summaryrefslogtreecommitdiff
path: root/src/internal/dag/parse.go
blob: 9d5b918b1147cc45d99c83734172ca2256d2aaf5 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
// Copyright 2022 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

// Package dag implements a language for expressing directed acyclic
// graphs.
//
// The general syntax of a rule is:
//
//	a, b < c, d;
//
// which means c and d come after a and b in the partial order
// (that is, there are edges from c and d to a and b),
// but doesn't provide a relative order between a vs b or c vs d.
//
// The rules can chain together, as in:
//
//	e < f, g < h;
//
// which is equivalent to
//
//	e < f, g;
//	f, g < h;
//
// Except for the special bottom element "NONE", each name
// must appear exactly once on the right-hand side of any rule.
// That rule serves as the definition of the allowed successor
// for that name. The definition must appear before any uses
// of the name on the left-hand side of a rule. (That is, the
// rules themselves must be ordered according to the partial
// order, for easier reading by people.)
//
// Negative assertions double-check the partial order:
//
//	i !< j
//
// means that it must NOT be the case that i < j.
// Negative assertions may appear anywhere in the rules,
// even before i and j have been defined.
//
// Comments begin with #.
package dag

import (
	"fmt"
	"sort"
	"strings"
)

type Graph struct {
	Nodes   []string
	byLabel map[string]int
	edges   map[string]map[string]bool
}

func newGraph() *Graph {
	return &Graph{byLabel: map[string]int{}, edges: map[string]map[string]bool{}}
}

func (g *Graph) addNode(label string) bool {
	if _, ok := g.byLabel[label]; ok {
		return false
	}
	g.byLabel[label] = len(g.Nodes)
	g.Nodes = append(g.Nodes, label)
	g.edges[label] = map[string]bool{}
	return true
}

func (g *Graph) AddEdge(from, to string) {
	g.edges[from][to] = true
}

func (g *Graph) DelEdge(from, to string) {
	delete(g.edges[from], to)
}

func (g *Graph) HasEdge(from, to string) bool {
	return g.edges[from] != nil && g.edges[from][to]
}

func (g *Graph) Edges(from string) []string {
	edges := make([]string, 0, 16)
	for k := range g.edges[from] {
		edges = append(edges, k)
	}
	sort.Slice(edges, func(i, j int) bool { return g.byLabel[edges[i]] < g.byLabel[edges[j]] })
	return edges
}

// Parse parses the DAG language and returns the transitive closure of
// the described graph. In the returned graph, there is an edge from "b"
// to "a" if b < a (or a > b) in the partial order.
func Parse(dag string) (*Graph, error) {
	g := newGraph()
	disallowed := []rule{}

	rules, err := parseRules(dag)
	if err != nil {
		return nil, err
	}

	// TODO: Add line numbers to errors.
	var errors []string
	errorf := func(format string, a ...any) {
		errors = append(errors, fmt.Sprintf(format, a...))
	}
	for _, r := range rules {
		if r.op == "!<" {
			disallowed = append(disallowed, r)
			continue
		}
		for _, def := range r.def {
			if def == "NONE" {
				errorf("NONE cannot be a predecessor")
				continue
			}
			if !g.addNode(def) {
				errorf("multiple definitions for %s", def)
			}
			for _, less := range r.less {
				if less == "NONE" {
					continue
				}
				if _, ok := g.byLabel[less]; !ok {
					errorf("use of %s before its definition", less)
				} else {
					g.AddEdge(def, less)
				}
			}
		}
	}

	// Check for missing definition.
	for _, tos := range g.edges {
		for to := range tos {
			if g.edges[to] == nil {
				errorf("missing definition for %s", to)
			}
		}
	}

	// Complete transitive closure.
	for _, k := range g.Nodes {
		for _, i := range g.Nodes {
			for _, j := range g.Nodes {
				if i != k && k != j && g.HasEdge(i, k) && g.HasEdge(k, j) {
					if i == j {
						// Can only happen along with a "use of X before deps" error above,
						// but this error is more specific - it makes clear that reordering the
						// rules will not be enough to fix the problem.
						errorf("graph cycle: %s < %s < %s", j, k, i)
					}
					g.AddEdge(i, j)
				}
			}
		}
	}

	// Check negative assertions against completed allowed graph.
	for _, bad := range disallowed {
		for _, less := range bad.less {
			for _, def := range bad.def {
				if g.HasEdge(def, less) {
					errorf("graph edge assertion failed: %s !< %s", less, def)
				}
			}
		}
	}

	if len(errors) > 0 {
		return nil, fmt.Errorf("%s", strings.Join(errors, "\n"))
	}

	return g, nil
}

// A rule is a line in the DAG language where "less < def" or "less !< def".
type rule struct {
	less []string
	op   string // Either "<" or "!<"
	def  []string
}

type syntaxError string

func (e syntaxError) Error() string {
	return string(e)
}

// parseRules parses the rules of a DAG.
func parseRules(rules string) (out []rule, err error) {
	defer func() {
		e := recover()
		switch e := e.(type) {
		case nil:
			return
		case syntaxError:
			err = e
		default:
			panic(e)
		}
	}()
	p := &rulesParser{lineno: 1, text: rules}

	var prev []string
	var op string
	for {
		list, tok := p.nextList()
		if tok == "" {
			if prev == nil {
				break
			}
			p.syntaxError("unexpected EOF")
		}
		if prev != nil {
			out = append(out, rule{prev, op, list})
		}
		prev = list
		if tok == ";" {
			prev = nil
			op = ""
			continue
		}
		if tok != "<" && tok != "!<" {
			p.syntaxError("missing <")
		}
		op = tok
	}

	return out, err
}

// A rulesParser parses the depsRules syntax described above.
type rulesParser struct {
	lineno   int
	lastWord string
	text     string
}

// syntaxError reports a parsing error.
func (p *rulesParser) syntaxError(msg string) {
	panic(syntaxError(fmt.Sprintf("parsing graph: line %d: syntax error: %s near %s", p.lineno, msg, p.lastWord)))
}

// nextList parses and returns a comma-separated list of names.
func (p *rulesParser) nextList() (list []string, token string) {
	for {
		tok := p.nextToken()
		switch tok {
		case "":
			if len(list) == 0 {
				return nil, ""
			}
			fallthrough
		case ",", "<", "!<", ";":
			p.syntaxError("bad list syntax")
		}
		list = append(list, tok)

		tok = p.nextToken()
		if tok != "," {
			return list, tok
		}
	}
}

// nextToken returns the next token in the deps rules,
// one of ";" "," "<" "!<" or a name.
func (p *rulesParser) nextToken() string {
	for {
		if p.text == "" {
			return ""
		}
		switch p.text[0] {
		case ';', ',', '<':
			t := p.text[:1]
			p.text = p.text[1:]
			return t

		case '!':
			if len(p.text) < 2 || p.text[1] != '<' {
				p.syntaxError("unexpected token !")
			}
			p.text = p.text[2:]
			return "!<"

		case '#':
			i := strings.Index(p.text, "\n")
			if i < 0 {
				i = len(p.text)
			}
			p.text = p.text[i:]
			continue

		case '\n':
			p.lineno++
			fallthrough
		case ' ', '\t':
			p.text = p.text[1:]
			continue

		default:
			i := strings.IndexAny(p.text, "!;,<#\n \t")
			if i < 0 {
				i = len(p.text)
			}
			t := p.text[:i]
			p.text = p.text[i:]
			p.lastWord = t
			return t
		}
	}
}