Implement processing of conditions (#276)

Implement processing of conditions

This is not a full support for conditions, i.e. it doesn't re-solve #231 where an entire section can be conditionalized (however, the new process_conditions() function is generic enough to be used as a base for the resolution of that issue), nevertheless I think it's quite an improvement.
Note that macro definitions have to be parsed twice, because a macro definition can contain a condition (see https://src.fedoraproject.org/rpms/kernel/blob/rawhide/f/kernel.spec for examples), and vice versa, a condition can encapsulate a macro definition.
Macro definitions and tags gained a valid attribute that can be used to determine if that particular macro definition/tag is valid and can affect other entities in the spec file or if it would be ignored when parsing the spec file with RPM. This will be used to fix Specfile.update_value() and Specfile.update_tag(), but it could be beneficial for other use cases as well.
Related to packit/packit#2033.
RELEASE NOTES BEGIN
Macro definitions and tags gained a new valid attribute. A macro definition/tag is considered valid if it doesn't appear in a false branch of any condition appearing in the spec file.
RELEASE NOTES END

Reviewed-by: František Lachman <[email protected]>
Reviewed-by: Nikola Forró

Author: softwarefactory-project-zuul[bot]

specfile/conditions.py

@@ -0,0 +1,137 @@
+# Copyright Contributors to the Packit project.
+# SPDX-License-Identifier: MIT
+
+import re
+from typing import TYPE_CHECKING, List, Optional, Tuple
+
+from specfile.exceptions import RPMException
+from specfile.macros import Macros
+
+if TYPE_CHECKING:
+    from specfile.macro_definitions import MacroDefinitions
+    from specfile.specfile import Specfile
+
+
+def resolve_expression(
+    keyword: str, expression: str, context: Optional["Specfile"] = None
+) -> bool:
+    """
+    Resolves a RPM expression.
+
+    Args:
+        keyword: Condition keyword, e.g. `%if` or `%ifarch`.
+        expression: Expression string or a whitespace-delimited list
+          of arches/OSes in case keyword is a variant of `%ifarch`/`%ifos`.
+        context: `Specfile` instance that defines the context for macro expansions.
+
+    Returns:
+        Resolved expression as a boolean value.
+    """
+
+    def expand(s):
+        if not context:
+            return Macros.expand(s)
+        result = context.expand(s, skip_parsing=getattr(expand, "skip_parsing", False))
+        # parse only once
+        expand.skip_parsing = True
+        return result
+
+    if keyword in ("%if", "%elif"):
+        try:
+            result = expand(f"%{{expr:{expression}}}")
+        except RPMException:
+            return False
+        try:
+            return int(result) != 0
+        except ValueError:
+            return True
+    elif keyword.endswith("arch"):
+        target_cpu = expand("%{_target_cpu}")
+        match = any(t for t in expression.split() if t == target_cpu)
+        return not match if keyword == "%ifnarch" else match
+    elif keyword.endswith("os"):
+        target_os = expand("%{_target_os}")
+        match = any(t for t in expression.split() if t == target_os)
+        return not match if keyword == "%ifnos" else match
+    return False
+
+
+def process_conditions(
+    lines: List[str],
+    macro_definitions: Optional["MacroDefinitions"] = None,
+    context: Optional["Specfile"] = None,
+) -> List[Tuple[str, bool]]:
+    """
+    Processes conditions in a spec file. Takes a list of lines and returns the same
+    list of lines extended with information about their validity. A line is considered
+    valid if it doesn't appear in a false branch of any condition.
+
+    Args:
+        lines: List of lines in a spec file.
+        macro_definitions: Parsed macro definitions to be used to prevent parsing conditions
+          inside their bodies (and most likely failing).
+        context: `Specfile` instance that defines the context for macro expansions.
+
+    Returns:
+        List of tuples in the form of (line, validity).
+    """
+    excluded_lines = []
+    for md in macro_definitions or []:
+        position = md.get_position(macro_definitions)
+        excluded_lines.append(range(position, position + len(md.body.splitlines())))
+    condition_regex = re.compile(
+        r"""
+        ^
+        \s*                                           # optional preceding whitespace
+        (?P<kwd>%((el)?if(n?(arch|os))?|endif|else))  # keyword
+        \s*
+        (
+            \s+
+            (?P<expr>.*?)                             # expression
+            (?P<end>\s*|\\)                           # optional following whitespace
+                                                      # or a backslash indicating
+                                                      # that the expression continues
+                                                      # on the next line
+        )?
+        $
+        """,
+        re.VERBOSE,
+    )
+    result = []
+    branches = [True]
+    indexed_lines = list(enumerate(lines))
+    while indexed_lines:
+        index, line = indexed_lines.pop(0)
+        # ignore conditions inside macro definition body
+        if any(index in r for r in excluded_lines):
+            result.append((line, branches[-1]))
+            continue
+        m = condition_regex.match(line)
+        if not m:
+            result.append((line, branches[-1]))
+            continue
+        keyword = m.group("kwd")
+        if keyword == "%endif":
+            result.append((line, branches[-2]))
+            branches.pop()
+        elif keyword.startswith("%el"):
+            result.append((line, branches[-2]))
+            branches[-1] = not branches[-1]
+        else:
+            result.append((line, branches[-1]))
+        expression = m.group("expr")
+        if expression:
+            if m.group("end") == "\\":
+                expression += "\\"
+            while expression.endswith("\\") and indexed_lines:
+                _, line = indexed_lines.pop(0)
+                result.append((line, branches[-1]))
+                expression = expression[:-1] + line
+            branch = (
+                False if not branches[-1] else resolve_expression(keyword, expression)
+            )
+            if keyword.startswith("%el"):
+                branches[-1] = branch
+            else:
+                branches.append(branch)
+    return result

specfile/macro_definitions.py

@@ -4,11 +4,15 @@
 import collections
 import copy
 import re
-from typing import List, Optional, Tuple, Union, overload
+from typing import TYPE_CHECKING, List, Optional, Tuple, Union, overload
 
+from specfile.conditions import process_conditions
 from specfile.formatter import formatted
 from specfile.types import SupportsIndex
 
+if TYPE_CHECKING:
+    from specfile.specfile import Specfile
+
 
 class MacroDefinition:
     def __init__(
@@ -17,12 +21,14 @@ def __init__(
         body: str,
         is_global: bool,
         whitespace: Tuple[str, str, str, str],
+        valid: bool = True,
         preceding_lines: Optional[List[str]] = None,
     ) -> None:
         self.name = name
         self.body = body
         self.is_global = is_global
         self._whitespace = whitespace
+        self.valid = valid
         self._preceding_lines = (
             preceding_lines.copy() if preceding_lines is not None else []
         )
@@ -42,7 +48,7 @@ def __eq__(self, other: object) -> bool:
     def __repr__(self) -> str:
         return (
             f"MacroDefinition({self.name!r}, {self.body!r}, {self.is_global!r}, "
-            f"{self._whitespace!r}, {self._preceding_lines!r})"
+            f"{self._whitespace!r}, {self.valid!r}, {self._preceding_lines!r})"
         )
 
     def __str__(self) -> str:
@@ -189,7 +195,9 @@ def find(self, name: str) -> int:
         raise ValueError
 
     @classmethod
-    def parse(cls, lines: List[str]) -> "MacroDefinitions":
+    def _parse(
+        cls, lines: Union[List[str], List[Tuple[str, bool]]]
+    ) -> "MacroDefinitions":
         """
         Parses given lines into macro defintions.
 
@@ -200,6 +208,13 @@ def parse(cls, lines: List[str]) -> "MacroDefinitions":
             Constructed instance of `MacroDefinitions` class.
         """
 
+        def pop(lines):
+            line = lines.pop(0)
+            if isinstance(line, str):
+                return line, True
+            else:
+                return line
+
         def count_brackets(s):
             bc = pc = 0
             chars = list(s)
@@ -248,7 +263,7 @@ def count_brackets(s):
         buffer: List[str] = []
         lines = lines.copy()
         while lines:
-            line = lines.pop(0)
+            line, valid = pop(lines)
             m = md_regex.match(line)
             if m:
                 ws0, macro, ws1, name, ws2, body, ws3 = m.groups()
@@ -257,7 +272,7 @@ def count_brackets(s):
                     ws3 = ""
                 bc, pc = count_brackets(body)
                 while (bc > 0 or pc > 0 or body.endswith("\\")) and lines:
-                    line = lines.pop(0)
+                    line, _ = pop(lines)
                     body += "\n" + line
                     bc, pc = count_brackets(body)
                 tokens = re.split(r"(\s+)$", body, maxsplit=1)
@@ -268,14 +283,43 @@ def count_brackets(s):
                     ws3 = ws + ws3
                 data.append(
                     MacroDefinition(
-                        name, body, macro == "%global", (ws0, ws1, ws2, ws3), buffer
+                        name,
+                        body,
+                        macro == "%global",
+                        (ws0, ws1, ws2, ws3),
+                        valid,
+                        buffer,
                     )
                 )
                 buffer = []
             else:
                 buffer.append(line)
         return cls(data, buffer)
 
+    @classmethod
+    def parse(
+        cls,
+        lines: List[str],
+        with_conditions: bool = False,
+        context: Optional["Specfile"] = None,
+    ) -> "MacroDefinitions":
+        """
+        Parses given lines into macro defintions.
+
+        Args:
+            lines: Lines to parse.
+            with_conditions: Whether to process conditions before parsing and populate
+              the `valid` attribute.
+            context: `Specfile` instance that defines the context for macro expansions.
+
+        Returns:
+            Constructed instance of `MacroDefinitions` class.
+        """
+        result = cls._parse(lines)
+        if not with_conditions:
+            return result
+        return cls._parse(process_conditions(lines, result, context))
+
     def get_raw_data(self) -> List[str]:
         result = []
         for macro_definition in self.data:

specfile/sources.py

@@ -458,6 +458,21 @@ def _get_initial_tag_setup(self, number: int = 0) -> Tuple[int, str, str]:
             suffix = f"{number:0{self._default_source_number_digits}}"
         return len(self._tags) if self._tags else 0, f"{self.prefix}{suffix}", ": "
 
+    def _get_tag_validity(self, reference: Optional[TagSource] = None) -> bool:
+        """
+        Determines validity of a new source tag based on a reference tag, if specified,
+        or the last tag in the spec file. Defaults to True.
+
+        Args:
+            reference: Optional reference tag source.
+
+        Returns:
+            Whether the new source tag is valid or not.
+        """
+        if reference is not None:
+            return reference._tag.valid
+        return self._tags[-1].valid if self._tags else True
+
     def _deduplicate_tag_names(self, start: int = 0) -> None:
         """
         Eliminates duplicate numbers in source tag names.
@@ -505,9 +520,17 @@ def insert(self, i: int, location: str) -> None:
                 number = source.number
             if isinstance(source, self.tag_class):
                 name, separator = self._get_tag_format(cast(TagSource, source), number)
+                valid = self._get_tag_validity(cast(TagSource, source))
                 container.insert(
                     index,
-                    Tag(name, location, separator, Comments(), context=self._context),
+                    Tag(
+                        name,
+                        location,
+                        separator,
+                        Comments(),
+                        valid,
+                        context=self._context,
+                    ),
                 )
                 self._deduplicate_tag_names(i)
             else:
@@ -523,9 +546,12 @@ def insert(self, i: int, location: str) -> None:
             )
         else:
             index, name, separator = self._get_initial_tag_setup()
+            valid = self._get_tag_validity()
             self._tags.insert(
                 index,
-                Tag(name, location, separator, Comments(), context=self._context),
+                Tag(
+                    name, location, separator, Comments(), valid, context=self._context
+                ),
             )
 
     def insert_numbered(self, number: int, location: str) -> int:
@@ -555,11 +581,14 @@ def insert_numbered(self, number: int, location: str) -> int:
                 i += 1
                 index += 1
             name, separator = self._get_tag_format(source, number)
+            valid = self._get_tag_validity(source)
         else:
             i = 0
             index, name, separator = self._get_initial_tag_setup(number)
+            valid = self._get_tag_validity()
         self._tags.insert(
-            index, Tag(name, location, separator, Comments(), context=self._context)
+            index,
+            Tag(name, location, separator, Comments(), valid, context=self._context),
         )
         self._deduplicate_tag_names(i)
         return i

specfile/specfile.py

@@ -223,7 +223,9 @@ def macro_definitions(self) -> Generator[MacroDefinitions, None, None]:
             Macro definitions in the spec file as `MacroDefinitions` object.
         """
         with self.lines() as lines:
-            macro_definitions = MacroDefinitions.parse(lines)
+            macro_definitions = MacroDefinitions.parse(
+                lines, with_conditions=True, context=self
+            )
             try:
                 yield macro_definitions
             finally: