#!/usr/bin/env bash
# verify-fork-functions.sh — Verify that local forks retain ALL upstream functions.
#
# This is the FUNCTION-LEVEL verification that prevents the silent code-loss bug
# that occurred when git merge conflict resolution dropped upstream functions
# (kfdwrite, bulk_insert_files, resize, read_with_timeout, etc.) from the kernel fork.
#
# For each fork with an upstream remote, this script:
#   1. Extracts all function definitions (fn <name>) from upstream source files
#   2. Checks that each function exists in the local fork
#   3. Reports any missing functions with file paths and line numbers
#
# This catches the class of bugs that file-level verification CANNOT detect:
# when a file is touched by RB patches, file-level diff allows ANY difference,
# including silently dropped upstream functions.
#
# Usage:
#   ./local/scripts/verify-fork-functions.sh                    # Check all forks
#   ./local/scripts/verify-fork-functions.sh kernel relibc      # Check specific forks
#   ./local/scripts/verify-fork-functions.sh --no-fetch         # Skip network fetch
#   ./local/scripts/verify-fork-functions.sh --quiet            # Only print violations
#
# Exit codes:
#   0 = All upstream functions present in all forks
#   1 = One or more upstream functions are missing from forks

set -uo pipefail

SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"

NO_FETCH=0
QUIET=0
declare -a TARGET_FORKS=()

while [[ $# -gt 0 ]]; do
    case "$1" in
        --no-fetch) NO_FETCH=1 ;;
        --quiet)    QUIET=1 ;;
        -h|--help)
            echo "Usage: $0 [--no-fetch] [--quiet] [fork1 fork2 ...]"
            echo ""
            echo "Verifies that local forks retain ALL upstream functions."
            echo "Catches silent code loss from bad merge conflict resolution."
            exit 0
            ;;
        --*) echo "Unknown: $1" >&2; exit 1 ;;
        *) TARGET_FORKS+=("$1") ;;
    esac
    shift
done

cd "$PROJECT_ROOT"

# Discover forks if not specified
if [[ ${#TARGET_FORKS[@]} -eq 0 ]]; then
    while IFS= read -r line; do
        TARGET_FORKS+=("$line")
    done < <(for d in local/sources/*/; do
        [[ -d "$d/.git" ]] || continue
        name=$(basename "$d")
        git -C "$d" remote get-url upstream >/dev/null 2>&1 && echo "$name"
    done | sort)
fi

VIOLATIONS=0
TOTAL_MISSING=0

for fork in "${TARGET_FORKS[@]}"; do
    fork_dir="local/sources/${fork}"

    if [[ ! -d "$fork_dir/.git" ]]; then
        continue
    fi

    upstream_url=$(cd "$fork_dir" && git remote get-url upstream 2>/dev/null || true)
    if [[ -z "$upstream_url" ]]; then
        continue
    fi

    # Determine upstream branch
    upstream_branch="master"
    if ! (cd "$fork_dir" && git rev-parse --verify upstream/master >/dev/null 2>&1); then
        if (cd "$fork_dir" && git rev-parse --verify upstream/main >/dev/null 2>&1); then
            upstream_branch="main"
        else
            if [[ "$NO_FETCH" -eq 0 ]]; then
                (cd "$fork_dir" && git fetch upstream --depth=200 --quiet 2>/dev/null) || true
            fi
            if (cd "$fork_dir" && git rev-parse --verify upstream/master >/dev/null 2>&1); then
                upstream_branch="master"
            elif (cd "$fork_dir" && git rev-parse --verify upstream/main >/dev/null 2>&1); then
                upstream_branch="main"
            else
                continue
            fi
        fi
    fi

    # Fetch if needed
    if [[ "$NO_FETCH" -eq 0 ]]; then
        (cd "$fork_dir" && git fetch upstream --depth=200 --quiet 2>/dev/null) || true
    fi

    upstream_ref="upstream/$upstream_branch"

    # Get list of source files that exist in BOTH upstream and our fork
    upstream_files=$(cd "$fork_dir" && git ls-tree -r --name-only "$upstream_ref" 2>/dev/null | grep '\.rs$' | sort)
    local_files=$(cd "$fork_dir" && find . -name '*.rs' -not -path './.git/*' -not -path './target/*' -not -path './stage/*' | sed 's|^\./||' | sort)
    mapfile -t shared_files < <(comm -12 <(echo "$upstream_files") <(echo "$local_files"))

    fork_missing=0
    declare -a missing_details=()

    for f in "${shared_files[@]}"; do
        # Extract function names from upstream version of the file
        upstream_fns=$(cd "$fork_dir" && git show "${upstream_ref}:${f}" 2>/dev/null | \
            grep -oP '(?:pub )?(?:async )?(?:unsafe )?fn \w+' | \
            sed 's/fn //' | sed 's/ *$//' | sort -u)

        [[ -z "$upstream_fns" ]] && continue

        # Extract function names from our version of the file
        local_fns=$(cd "$fork_dir" && grep -oP '(?:pub )?(?:async )?(?:unsafe )?fn \w+' "$f" 2>/dev/null | \
            sed 's/fn //' | sed 's/ *$//' | sort -u)

        # Load per-fork exclusion list for intentionally removed/replaced functions
        exclude_file="${fork_dir}/.verify-fork-functions.exclude"
        declare -A excluded=()
        if [[ -f "$exclude_file" ]]; then
            while IFS= read -r line; do
                [[ "$line" =~ ^# ]] && continue
                [[ -z "$line" ]] && continue
                excluded["$line"]=1
            done < "$exclude_file"
        fi

        # Find functions in upstream but not in our fork
        missing=$(comm -23 <(echo "$upstream_fns") <(echo "$local_fns") 2>/dev/null)

        if [[ -n "$missing" ]]; then
            while IFS= read -r fn; do
                [[ -z "$fn" ]] && continue
                # Check exclusion list: "$f:$fn"
                if [[ -n "${excluded["$f:$fn"]:-}" ]]; then
                    [[ "$QUIET" -eq 0 ]] && missing_details+=("  $f: fn $fn → EXCLUDED (RB intentional)")
                    continue
                fi
                # Strip modifiers to get bare function name for cross-file search
                bare_fn=$(echo "$fn" | sed -E 's/^(pub |async |unsafe )+//')
                found_elsewhere=$(cd "$fork_dir" && \
                    grep -rlP "(?:pub )?(?:async )?(?:unsafe )?fn ${bare_fn}\b" \
                        --include='*.rs' --exclude-dir='.git' --exclude-dir='target' --exclude-dir='stage' . 2>/dev/null | \
                    grep -v "^\./${f}$" | head -1 | sed 's|^\./||')
                if [[ -n "$found_elsewhere" ]]; then
                    [[ "$QUIET" -eq 0 ]] && missing_details+=("  $f: fn $fn → MOVED to $found_elsewhere")
                else
                    missing_details+=("  $f: fn $fn")
                    fork_missing=$((fork_missing + 1))
                    TOTAL_MISSING=$((TOTAL_MISSING + 1))
                fi
            done <<< "$missing"
        fi
    done

    if [[ "$fork_missing" -gt 0 ]]; then
        if [[ "$QUIET" -eq 0 ]] || [[ "$fork_missing" -gt 0 ]]; then
            echo ""
            echo -e "\033[1;31mFAIL: $fork is missing $fork_missing upstream function(s):\033[0m"
            for detail in "${missing_details[@]}"; do
                echo "$detail"
            done
            echo ""
            echo "  These functions exist in upstream $upstream_ref but are MISSING from the local fork."
            echo "  This likely means a bad merge or cherry-pick dropped them silently."
            echo "  Fix: ./local/scripts/upgrade-forks.sh $fork"
        fi
        VIOLATIONS=$((VIOLATIONS + 1))
    elif [[ "$QUIET" -eq 0 ]]; then
        echo -e "\033[1;32mOK: $fork — all upstream functions present\033[0m"
    fi
done

echo ""
if [[ "$VIOLATIONS" -gt 0 ]]; then
    echo -e "\033[1;31mFUNCTION VERIFICATION FAILED: $VIOLATIONS fork(s) missing $TOTAL_MISSING upstream function(s).\033[0m"
    exit 1
fi

echo -e "\033[1;32mAll forks retain all upstream functions.\033[0m"
exit 0