#!/usr/bin/env bash
# okf-validate.sh — validador OKF v0.1 para o Hub Obsidian
# Uso: okf-validate.sh [--all] [--strict] [file ...]
# --all   valida todos os .md do Hub (exclui exclusões)
# --strict aplica regras layer:wiki (description≥45ch, tags≥2)
# Sem args + stdin: valida ficheiros staged (pre-commit hook)

set -euo pipefail

HUB_ROOT="/media/ealmeida/Dados/Hub"
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"

# --- Exclusion patterns (relative to HUB_ROOT) ---
EXCLUDE_PATTERNS=(
    "99-Arquivo/"
    ".stversions/"
    ".obsidian/"
    ".ijfw/"
    "_templates/"
    "90-Templates/"
    ".git/"
    "node_modules/"
    ".github/"
    ".wayland/"
    ".hermes/"
)

# Exclude specific filenames
EXCLUDE_FILES=(
    "MEMORY.md"
    "CLAUDE.md"
    "GEMINI.md"
    "AGENTS.md"
    "copilot-instructions.md"
)

# Canonical type vocabulary (§5 of OKF-Compliance-Plan-Global.md)
CANONICAL_TYPES=(
    "Playbook"
    "Reference"
    "Specification"
    "Status"
    "Changelog"
    "Index"
    "Template"
    "Client Profile"
    "Document"
    "note"
)

ERRORS=0
WARNINGS=0
CHECKED=0
STRICT=false

RED='\033[0;31m'
YEL='\033[0;33m'
GRN='\033[0;32m'
NC='\033[0m'

is_excluded() {
    local f="$1"
    local rel="${f#$HUB_ROOT/}"

    # Check directory patterns
    for pat in "${EXCLUDE_PATTERNS[@]}"; do
        if [[ "$rel" == *"$pat"* ]]; then
            return 0
        fi
    done

    # Check filename patterns
    local base
    base="$(basename "$rel")"
    for pat in "${EXCLUDE_FILES[@]}"; do
        if [[ "$base" == "$pat" ]]; then
            return 0
        fi
    done

    return 1
}

is_index_subfolder() {
    # index.md in subfolders don't need frontmatter (anti-pattern)
    local f="$1"
    local base
    base="$(basename "$f")"
    local dir
    dir="$(dirname "$f")"
    local parent_base
    parent_base="$(basename "$dir")"

    # Root index.md DOES need frontmatter
    if [[ "$dir" == "$HUB_ROOT" ]]; then
        return 1
    fi

    if [[ "$base" == "index.md" ]]; then
        return 0
    fi
    return 1
}

is_log_file() {
    local base
    base="$(basename "$1")"
    [[ "$base" == "log.md" ]]
}

validate_file() {
    local f="$1"
    local rel="${f#$HUB_ROOT/}"
    local content

    if [[ ! -f "$f" ]]; then
        echo -e "${RED}ERROR${NC} $rel: file not found"
        ((ERRORS++)) || true
        return
    fi

    ((CHECKED++)) || true

    # Reservados OKF (index.md em qualquer nível, log.md) — isentos de todas as verificações
    local base; base="$(basename "$f")"
    if [[ "$base" == "index.md" || "$base" == "log.md" ]]; then
        return
    fi

    # Check frontmatter
    if ! head -1 "$f" | grep -q '^---$'; then
        # log.md and subfolder index.md are exempt
        if is_log_file "$f" || is_index_subfolder "$f"; then
            return
        fi
        echo -e "${RED}ERROR${NC} $rel: missing frontmatter"
        ((ERRORS++)) || true
        return
    fi

    # Extract frontmatter using awk (robust: handles leading whitespace, multiline values)
    content="$(awk 'BEGIN{n=0} /^---[[:space:]]*$/{n++; if(n==2) exit} n==1 && n>0{print}' "$f")"

    if [[ -z "$content" ]]; then
        echo -e "${RED}ERROR${NC} $rel: empty or invalid frontmatter"
        ((ERRORS++)) || true
        return
    fi

    # Parse YAML fields (simple grep — no Python dependency)
    local has_type has_title has_timestamp has_desc has_tags has_layer

    has_type=$(echo "$content" | grep -c '^type:' || true)
    has_title=$(echo "$content" | grep -c '^title:' || true)
    has_timestamp=$(echo "$content" | grep -c '^timestamp:' || true)
    has_desc=$(echo "$content" | grep -c '^description:' || true)
    has_tags=$(echo "$content" | grep -c '^tags:' || true)
    has_layer=$(echo "$content" | grep -c '^layer:' || true)

    # Required for all
    if [[ "$has_type" -eq 0 ]]; then
        echo -e "${RED}ERROR${NC} $rel: missing 'type'"
        ((ERRORS++)) || true
    fi
    if [[ "$has_title" -eq 0 ]]; then
        echo -e "${RED}ERROR${NC} $rel: missing 'title'"
        ((ERRORS++)) || true
    fi
    if [[ "$has_timestamp" -eq 0 ]]; then
        echo -e "${RED}ERROR${NC} $rel: missing 'timestamp'"
        ((ERRORS++)) || true
    fi

    # Strict checks (layer:wiki or --strict flag)
    if [[ "$STRICT" == "true" ]] || [[ "$has_layer" -gt 0 ]] && echo "$content" | grep -q '^layer: wiki'; then
        if [[ "$has_desc" -eq 0 ]]; then
            echo -e "${RED}ERROR${NC} $rel [wiki]: missing 'description'"
            ((ERRORS++)) || true
        else
            # Check description length (≥45 chars, excluding YAML key prefix)
            # Comprimento da description (>=45 ch; trata folded scalars >- / |)
            local desc_len
            desc_len=$(echo "$content" | awk '
                /^description:/{f=1; s=$0; sub(/^description:[[:space:]]*[>|+-]*[[:space:]]*/,"",s); buf=s; next}
                f && /^[[:space:]]+[^[:space:]]/{t=$0; sub(/^[[:space:]]+/,"",t); buf=buf " " t; next}
                f && /^[^[:space:]]/{f=0}
                END{gsub(/[>|]/,"",buf); print length(buf)}
            ')
            if [[ "$desc_len" -lt 45 ]]; then
                echo -e "${RED}ERROR${NC} $rel [wiki]: description too short ($desc_len < 45 chars)"
                ((ERRORS++)) || true
            fi
        fi
        if [[ "$has_tags" -eq 0 ]]; then
            echo -e "${RED}ERROR${NC} $rel [wiki]: missing 'tags'"
            ((ERRORS++)) || true
        fi
    fi

    # Warnings (non-blocking)
    if [[ "$has_desc" -eq 0 ]]; then
        echo -e "${YEL}WARN${NC}  $rel: missing 'description'"
        ((WARNINGS++)) || true
    fi
    if [[ "$has_tags" -eq 0 ]]; then
        echo -e "${YEL}WARN${NC}  $rel: missing 'tags'"
        ((WARNINGS++)) || true
    fi

    # Check for wikilinks in body (wiki layer)
    if echo "$content" | grep -q '^layer: wiki'; then
        local body
        body="$(sed -n '/^---$/,/^---$/d; p' "$f")"
        if echo "$body" | grep -q '\[\['; then
            echo -e "${YEL}WARN${NC}  $rel [wiki]: contains wikilinks [[ ]] — convert to [text](path)"
            ((WARNINGS++)) || true
        fi
    fi
}

# --- Main ---
FILES=()
ALL=false

for arg in "$@"; do
    case "$arg" in
        --all) ALL=true ;;
        --strict) STRICT=true ;;
        *) FILES+=("$arg") ;;
    esac
done

if [[ "$ALL" == "true" ]]; then
    while IFS= read -r -d '' f; do
        if ! is_excluded "$f"; then
            FILES+=("$f")
        fi
    done < <(find "$HUB_ROOT" -name '*.md' -not -path '*/.git/*' -not -path '*/node_modules/*' -print0)
elif [[ ${#FILES[@]} -eq 0 ]]; then
    # Pre-commit mode: read staged files from git
    while IFS= read -r f; do
        if [[ "$f" == *.md ]] && ! is_excluded "$HUB_ROOT/$f"; then
            FILES+=("$HUB_ROOT/$f")
        fi
    done < <(git -C "$HUB_ROOT" diff --cached --name-only --diff-filter=ACM 2>/dev/null || true)
fi

if [[ ${#FILES[@]} -eq 0 ]]; then
    echo "No .md files to validate."
    exit 0
fi

echo "Validating ${#FILES[@]} files..."
echo ""

for f in "${FILES[@]}"; do
    validate_file "$f"
done

echo ""
echo "────────────────────────────────────"
echo -e "Checked: $CHECKED | ${GRN}OK: $((CHECKED - ERRORS - WARNINGS))${NC} | ${YEL}Warnings: $WARNINGS${NC} | ${RED}Errors: $ERRORS${NC}"

if [[ "$ERRORS" -gt 0 ]]; then
    echo -e "${RED}VALIDATION FAILED${NC}"
    exit 1
else
    echo -e "${GRN}VALIDATION PASSED${NC}"
    exit 0
fi