srcinfo-cache: include extra metadata extracted from the PKGBUILDs

Bash variables starting with msys2_ or mingw_ in the PKGBUILD will
now be included in the srcinfo cache under "extra". This supports
both strings and arrays of strings.

msys2-srcinfo-cache

@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+
+from msys2_devtools import srcinfo_cache
+
+srcinfo_cache.run()

msys2_devtools/pkgbuild.py

@@ -0,0 +1,27 @@
+from typing import Collection, Dict, Union
+import json
+import subprocess
+import os
+
+PKGBUILD2JSON = os.path.join(os.path.dirname(__file__), "pkgbuild2json.sh")
+
+
+def get_extra_meta_for_pkgbuild(msys2_root: str, pkgbuild_path: str) -> \
+        Dict[str, Union[str, Collection[str], Dict[str, Union[str, None]]]]:
+    """Returns a dict with the MSYS2 specific metadata from the PKGBUILD file"""
+
+    executable = os.path.join(msys2_root, 'usr', 'bin', 'bash.exe')
+    prefixes = ["mingw_", "msys2_"]
+    out = subprocess.check_output(
+        [executable, PKGBUILD2JSON, pkgbuild_path] + prefixes,
+        text=True, encoding="utf-8")
+    data = json.loads(out)
+
+    meta = {}
+    for key, value in data.items():
+        if key == "mingw_arch":
+            key = "msys2_environments"
+        if key.startswith("msys2_"):
+            key = key.split("_", 1)[-1]
+            meta[key] = value
+    return meta

msys2_devtools/pkgbuild2json.sh

@@ -0,0 +1,84 @@
+#!/bin/bash
+# Sources a PKGBUILD and outputs a json representation of all variables in it.
+# Usage: ./src2json.sh PKGBUILD [prefix1 prefix2...]
+# Example Output:
+# {
+#   "prefix1_name1": "value1",
+#   "prefix2_name2": ["element1", "element2", ...]
+# }
+
+set -e
+
+escape_json_string() {
+    local json_string
+    json_string="${1//\\/\\\\}"
+    json_string="${json_string//\"/\\\"}"
+    json_string="${json_string//$'\n'/\\n}"
+    json_string="${json_string//$'\t'/\\t}"
+    json_string="${json_string//$'\r'/\\r}"
+    echo -n "$json_string"
+}
+
+# Given a variable name returns a json value representation of it
+get_variable_as_json() {
+    local var_name="$1"
+
+    if [[ "$(declare -p "$var_name" 2>/dev/null)" =~ "declare -a" ]]; then
+        local -n var_ref="$var_name"
+        local elements=""
+        for element in "${var_ref[@]}"; do
+            local escaped_element
+            escaped_element="$(escape_json_string "$element")"
+            elements+="\"$escaped_element\","
+        done
+        elements="${elements%,*}"
+        echo "[$elements]"
+    else
+        local var_value
+        local escape_value
+        eval "var_value=\$$var_name"
+        escape_value="$(escape_json_string "$var_value")"
+        echo "\"$escape_value\""
+    fi
+}
+
+check_prefixes() {
+    local variable="$1"
+    shift
+    for prefix in "$@"; do
+        if [[ "$variable" == "$prefix"* ]]; then
+            return 0
+        fi
+    done
+    return 1
+}
+
+# Given a file name returns a json representation of all variables in it.
+# Supports strings and arrays of strings.
+get_variables_as_json() {
+    set +e
+    source "$1"
+    set -e
+    shift
+    local variable_names
+    variable_names=$(compgen -v)
+    local json=""
+    for var in $variable_names; do
+        # the main reason we don't just extract all is to make it faster
+        # with cygwin bash
+        if check_prefixes "$var" "$@"; then
+            json+="\"$(escape_json_string "$var")\": $(get_variable_as_json "$var"),"
+            continue
+        fi
+    done
+    json="{${json%,*}}"
+    echo -n "$json"
+}
+
+main() {
+    local pkgbuild="$1"
+    shift
+    get_variables_as_json "$pkgbuild" "$@"
+}
+
+main "$@"

msys2_devtools/srcinfo_cache.py

@@ -12,6 +12,7 @@ from concurrent.futures import ThreadPoolExecutor
 from functools import partial
 
 from typing import List, Iterator, Tuple, Dict, Optional, Union, Collection, Sequence, Any
+from .pkgbuild import get_extra_meta_for_pkgbuild
 
 
 CacheEntry = Dict[str, Union[str, Collection[str]]]
@@ -127,7 +128,9 @@ def get_srcinfo_for_pkgbuild(msys2_root: str, args: Tuple[str, str]) -> Optional
             ["git", "log", "-1", "--format=%aI", git_path],
             cwd=git_cwd).decode("utf-8").strip()
 
-        meta = {"repo": repo, "path": relpath, "date": date, "srcinfo": srcinfos}
+        extra_meta = get_extra_meta_for_pkgbuild(msys2_root, pkgbuild_path)
+
+        meta = {"repo": repo, "path": relpath, "date": date, "srcinfo": srcinfos, "extra": extra_meta}
     except subprocess.CalledProcessError as e:
         print("ERROR: %s %s" % (pkgbuild_path, e.output.splitlines()))
         return None