feat: pull out most things as attachments

pkg/message/parser.go

@@ -68,19 +68,46 @@ func Parse(r io.Reader, key, keyName string) (m *pmapi.Message, mimeMessage, pla
 }
 
 func collectAttachments(p *parser.Parser) (atts []*pmapi.Attachment, data []io.Reader, err error) {
-	w := p.NewWalker()
+	w := p.NewWalker().
-
+		RegisterContentDispositionHandler("attachment", func(p *parser.Part) error {
-	w.RegisterContentDispositionHandler("attachment").
-		OnEnter(func(p *parser.Part, _ parser.PartHandlerFunc) (err error) {
 			att, err := parseAttachment(p.Header)
 			if err != nil {
-				return
+				return err
 			}
 
 			atts = append(atts, att)
 			data = append(data, bytes.NewReader(p.Body))
 
-			return
+			return nil
+		}).
+		RegisterContentTypeHandler("text/calendar", func(p *parser.Part) error {
+			att, err := parseAttachment(p.Header)
+			if err != nil {
+				return err
+			}
+
+			atts = append(atts, att)
+			data = append(data, bytes.NewReader(p.Body))
+
+			return nil
+		}).
+		RegisterContentTypeHandler("text/.*", func(p *parser.Part) error {
+			return nil
+		}).
+		RegisterDefaultHandler(func(p *parser.Part) error {
+			if len(p.Children()) > 0 {
+				return nil
+			}
+
+			att, err := parseAttachment(p.Header)
+			if err != nil {
+				return err
+			}
+
+			atts = append(atts, att)
+			data = append(data, bytes.NewReader(p.Body))
+
+			return nil
 		})
 
 	if err = w.Walk(); err != nil {
@@ -118,7 +145,7 @@ func buildBodies(p *parser.Parser) (richBody, plainBody string, err error) {
 // collectBodyParts collects all body parts in the parse tree, preferring
 // parts of the given content type if alternatives exist.
 func collectBodyParts(p *parser.Parser, preferredContentType string) (parser.Parts, error) {
-	v := parser.
+	v := p.
 		NewVisitor(func(p *parser.Part, visit parser.Visit) (interface{}, error) {
 			childParts, err := collectChildParts(p, visit)
 			if err != nil {
@@ -142,7 +169,7 @@ func collectBodyParts(p *parser.Parser, preferredContentType string) (parser.Par
 			return parser.Parts{p}, nil
 		})
 
-	res, err := v.Visit(p.Root())
+	res, err := v.Visit()
 	if err != nil {
 		return nil, err
 	}
@@ -203,12 +230,10 @@ func allHaveContentType(parts parser.Parts, contentType string) bool {
 }
 
 func determineMIMEType(p *parser.Parser) (string, error) {
-	w := p.NewWalker()
-
 	var isHTML bool
 
-	w.RegisterContentTypeHandler("text/html").
+	w := p.NewWalker().
-		OnEnter(func(p *parser.Part) (err error) {
+		RegisterContentTypeHandler("text/html", func(p *parser.Part) (err error) {
 			isHTML = true
 			return
 		})
@@ -349,6 +374,8 @@ func parseAttachment(h message.Header) (att *pmapi.Attachment, err error) {
 		att.Name = dispParams["filename"]
 	}
 
+	att.ContentID = strings.Trim(h.Get("Content-Id"), " <>")
+
 	// TODO: Set att.Header
 
 	return

pkg/message/parser/handler.go

@@ -0,0 +1,40 @@
+package parser
+
+import "regexp"
+
+type HandlerFunc func(*Part) error
+
+type handler struct {
+	typeRegExp, dispRegExp string
+	fn                     HandlerFunc
+}
+
+func (h *handler) matchPart(p *Part) bool {
+	return h.matchType(p) || h.matchDisp(p)
+}
+
+func (h *handler) matchType(p *Part) bool {
+	if h.typeRegExp == "" {
+		return false
+	}
+
+	t, _, err := p.Header.ContentType()
+	if err != nil {
+		t = ""
+	}
+
+	return regexp.MustCompile(h.typeRegExp).MatchString(t)
+}
+
+func (h *handler) matchDisp(p *Part) bool {
+	if h.dispRegExp == "" {
+		return false
+	}
+
+	disp, _, err := p.Header.ContentDisposition()
+	if err != nil {
+		disp = ""
+	}
+
+	return regexp.MustCompile(h.dispRegExp).MatchString(disp)
+}

pkg/message/parser/handlers.go

@@ -1,72 +0,0 @@
-package parser
-
-type PartHandlerFunc func(*Part) error
-type DispHandlerFunc func(*Part, PartHandlerFunc) error
-
-type PartHandler struct {
-	enter, exit PartHandlerFunc
-}
-
-func NewPartHandler() *PartHandler {
-	return &PartHandler{
-		enter: partNoop,
-		exit:  partNoop,
-	}
-}
-
-func (h *PartHandler) OnEnter(fn PartHandlerFunc) *PartHandler {
-	h.enter = fn
-	return h
-}
-
-func (h *PartHandler) OnExit(fn PartHandlerFunc) *PartHandler {
-	h.exit = fn
-	return h
-}
-
-func (h *PartHandler) handleEnter(_ *Walker, p *Part) error {
-	return h.enter(p)
-}
-
-func (h *PartHandler) handleExit(_ *Walker, p *Part) error {
-	return h.exit(p)
-}
-
-type DispHandler struct {
-	enter, exit DispHandlerFunc
-}
-
-func NewDispHandler() *DispHandler {
-	return &DispHandler{
-		enter: dispNoop,
-		exit:  dispNoop,
-	}
-}
-
-func (h *DispHandler) OnEnter(fn DispHandlerFunc) *DispHandler {
-	h.enter = fn
-	return h
-}
-
-func (h *DispHandler) OnExit(fn DispHandlerFunc) *DispHandler {
-	h.exit = fn
-	return h
-}
-
-func (h *DispHandler) handleEnter(w *Walker, p *Part) error {
-	// NOTE: This is hacky -- is there a better solution?
-	return h.enter(p, func(p *Part) error {
-		return w.getTypeHandler(p).handleEnter(w, p)
-	})
-}
-
-func (h *DispHandler) handleExit(w *Walker, p *Part) error {
-	// NOTE: This is hacky -- is there a better solution?
-	return h.exit(p, func(p *Part) error {
-		return w.getTypeHandler(p).handleExit(w, p)
-	})
-}
-
-func partNoop(*Part) error { return nil }
-
-func dispNoop(*Part, PartHandlerFunc) error { return nil }

pkg/message/parser/parser.go

@@ -33,6 +33,10 @@ func (p *Parser) NewWalker() *Walker {
 	return newWalker(p.root)
 }
 
+func (p *Parser) NewVisitor(defaultRule VisitorRule) *Visitor {
+	return newVisitor(p.root, defaultRule)
+}
+
 func (p *Parser) NewWriter() *Writer {
 	return newWriter(p.root)
 }

pkg/message/parser/visitor.go

@@ -3,27 +3,29 @@ package parser
 import "regexp"
 
 type Visitor struct {
-	rules    []*rule
+	root        *Part
-	fallback Rule
+	rules       []*visitorRule
+	defaultRule VisitorRule
 }
 
-func NewVisitor(fallback Rule) *Visitor {
+func newVisitor(root *Part, defaultRule VisitorRule) *Visitor {
 	return &Visitor{
-		fallback: fallback,
+		root:        root,
+		defaultRule: defaultRule,
 	}
 }
 
 type Visit func(*Part) (interface{}, error)
 
-type Rule func(*Part, Visit) (interface{}, error)
+type VisitorRule func(*Part, Visit) (interface{}, error)
 
-type rule struct {
+type visitorRule struct {
 	re string
-	fn Rule
+	fn VisitorRule
 }
 
-func (v *Visitor) RegisterRule(contentTypeRegex string, fn Rule) *Visitor {
+func (v *Visitor) RegisterRule(contentTypeRegex string, fn VisitorRule) *Visitor {
-	v.rules = append(v.rules, &rule{
+	v.rules = append(v.rules, &visitorRule{
 		re: contentTypeRegex,
 		fn: fn,
 	})
@@ -31,20 +33,24 @@ func (v *Visitor) RegisterRule(contentTypeRegex string, fn Rule) *Visitor {
 	return v
 }
 
-func (v *Visitor) Visit(p *Part) (interface{}, error) {
+func (v *Visitor) Visit() (interface{}, error) {
+	return v.visit(v.root)
+}
+
+func (v *Visitor) visit(p *Part) (interface{}, error) {
 	t, _, err := p.Header.ContentType()
 	if err != nil {
 		return nil, err
 	}
 
 	if rule := v.getRuleForContentType(t); rule != nil {
-		return rule.fn(p, v.Visit)
+		return rule.fn(p, v.visit)
 	}
 
-	return v.fallback(p, v.Visit)
+	return v.defaultRule(p, v.visit)
 }
 
-func (v *Visitor) getRuleForContentType(contentType string) *rule {
+func (v *Visitor) getRuleForContentType(contentType string) *visitorRule {
 	for _, rule := range v.rules {
 		if regexp.MustCompile(rule.re).MatchString(contentType) {
 			return rule

pkg/message/parser/walker.go

@@ -3,96 +3,64 @@ package parser
 type Walker struct {
 	root *Part
 
-	defaultHandler handler
+	handlers       []*handler
-	typeHandlers   map[string]handler
+	defaultHandler HandlerFunc
-	dispHandlers   map[string]handler
-}
-
-type handler interface {
-	handleEnter(*Walker, *Part) error
-	handleExit(*Walker, *Part) error
 }
 
 func newWalker(root *Part) *Walker {
 	return &Walker{
 		root:           root,
-		defaultHandler: NewPartHandler(),
+		defaultHandler: func(*Part) error { return nil },
-		typeHandlers:   make(map[string]handler),
-		dispHandlers:   make(map[string]handler),
 	}
 }
 
 func (w *Walker) Walk() (err error) {
-	return w.visitPart(w.root)
+	return w.walkOverPart(w.root)
 }
 
-func (w *Walker) visitPart(p *Part) (err error) {
+func (w *Walker) walkOverPart(p *Part) error {
-	hdl := w.getHandler(p)
+	if err := w.getHandlerFunc(p)(p); err != nil {
-
+		return err
-	if err = hdl.handleEnter(w, p); err != nil {
-		return
 	}
 
 	for _, child := range p.children {
-		if err = w.visitPart(child); err != nil {
+		if err := w.walkOverPart(child); err != nil {
-			return
+			return err
 		}
 	}
 
-	return hdl.handleExit(w, p)
+	return nil
 }
 
-func (w *Walker) WithDefaultHandler(handler handler) *Walker {
+func (w *Walker) RegisterDefaultHandler(fn HandlerFunc) *Walker {
-	w.defaultHandler = handler
+	w.defaultHandler = fn
 	return w
 }
-func (w *Walker) RegisterContentTypeHandler(contType string) *PartHandler {
-	hdl := NewPartHandler()
 
-	w.typeHandlers[contType] = hdl
+func (w *Walker) RegisterContentTypeHandler(typeRegExp string, fn HandlerFunc) *Walker {
+	w.handlers = append(w.handlers, &handler{
+		typeRegExp: typeRegExp,
+		fn:         fn,
+	})
 
-	return hdl
+	return w
 }
 
-func (w *Walker) RegisterContentDispositionHandler(contDisp string) *DispHandler {
+func (w *Walker) RegisterContentDispositionHandler(dispRegExp string, fn HandlerFunc) *Walker {
-	hdl := NewDispHandler()
+	w.handlers = append(w.handlers, &handler{
+		dispRegExp: dispRegExp,
+		fn:         fn,
+	})
 
-	w.dispHandlers[contDisp] = hdl
+	return w
-
-	return hdl
 }
 
-func (w *Walker) getHandler(p *Part) handler {
+func (w *Walker) getHandlerFunc(p *Part) HandlerFunc {
-	if dispHandler := w.getDispHandler(p); dispHandler != nil {
+	for _, hdl := range w.handlers {
-		return dispHandler
+		if hdl.matchPart(p) {
+			return hdl.fn
+		}
 	}
 
-	return w.getTypeHandler(p)
-}
-
-// getTypeHandler returns the appropriate PartHandler to handle the given part.
-// If no specialised handler exists, it returns the default handler.
-func (w *Walker) getTypeHandler(p *Part) handler {
-	t, _, err := p.Header.ContentType()
-	if err != nil {
 	return w.defaultHandler
-	}
-
-	hdl, ok := w.typeHandlers[t]
-	if !ok {
-		return w.defaultHandler
-	}
-
-	return hdl
-}
-
-// getDispHandler returns the appropriate DispHandler to handle the given part.
-// If no specialised handler exists, it returns nil.
-func (w *Walker) getDispHandler(p *Part) handler {
-	t, _, err := p.Header.ContentDisposition()
-	if err != nil {
-		return nil
-	}
-
-	return w.dispHandlers[t]
 }

pkg/message/parser/walker_test.go

@@ -11,14 +11,13 @@ func TestWalker(t *testing.T) {
 
 	allBodies := [][]byte{}
 
-	walker := p.
+	walker := p.NewWalker().
-		NewWalker().
+		RegisterDefaultHandler(func(p *Part) (err error) {
-		WithDefaultHandler(NewPartHandler().OnEnter(func(p *Part) (err error) {
 			if p.Body != nil {
 				allBodies = append(allBodies, p.Body)
 			}
 			return
-		}))
+		})
 
 	assert.NoError(t, walker.Walk())
 	assert.ElementsMatch(t, [][]byte{
@@ -32,10 +31,8 @@ func TestWalkerTypeHandler(t *testing.T) {
 
 	html := [][]byte{}
 
-	walker := p.NewWalker()
+	walker := p.NewWalker().
-
+		RegisterContentTypeHandler("text/html", func(p *Part) (err error) {
-	walker.RegisterContentTypeHandler("text/html").
-		OnEnter(func(p *Part) (err error) {
 			html = append(html, p.Body)
 			return
 		})
@@ -51,10 +48,8 @@ func TestWalkerDispositionHandler(t *testing.T) {
 
 	attachments := [][]byte{}
 
-	walker := p.NewWalker()
+	walker := p.NewWalker().
-
+		RegisterContentDispositionHandler("attachment", func(p *Part) (err error) {
-	walker.RegisterContentDispositionHandler("attachment").
-		OnEnter(func(p *Part, hdl PartHandlerFunc) (err error) {
 			attachments = append(attachments, p.Body)
 			return
 		})
@@ -65,22 +60,22 @@ func TestWalkerDispositionHandler(t *testing.T) {
 	}, attachments)
 }
 
-func TestWalkerDispositionAndTypeHandler(t *testing.T) {
+func TestWalkerDispositionAndTypeHandler_TypeDefinedFirst(t *testing.T) {
 	p := newTestParser(t, "text_html_octet_attachment.eml")
 
-	walker := p.NewWalker()
+	var typeCalled, dispCalled bool
 
-	var enter, exit int
+	walker := p.NewWalker().
-
+		RegisterContentTypeHandler("application/octet-stream", func(p *Part) (err error) {
-	walker.RegisterContentTypeHandler("application/octet-stream").
+			typeCalled = true
-		OnEnter(func(p *Part) (err error) { enter++; return }).
+			return
-		OnExit(func(p *Part) (err error) { exit--; return })
+		}).
-
+		RegisterContentDispositionHandler("attachment", func(p *Part) (err error) {
-	walker.RegisterContentDispositionHandler("attachment").
+			dispCalled = true
-		OnEnter(func(p *Part, hdl PartHandlerFunc) (err error) { _ = hdl(p); _ = hdl(p); return }).
+			return
-		OnExit(func(p *Part, hdl PartHandlerFunc) (err error) { _ = hdl(p); _ = hdl(p); return })
+		})
 
 	assert.NoError(t, walker.Walk())
-	assert.Equal(t, 2, enter)
+	assert.True(t, typeCalled)
-	assert.Equal(t, -2, exit)
+	assert.False(t, dispCalled)
 }