Handle multi-line tag values (#412)

Handle multi-line tag values

Fixes #410.
RELEASE NOTES BEGIN
specfile can now handle multi-line tag values (enclosed in a macro body, e.g. %shrink).
RELEASE NOTES END

Reviewed-by: Maja Massarini
Reviewed-by: Nikola Forró

Author: softwarefactory-project-zuul[bot]

specfile/macro_definitions.py

@@ -9,7 +9,7 @@
 from specfile.conditions import process_conditions
 from specfile.formatter import formatted
 from specfile.types import SupportsIndex
-from specfile.utils import UserList
+from specfile.utils import UserList, count_brackets
 
 if TYPE_CHECKING:
     from specfile.specfile import Specfile
@@ -303,35 +303,6 @@ def pop(lines):
             else:
                 return line
 
-        def count_brackets(s):
-            bc = pc = 0
-            chars = list(s)
-            while chars:
-                c = chars.pop(0)
-                if c == "\\" and chars:
-                    chars.pop(0)
-                    continue
-                if c == "%" and chars:
-                    c = chars.pop(0)
-                    if c == "{":
-                        bc += 1
-                    elif c == "(":
-                        pc += 1
-                    continue
-                if c == "{" and bc > 0:
-                    bc += 1
-                    continue
-                if c == "}" and bc > 0:
-                    bc -= 1
-                    continue
-                if c == "(" and pc > 0:
-                    pc += 1
-                    continue
-                if c == ")" and pc > 0:
-                    pc -= 1
-                    continue
-            return bc, pc
-
         md_regex = re.compile(
             r"""
             ^

specfile/tags.py

@@ -23,7 +23,7 @@
 from specfile.macros import Macros
 from specfile.sections import Section
 from specfile.types import SupportsIndex
-from specfile.utils import UserList, split_conditional_macro_expansion
+from specfile.utils import UserList, count_brackets, split_conditional_macro_expansion
 
 if TYPE_CHECKING:
     from specfile.specfile import Specfile
@@ -489,6 +489,13 @@ def parse(cls, section: Section, context: Optional["Specfile"] = None) -> "Tags"
             New instance of `Tags` class.
         """
 
+        def pop(lines):
+            line = lines.pop(0)
+            if isinstance(line, str):
+                return line, True
+            else:
+                return line
+
         def regex_pattern(tag):
             name_regex = get_tag_name_regex(tag)
             return rf"^(?P<n>{name_regex})(?P<s>\s*:\s*)(?P<v>.+)"
@@ -498,7 +505,8 @@ def regex_pattern(tag):
         tag_regexes = [re.compile(regex_pattern(t), re.IGNORECASE) for t in TAG_NAMES]
         data = []
         buffer: List[str] = []
-        for line, valid in lines:
+        while lines:
+            line, valid = pop(lines)
             ws = ""
             tokens = re.split(r"([^\S\n]+)$", line, maxsplit=1)
             if len(tokens) > 1:
@@ -507,10 +515,21 @@ def regex_pattern(tag):
             # find out if there is a match for one of the tag regexes
             m = next((m for m in (r.match(line) for r in tag_regexes) if m), None)
             if m:
+                value = m.group("v")
+                if not suffix:
+                    bc, pc = count_brackets(value)
+                    while (bc > 0 or pc > 0) and lines:
+                        value += ws
+                        line, _ = pop(lines)
+                        tokens = re.split(r"([^\S\n]+)$", line, maxsplit=1)
+                        if len(tokens) > 1:
+                            line, ws, _ = tokens
+                        value += "\n" + line
+                        bc, pc = count_brackets(value)
                 data.append(
                     Tag(
                         m.group("n"),
-                        m.group("v"),
+                        value,
                         m.group("s"),
                         Comments.parse(buffer),
                         valid,
@@ -534,8 +553,10 @@ def get_raw_section_data(self) -> List[str]:
         result = []
         for tag in self.data:
             result.extend(tag.comments.get_raw_data())
-            result.append(
-                f"{tag._prefix}{tag.name}{tag._separator}{tag.value}{tag._suffix}"
+            result.extend(
+                f"{tag._prefix}{tag.name}{tag._separator}{tag.value}{tag._suffix}".split(
+                    "\n"
+                )
             )
         result.extend(self._remainder)
         return result

specfile/utils.py

@@ -254,6 +254,45 @@ def get_filename_from_location(location: str) -> str:
     return location[slash + 1 :].split("=")[-1]
 
 
+def count_brackets(string: str) -> Tuple[int, int]:
+    """
+    Counts non-pair brackets in %{...} and %(...) expressions appearing in the given string.
+
+    Args:
+        string: Input string.
+
+    Returns:
+        The count of non-pair curly braces and the count of non-pair parentheses.
+    """
+    bc = pc = 0
+    chars = list(string)
+    while chars:
+        c = chars.pop(0)
+        if c == "\\" and chars:
+            chars.pop(0)
+            continue
+        if c == "%" and chars:
+            c = chars.pop(0)
+            if c == "{":
+                bc += 1
+            elif c == "(":
+                pc += 1
+            continue
+        if c == "{" and bc > 0:
+            bc += 1
+            continue
+        if c == "}" and bc > 0:
+            bc -= 1
+            continue
+        if c == "(" and pc > 0:
+            pc += 1
+            continue
+        if c == ")" and pc > 0:
+            pc -= 1
+            continue
+    return bc, pc
+
+
 def split_conditional_macro_expansion(value: str) -> Tuple[str, str, str]:
     """
     Splits conditional macro expansion into its body and prefix and suffix of it.

tests/unit/test_tags.py

@@ -43,6 +43,11 @@ def test_parse():
                 "Epoch:   1",
                 "%endif",
                 "",
+                "License: %{shrink:",
+                "             MIT AND",
+                "             (MIT OR Apache-2.0)",
+                "          }",
+                "",
                 "Requires:          make ",
                 "Requires(post):    bash",
                 "",
@@ -62,6 +67,13 @@ def test_parse():
     assert not tags[1].comments
     assert tags.release.comments[0].prefix == "  # "
     assert tags.epoch.name == "Epoch"
+    assert tags[-6].name == "License"
+    assert (
+        tags[-6].value == "%{shrink:\n"
+        "             MIT AND\n"
+        "             (MIT OR Apache-2.0)\n"
+        "          }"
+    )
     assert tags.requires.value == "make"
     assert "requires(post)" in tags
     assert tags[-4].name == "Requires(post)"
@@ -102,11 +114,20 @@ def test_get_raw_section_data():
                 Comments([Comment("this is a valid comment", "  # ")]),
             ),
             Tag("Epoch", "1", ":   ", Comments([], ["", "%if 0"])),
+            Tag(
+                "License",
+                "%{shrink:\n"
+                "             MIT AND\n"
+                "             (MIT OR Apache-2.0)\n"
+                "          }",
+                ": ",
+                Comments([], ["%endif", ""]),
+            ),
             Tag(
                 "Requires",
                 "make",
                 ":          ",
-                Comments([], ["%endif", ""]),
+                Comments([], [""]),
                 True,
                 "",
                 " ",
@@ -141,6 +162,11 @@ def test_get_raw_section_data():
         "Epoch:   1",
         "%endif",
         "",
+        "License: %{shrink:",
+        "             MIT AND",
+        "             (MIT OR Apache-2.0)",
+        "          }",
+        "",
         "Requires:          make ",
         "Requires(post):    bash",
         "",

tests/unit/test_utils.py

@@ -3,7 +3,7 @@
 
 import pytest
 
-from specfile.utils import EVR, NEVR, NEVRA, get_filename_from_location
+from specfile.utils import EVR, NEVR, NEVRA, count_brackets, get_filename_from_location
 
 
 @pytest.mark.parametrize(
@@ -32,6 +32,23 @@ def test_get_filename_from_location(location, filename):
     assert get_filename_from_location(location) == filename
 
 
+@pytest.mark.parametrize(
+    "string, count",
+    [
+        ("", (0, 0)),
+        ("%macro", (0, 0)),
+        ("%{macro}", (0, 0)),
+        ("%{{macro}}", (0, 0)),
+        ("%{{macro}", (1, 0)),
+        ("%{macro:", (1, 0)),
+        ("%(echo %{v}", (0, 1)),
+        ("%(echo %{v} | cut -d. -f3)", (0, 0)),
+    ],
+)
+def test_count_brackets(string, count):
+    assert count_brackets(string) == count
+
+
 def test_EVR_compare():
     assert EVR(version="0") == EVR(version="0")
     assert EVR(version="0", release="1") != EVR(version="0", release="2")