Implement draft/cnm-selectorv0.3.0

author: clsr <clsr@clsr.net> 2017-08-25 15:06:28 +0200
committer: clsr <clsr@clsr.net> 2017-08-25 15:06:28 +0200
commit: 6e909014e6826e14d629e4e49c51394b27a9094e (patch)
tree: 05066ca837df9c0bec6729b5b7aa00bdc95a3203
parent: 4c0470d24b00445b67a5a0accde8538d8c810730 (diff)
download: cnm-go-6e909014e6826e14d629e4e49c51394b27a9094e.tar.gz
cnm-go-6e909014e6826e14d629e4e49c51394b27a9094e.zip
2 files changed, 582 insertions, 0 deletions
diff --git a/selector.go b/selector.go
new file mode 100644
index 0000000..5259df2
--- /dev/null
+++ b/selector.go
@@ -0,0 +1,277 @@
+package cnm
+
+import (
+	"errors"
+	"net/url"
+	"reflect"
+	"strconv"
+	"strings"
+)
+
+// Select selects content from the document based on an extended CNM selector.
+//
+// See ContNet draft/cnm-selector for more information about content selectors.
+//
+// This function may create copies of the document and all container content
+// blocks, but includes non-container blocks untouched in the copy.
+func (doc *Document) Select(selector string) (*Document, error) {
+	deep := true
+	if len(selector) > 0 && selector[0] == '!' {
+		deep = false
+		selector = selector[1:]
+	}
+
+	if len(selector) == 0 {
+		return duplicateBlock(doc, true, deep).(*Document), nil
+	}
+
+	indices, block, err := doc.SelectBlock(selector)
+	if err != nil || block == nil {
+		return nil, err
+	}
+
+	d := NewDocument()
+	if block := selectContent(doc.Content, indices, deep); block != nil {
+		d.Content = block.(*ContentBlock)
+	}
+
+	return d, nil
+}
+
+// SelectIndex selects a content block based on an index path.
+//
+// For example, indices=[]int{1, 4, 2} selects
+// doc.Content.Children()[1].Children()[4].Children(2) if it exists, otherwise
+// returns nil. Note that this is *not* the same as a CNM index selector.
+func (doc *Document) SelectIndex(indices ...int) (block Block) {
+	return SelectIndex(doc.Content, indices...)
+}
+
+// SelectBlock selects section blocks from the document with a selector query.
+//
+// Returns the block index path within content blocks, the target block and, if
+// the query was invalid, an error. All are nil when no block was found with a
+// valid selector.
+//
+// See ContNet draft/cnm-selector for more information.
+func (doc *Document) SelectBlock(selector string) (indices []int, block Block, err error) {
+	if len(selector) == 0 {
+		block = doc
+		indices = []int{}
+		return
+	}
+
+	switch selector[0] {
+	case '/':
+		indices, block, err = doc.selectPath(selector[1:])
+	case '#':
+		indices, block, err = doc.selectTitle(selector[1:])
+	case '$':
+		indices, block, err = doc.selectSectionIndex(selector[1:])
+	default:
+		err = errors.New("cnm-go: invalid selector")
+	}
+
+	return
+}
+
+func (doc *Document) selectPath(selector string) (indices []int, block Block, err error) {
+	ss := []string{}
+	if selector != "" {
+		ss = strings.Split(selector, "/")
+		for i := range ss {
+			ss[i], err = url.PathUnescape(ss[i])
+			if err != nil {
+				return
+			}
+		}
+	}
+	indices, block = selectSearch(doc.Content, ss, false)
+	return
+}
+
+func (doc *Document) selectTitle(selector string) (indices []int, block Block, err error) {
+	if strings.ContainsRune(selector, '/') {
+		err = errors.New("cnm-go: invalid title selector")
+		return
+	}
+	selector, err = url.PathUnescape(selector)
+	if err != nil {
+		return
+	}
+	ss := []string{}
+	if selector != "" {
+		ss = []string{selector}
+	}
+	indices, block = selectSearch(doc.Content, ss, true)
+	return
+}
+
+func (doc *Document) selectSectionIndex(selector string) (indices []int, block Block, err error) {
+	secs := []int{}
+	if selector != "" {
+		for _, s := range strings.Split(selector, ".") {
+			var n uint64
+			n, err = strconv.ParseUint(s, 10, 31)
+			if err != nil {
+				return
+			}
+			secs = append(secs, int(n))
+		}
+	}
+	indices, block = selectSectionIndex(doc.Content, secs)
+	return
+}
+
+func selectSearch(block Block, titles []string, any bool) ([]int, Block) {
+	if len(titles) == 0 {
+		return []int{}, block
+	}
+
+	if bl, ok := block.(*SectionBlock); ok {
+		if len(titles) == 1 && bl.Title() == titles[0] {
+			return []int{}, bl
+		}
+		if any || bl.Title() == titles[0] {
+			if !any {
+				titles = titles[1:]
+			}
+			for i, c := range bl.Children() {
+				if p, b := selectSearch(c, titles, any); b != nil {
+					return append([]int{i}, p...), b
+				}
+			}
+		}
+	} else if bl, ok := block.(ContainerBlock); ok {
+		for i, c := range bl.Children() {
+			if p, b := selectSearch(c, titles, any); b != nil {
+				return append([]int{i}, p...), b
+			}
+		}
+	}
+
+	return nil, nil
+}
+
+func selectSectionIndex(block Block, secs []int) ([]int, Block) {
+	if len(secs) == 0 {
+		return []int{}, block
+	}
+	if bl, ok := block.(ContainerBlock); ok {
+		for i, c := range bl.Children() {
+			if secs[0] < 1 {
+				return nil, nil
+			}
+			if cb, ok2 := c.(*SectionBlock); ok2 {
+				secs[0]-- // starts with 1 for first section
+				if secs[0] == 0 {
+					if len(secs) == 1 {
+						return []int{i}, cb
+					}
+					p, b := selectSectionIndex(cb, secs[1:])
+					if b == nil {
+						return nil, nil
+					}
+					return append([]int{i}, p...), b
+				}
+			} else if _, ok2 := c.(ContainerBlock); ok2 {
+				p, b := selectSectionIndex(c, secs)
+				if b != nil {
+					return append([]int{i}, p...), b
+				}
+			}
+		}
+	}
+	return nil, nil
+}
+
+func selectContent(block Block, indices []int, deep bool) Block {
+	if indices == nil {
+		return nil
+	}
+	if len(indices) == 0 {
+		return duplicateBlock(block, true, deep)
+	}
+	if bl, ok := block.(ContainerBlock); ok {
+		ch := bl.Children()
+		if len(ch) <= indices[0] {
+			return nil
+		}
+		b := duplicateBlock(block, false, true).(ContainerBlock)
+		c := selectContent(ch[indices[0]], indices[1:], deep)
+		if c != nil {
+			b.AppendChild(c)
+		}
+		return b
+	}
+	return nil
+}
+
+func duplicateBlock(block Block, deep, sections bool) Block {
+	if bl, ok := block.(*Document); ok {
+		doc := &Document{
+			Title: bl.Title,
+			Links: bl.Links,
+			Site:  bl.Site,
+		}
+		if deep {
+			doc.Content = duplicateBlock(bl.Content, deep, sections).(*ContentBlock)
+		}
+		return doc
+	}
+	if _, ok := block.(ContainerBlock); !ok {
+		return block
+	}
+
+	var cb ContainerBlock
+
+	switch bl := block.(type) {
+	case *ContentBlock:
+		cb = NewContentBlock(bl.Name(), bl.Args()...)
+	case *HeaderBlock:
+		cb = NewHeaderBlock()
+	case *ListBlock:
+		cb = NewListBlock(bl.Ordered())
+	case *RowBlock:
+		cb = NewRowBlock()
+	case *SectionBlock:
+		cb = NewSectionBlock(bl.Title())
+	case *TableBlock:
+		cb = NewTableBlock()
+	default: // XXX
+		//return nil
+		val := reflect.New(reflect.TypeOf(bl).Elem())
+		cb = val.Interface().(ContainerBlock)
+	}
+
+	if cb != nil && deep {
+		for _, ch := range block.(ContainerBlock).Children() {
+			_, isSection := ch.(*SectionBlock)
+			cb.AppendChild(duplicateBlock(ch, sections || !isSection, sections))
+		}
+	}
+
+	return cb
+}
+
+// SelectIndex selects a child block based on an index path.
+//
+// Note that this is *not* the same as a CNM section index selector.
+func SelectIndex(block Block, indices ...int) Block {
+	if indices == nil {
+		return nil
+	}
+	for len(indices) > 0 {
+		if bl, ok := block.(ContainerBlock); ok {
+			ch := bl.Children()
+			if len(ch) <= indices[0] {
+				return nil
+			}
+			block = ch[indices[0]]
+			indices = indices[1:]
+		} else {
+			return nil
+		}
+	}
+	return block
+}
diff --git a/selector_test.go b/selector_test.go
new file mode 100644
index 0000000..0c714dc
--- /dev/null
+++ b/selector_test.go
@@ -0,0 +1,305 @@
+package cnm
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+)
+
+var (
+	sec11 = &SectionBlock{ContentBlock{
+		name: "section",
+		args: []string{"foo/bar"},
+		children: []Block{
+			&TextBlock{
+				Format:   "pre",
+				Contents: TextPreContents{"qwe"},
+			},
+		},
+	}}
+	sec1 = &SectionBlock{ContentBlock{
+		name: "section",
+		args: []string{"bar"},
+		children: []Block{
+			sec11,
+			&TextBlock{
+				Format:   "pre",
+				Contents: TextPreContents{"..."},
+			},
+		},
+	}}
+	sec31 = &SectionBlock{ContentBlock{
+		name: "section",
+		args: []string{"baz"},
+		children: []Block{
+			&TextBlock{
+				Format:   "pre",
+				Contents: TextPreContents{"zxc"},
+			},
+		},
+	}}
+	sec322 = &SectionBlock{ContentBlock{
+		name: "section",
+		args: []string{"baz"},
+		children: []Block{
+			&TextBlock{
+				Format:   "pre",
+				Contents: TextPreContents{"xyz"},
+			},
+		},
+	}}
+	sec32 = &SectionBlock{ContentBlock{
+		name: "section",
+		args: []string{"quux"},
+		children: []Block{
+			&TextBlock{
+				Format:   "pre",
+				Contents: TextPreContents{"123"},
+			},
+			sec322,
+		},
+	}}
+	sec4 = &SectionBlock{ContentBlock{
+		name: "section",
+		args: []string{"foo/bar"},
+		children: []Block{
+			&TextBlock{
+				Format:   "pre",
+				Contents: TextPreContents{"abc"},
+			},
+		},
+	}}
+	doc1 = &Document{
+		Title: "test title",
+		Content: &ContentBlock{
+			name: "content",
+			children: []Block{
+				sec1,
+				&TextBlock{
+					Format:   "pre",
+					Contents: TextPreContents{"asd"},
+				},
+				&ListBlock{ContentBlock{
+					name: "list",
+					children: []Block{
+						sec31,
+						sec32,
+						&TextBlock{
+							Format:   "pre",
+							Contents: TextPreContents{"rty"},
+						},
+					},
+				}},
+				sec4,
+			},
+		},
+	}
+
+	doc1Shallow = &Document{
+		Title: doc1.Title,
+		Content: &ContentBlock{
+			name: "content",
+			children: []Block{
+				&SectionBlock{ContentBlock{
+					name: "section",
+					args: sec1.Args(),
+				}},
+				doc1.Content.Children()[1],
+				&ListBlock{ContentBlock{
+					name: "list",
+					children: []Block{
+						&SectionBlock{ContentBlock{
+							name: "section",
+							args: sec31.Args(),
+						}},
+						&SectionBlock{ContentBlock{
+							name: "section",
+							args: sec32.Args(),
+						}},
+						doc1.Content.Children()[2].(ContainerBlock).Children()[2],
+					},
+				}},
+				&SectionBlock{ContentBlock{
+					name: "section",
+					args: sec4.Args(),
+				}},
+			},
+		},
+	}
+
+	sec1Shallow = &Document{Content: &ContentBlock{
+		name: "content",
+		children: []Block{
+			&SectionBlock{ContentBlock{
+				name: sec1.Name(),
+				args: sec1.Args(),
+				children: []Block{
+					&SectionBlock{ContentBlock{
+						name: "section",
+						args: sec11.Args(),
+					}},
+					sec1.Children()[1],
+				},
+			}},
+		},
+	}}
+)
+
+var selectTests = []struct {
+	sel string
+	doc *Document
+	res *Document
+}{
+	{"", doc1, doc1},
+	{"/", doc1, &Document{Content: doc1.Content}},
+	{"$", doc1, &Document{Content: doc1.Content}},
+	{"#", doc1, &Document{Content: doc1.Content}},
+
+	{"!", doc1, doc1Shallow},
+	{"!/", doc1, &Document{Content: doc1Shallow.Content}},
+	{"!$", doc1, &Document{Content: doc1Shallow.Content}},
+	{"!#", doc1, &Document{Content: doc1Shallow.Content}},
+
+	{"/bar", doc1, &Document{Content: &ContentBlock{
+		name: "content",
+		children: []Block{
+			sec1,
+		},
+	}}},
+	{"!/bar", doc1, sec1Shallow},
+	{"#bar", doc1, &Document{Content: &ContentBlock{
+		name:     "content",
+		children: []Block{sec1},
+	}}},
+	{"!#bar", doc1, sec1Shallow},
+	{"$1", doc1, &Document{Content: &ContentBlock{
+		name:     "content",
+		children: []Block{sec1},
+	}}},
+	{"!$1", doc1, sec1Shallow},
+}
+
+func TestSelect(t *testing.T) {
+	for _, v := range selectTests {
+		t.Run(strings.Replace(v.sel, "/", "_", -1), func(t *testing.T) {
+			res, err := v.doc.Select(v.sel)
+			if err != nil {
+				printTestDoc(t, v.doc)
+				t.Fatalf("Select(%q) on %#v:\nerror: %v", v.sel, v.doc, err)
+			}
+			if !documentEqual(t, res, v.res) {
+				t.Log("=============== doc")
+				printTestDoc(t, v.doc)
+				t.Log("=============== expect")
+				printTestDoc(t, v.res)
+				t.Log("=============== got")
+				printTestDoc(t, res)
+				t.Fatalf("Select(%q) on %#v:\nexpected: %#v\n     got: %#v", v.sel, v.doc, v.res, res)
+			}
+		})
+	}
+}
+
+var selectBlockTests = []struct {
+	sel string
+	doc *Document
+	res Block
+	ind []int
+}{
+	{"", doc1, doc1, []int{}},
+
+	{"/", doc1, doc1.Content, []int{}},
+	{"/qwe", doc1, nil, nil},
+	{"/bar/qwe", doc1, nil, nil},
+	{"/bar/", doc1, nil, nil},
+	{"//", doc1, nil, nil},
+	{"/./", doc1, nil, nil},
+	{"/../", doc1, nil, nil},
+	{"/.", doc1, nil, nil},
+	{"/..", doc1, nil, nil},
+	{"/foo%2fbar", doc1, sec4, []int{3}},
+	{"/bar/foo%2fbar", doc1, sec11, []int{0, 0}},
+	{"/baz", doc1, sec31, []int{2, 0}},
+	{"/quux", doc1, sec32, []int{2, 1}},
+	{"/quux/baz", doc1, sec322, []int{2, 1, 1}},
+	{"/bar/baz", doc1, nil, nil},
+
+	{"$", doc1, doc1.Content, []int{}},
+	{"$0", doc1, nil, nil},
+	{"$1", doc1, sec1, []int{0}},
+	{"$1.1", doc1, sec11, []int{0, 0}},
+	{"$2", doc1, sec31, []int{2, 0}},
+	{"$3", doc1, sec32, []int{2, 1}},
+	{"$4", doc1, sec4, []int{3}},
+	{"$5", doc1, nil, nil},
+	{"$2.1", doc1, nil, nil},
+	{"$3.1", doc1, sec322, []int{2, 1, 1}},
+	{"$3.0", doc1, nil, nil},
+	{"$3.2", doc1, nil, nil},
+
+	{"#", doc1, doc1.Content, []int{}},
+	{"#qwe", doc1, nil, nil},
+	{"#bar", doc1, sec1, []int{0}},
+	{"#foo%2fbar", doc1, sec11, []int{0, 0}},
+	{"#baz", doc1, sec31, []int{2, 0}},
+	{"#quux", doc1, sec32, []int{2, 1}},
+}
+
+func TestSelectBlock(t *testing.T) {
+	for _, v := range selectBlockTests {
+		t.Run(strings.Replace(v.sel, "/", "_", -1), func(t *testing.T) {
+			ind, res, err := v.doc.SelectBlock(v.sel)
+			if err != nil {
+				printTestDoc(t, v.doc)
+				t.Fatalf("SelectBlock(%q) on %#v:\nerror: %v", v.sel, v.doc, err)
+			}
+			if !indicesEqual(t, ind, v.ind) {
+				printTestDoc(t, v.doc)
+				t.Fatalf("SelectBlock(%q) on %#v:\nexpected indices %#v, got %#v", v.sel, v.doc, v.ind, ind)
+			}
+			if !blockEqual(t, res, v.res) {
+				printTestDoc(t, v.doc)
+				t.Fatalf("SelectBlock(%q) on %#v:\nexpected: %#v\n     got: %#v", v.sel, v.doc, v.res, res)
+			}
+			if doc, ok := res.(*Document); ok {
+				res = doc.Content
+			}
+			b := v.doc.SelectIndex(ind...)
+			if !blockEqual(t, b, res) {
+				printTestDoc(t, v.doc)
+				t.Fatalf("SelectIndex(%#v) (for %q):\nexpected: %#v\n     got: %#v", ind, v.sel, res, b)
+			}
+		})
+	}
+}
+
+func indicesEqual(t *testing.T, a, b []int) bool {
+	if a == nil && b == nil {
+		return true
+	}
+	if a == nil || b == nil {
+		t.Log("indice slice nilness differs")
+		return false
+	}
+	if len(a) != len(b) {
+		t.Log("indice slice lengths differ")
+		return false
+	}
+	for i := range a {
+		if a[i] != b[i] {
+			t.Log("indice slice element differs")
+			return false
+		}
+	}
+	return true
+}
+
+func printTestDoc(t *testing.T, doc *Document) {
+	if doc == nil {
+		t.Log(doc)
+		return
+	}
+	var buf bytes.Buffer
+	doc.Write(&buf)
+	t.Log("\n" + buf.String())
+}
author	clsr <clsr@clsr.net>	2017-08-25 15:06:28 +0200
committer	clsr <clsr@clsr.net>	2017-08-25 15:06:28 +0200
commit	6e909014e6826e14d629e4e49c51394b27a9094e (patch)
tree	05066ca837df9c0bec6729b5b7aa00bdc95a3203
parent	4c0470d24b00445b67a5a0accde8538d8c810730 (diff)
download	cnm-go-6e909014e6826e14d629e4e49c51394b27a9094e.tar.gz cnm-go-6e909014e6826e14d629e4e49c51394b27a9094e.zip