path: root/internal/engine/engine.go

// SPDX-License-Identifier: GPL-3.0-or-later

// Package engine is what every krino front end calls: it loads and compiles
// the configuration, matches files against rules, and (from plan 3) plans
// and applies actions. It returns data; front ends only render it.
package engine

import (
	"fmt"
	"os"
	"strings"
	"time"

	"krino/internal/cond"
	"krino/internal/config"
	"krino/internal/extract"
	"krino/internal/ignore"
	"krino/internal/plan"
)

// Engine holds a loaded, compiled configuration: everything a front end
// needs to check, match and (from plan 3) act.
type Engine struct {
	Config   *config.Config
	Dirs     []*Dir
	Extract  *extract.Extractor
	Now      func() time.Time // time.Now; tests replace it
	MainFile string
}

// Dir is one configured directory, with its ignore matcher and rules
// compiled.
type Dir struct {
	Name     string
	Root     string // absolute
	Conf     *config.Dir
	Settings config.Resolved // built-in, then defaults, then the directory
	Ignore   *ignore.Matcher
	Rules    []*Rule

	// ContentVariants is the distinct (IgnoreCase, Fold) pairs any of
	// Rules' content tests evaluate under, in first-seen order. B2: when
	// this holds exactly one variant, facts.Content releases a file's raw
	// extracted text once that variant's normalised copy exists, since no
	// other variant will ever be asked for; with more than one, both must
	// stay memoised, as before.
	ContentVariants []cond.Options

	// DupScopes is every distinct directory list the duplicate tests of
	// Rules use, in first-seen order, the plain (duplicate) as an empty
	// list. Spec §5.5 rule 2 looks a file up under each of them before any
	// rule may delete it.
	DupScopes [][]string
}

// Rule is one directory's rule, with its condition compiled.
type Rule struct {
	Name     string
	Conf     *config.Rule
	Settings config.Resolved // the rule's own settings over its directory's
	Cond     *cond.Cond
}

// Load reads and compiles everything. Any problem anywhere returns a nil
// Engine and every diagnostic: krino never acts on a configuration it only
// partly understood. Duplicate names are ignored after their first use.
func Load(mainFile string, names ...string) (*Engine, []*config.Diag) {
	cfg, errs := config.Load(mainFile, dedupeNames(names)...)
	if cfg == nil {
		return nil, errs
	}

	var dirs []*Dir
	for _, d := range cfg.Dirs {
		dir := &Dir{
			Name:     d.Name,
			Root:     d.Path,
			Conf:     d,
			Settings: cfg.Resolved(d),
		}
		if m, err := ignore.New(d.Ignore); err != nil {
			errs = append(errs, &config.Diag{File: d.File, Msg: err.Error()})
		} else {
			dir.Ignore = m
		}
		for _, r := range d.Rules {
			rs := r.Settings.Over(dir.Settings)
			c, cerrs := cond.Compile(d.File, r.When, cond.Options{
				IgnoreCase: rs.Case == config.CaseIgnore,
				Fold:       rs.Fold,
			})
			if len(cerrs) > 0 {
				errs = append(errs, cerrs...)
				continue
			}
			if diag := checkCaptures(d.File, r, c); diag != nil {
				errs = append(errs, diag)
				continue
			}
			if diag := checkDuplicateDelete(d.File, r, c); diag != nil {
				errs = append(errs, diag)
				continue
			}
			dir.Rules = append(dir.Rules, &Rule{Name: r.Name, Conf: r, Settings: rs, Cond: c})
		}
		dir.ContentVariants = contentVariants(dir.Rules)
		dir.DupScopes = dupScopes(dir.Rules)
		dirs = append(dirs, dir)
	}

	if len(errs) > 0 {
		return nil, errs
	}
	return &Engine{
		Config:   cfg,
		Dirs:     dirs,
		Extract:  extract.New(),
		Now:      time.Now,
		MainFile: mainFile,
	}, nil
}

// checkCaptures validates a compiled rule's actions against the capture
// groups its own name tests can supply (spec 7.3): a rule using {N} needs a
// name test at all, and every name test in it needs at least N groups. It
// reports only the first offending action, so one config mistake yields one
// diagnostic.
func checkCaptures(file string, r *config.Rule, c *cond.Cond) *config.Diag {
	groups := c.NameGroups()
	for _, a := range r.Actions {
		n, err := plan.MaxIndex(a.Arg)
		if err != nil || n == 0 {
			continue
		}
		if len(groups) == 0 {
			return &config.Diag{File: file, Pos: a.Pos, Msg: fmt.Sprintf("rule %q: {%d} needs a name test to capture from", r.Name, n)}
		}
		for _, g := range groups {
			if g < n {
				return &config.Diag{File: file, Pos: a.Pos, Msg: fmt.Sprintf("rule %q: {%d} but a name test has only %s", r.Name, n, captureGroups(g))}
			}
		}
	}
	return nil
}

// checkDuplicateDelete refuses a rule that combines a duplicate test with a
// delete action (spec §4.5, §5.5): duplicates are found, never deleted.
// Cond.DupDirs records every duplicate test compiled, inside or and not
// too, so a test anywhere in the condition counts. It reports only the
// first delete action, so one config mistake yields one diagnostic.
func checkDuplicateDelete(file string, r *config.Rule, c *cond.Cond) *config.Diag {
	if len(c.DupDirs) == 0 {
		return nil
	}
	for _, a := range r.Actions {
		if a.Kind == config.Delete || a.Kind == config.DeletePermanent {
			return &config.Diag{File: file, Pos: a.Pos, Msg: fmt.Sprintf("rule %q: (duplicate) cannot be combined with (%s): duplicates are never deleted, move them aside instead", r.Name, a.Kind)}
		}
	}
	return nil
}

// captureGroups renders a capture-group count with correct singular/plural.
func captureGroups(n int) string {
	if n == 1 {
		return "1 capture group"
	}
	return fmt.Sprintf("%d capture groups", n)
}

// dedupeNames returns names with every repeat after its first occurrence
// removed, order preserved.
func dedupeNames(names []string) []string {
	var out []string
	seen := map[string]bool{}
	for _, n := range names {
		if seen[n] {
			continue
		}
		seen[n] = true
		out = append(out, n)
	}
	return out
}

// contentVariants returns the distinct (IgnoreCase, Fold) pairs any of
// rules' content tests evaluate under, in first-seen order — B2's per-Dir
// ContentVariants. A rule whose condition has no content test at all
// (Cond.UsesContent false) never calls facts.Content, so its resolved
// case/fold settings contribute no variant here.
func contentVariants(rules []*Rule) []cond.Options {
	var out []cond.Options
	seen := map[cond.Options]bool{}
	for _, r := range rules {
		if !r.Cond.UsesContent {
			continue
		}
		opt := cond.Options{IgnoreCase: r.Settings.Case == config.CaseIgnore, Fold: r.Settings.Fold}
		if seen[opt] {
			continue
		}
		seen[opt] = true
		out = append(out, opt)
	}
	return out
}

// dupScopes returns the distinct Cond.DupDirs lists of rules, in first-seen
// order. Lists are compared as written, with their length in the key so
// (duplicate) and (duplicate "") stay apart; two spellings of one directory
// stay two entries, which costs a second lookup but never a wrong answer,
// since facts.Duplicate resolves and shares the index itself.
func dupScopes(rules []*Rule) [][]string {
	var out [][]string
	seen := map[string]bool{}
	for _, r := range rules {
		for _, dirs := range r.Cond.DupDirs {
			key := fmt.Sprintf("%d\x00%s", len(dirs), strings.Join(dirs, "\x00"))
			if seen[key] {
				continue
			}
			seen[key] = true
			out = append(out, dirs)
		}
	}
	return out
}

// Report is what Check reports: the files involved and each directory's
// state.
type Report struct {
	MainFile string
	LogFile  string
	Dirs     []DirReport
	Tools    []extract.Tool
}

// DirReport is one directory's state in a Report.
type DirReport struct {
	Dir     *Dir
	Missing bool // the root is not a directory right now
}

// Check reports the engine's configuration files, each directory's current
// state and the external tools found for content extraction.
func (e *Engine) Check() Report {
	r := Report{
		MainFile: e.MainFile,
		LogFile:  e.Config.LogFile(),
		Tools:    e.Extract.Tools(),
	}
	for _, d := range e.Dirs {
		fi, err := os.Stat(d.Root)
		r.Dirs = append(r.Dirs, DirReport{Dir: d, Missing: err != nil || !fi.IsDir()})
	}
	return r
}