 | 
|  | 141 | +                if var in visited:
 | 
|  | 142 | +                    raise LoadError(LoadErrorReason.RECURSIVE_VARIABLE,
 | 
|  | 143 | +                                    "{}: ".format(_yaml.node_get_provenance(self.original, var)) +
 | 
|  | 144 | +                                    ("Variable '{}' expands to contain a reference to itself. " +
 | 
|  | 145 | +                                     "Perhaps '{}' contains '%{{{}}}").format(var, visited[-1], var))
 | 
|  | 146 | +                visited.append(var)
 | 
|  | 147 | +                cycle_check(self.newexp[var], visited, cleared)
 | 
|  | 148 | +                visited.pop()
 | 
|  | 149 | +                cleared.add(var)
 | 
|  | 150 | +
 | 
|  | 151 | +        cleared = set()
 | 
|  | 152 | +        for k, es in self.newexp.items():
 | 
|  | 153 | +            if k not in cleared:
 | 
|  | 154 | +                cycle_check(es, [k], cleared)
 | 
|  | 155 | +
 | 
|  | 156 | +    # resolved():
 | 
|  | 157 | +    #
 | 
|  | 158 | +    # Perform any substitutions necessary and return a dictionary of all
 | 
|  | 159 | +    # inputs fully resolved.
 | 
|  | 160 | +    #
 | 
|  | 161 | +    # Returns:
 | 
|  | 162 | +    #    (dict): A dictionary mapping variable name to resolved content
 | 
|  | 163 | +    #
 | 
|  | 164 | +    def resolved(self):
 | 
|  | 165 | +        # Make a fresh dict of the resolved expansions
 | 
|  | 166 | +        ret = {}
 | 
|  | 167 | +        for k, es in self.newexp.items():
 | 
|  | 168 | +            ret[k] = _expand_expstr(self.newexp, es)
 | 
|  | 169 | +        return ret
 | 
| 225 | 170 |  
 | 
| 226 |  | -def __parse_expstr(s):
 | 
| 227 |  | -    spl = PARSE_EXPANSION.split(s)
 | 
| 228 |  | -    if spl[-1] == '':
 | 
| 229 |  | -        spl = spl[:-1]
 | 
| 230 |  | -    return (len(spl), [sys.intern(s) for s in spl])
 | 
| 231 | 171 |  
 | 
|  | 172 | +# Cache for the parsed expansion strings.  While this is nominally
 | 
|  | 173 | +# something which might "waste" memory, in reality each of these
 | 
|  | 174 | +# will live as long as the element which uses it, which is the
 | 
|  | 175 | +# vast majority of the memory usage across the execution of BuildStream.
 | 
| 232 | 176 |  PARSE_CACHE = {}
 | 
| 233 | 177 |  
 | 
| 234 |  | -def _parse_expstr(s):
 | 
|  | 178 | +
 | 
|  | 179 | +# Helper to parse a string into an expansion string tuple, caching
 | 
|  | 180 | +# the results so that future parse requests don't need to think about
 | 
|  | 181 | +# the string
 | 
|  | 182 | +def _parse_expstr(instr):
 | 
| 235 | 183 |      try:
 | 
| 236 |  | -        return PARSE_CACHE[s]
 | 
|  | 184 | +        return PARSE_CACHE[instr]
 | 
| 237 | 185 |      except KeyError:
 | 
| 238 |  | -        PARSE_CACHE[s] = __parse_expstr(s)
 | 
| 239 |  | -        return PARSE_CACHE[s]
 | 
|  | 186 | +        spl = PARSE_EXPANSION.split(instr)
 | 
|  | 187 | +        if spl[-1] == '':
 | 
|  | 188 | +            spl = spl[:-1]
 | 
|  | 189 | +        PARSE_CACHE[instr] = (len(spl), [sys.intern(s) for s in spl])
 | 
|  | 190 | +        return PARSE_CACHE[instr]
 | 
| 240 | 191 |  
 | 
| 241 |  | -#def _expand_expstr(content, value):
 | 
| 242 |  | -#    ret = []
 | 
| 243 |  | -#    expstack = [(value,0)]
 | 
| 244 |  | -#    while expstack:
 | 
| 245 |  | -#        ((elen,expanded), idx) = expstack.pop()
 | 
| 246 |  | -#        ret.append(expanded[idx])
 | 
| 247 |  | -#        idx += 1
 | 
| 248 |  | -#        if idx < elen:
 | 
| 249 |  | -#            if elen > idx+1:
 | 
| 250 |  | -#                expstack.append(((elen,expanded), idx+1))
 | 
| 251 |  | -#            expstack.append((content[expanded[idx]], 0))
 | 
| 252 |  | -#    return "".join(ret)
 | 
| 253 | 192 |  
 | 
| 254 |  | -def __expand(content, value):
 | 
| 255 |  | -    (elen, bits) = value
 | 
| 256 |  | -    idx = 0
 | 
| 257 |  | -    while idx < elen:
 | 
| 258 |  | -        yield bits[idx]
 | 
| 259 |  | -        idx += 1
 | 
| 260 |  | -        if idx < elen:
 | 
| 261 |  | -            yield from __expand(content, content[bits[idx]])
 | 
| 262 |  | -        idx += 1
 | 
| 263 |  | -
 | 
| 264 |  | -def _expand_expstr(content, value):
 | 
| 265 |  | -    return "".join(__expand(content, value)) | 
|  | 193 | +# Helper to expand a given top level expansion string tuple in the context
 | 
|  | 194 | +# of the given dictionary of expansion strings.
 | 
|  | 195 | +#
 | 
|  | 196 | +# Note: Will raise KeyError if any expansion is missing
 | 
|  | 197 | +def _expand_expstr(content, topvalue):
 | 
|  | 198 | +    def __expand(value):
 | 
|  | 199 | +        (elen, bits) = value
 | 
|  | 200 | +        idx = 0
 | 
|  | 201 | +        while idx < elen:
 | 
|  | 202 | +            yield bits[idx]
 | 
|  | 203 | +            idx += 1
 | 
|  | 204 | +            if idx < elen:
 | 
|  | 205 | +                yield from __expand(content[bits[idx]])
 | 
|  | 206 | +            idx += 1
 | 
|  | 207 | +
 | 
|  | 208 | +    return "".join(__expand(topvalue)) |